Light therapies for acne

  • Review
  • Intervention

Authors


Abstract

Background

Acne vulgaris is a very common skin problem that presents with blackheads, whiteheads, and inflamed spots. It frequently results in physical scarring and may cause psychological distress. The use of oral and topical treatments can be limited in some people due to ineffectiveness, inconvenience, poor tolerability or side-effects. Some studies have suggested promising results for light therapies.

Objectives

To explore the effects of light treatment of different wavelengths for acne.

Search methods

We searched the following databases up to September 2015: the Cochrane Skin Specialised Register, CENTRAL, MEDLINE, Embase and LILACS. We searched ISI Web of Science and Dissertation Abstracts International (from inception). We also searched five trials registers, and grey literature sources. We checked the reference lists of studies and reviews and consulted study authors and other experts in the field to identify further references to relevant randomised controlled trials (RCTs). We updated these searches in July 2016 but these results have not yet been incorporated into the review.

Selection criteria

We included RCTs of light for treatment of acne vulgaris, regardless of language or publication status.

Data collection and analysis

We used standard methodological procedures expected by Cochrane.

Main results

We included 71 studies, randomising a total of 4211 participants.

Most studies were small (median 31 participants) and included participants with mild to moderate acne of both sexes and with a mean age of 20 to 30 years. Light interventions differed greatly in wavelength, dose, active substances used in photodynamic therapy (PDT), and comparator interventions (most commonly no treatment, placebo, another light intervention, or various topical treatments). Numbers of light sessions varied from one to 112 (most commonly two to four). Frequency of application varied from twice daily to once monthly.

Selection and performance bias were unclear in the majority of studies. Detection bias was unclear for participant-assessed outcomes and low for investigator-assessed outcomes in the majority of studies. Attrition and reporting bias were low in over half of the studies and unclear or high in the rest. Two thirds of studies were industry-sponsored; study authors either reported conflict of interest, or such information was not declared, so we judged the risk of bias as unclear.

Comparisons of most interventions for our first primary outcome 'Participant's global assessment of improvement' were not possible due to the variation in the interventions and the way the studies' outcomes were measured. We did not combine the effect estimates but rated the quality of the evidence as very low for the comparison of light therapies, including PDT to placebo, no treatment, topical treatment or other comparators for this outcome. One study which included 266 participants with moderate to severe acne showed little or no difference in effectiveness for this outcome between 20% aminolevulinic acid (ALA)-PDT (activated by blue light) versus vehicle plus blue light (risk ratio (RR) 0.87, 95% confidence interval (CI) 0.72 to 1.04, low-quality evidence). A study (n = 180) of a comparison of ALA-PDT (activated by red light) concentrations showed 20% ALA was no more effective than 15% (RR 1.05, 95% CI 0.96 to 1.15) but better than 10% ALA (RR 1.22, 95% CI 1.05 to 1.42) and 5% ALA (RR 1.47, 95% CI 1.19 to 1.81). The number needed to treat for an additional beneficial outcome (NNTB) was 6 (95% CI 3 to 19) and 4 (95% CI 2 to 6) for the comparison of 20% ALA with 10% and 5% ALA, respectively.

For our second primary outcome 'Investigator-assessed changes in lesion counts', we combined three RCTs, with 360 participants with moderate to severe acne and found methyl aminolevulinate (MAL) PDT (activated by red light) was no different to placebo cream plus red light with regard to change in inflamed lesions (ILs) (mean difference (MD) -2.85, 95% CI -7.51 to 1.81), percentage change in ILs (MD -10.09, 95% CI -20.25 to 0.06), change in non-inflamed lesions (NILs) (MD -2.01, 95% CI -7.07 to 3.05), or in percentage change in NILs (MD -8.09, 95% CI -21.51 to 5.32). We assessed the evidence as moderate quality for these outcomes meaning that there is little or no clinical difference between these two interventions for lesion counts.

Studies comparing the effects of other interventions were inconsistent or had small samples and high risk of bias. We performed only narrative synthesis for the results of the remaining trials, due to great variation in many aspects of the studies, poor reporting, and failure to obtain necessary data. Several studies compared yellow light to placebo or no treatment, infrared light to no treatment, gold microparticle suspension to vehicle, and clindamycin/benzoyl peroxide combined with pulsed dye laser to clindamycin/benzoyl peroxide alone. There were also several other studies comparing MAL-PDT to light-only treatment, to adapalene and in combination with long-pulsed dye laser to long-pulsed dye laser alone. None of these showed any clinically significant effects.

Our third primary outcome was 'Investigator-assessed severe adverse effects'. Most studies reported adverse effects, but not adequately with scarring reported as absent, and blistering reported only in studies on intense pulsed light, infrared light and photodynamic therapies. We rated the quality of the evidence as very low, meaning we were uncertain of the adverse effects of the light therapies.

Although our primary endpoint was long-term outcomes, less than half of the studies performed assessments later than eight weeks after final treatment. Only a few studies assessed outcomes at more than three months after final treatment, and longer-term assessments are mostly not covered in this review.

Authors' conclusions

High-quality evidence on the use of light therapies for people with acne is lacking. There is low certainty of the usefulness of MAL-PDT (red light) or ALA-PDT (blue light) as standard therapies for people with moderate to severe acne.

Carefully planned studies, using standardised outcome measures, comparing the effectiveness of common acne treatments with light therapies would be welcomed, together with adherence to the Consolidated Standards of Reporting Trials (CONSORT) guidelines.

Résumé scientifique

Les traitements de l'acné par la lumière

Contexte

L'acné est un problème de peau très courant qui se manifeste par des comédons blancs et noirs et des boutons enflammés. Elle entraîne souvent physiquement, des cicatrices et peut entraîner une détresse psychologique. L'utilisation de traitements oraux et topiques peut être limitée chez certaines personnes en raison d'une inefficacité, d'un inconfort, d'une mauvaise tolérance ou d'effets secondaires. Certaines études ont suggéré des résultats prometteurs pour les thérapies par la lumière.

Objectifs

Pour étudier les effets du traitement par la lumière de différentes longueurs d'onde pour l'acné.

Stratégie de recherche documentaire

Nous avons effectué des recherches dans les bases de données suivantes jusqu'à septembre 2015 : le registre spécialisé du groupe Cochrane CENTRAL, MEDLINE, Embase et LILACS. nous avons consulté l'ISI Web of Science et Dissertation Abstracts International (depuis leur création). Nous avons également effectué des recherches dans cinq registres d'essais cliniques et dans les sources de littérature grise. Nous avons examiné les références bibliographiques des études et des revues et contacté des auteurs d'études et d'autres experts dans le domaine afin d'identifier d'autres références à des essais contrôlés randomisés (ECR) pertinents. Nous avons mis à jour ces recherches en juillet 2016, mais ces résultats n'ont pas encore été incorporés dans la revue.

Critères de sélection

Nous avons inclus des ECR portant sur la lumière comme traitement de l'acné, indépendamment de la langue ou du statut de la publication.

Recueil et analyse des données

Nous avons utilisé les procédures méthodologiques standard prévues par Cochrane.

Résultats principaux

Nous avons inclus 71 études ayant randomisé un total de 4211 participants.

La plupart des études étaient de petite taille (médiane de 31 participants) et incluaient des participants atteints d'acné légère à modérée des deux sexes et d'un âge moyen de 20 à 30 ans. Les interventions lumineuses différaient considérablement en matière de longueur d'onde, de dose, de substance active utilisée dans la thérapie photodynamique (TPD) et d'interventions de comparaison (le plus souvent une absence de traitement, un placebo, une autre intervention de lumière ou divers traitements topiques). Le nombre de séances de thérapies par la lumière variait de un à 112 (le plus souvent de deux à quatre). La fréquence d'application variait de deux fois par jour à une fois par mois.

Les biais de sélection et de performance étaient incertains dans la majorité des études. Dans la majorité des études, la détection des biais n'était pas claire pour les critères de jugement relatifs aux évaluations faites par les patients et faible pour les critères de jugement relatifs aux évaluations faites par les investigateurs. Les biais d'attrition et de notification étaient faibles dans plus de la moitié des études et incertains ou élevés dans les autres. Deux tiers des études ont été financées par l'industrie ; les auteurs des études ont soit rapporté des conflits d'intérêts, soit ces informations n'ont pas été déclarées, de sorte que nous avons jugé le risque de biais comme incertain.

Les comparaisons de la plupart des interventions pour notre critère de jugement principal « évaluation globale de l'amélioration par les participants » n'ont pas été possibles en raison de la diversité des interventions et de la manière dont les critères de jugement des études ont été mesurés. Nous n'avons pas regroupé les estimations d'effet, mais nous avons évalué la qualité des preuves comme étant très faible pour la comparaison des thérapies par la lumière, y compris la TPD, versus placebo, l'absence de traitement, un traitement topique ou d'autres comparateurs pour ce critère de jugement. Pour ce critère de jugement, une étude portant sur 266 participants présentant une acné modérée à sévère a montré peu ou pas de différence en termes d'efficacité entre un 20 % d'acide aminolévulinique (AAL)-TPD activé par lumière bleue par rapport à l'excipient associé à la lumière bleue (risque relatif (RR) 0,87, intervalle de confiance à 95 % (IC) 0,72 à 1,04, preuves de faible qualité). Une étude (n = 180) d'une comparaison de concentrations d'AAL-TPD (activé par lumière rouge) a montré que la concentration de 20 % d'AAL n'était pas plus efficace que celle de 15 % (RR 1,05, IC à 95 % 0,96 à 1,15), mais était meilleure que celle de 10 % d'AAL (RR 1,22, IC à 95 % 1,05 à 1,42) et de 5 % d'AAL (RR 1,47, IC à 95 % 1,19 à 1,81). Le nombre de sujets à traiter pour obtenir un résultat bénéfique supplémentaire (NSTB) était de 6 (IC à 95 % 3 à 19) et de 4 (IC à 95 % 2 à 6) quand l'AAL 20 % était comparé à l'AAL 10 % et à l'AAL 5 %, respectivement.

Pour notre deuxième critère de jugement principal « modifications dans le nombre de lésions évaluées par l'investigateur », nous avons combiné trois ECR, avec 360 participants atteints d'acné modérée à sévère et nous avons trouvé que l'aminolévulinate de méthyle (ALM)-TPD (activé par lumière rouge) n'était pas différent d'une crème placebo plus lumière rouge en ce qui concerne les modifications dans les lésions inflammatoires (LI) (différence moyenne (DM) -2,85, IC à 95 % -7,51 à 1,81), le pourcentage de changement des LI (DM -10,09, IC à 95 % -20,25 à 0,06), les modifications dans les lésions non inflammatoires (LNI) (DM -2,01, IC à 95 % -7,07 à 3,05) ou le pourcentage de changement des LNI (DM -8,09, IC à 95 % -21,51 à 5,32). Nous avons évalué les preuves comme étant de qualité modérée pour ces critères de jugement, ce qui signifie qu'il n'existe que peu ou pas de différence clinique entre ces deux interventions en ce qui concerne le nombre de lésions.

Les études comparant les effets d'autres interventions étaient contradictoires ou concernaient des échantillons de petite taille ou présentaient un risque élevé de biais. Nous n'avons effectué qu'une synthèse narrative pour les résultats des essais restants, en raison d'une grande variation dans de nombreux aspects de ces études, de la notification médiocre et de l'échec à obtenir les données nécessaires. Plusieurs études comparaient la lumière jaune à un placebo ou à l'absence de traitement, la lumière infrarouge à l'absence de traitement, une suspension de microparticules d'or à l'excipient ou encore la clindamycine/peroxyde de benzoyle associée au laser pulsé à colorant à la clindamycine/peroxyde de benzoyle seule. Il y avait également plusieurs autres études comparant l'ALM-TPD à la thérapie par la lumière seule, à l'adapalène ou encore associé au laser à colorant à longue durée d'impulsion au laser à colorant à longue durée d'impulsion seul. Aucun de ces essais n'a montré d'effets cliniquement significatifs.

Notre troisième critère principal de jugement était « les effets indésirables graves évalués par l'investigateur ». La plupart des études ont rapporté des effets indésirables, mais pas de façon adéquate avec des cicatrices rapportées comme absentes et des cloques rapportées seulement dans des études sur la lumière pulsée intense, la lumière infrarouge et les thérapies photodynamiques. Nous avons évalué la qualité des preuves comme étant très médiocre, ce qui signifie que nous avons été incertains sur les effets indésirables des thérapies par la lumière.

Bien que notre critère de temps principal concernait des résultats de long terme, moins de la moitié des études ont procédé à des évaluations à plus de huit semaines après le traitement final. Seules quelques études ont évalué les résultats à plus de trois mois après le traitement final et les évaluations à plus long terme ne sont, pour la plupart, pas abordées dans cette revue.

Conclusions des auteurs

Des preuves de bonne qualité manquent sur l'utilisation de thérapies par la lumière pour les personnes souffrant d'acné. Il y a une faible certitude de l'utilité de l'ALM-TPD (lumière rouge) et de l'ALA-TPD (lumière bleue) en tant que traitement standard pour les personnes souffrant d'acné modérée à sévère.

Des études soigneusement organisées, utilisant des mesures de résultats standardisées, comparant l'efficacité des traitements courants de l'acné avec des thérapies par la lumière seraient bienvenues, ainsi que le respect des directives de CONSORT (Consolidated Standards of Reporting Trials).

Notes de traduction

Traduction réalisée par Daniel Pinchenzon et révisée par Cochrane France

摘要

痤疮的光治疗

研究背景

寻常痤疮(acne vulgaris)是一种常见的皮肤问题表现为黒头、白头和点状炎症。其通常留下实体瘢㾗并可能引起心理郁闷。因治疗无效、便捷性差、耐受性差或副作用,只有部分患者使用口服和局部治疗。一些研究显示光治疗有良好效果。

研究目的

为研究不同波长光治疗对痤疮的效果。

检索策略

我们检索了截止2015年9月的下列数据库:Cochrane皮肤特异性注册中心、CENTRAL、MEDLINE、Embase和LILACS。我们检索了ISI Web of Science和国际学位论文摘要数据库(Dissertation Abstracts International)(从起始日期)。我们同时检索了5个临床试验注册中心和灰色源文献。我们检查了研究和综述的参考文献并咨询了研究作者和该领域的其它专家以寻找进一步扩展可供参考的相关性随机化对照研究(RCTs)。我们在2016年7月更新了这些检索,但新的检索结果尚未加入综述中。

标准/纳入排除标准

我们纳入了光治疗寻常痤疮的随机化对照试验(RCTs),不论其语种和是否发表。

数据收集与分析

我们使用了Cochrane推荐的标准方法流程。

主要结果

我们纳入71项研究,随机化分组的受试者数共达4211人。

大多数是小型研究(中位数参与者这31人),包含了不同性别、平均年龄20-30岁的轻度至中度的痤疮病例。光治疗研究中,所使用波长、剂量、光动力学治疗(photodynamic therapy, PDT)的光敏化学物和对照处理(最常见的对照是无处理、安慰剂、另一种光治疗或各种局部治疗)有很大差异。光治疗1-122个周期(绝大多数2-4个周期)。治疗频率从每天两次至每月一次不等。

大多数研究未说明选择偏倚和执行偏倚情况。大多数参与者评价的研究中的终点测量偏倚未知但研究究者评价的终点测量偏倚较低。过半的研究中病例退出(attrition)和报告的偏倚较低,其余研究中偏倚未知或较高。2/3的研究由企业资助;部分研究者报告了利益冲突(conflict of interest, COI),部分研究者未声明,故我们判定偏倚风险为未知。

我们的首要研究终点“参与者改善的总体评价”在大部分干预研究中的比较无法实现,因为干预内容和研究终点测量的不一致。我们未估计合并效应但评定光治疗方法比较研究结果的质量为低级证据,包括了PDT组与安慰剂、无处理、局部治疗或其它对照组的比较。一项研究中包含266例中度至重度痤疮参与者,在该研究终点上未显示20%氨基乙酰丙酸-光动力学治疗(aminolevulinic acid-PDT, ALA-PDT)(通过蓝光激活)与溶剂对照加蓝光照射有差异(危险比(risk ratio, RR)为0.87, 95% CI为0.72-1.04, 低质量证据)。一项比较不同ALA-PDT(红光激活)浓度的研究(n=180)显示,20% ALA不比15% ALA更有效(RR 1.05, 95% CI 0.96-1.15),但优于10% ALA (RR 1.22, 95% CI 1.05-1.42)和5% ALA (RR 1.47, 95% CI 1.19-1.81)。20% ALA与10% ALA及5% ALA相比,为获得额外收益需治疗的人数(number needed to treat for an additional beneficial outcome, NNTB, 有时缩写为需治人数)分别为6 (95% CI 3-19)和4 (95% CI 2-6)。

我们第二个主要终点是“研究者评估的病灶数量变化”,合并了3个随机化临床试验,有360名中度至重度痤疮的参与者,发现氨基乙酰丙酸甲酯(methyl aminolevulate, MAL)作为光敏剂的光动力学治疗(MAL-PDT, 由红光激活)与安慰膏剂加红光照射在炎症病灶(inflamed lesions, ILs)改变方面没有差异,均数之差(MD)为-2.85, 95% CI -7.51 - 1.81,ILs改变百分比均值的差异为-10.09, 95% CI -20.25 - 0.06,非炎性病灶(non-inflamed lesions, NILs)改变的均数之差为-2.01, 95% CI -7.07 - 3.05,NILs改变百分比均值的差异为-8.09, 95% CI -21.51 - 5.32。这些结果我们评定为中度质量的证据,这意味着这两种治疗方案对病灶数量改变几乎没有差异。

其它比较疗效的干预或为结果不一致或为小样本或为高偏倚风险的研究。对于其它试验结果我们只进行了描述性总结,因为研究本身的巨大差异、报告不完整或未获得必要的数据。有几个比较黄光源与安慰剂或无处理、红外线与无处理、金微粒子悬液与溶剂、克林霉素苯甲酰软膏(clindamycin/benzoyl peroxide)联合脉冲染料激光(pulsed dye laser)与单用克林霉素苯甲酰软膏的研究。还有几个其它研究比较了MAL-PDT与只用光照治疗、与阿达帕林(三代维A酸衍生物)、联合长脉冲染料激光与单用长脉冲染料激光。这些研究均未显示有任何临床疗效。

1/3的主要终点是“研究者评价的严重不良反应”。大多数研究报道了不良反应(adverse effects),但对于报道未观察到瘢痕方面不充分,且只在使用强化脉冲光照、红外线和光动力学治疗组报道了水泡形成。我们评定该为极低质量证据,意味着对光治疗的不良反应是不确定的。

虽然我们的主要终点是长期结局,不到一半的研究在结束治疗八周后进行了评价。只有少数研究在结束治疗三个月后评估了结局,长期评价在本分析中大多未涉及。

作者结论

使用光治疗痤疮者方面缺乏高质量的证据。对中度至重度痤疮者使用MAL-PDT(红光)或ALA-PDT(蓝光)作为标准疗法的有效性存在很大的不确定性。

亟须同时遵循随机对照试验报告指南CONSORT声明(Consolidated Standards of Reporting Trials),又精心设计、使用标准化终点指标、与常用痤疮治疗方法相比较的光治疗有效性的研究。

翻译注解

译者:张国州; 审校:李迅(北京中医药大循证医学中心)2017年2月6日

Plain language summary

The use of light as a therapy for acne

What is the aim of this review?

The aim of this Cochrane Review was to find out whether treatment using lasers and other light sources improves the whiteheads and blackheads, and inflamed spots that people with acne have. We also wanted to know how people with acne assessed their own improvement, and whether they found that these therapies caused unpleasant effects like blistering or scarring. Cochrane researchers collected and analysed all relevant studies to answer these questions and found 71 studies, with a total of 4211 participants.

What was studied in this review?

Acne is a common skin problem. It causes blackheads, whiteheads and inflamed spots, and may lead to scarring. Current treatment options are limited in their effectiveness and convenience, and may cause side-effects. We investigated lasers and other light sources, which are used as an alternative therapy, either on their own or in combination with a chemical that makes the skin more sensitive to the light source (photodynamic therapy (PDT)). We compared different light therapies with other treatment options, no treatment, or placebo.

Most studies included people with mild to moderate acne in their twenties. Light treatments in these studies varied greatly in many important aspects, such as wavelength of light used, duration of treatment, chemicals used in photodynamic therapy, and others.

Over half of the studies were industry sponsored; study authors reported either conflict of interest, or such information was not declared.

Key messages

We are unable to draw firm conclusions from the results of our review, as it was not clear whether the light therapies (including PDT) assessed in these studies were more effective than the other comparators tested such as placebo, no treatment, or treatments rubbed on the skin, nor how long the possible benefits lasted.

What are the main results of this review?

We investigated how people with acne assessed their own improvement, but it was not clear whether the light therapies in the studies had a beneficial effect. Evidence on how investigators assessed changes in numbers of blackheads, whiteheads and inflamed spots in people with acne was also limited for most types of light therapies, due to variation in the way the studies were conducted and measured.

Most studies reported side-effects, but not adequately. Scarring was reported as absent, and blistering was reported in studies on intense pulsed light, infrared light and on PDT.

Three studies, with a total of 360 participants with moderate to severe acne, showed that photodynamic therapy with methyl aminolevulinate (MAL), activated by red light, had a similar effect on changes in numbers of blackheads, whiteheads and inflamed spots when compared with placebo cream with red light. We judged the quality of this evidence moderate.

Future well planned studies comparing the effectiveness of common acne treatments with light therapies are needed to assess the true clinical effects and side-effects of light therapies for acne.

How up to date is this review?

This review included studies up to September 2015.

Резюме на простом языке

Применение света для лечения акне

Какова цель этого обзора?

Целью этого Кокрейновского обзора была оценка эффективности применения лазеров и других источников света в лечении белых и черных угрей, а также участков воспаления, которые наблюдаются у людей с акне. Также мы хотели выяснить, как сами пациенты с акне оценивают улучшение и не обнаружили ли они неблагоприятных эффектов от лечения, таких как образование пузырей или рубцевание. Для ответа на эти вопросы исследователи Кокрейн собрали и проанализировали все соответствующие исследования и обнаружили 71 исследование с 4211 участниками.

Что изучалось в этом обзоре?

Акне – распространенная кожная проблема. Акне вызывает образование черных и белых угрей, а также участков воспаления и может привести к рубцеванию. Настоящие методы лечения имеют ограниченные эффективность и удобство и могут приводить к развитию побочных эффектов. Мы изучили лазеры и другие источники света, применяемые в качестве альтернативной терапии как самостоятельно, так и в комбинации с химическими препаратами, повышающими чувствительность кожи к свету (фотодинамическая терапия (ФДТ)). Мы сравнили различные виды светотерапии с другими вариантами лечения, отсутствием лечения или плацебо.

Большинство исследований включали участников с акне легкой или умеренной степени тяжести в возрасте от 20 до 29 лет. Варианты светотерапии в этих исследованиях крайне отличались по многим важным аспектам, таким как длина волны используемого света, продолжительность лечения, химические препараты для фотодинамической терапии и т. д.

Более половины исследований были спонсированы индустрией; либо авторы исследований сообщали о конфликтах интересов, либо же такая информация представлена не была.

Ключевая информация

Мы не можем сделать каких-либо однозначных выводов по результатам обзора, так как неясно насколько эффективными были различные виды светотерапии (включая ФДТ) в сравнении с плацебо, отсутствием лечения или препаратами, наносимыми на кожу, а также как долго сохранялись возможные положительные эффекты.

Каковы основные результаты этого обзора?

Мы изучили, как сами пациенты с акне оценивали улучшение, однако осталось неясным, оказывала ли светотерапия в исследованиях какой-либо положительный эффект. Доказательства относительно того, каким образом исследователи оценивали изменение числа черных и белых угрей, а также участков воспаления у пациентов с акне, были ограничены для большинства видов терапии светом, ввиду различий в проведении исследований и методов измерения результатов.

В большинстве исследований сообщалось о побочных эффектах, однако не вполне адекватно. О рубцевании не сообщалось, а об образовании пузырей сообщалось в исследованиях, посвященных интенсивному импульсному свету, инфракрасному излучению и ФДТ.

В 3 исследованиях с 360 участниками, страдающими от акне умеренной или тяжелой степени, было показано, что фотодинамическая терапия с метиламинолевулинатом (МАЛ), активированным красным светом, оказала такой же эффект на число черных и белых угрей, а также участков воспаления, как и крем плацебо в сочетании с красным светом. Мы оценили качество этих доказательств как умеренное.

Для должной оценки клинической эффективности и побочных эффектов светотерапии необходимы дальнейшие хорошо спланированные исследования, посвященные сравнению распространенных методов лечения акне со светотерапией.

Насколько актуален этот обзор?

В этот обзор были включены исследования, опубликованные по сентябрь 2015 года.

Заметки по переводу

Перевод: Кукушкин Михаил Евгеньевич. Редактирование: Зиганшина Лилия Евгеньевна. Координация проекта по переводу на русский язык: Cochrane Russia - Кокрейн Россия (филиал Северного Кокрейновского Центра на базе Казанского федерального университета). По вопросам, связанным с этим переводом, пожалуйста, обращайтесь к нам по адресу: cochrane.russia.kpfu@gmail.com; cochranerussia@kpfu.ru

Laički sažetak

Korištenje svjetla za liječenje akni

Koji je cilj ovog pregleda?

Cilj ovog Cochrane sustavnog pregleda literture je bio saznati poboljšava li liječenje pomoću lasera i ostalih izvora svjetla komedone, mitesere i upalne promjene u osoba s aknama. Također smo željeli saznati kako su ljudi s aknama sami procjenjivali poboljšanje, te je li ovakvo liječenje uzrokovalo neugodne učinke poput stvaranja opeklinskih mjehurića ili ožiljaka. Cochrane istraživači su prikupili i analizirali sve relevantne studije kako bi odgovorili na ova pitanja i našli 71 studiju, s ukupno 4211 sudionika.

Što je istraženo u ovom pregledu literature?

Akne su čest kožni problem. Uzrokuju mitesere, komedone i upalne promjene, a mogu dovesti i do stvaranja ožiljaka. Sadašnje terapijske mogućnosti imaju ograničenu učinkovitost i praktičnost korištenja, a mogu uzrokovati i nuspojave. Istraživali smo lasere i druge izvore svjetlosti, koji se koriste kao zamjenska mogućnost liječenja, samostalno, ili u kombinaciji s kemikalijom koja kožu čini osjetljivijom na svjetlost (fotodinamička terapija, PDT). Usporedili smo različite terapije svjetlom s drugim mogućnostima liječenja, nikakvim liječenjem, ili placebom.

Većina studija je uključila ljude s blagim do umjerenim aknama u dvadesetim godinama života. U navedenim studijama, terapije svjetlom su varirale u mnogim važnim aspektima, poput valnih duljina svjetlosti koja je korištena, trajanju tretmana, kemikalija koje su korištene kod fotodinamičkih terapija, i ostalih.

Više od polovine studija je sponzorirala industrija, autori studija su ili prijavili sukob interesa, ili takve informacije nisu navedene.

Ključne poruke

Nismo mogli donijeti čvrste zaključke iz rezultata našeg pregleda literature jer nije bilo jasno jesu li terapije svjetlom (uključujući PDT) koje su procijenjene u studijama bile učinkovitije od onih s kojima ih se uspoređivalo, poput placeba, nikakvog liječenja, terapija primijenjenih na kožu, kao niti koliko su dugo mogući korisni učinci trajali.

Koji su glavni rezultati ovog pregleda?

Istraživali smo kako su osobe s aknama procjenjivale poboljšanje svojih akni, ali nije bilo jasno jesu li terapije svjetlom u studijama imale korisni učinak. Dokazi o tome kako su istraživači procjenjivali promjene u broju mitesera, komedona i upalnih promjena kod ljudi s aknama su također ograničeni za većinu tipova svjetlosnih terapija, zbog razlika u načinu na koji su studije, kao i mjerenja u njima provedeni.

U većini studija su prikazani podatci o nuspojavama, ali ne na prikladan način. Navedeno je da ožiljaka kao nuspojava nije bilo. Opisana je pojava mjehurića uslijed opeklina u studijama o intenzivnom pulsnom svjetlu (IPL-u), infracrvenom svjetlu i studijama o PDT-u.

Tri studije, s ukupno 360 sudionika s umjerenim do teškim aknama, pokazale su da je fotodinamička terapija s metil-aminolevulinatom (MAL) aktiviranim crvenim svjetlom, imala sličan učinak na promjene broja mitesera, komedona i upalnih promjena u usporedbi s placebo kremom i crvenim svjetlom. Kvalitetu navedenih dokaza smo procijenili kao umjerenu.

Potrebne su buduće dobro planirane studije koje uspoređuju učinkovitost uobičajenih oblika liječenja za akne s liječenjem koje se temelji na svjetlu kako bi se procijenili stvarni klinički učinci i nuspojave svjetlosnih terapija za akne.

Za koje razdoblje je pretražena literatura korištena u ovom sustavnom pregledu?

Ovaj pregled je uključio studije objavljene do rujna 2015.

Bilješke prijevoda

Hrvatski Cochrane
Prevela: Jelena Barbarić
Ovaj sažetak preveden je u okviru volonterskog projekta prevođenja Cochrane sažetaka. Uključite se u projekt i pomozite nam u prevođenju brojnih preostalih Cochrane sažetaka koji su još uvijek dostupni samo na engleskom jeziku. Kontakt: cochrane_croatia@mefst.hr

Streszczenie prostym językiem

Wykorzystanie światła w leczeniu trądziku

Jaki był cel tego przeglądu?

Celem tego przeglądu Cochrane było sprawdzenie, czy leczenie za pomocą laserów oraz innych źródeł światła poprawia stan skóry osób zmagających się z trądzikiem wpływając na obecność otwartych i zamkniętych zaskórników oraz wykwitów zapalnych. Chcieliśmy również dowiedzieć się, jak osoby borykające się z trądzikiem oceniają poprawę stanu swojej skóry i czy stosowanie leczenia z wykorzystaniem światła wiąże się z wystąpieniem nieprzyjemnych skutków ubocznych, takich jak powstawanie pęcherzy czy blizn. Naukowcy z Cochrane, chcąc znaleźć odpowiedzi na te pytania, zgromadzili i przeanalizowali wszystkie odpowiadające tematowi badania. W wyniku wyszukiwania znaleziono 71 badań, w których łącznie udział brało 4211 uczestników.

Co badano w tym przeglądzie?

Trądzik jest powszechnym problemem skórnym. W jego przebiegu na skórze pojawiają się otwarte i zamknięte zaskórniki oraz wykwity zapalne, które mogą skutkować powstawaniem blizn. Obecnie dostępne metody leczenia trądziku posiadają ograniczenia co do swojej skuteczności oraz wygody stosowania i mogą powodować wystąpienie skutków ubocznych. Przyjrzeliśmy się leczeniu laserem oraz innymi źródłami światła, które to stosowane samodzielnie lub też w skojarzeniu z substancją chemiczną, czyniącą skórę bardziej podatną na oddziaływanie światła (terapia fotodynamiczna (PDT)), wykorzystywane są jako alternatywna metoda leczenia trądziku. Porównaliśmy różne rodzaje terapii światłem z innymi opcjami terapeutycznymi, brakiem leczenia lub placebo.

Uczestnikami większości badań były osoby dwudziestokuletnie doświadczające trądziku o nasileniu łagodnym do umiarkowanego. Analizowane terapie światłem różniły się od siebie znacznie w wielu ważnych aspektach, takich jak między innymi długość fali użytego światła, czas stosowania terapii, substancja chemiczna użyta podczas terapii fotodynamicznej.

Ponad połowa badań sponsorowana była przez przemysł; autorzy badań odnotowywali konflikt interesów, bądź też nie zamieszczono informacji w tej kwestii.

Główne wnioski

Nie jesteśmy w stanie wyciągnąć jednoznacznych wniosków w oparciu o wyniki przeglądu, gdyż nie jest jednoznaczne czy terapia z użyciem światła (włączając PDT) ocenione w przeanalizowanych badaniach były bardziej skuteczne niż inne metody, z którymi je porównywano, tj.: placebo, brak leczenia, zabiegi polegające na wcieraniu określonych substancji w skórę. Nie jest też jasne jak długotrwałe były zaobserwowane korzyści.

Jakie są główne wyniki przeglądu?

Zbadaliśmy jak osoby z trądzikiem oceniają poprawę stanu swojej skóry, jakkolwiek nie zostało jasno udowodnione, że leczenie światłem przyniosło poprawę w tym zakresie. Dane naukowe dotyczące dokonanej przez badaczy oceny liczby otwartych i zamkniętych zaskórników oraz wykwitów zapalnych u osób z trądzikiem, w przypadku większości rodzajów terapii światłem były ograniczone ze względu na różnice w sposobie przeprowadzania badań i dokonywania pomiarów pomiędzy badaniami.

W większości badań odnotowywano występowanie działań ubocznych, jednakże w sposób niewystarczający. Nie odnotowano, aby badane terapie powodowały powstawanie blizn. Natomiast tworzenie się pęcherzy zaobserwowano w badaniach, w których pacjenci leczeni byli za pomocą terapii wykorzystujących intensywnie pulsujące światło, światło podczerwone, jak również w wyniku terapii fotodynamicznej.

W trzech badaniach, w których udział brało łącznie 360 uczestników zmagających się z trądzikiem o umiarkowanym lub ciężkim nasileniu, wykazano, że terapia fotodynamiczna z użyciem aminolewulinianu metylu (MAL) aktywowanego podczas naświetlania światłem czerwonym przynosiła podobne efekty co placebo - terapia polegająca na aplikacji kremu niezawierającego substancji czynnej i naświetlanie światłem czerwonym (oceniano zmianę w liczbie otwartych i zamkniętych zaskórników oraz wykwitów zapalnych). Autorzy przeglądu ocenili jakość tych danych jako umiarkowaną.

Aby ocenić skuteczność stosowanych w leczeniu trądziku terapii światłem oraz ich skutki uboczne potrzebne są dalsze, dobrze zaplanowane badania porównujące je z powszechnie stosowanymi metodami leczenia.

Jak aktualny jest ten przegląd?

Przegląd ten zawiera badania opublikowane do września 2015 r.

Uwagi do tłumaczenia

Tłumaczenie: Joanna Błońska, Redakcja: Małgorzata Bała

Laienverständliche Zusammenfassung

Verwendung von Licht als Therapie bei Akne

Was ist das Ziel dieses Reviews?

Das Ziel dieses Cochrane Reviews war es herauszufinden, ob Laser- oder andere Lichtbehandlungen, Mitesser und entzündete Pickel, die Menschen mit Akne haben, verbessern. Zusätzlich wollten wir wissen, wie Menschen mit Akne ihre eigene Verbesserung beurteilen und ob ihrer Meinung nach, die Therapien unangenehme Wirkungen aufzeigen, wie Blasenbildung oder Narben. Cochrane-Autoren sammelten und untersuchten alle relevanten Studien, um diese Fragen zu beantworten und fanden 71 Studien, mit einer Gesamtzahl von 4211 Teilnehmern.

Was wurde in diesem Review untersucht?

Akne ist ein häufiges Hautproblem. Sie verursacht Mitesser, entzündete Pickel und könnte zur Narbenbildung führen. Aktuelle Behandlungsmöglichkeiten sind in ihrer Wirksamkeit und ihrer Zweckmäßigkeit begrenzt und könnten Nebenwirkungen verursachen. Wir untersuchten Laser und andere Lichtquellen, welche als eine alternative Behandlung verwendet werden. Sie werden entweder einzeln oder in Kombination mit einer chemischen Ergänzung, die die Haut sensibler für die Lichtquelle machen soll, angewendet (Photodynamische Therapie (PDT)). Wir verglichen verschiedene Lichtbehandlungen mit anderen Behandlungsoptionen, keiner Behandlung oder einem Placebo.

Die meisten Studien beinhalten Personen in ihren zwanzigern mit leichter bis moderater Akne. In diesen Studien variierten die Lichttherapien in vielen wichtigen Aspekten stark. Dies umfasst den Gebrauch von Wellenlängen während der Therapie, Gebrauch von chemischen Ergänzungen in photodynamischen Therapien und vieles mehr.

Über die Hälfte der Studien wurden von der Industrie finanziert; Studienautoren berichteten entweder einen Interessenkonflikt oder es wurden keine Angaben dazu gemacht.

Hauptaussagen

Wir sind nicht in der Lage aus den Ergebnissen unseres Reviews fundierte Schlussfolgerungen zu ziehen, da es nicht eindeutig war, ob die Lichttherapien (PDT mit eingeschlossen), die in diesen Studien untersucht wurden, wirksamer waren als Placebo, keine Behandlung oder Behandlungen, die auf die Haut aufgetragen wurden. Auch war es nicht eindeutig, wie lange der mögliche Nutzen bestehen bleibt.

Was sind die Hauptergebnisse dieses Reviews?

Wir untersuchten wie Personen mit Akne ihre eigene Verbesserung beurteilten. Es war jedoch nicht eindeutig, ob die Lichttherapien in den Studien eine nützliche Wirkung hatten. Die Evidenz darüber, wie die Anzahl von Mitessern und entzündeten Pickeln bei Personen mit Akne beurteilt wurde, war für die meisten Lichttherapien ebenfalls begrenzt. Dies lag an Abweichungen, wie die Studien durchgeführt und gemessen wurden.

Die meisten Studien berichteten Nebenwirkungen, jedoch nicht ausreichend. Narbenbildung wurden als ausbleibend berichtet und Blasenbildung wurde in Studien mit intensiv gepulstem Licht, Infrarotlicht und bei PDT berichtet.

Drei Studien mit einer Gesamtzahl von 360 Teilnehmern mit moderater bis starker Akne zeigten, dass photodynamsiche Therapien mit Methyl Aminolaevulinat (MAL), das bei rotem Licht aktiviert wird, eine ähnliche Wirkung auf die Anzahl der Mitesser und entzündeten Pickel hat wie eine Placebo Creme und rotes Licht. Wir bewerteten die Qualität dieser Evidenz als moderat.

Zukünftige gut geplante Studien, die die Alltagswirksamkeit gängiger Aknebehandlungen mit Lichttherapien vergleichen, werden benötigt, um die wahren klinischen Wirkungen und Nebenwirkungen der Lichttherapien bei Akne zu beurteilen.

Wie aktuell ist dieses Reviews?

Dieses Review hat bis September 2015 Studien eingeschlossen.

Anmerkungen zur Übersetzung

A. Ioannaki, freigegeben durch Cochrane Deutschland.

Résumé simplifié

L'utilisation de la lumière comme traitement de l'acné

Quel est l'objectif de cette revue ?

L'objectif de cette revue Cochrane a été de découvrir si un traitement par laser ou d'autres sources de lumière améliore les points blancs et points noirs et les boutons enflammés présentés par les personnes souffrant d'acné. Nous avons également voulu savoir comment les personnes souffrant d'acné ont évalué leur propre amélioration et si elles ont trouvé que ces thérapies provoquaient des effets désagréables tels que la formation de cloques ou de cicatrices. Les chercheurs de Cochrane ont recueilli et analysé toutes les études pertinentes pour répondre à ces questions, ils ont trouvé 71 études, avec un total de 4211 participants.

Qu'est-ce qui a été étudié dans cette revue ?

L'acné est un problème courant de la peau. Elle provoque des comédons, noirs et blancs et des boutons enflammés pouvant conduire à des cicatrices. Les options actuelles de traitement sont limitées dans leur efficacité et leur commodité et peuvent entraîner des effets secondaires. Nous avons étudié les lasers et d'autres sources de lumière, utilisés comme traitement alternatif, soit seuls ou associés à un produit chimique qui rend la peau plus sensible à la source lumineuse (thérapie photodynamique (TPD)). Nous avons comparé différentes thérapies lumineuses à d'autres options de traitement, à l'absence de traitement ou à un placebo.

La plupart des études portaient sur des personnes d'une vingtaine d'années présentant une acné légère à modérée. Les traitements par la lumière dans ces études variaient considérablement dans de nombreux aspects importants, tels que la longueur d'onde de la lumière utilisée, la durée du traitement, les substances chimiques utilisées dans la thérapie photodynamique et d'autres aspects.

Plus de la moitié des études étaient financées par l'industrie ; les auteurs des études, ont soit rapporté des conflits d'intérêt, soit que de telles informations n'ont pas été déclarées.

Éléments clés

Nous ne sommes pas en mesure de tirer des conclusions définitives d'après les résultats de notre revue, car il n'était pas clair que les thérapies par la lumière (y compris la TPD) évaluées dans ces études aient été plus efficaces que les autres comparateurs testés, tels qu'un placebo, une absence de traitement ou des traitements appliqués sur la peau, ni combien de temps les bénéfices potentiels ont duré.

Quels sont les principaux résultats de cette revue ?

Nous avons étudié la manière dont les personnes souffrant d'acné évaluaient leur propre amélioration, mais il n'était pas clair que les thérapies par la lumière, dans les études, aient présenté un effet bénéfique. Les données concernant la manière dont les investigateurs ont évalué les changements dans le nombre des comédons noirs et blancs, et des boutons enflammés chez les personnes souffrant d'acné étaient également limitées pour la plupart des types de thérapies par la lumière, en raison des variations dans la façon dont les études avaient été réalisées et mesurées.

La plupart des études ont rapporté des effets secondaires, mais pas de manière adéquate. Les cicatrices ont été rapportées comme absentes et la formation de cloques a été rapportée dans les études sur la lumière pulsée intense, la lumière infrarouge et la TPD.

Trois études, avec un total de 360 participants atteints d'acné modérée à sévère, ont montré que la thérapie photodynamique avec de l'aminolévulinate de méthyle, activé par lumière rouge, a eu un effet similaire sur les changements dans le nombre de comédons noirs et blancs, et de boutons enflammés, par rapport à une crème placebo et à de la lumière rouge. Nous avons estimé que la qualité de ces preuves était modérée.

De futures études bien planifiées comparant l'efficacité des traitements courants de l'acné avec des thérapies par la lumière sont nécessaires pour évaluer les véritables effets cliniques et les effets secondaires des thérapies par la lumière pour l'acné.

Cette revue est-elle à jour ?

Cette revue a inclus des études jusqu'à septembre 2015.

Notes de traduction

Traduction réalisée par Daniel Pinchenzon et révisée par Cochrane France

概要

使用光作为痤疮治疗的一种方法

此综述的目的是什么?

该Cochrane综述的目的是找出利用激光和其它光源治疗是否可改善痤疮患者出现的白头(whitehead)、黑头(blackhead)和点状炎症(inflamed spot)。我们也想知道痤疮患者如何评价其病情改善,他们是否发现这些治疗引起了不快的效果如水疱或瘢㾗。Cochrane的研究者收集并分析了所有相关的研究以回答这些问题,发现有71个研究总参与者达4211例。

该综述研究了什么?

痤疮是一种常见的皮肤问题。它引起黒头、白头和点状炎症并可能导致瘢㾗。当前的治疗选择仅限于有效性和便利性,有可能引起副作用。我们研究了作为替代治疗的激光和其它光源单独或与增强皮肤对光照敏感性化学物联合使用(光动力学治疗, photodynamic therapy, PDT)。我们比较了不同光动力学治疗与其它治疗、未治疗或安慰剂的作用。

大多数研究的参与者是20-30岁轻度至中度的痤疮患者。这些研究中的光治疗在许多重要方面差异很大,如使用的光源波长、治疗持续时间、光动力学治疗中使用的化学物及其它条件。

半数以上的研究由工业界资助;研究者公布了存在利益冲突或未公布。

关键信息

我们的综述结果无法得出肯定的结论,因为不清楚这些研究中的光治疗(包括光动力学治疗)是否较其它对照治疗方法如安慰剂、未治疗和皮肤摩擦更有效,也不清楚可能的收益会持续多久。

该综述主要的结果是什么?

我们研究了痤疮患者如何评价其自身的改善,但不清楚光治疗是否有益。对于大多数种类的痤疮患者光治疗,因其进行研究方式和测量的不同,其研穹者如何评价黑头、白头和点状炎症变化方面的证据也有限。

大多数研究报道了副作用,但不充分。未有瘢㾗的报道,但在强化脉冲光疗、红外线和光动力学治疗中报道有水泡。

共含360名中度至重度的痤疮参与者的三项研究显示,利用红光激活、使用氨基乙酰丙酸甲酯(methyl aminolevulinate, MAL)的光动力学治疗,与安慰剂软膏加红光照射对于黒头、白头和点状炎症病灶数量改变的效果类似。我们判定该证据的质量为中度。

进一步的计划周详的研究需比较痤疮常规治疗与光治疗的有效性,以评价痤疮光治疗真正的临床效应和副作用。

该综述的时效性如何?

该综述包含截止2015年9月的研究。

翻译注解

译者:张国州; 审校:李迅(北京中医药大循证医学中心)2017年2月6日

Ringkasan bahasa mudah

Penggunaan cahaya sebagai terapi untuk jerawat

Apakah matlamat ulasan ini?

Tujuan Ulasan Cochrane ini adalah untuk mengetahui sama ada rawatan menggunakan laser dan sumber cahaya yang lain menambah baik bintik puteh dan bintik hitam, dan tempat-tempat radang yang ada pada mereka yang mempunyai jerawat. Kami juga mahu mengetahui bagaimana mereka yang mempunyai jerawat menilai pemulihan mereka sendiri, dan sama ada mereka mendapati bahawa terapi ini menyebabkan kesan yang tidak menyenangkan seperti melepuh dan parut. Para penyelidik Cochrane mengumpul dan menganalisis semua kajian yang berkaitan untuk menjawab soalan-soalan ini dan mendapati 71 kajian, dengan sejumlah 4211 peserta.

Apakah yang dikaji dalam ulasan ini?

Jerawat merupakan masalah kulit yang biasa. Ia menyebabkan bintik hitam, bintik putih dan bintik radang, dan boleh menyebabkan parut. Pilihan rawatan semasa adalah terhad keberkesanan dan keselesaannya, dan boleh menyebabkan kesan sampingan. Kami telah menyiasat laser dan sumber cahaya yang lain, yang digunakan sebagai terapi alternatif, sama ada secara bersendirian atau dalam kombinasi dengan bahan kimia yang menjadikan kulit lebih sensitif kepada sumber cahaya (terapi photodynamic (PDT)). Kami berbanding terapi cahaya yang berbeza dengan pilihan rawatan lain, tiada rawatan, atau plasebo.

Kebanyakan kajian melibatkan mereka yang dalam umur dua puluhan dyang mempunyai jerawat yang ringan hingga sederhana. Rawatan cahaya dalam kajian ini amat berbeza dalam banyak aspek penting, seperti panjang gelombang cahaya yang digunakan, tempoh rawatan, bahan kimia yang digunakan dalam terapi photodynamic, dan lain-lain.

Lebih separuh daripada kajian adalah tajaan industri; pengarang kajian melaporkan sama ada konflik kepentingan, atau maklumat yang tidak diisytiharkan.

Keputusan utama

Kami tidak dapat membuat kesimpulan yang kukuh dari hasil kajian kami, kerana ia tidak jelas sama ada terapi cahaya (termasuk PDT) dinilai dalam kajian ini adalah lebih berkesan daripada pembanding lain yang diuji seperti plasebo, tiada rawatan atau rawatan disapu pada kulit, dan berapa lama kemungkinan manfaat bertahan.

Apakah matlamat yang utama bagi ulasan itu?

Kami menyiasat bagaimana orang dengan jerawat menilai pemulihan mereka sendiri, tetapi ia tidak jelas sama ada terapi cahaya dalam kajian mempunyai kesan yang baik. Bukti bagaimana penyiasat menilai perubahan dalam jumlah bintik hitam, bintik putih dan bintik radang pada orang dengan jerawat juga terhad untuk kebanyakan jenis terapi cahaya, disebabkan oleh perubahan dalam cara kajian telah dijalankan dan diukur.

Kebanyakan kajian melaporkan kesan sampingan, tetapi tidak mencukupi. Tiada parut dilaporkan, dan melepuh dilaporkan dalam kajian mengenai cahaya 'intense pulsed', cahaya inframerah dan PDT.

Tiga kajian, dengan sejumlah 360 peserta dengan sederhana kepada jerawat yang teruk, menunjukkan bahawa terapi photodynamic dengan methyl aminolevulinate (MAL), diaktifkan oleh cahaya merah, mempunyai kesan yang sama ke atas perubahan dalam jumlah bintik hitam, bintik putih dan bintik radang berbanding dengan krim tiruan dengan lampu merah. Kami menilai kualiti bukti sebagai sangat rendah.

Kajian akan datang yang rancang membandingkan keberkesanan rawatan jerawat biasa dengan terapi cahaya yang diperlukan untuk menilai kesan klinikal benar dan kesan sampingan terapi cahaya untuk jerawat.

Bagaimanakah cara untuk mengemaskinikan ulasan ini?

Ulasan ini melibatkan kajian sehingga September 2015.

Catatan terjemahan

Diterjemah oleh Nila Pillai (Penang Medical College). Disunting oleh Sharifah Halimah Jaafar (Regency Specialist Hospital Sdn Bhd, JB). Untuk sebarang pertanyaan mengenai terjemahan ini sila hubungi nila@pmc.edu.my.

Summary of findings(Explanation)

Summary of findings for the main comparison. Light therapies (including photodynamic therapy) compared to placebo, no treatment, topical treatment and other comparators for acne vulgaris
  1. 1 We downgraded by one level because of risk of bias: unclear to high overall risk of bias in the majority of studies.
    2 We downgraded by one level because of indirectness: lack of comparisons with conventional treatments. Limited generalisation due to variation of participants (such as Fitzpatrick skin types, severity of acne etc.).
    3 We downgraded by one level because of imprecision: small sample sizes (median of 24 for 'Participant's global assessment of improvement', and median of 30 for studies on each of the other two outcomes), power calculations not reported, often unclear assignment to groups or face sides.
    4 We have not downgraded further because of inconsistency, but there was heterogeneity across studies due to diversity of populations, interventions, comparators and methods of outcome assessment.
    5 We have not downgraded further because of publication bias, however our searches identified considerable number of unpublished studies, but with no available data.
    6 Three split-face trials; one included two reports on the infrared treated sides 2/46 (4.3%) and no reports on the untreated sides (0%); one included one report on the single pass 1450 nm laser-treated side 1/11 (9%) and no reports on the double pass 1450 nm laser-treated sides (0%); one study included one report on the intense pulsed light (IPL)-treated sides 1/10 (10%) and no reports on the untreated sides (0%).
    7 Three studies on methyl aminolevulinate (MAL)-PDT, (one of which is presented in Summary of findings table 2), the second was a split-face within parallel-group trial included one report on the 37 J/cm² 80 mg/g MAL-PDT with occlusion 1/22 (4.5%) sides and no reports on the 37 J/cm² 80 mg/g MAL-PDT without occlusion sides (0%), nor on the 25 J/cm² 80 mg/g MAL-PDT with or without occlusion sides (0%). Further split-face study included one report on 160 mg/g MAL-PDT sides 1/30 (3%), and no reports on red-light-only control sides. Four 20% aminolevulinic acid (ALA )-PDT studies: one split-face trial included one report 1/44 (2.3%) on the sides with pulsed dye laser (PDL) used for activation and no reports on the untreated sides. One split-back within parallel-group included one report 1/11 (9%) in the single-treatment group on back sites with 550–700 nm light used for activation, and no reports in the multiple treatment groups on the ALA-PDT, nor ALA alone, light alone or untreated back sites in any of the groups. One parallel-group trial included one report in the arm which used a combination of IPL of 580–980 nm and bipolar radiofrequency energies for activation, and no reports in the arms which used 517 nm light or IPL-alone (600–850 nm) for activation; the number of participants per group unclear. One parallel-group trial included one report in the arm which used 20% ALA 1/45 (2%) and no reports (0%) in arms with 5%, 10% nor 15% ALA activated by 633 nm light.

Light therapies (including photodynamic therapy) for acne vulgaris

Patient or population: Mild, moderate and severe acne vulgaris
Settings: Single and multicentre, worldwide
Intervention: Light therapies including photodynamic therapy

Comparison: Placebo, no treatment, topical treatment and other comparators

OutcomesIllustrative comparative risks* (95% CI)Relative effect
(95% CI)
No of participants
(studies)
Quality of the evidence
(GRADE)
Comments
Assumed riskCorresponding risk
Control Light therapies
Participant's global assessment of improvement
Non-standardised scales
Follow-up: up to 24 weeks after final treatment
See commentSee commentNot estimable1033
(23 studies)
⊕⊝⊝⊝
very low 1,2,3

We decided not to combine the effect estimates from the different interventions. We instead rated the quality of the evidence based on the GRADE considerations. The direction and size of effect across the individual study results across the 38 different comparisons were inconsistent.

13 studies used Likert or Likert-like scales, 5 visual analogue scales, 3 other methods and in 2 studies it was unclear which method was used. In many studies last evaluation at final treatment, timing of assessment unclearly reported or not reported. 13 studies had split-face design, 8 parallel-group design, 2 split faces within parallel-group design.4,5

Investigator-assessed change in lesion counts
Lesion counts
Follow-up: up to 12 months after final treatment
See commentSee commentNot estimable2242
(51 studies)
⊕⊝⊝⊝
very low 1,2,3

We decided not to combine the effect estimates from the different interventions. We instead rated the quality of the evidence based on the GRADE considerations. The direction and size of effect across the individual study results across the 76 different comparisons were inconsistent.

Different methods for lesion counting reported including change or percentage change from baseline in the number of individual or various aggregates of counts of inflamed lesions, non-inflamed lesions, nodules and cysts.

22 studies had split-face design, 1 split-face or back design, 2 split-back design, 19 parallel-group design, 7 split-face within parallel-group design.4,5

Investigator-assessed severe adverse effects
Blistering or scarring
Follow-up: up to 12 months after final treatment
See commentSee commentNot estimable3945
(66 studies)
⊕⊝⊝⊝
very low 1,2,3
We decided not to combine the effect estimates from the different interventions. We rated the quality of the evidence based on the GRADE considerations. In most studies it was reported that adverse effects were recorded, without stating explicit intent to record blistering and scarring. No reports of scarring in any of the studies. No reports of blistering in 56 studies with a total of 3378 participants. Blistering was reported in two studies on infrared light and one study on intense pulsed light6, as well as in seven studies on photodynamic therapies (PDT)7.
*The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).
CI: Confidence interval; RR: Risk ratio;
GRADE Working Group grades of evidence
High quality: We are very confident that the true effect lies close to that of the estimate of the effect
Moderate quality: We are moderately confident in the effect estimate: The true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different
Low quality: Our confidence in the effect estimate is limited: The true effect may be substantially different from the estimate of the effect
Very low quality: We have very little confidence in the effect estimate: The true effect is likely to be substantially different from the estimate of effect

Summary of findings 2 MAL-PDT compared to red light only for acne vulgaris

Summary of findings 2. MAL-PDT compared to red light only for acne vulgaris
  1. 1 We downgraded by one level because of indirectness: comparisons with no treatment, placebo or conventional treatments not included.
    2 We have not downgraded because of risk of bias. Please note that these were industry-sponsored studies, so we judged 'other bias' as unclear. NCT00594425 had high attrition and selective reporting bias. Low risk in all other bias domain for all three studies.
    3 We have not downgraded because of inconsistency. There was some clinical heterogeneity across studies to take into account, in one study only participants with severe acne were included, in the other two studies participants with both moderate and severe acne were included (less than 20% of the included participants had severe acne in those trials).
    4 The three studies included 53, 53,and 52 participants in the control group and 100, 54 and 48 participants in the intervention group respectively.
    5 1 = almost clear; 2 = mild severity; 3 = moderate severity; 4 = severe. Success defined as improvement of at least two grades from baseline.

MAL-PDT compared to red light only for acne vulgaris
Patient or population: Moderate and severe acne vulgaris
Settings: Multicentre, USA and Canada
Intervention: 80 mg/g methyl aminolevulinate (MAL) PDT activated by red light
Comparison: Placebo cream with red light
OutcomesIllustrative comparative risks* (95% CI)Relative effect
(95% CI)
No of Participants
(studies)
Quality of the evidence
(GRADE)
Comments
Assumed riskCorresponding risk
Red light only MAL-PDT
Paticipant's global assessment of improvement - Not measured--Not estimable---
Investigator-assessed change in inflamed lesions (ILs)
Lesion counts
Follow-up: 6 weeks after final treatment
Baseline mean ILs count in the red-light-only groups was 39.9; the mean investigator-assessed change in ILs in the red-light-only groups was -10.6Baseline mean ILs count in the MAL-PDT group was 39.2; the mean investigator-assessed change in ILs in the MAL-PDT groups was
2.85 lower
(7.51 lower to 1.81 higher)
-360
(3 studies)
⊕⊕⊕⊝
moderate 1

Two additional trials not included due to clinical and methodological heterogeneity.

Assumed risk is based on weighted average of the mean ILs counts in the control groups and the corresponding risk on weighted average of the mean ILs counts in the intervention groups of the three studies2,3,4

Investigator-assessed change in non-inflamed lesions (NILs)
Lesion counts
Follow-up: 6 weeks after final treatment
Baseline mean NILs count in the red-light-only groups was 47.6; the mean investigator-assessed change in NILs in the red-light-only groups was -10.8Baseline mean NILs count in the MAL-PDT group was 45.6; the mean investigator-assessed change in NILs in the MAL-PDT groups was
2.01 lower
(7.07 lower to 3.05 higher)
-360
(3 studies)
⊕⊕⊕⊝
moderate 1

Two additional trials not included due to clinical and methodological heterogeneity.

Assumed risk is based on weighted average of the mean NILs counts in the control groups and the corresponding risk on the weighted average of the mean NILs counts in the intervention groups of the three studies2,3,4

Investigator-assessed percentage change in ILs
Lesion counts
Follow-up: 6 weeks after final treatment
Baseline mean ILs count in the red-light-only groups was 39.9; the mean investigator-assessed percentage change in ILs in the red-light-only groups was -25.7%Baseline mean ILs count in the MAL-PDT group was 39.2; the mean investigator-assessed percentage change in ILs in the MAL-PDT groups was
10.09 lower
(20.25 lower to 0.06 higher)
-360
(3 studies)
⊕⊕⊕⊝
moderate 1

Two additional trials not included due to clinical and methodological heterogeneity.

Assumed risk is based on weighted average of the mean ILs counts in the control groups and the corresponding risk on the weighted average of the mean ILs counts in the intervention groups of the three studies2,3,4

Investigator-assessed percentage change in NILs
Lesion counts
Follow-up: 6 weeks after final treatment
Baseline mean NILs count in the red-light-only groups was 47.6; the mean investigator-assessed percentage change in NILs in the red-light-only groups was -16.6%Baseline mean ILs count in the MAL-PDT group was 45.6; the mean investigator-assessed percentage change in NILs in the MAL-PDT groups was
8.09 lower
(21.51 lower to 5.32 higher)
-360
(3 studies)
⊕⊕⊕⊝
moderate 1

Two additional trials not included due to clinical and methodological heterogeneity.

Assumed risk is based on weighted average of the mean NILs counts in the control groups and the corresponding risk on the weighted average of the mean NILs counts in the intervention groups of the three studies2,3,4

Investigator-assessed severe adverse effects

Application site blister

Follow-up: during whole study period

Study populationNot estimable360
(3 studies)
⊕⊕⊕⊝
moderate 1
Scarring was not reported. Two additional trials not included due to clinical and methodological heterogeneity. Due to the lack of events occurring in both groups, the relative risk is unreliable2,3,4
Application site blister rates in the red-light-only groups were 0/158 (0%)Application site blister rates in the MAL-PDT groups were 1/202 (0.5 %)
Investigator's global assessment (IGA) of improvement
Treatment 'success' as defined by IGA score decrease5
Follow-up: 6 weeks after final treatment
Study population RR 1.74
(1.11 to 2.74)
360
(3 studies)
⊕⊕⊕⊝
moderate 1

The absolute effect was 89 more per 1000 (95% CI 13 more to 209 more). The number needed to treat for an additional treatment 'success' was 7 (95% CI 5 to 11).2,3,4

An additional trial not included due to clinical and methodological heterogeneity.

120 per 1000 209 per 1000
(133 to 329)
*The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).
CI: Confidence interval; RR: Risk ratio;
GRADE Working Group grades of evidence
High quality: We are very confident that the true effect lies close to that of the estimate of the effect
Moderate quality: We are moderately confident in the effect estimate: The true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different
Low quality: Our confidence in the effect estimate is limited: The true effect may be substantially different from the estimate of the effect
Very low quality: We have very little confidence in the effect estimate: The true effect is likely to be substantially different from the estimate of effect

Summary of findings 3 ALA-PDT compared to blue light only for acne vulgaris

Summary of findings 3. ALA-PDT compared to blue light only for acne vulgaris
  1. 1 We downgraded by one level because of indirectness: comparisons with no treatment, placebo or conventional treatments not included.
    2 We have downgraded by one level because of risk of bias.
    3 We have downgraded by two levels because of risk of bias. Means and 95% CIs were not reported.
    4 We have downgraded by two levels because of risk of bias. There were no reports of application site blisters among adverse effects, however it is possible that some occurred, but it is impossible to separate those as they were reported together with oozing and crusting under "Oozing/ Vesiculation/Crusting".
    5 Excellent = very satisfied; good = moderately satisfied; fair = slightly satisfied; poor = not satisfied at all. Success defined as improvement of at least two grades from baseline.
    6 0 = clear skin with no ILs or NILs; almost clear; rare NILs with no more than a few small ILs; Mild; > Grade 1 = some NILs with some ILs (papules/pustules only; no nodules); Moderate; > Grade 2 = up to many NILs and a moderate number of ILs but no more than one small nodule; Severe; > Grade 3 = up to many NILs and ILs, but no more than a few nodules. Success was defined as a two-point or more improvement on the IGA scale since baseline'

ALA-PDT compared to blue light only for acne vulgaris
Patient or population: Moderate and severe acne vulgaris
Setting: Multicentre, USA
Intervention: 20% aminolevulinic acid (ALA) activated by 500 s and 1000 s blue light
Comparison: Vehicle plus 500 s and 1000 s blue light
OutcomesAnticipated absolute effects* (95% CI)Relative effect
(95% CI)
№ of participants
(studies)
Quality of the evidence
(GRADE)
Comments
Risk with blue light onlyRisk with ALA-PDT

Participant's global assessment of improvement

Non-standardised scale5
Follow up: 6 weeks

Study population RR 0.87
(0.72 to 1.04)
266
(1 study)

⊕⊕⊝⊝

low 1,2

Results for 500 s ALA and 1000 s ALA groups combined under 'Intervention', as our analyses found no statistically significant difference between them. 1000 s vehicle plus blue light and 500 s vehicle plus blue light groups combined in 'Comparison' , as our analyses found no statistically significant difference between them.
602 per 1000 523 per 1000
(433 to 626)
Investigator-assessed change in inflamed lesions (ILs)
Lesion counts
Follow up: 6 weeks
Not estimable. See comment.Not estimable. See comment.Not estimable266
(1 study)

⊕⊝⊝⊝

very low 1,3

Means not reported nor provided upon request. The median investigator-assessed change (standard deviation, SD) in ILs was -21.0 (23.63) in the vehicle 1000 s, -17.0 (26.71) in the vehicle 500 s group, -18.5 (30.15) in the ALA 1000 s and -13.0 (28.74) in the ALA 500 s group.

Investigator-assessed percentage change in ILs

Lesion counts
Follow up: 6 weeks

Not estimable. See comment.Not estimable. See comment.Not estimable266
(1 study)

⊕⊝⊝⊝

very low 1,3

Means not reported nor provided upon request. The median investigator-assessed percentage change (SD) in ILs was -48.4 (32.81) in the vehicle 1000 s, -45.2 (50.15) in the vehicle 500 s group, -34.4 (37.8) in the ALA 1000 s group and -29.0 (42.57) in the ALA 500 s group.

Investigator-assessed severe adverse effects

Application site blister

Follow-up: during whole study period

Study populationNot estimable266
(1 study)

⊕⊝⊝⊝

very low 1,4

"Oozing/Vesiculation/Crusting" were evaluated at baseline, and were then assessed pre- and post-treatment & 48 h after treatment at each treatment session, as well as 3 and 6 weeks after final treatment.
0 per 10000 per 1000
(0 to 0)
Investigator's global assessment (IGA) of improvement
Treatment 'success' as defined by IGA score decrease6
Follow up: 6 weeks
Study population RR 0.81
(0.51 to 1.29)
266
(1 study)

⊕⊕⊝⊝

low 1,2

Results for 1000 s ALA and 500 s ALA groups combined under 'Intervention', as our analyses found no statistically significant difference between them. 1000 s vehicle plus blue light and 500 s vehicle plus blue light groups combined in 'Comparison', as our analyses found no statistically significant difference between them.
195 per 1000 158 per 1000
(100 to 252)
*The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).
CI: Confidence interval; RR: Risk ratio;
GRADE Working Group grades of evidence
High quality: We are very confident that the true effect lies close to that of the estimate of the effect
Moderate quality: We are moderately confident in the effect estimate: The true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different
Low quality: Our confidence in the effect estimate is limited: The true effect may be substantially different from the estimate of the effect
Very low quality: We have very little confidence in the effect estimate: The true effect is likely to be substantially different from the estimate of effect

Background

Description of the condition

Acne is a very common inflammatory skin condition that affects the face of over 90% of people some point in their lives, the chest in 60% of people, and the back in 15% (Cunliffe 1989). The condition usually starts in adolescence and frequently resolves by the mid-twenties (Bhate 2013; Burton 1971). 

Acne is characterised by an increase in sebum production; the formation of lesions called open and closed comedones (which appear as blackheads and whiteheads); raised red spots, known as papules and pustules and in more severe cases nodules; deep pustules; and cysts (Degitz 2007; Nast 2012). Acne can range from a mild form, with a few of these lesions, to more severe forms embracing multiple lesions over the face and trunk (O'Brien 1998).

Mild acne is more prevalent than the severe form (Kilkenny 1998). In some cases, acne persists, or initially starts, in adulthood, and in this situation, it is seen more commonly in adult women than men (Choi 2011; Dreno 2013; Preneau 2012; Williams 2006).

Impact

Acne results in a significant burden. One study from the USA indicated that the prevalence by the mid-teens was virtually 100% (Stern 1992). A more recent European study estimated a rate to be 82.4% in 10 to 12 year olds and identified that over 40% of people sought treatment (Amado 2006).

The duration of acne can be anything from 5 to 10 years (Cunliffe 1979). In most people, acne has resolved by the age of 25 years (Cunliffe 1979). Between 7% and 17% of those affected have clinical acne beyond this time (Goulden 1997).

Acne can produce significant psychological and social problems, and those having acne may be affected by lower self-esteem, anxiety, depression, and low mood (Baldwin 2002; Tan 2004; Thomas 2004). Scarring is a very common problem, and treatment is extremely difficult (Jordan 2000; Layton 1994; Tan 2010); scarring can also result in significant psychological and social problems (Hayashi 2015).

The treatments available for acne may result in adverse effects, which may limit their use (Nast 2012; Williams 2012). The complex pathophysiology of acne often results in the need for multiple treatments within any given regimen, and this can have impact on adherence (Dreno 2010; Krejci-Manwaring 2006). There is increasing concern about the use of antibiotics in the management of acne due to emerging bacterial resistance (Coates 2002).

Causes

Acne usually presents around puberty and arises as a result of an increase in hormone levels, particularly androgen hormones (Thiboutot 2004; Zouboulis 2004). This leads to enlargement of the sebaceous (grease) glands and an increased cell turnover resulting in blockage and plugging of the duct that carries the sebum to the skin, which leads to the formation of a comedone (whiteheads and blackheads, Cunliffe 2004). Skin bacteria, in particular Propionibacterium acnes (P acnes), become trapped within the duct, and an intense inflammatory reaction ensues, which results in the inflamed skin lesions characteristic to acne, that is, the pustules, papules, and in the worst cases, nodules and cysts (Degitz 2007; Nast 2012). Insulin resistance is one factor implicated in the development of severe acne and is a common complaint of women with polycystic ovarian syndrome (Archer 2004; Pfeifer 2005).

Conventional treatments

First-line treatments in Europe include fixed combinations of benzoyl peroxide (BPO) with adapalene or clindamycin for mild-to-moderate papulopustular acne, whereas isotretinoin is recommended for more severe forms of acne (Nast 2012). Recent guidelines published by the American Academy of Dermatology (AAD) also recommend BPO or topical retinoid, or topical combination therapy including BPO with or without antibiotic for mild acne, however separate components, as well as fixed combination products may be prescribed (Zaenglein 2016). Topical combination therapy for moderate acne may also be prescribed together with an oral antibiotic for moderate and severe acne as a first line treatment (Zaenglein 2016). As in Europe, isotretinoin is only recommended for more severe forms of acne as a first line treatment (Zaenglein 2016). Systemic antibiotics in combination with adapalene, azelaic acid, or a fixed combination of adapalene and BPO are recommended for more severe forms of acne (Nast 2012).

For mild-to-moderate acne, second-line treatments in Europe include topical treatments such as azelaic acid, BPO, or topical retinoids; however, systemic antibiotics in combination with adapalene can also be considered (Nast 2012). Alternative treatment suggested by the AAD guidelines for mild forms of acne include adding topical retinoid or BPO if they have not been part of the combination already, and considering alternate retinoid or topical dapsone (Zaenglein 2016). Alternative treatment for moderate forms of acne include alternating combination therapies, whereas, for both moderate and severe acne, changes in oral antibiotics, adding combined oral contraceptive or oral spironolactone for women, as well as oral isotretinoin may be considered (Zaenglein 2016).

Topical treatments target the plugged follicle and the bacteria implicated in acne as well as inflammation (Nast 2012). It is now recommended that topical antibiotics should not be used alone as they can lead to antibiotic resistance (Nast 2012). All antibiotics employed for acne should be used alongside anti-resistant agents in the treatment of moderate acne, that is, agents that reduce antibiotic-resistant strains of P acnes and avoid emergence of novel resistant strains (Nast 2012).

Women with acne may be prescribed hormone therapies, which are also used as combined oral contraceptives (Arowojolu 2012; Zaenglein 2016). Oral isotretinoin, which is a synthetic form of vitamin A, is very effective for moderate nodular and severe papulopustular acne (Nast 2012). For the majority of people following a course of isotretinoin, their skin clears fully by the end of a course of therapy; however, in some cases, the acne will recur (White 1998). Side-effects from oral isotretinoin include dry lips, eyes, skin, and mucous membranes (Charakida 2004). Isotretinoin is also teratogenic, meaning that if a woman becomes pregnant whilst taking isotretinoin, it is likely to cause birth defects (Lammer 1985). This limits its use in women of childbearing age (Abroms 2006; Stern 1989).

Description of the intervention

Light therapies utilise light with different properties (wavelength, intensity, coherent or incoherent light) with the aim of achieving a beneficial result for those with acne (Haedersdal 2008a; Mariwalla 2005). Lasers (Light amplification by stimulated emission of radiation) (Leinwoll 1965) are the most common light sources that have been used for acne therapy. Lasers produce a high-energy beam of light of a precise wavelength range, which can be focused accurately (Haedersdal 2008a; Mariwalla 2005). Several different delivery systems are used, incorporating timing controls for safety, and cooling systems to reduce discomfort during treatment (Haedersdal 2008a; Hamilton 2009; Mariwalla 2005).

How the intervention might work

The exact mechanisms of action for light therapies are still not fully understood, but three components of the intervention are considered crucial: light, photosensitisers (i.e. molecules that absorb and are then activated by light), and oxidative stress resulting from their activation (Fritsch 1998; Mariwalla 2005; Sakamoto 2010). Photosensitisers can be produced endogenously or applied exogenously (Fritsch 1998). Probable biological consequences of oxidative stress include damaging bacteria and sebaceous glands, together with reduction of follicular obstruction and hyperkeratosis (Mariwalla 2005; Sakamoto 2010). Possible interference with the immunological response, not necessarily mediated by photosensitisers, are also believed to be important (Sakamoto 2010).

Different wavelengths have different effects on P acnes bacterial colonies in vitro (Cho 2006). However, the evidence on in vivo reduction of P acnes is limited, although different light therapies have had different effects on outcomes in clinical trials (Haedersdal 2008a; Hamilton 2009).

P acnes produces endogenous porphyrins, which absorb light to form a highly reactive singlet oxygen, which destroys the bacteria (Mariwalla 2005). The peak absorption occurs at blue light wavelengths, providing a rationale for selecting blue light as a logical wavelength when using physical therapy for acne (Mariwalla 2005). However, red light is also absorbed by porphyrins and can penetrate deeper into the skin where it may directly affect inflammatory mediators (Mariwalla 2005; Ross 2005). Other light therapies, including infra-red lasers, low energy pulsed-dye lasers (PDL), and radiofrequency devices (Mariwalla 2005), are directed towards damaging sebaceous glands, reducing their size and thus sebum output (Lloyd 2002). Photodynamic therapy (PDT) uses specific light-activating topical products, consisting of various porphyrin precursors, most commonly 5-aminolevulinic acid (ALA) and its methyl-ester methyl-aminolevulinate (MAL) (Sakamoto 2010a). These are absorbed into the skin and amplify the response to light therapy, but in so doing, tend to produce more side-effects (Sakamoto 2010a).

Since the 1970s the mechanism of action of PDT has been better known for the treatment of malignancies than for other uses in dermatology (Fritsch 1998; Sharma 2012). Photosensitisers used in PDT probably accumulate inside gram-positive bacteria (such as P acnes), and when activated, a type I reaction is induced, producing hydroxyl radicals, a leak-out of cellular contents, and death of the microbial cells (Sharma 2012). Differences in pharmacokinetic characteristics of drugs used in PDT, their incubation time, whether they were administered under occlusion or not, their ability to penetrate the intrafollicular duct, alongside wavelengths and doses of light used for activation, as well as care applied before and after the treatment, are all confounding factors likely to affect clinical results (Sakamoto 2010a). Sakamoto et al suggested two dose-related PDT mechanisms of action: 'low dose' PDT ('low drug concentration, low light fluence, short incubation time between drug application and light exposure, use of blue light with minimal penetration depth, and/or various pulsed source exposures') is probably mainly based on transient antimicrobial or immunomodulatory effects, whereas 'high dose' PDT ('prolonged application of high ALA concentration followed by high fluence red light') is based mainly on damaging sebaceous glands (Sakamoto 2010). Optimal regimens have not yet been established (Sakamoto 2010a). There is an ongoing debate on whether lack of selectivity of the photosensitisers could lead to substantial damage to the surrounding tissue and subsequent necrosis (Sharma 2012).

Why it is important to do this review

Current treatment options may be limited in effectiveness or acceptability due to adverse effects, poor tolerability and the inconvenience of using them on a regular and prolonged basis (Nast 2012; Williams 2012; Zaenglein 2016). Conventional treatments have limitations. Most oral and topical treatments are less effective than oral isotretinoin, but the latter has significant adverse effects (Nast 2012; Williams 2012). Combination regimens, which are required for the treatment of acne, are often complex for a person to use, are time-consuming, and can result in poor adherence (Dreno 2010). Increasing concern about the use of antibiotics for acne has emerged due to the rise in antibiotic-resistant bacteria (Nast 2012). If we were able to identify alternative therapies that addressed some of these issues, it would clearly be advantageous to patients, the wider community, and prescribers. This is highlighted by the fact that the Acne Priority Setting Partnership, which received responses from over 8000 clinicians, patients, and carers placed the question of safety and effectiveness of physical therapies, including lasers and other light-based treatments, in treating acne among the top 10 research priorities (Layton 2015). Light therapies seem to be increasingly popular, and many light sources are now offered for people to purchase directly using the Internet. Therefore, there is a lot of public interest in this treatment, as well as interest from health service commissioners.

To date, the evidence regarding the efficacy of light and laser interventions is not robust (Nast 2012; Zaenglein 2016 ). There have been few studies comparing lasers and light therapies with conventional acne treatments, or studies using physical therapies in severe acne, or any evaluation of the long-term benefit of these treatments (Hamilton 2009), and so there is still uncertainty and controversy (Sanclemente 2014; Williams 2012). European guidelines (Nast 2012) gave negative recommendations for artificial ultraviolet (UV) radiation in mild, moderate, and severe papulopustular acne and for visible light as monotherapy in severe papulopustular acne. Blue light monotherapy is recommended with a low strength of recommendation for treatment of mild to moderate papulopustular acne (Nast 2012). Because of a lack of evidence, Nast 2012 left recommendations open for visible light of other wavelengths as monotherapy, lasers with infrared wavelengths, intense pulsed light (IPL), and PDT for mild to moderate and severe papulopustular acne. This is somewhat contradictory to the European guidelines for topical PDT, where inflammatory and infectious dermatoses are seen as an "emerging indication", and acne has the highest strength of recommendation, with the evidence rated as of highest possible quality (Morton 2013). Recently updated American guidelines included lasers and PDT as a new clinical question, but are not explicit in stating the strength of their recommendation, nor levels of underlying evidence (Zaenglein 2016). The study authors concluded that there was "limited evidence to recommend the use and benefit of physical modalities for the routine treatment, including pulsed dye laser..." and that "Some laser and light devices may be beneficial for acne, but additional studies are needed" (Zaenglein 2016). Zaenglein 2016 have also included clinical trials of lasers and light-based therapies as one of the most important current research and knowledge gaps to address in acne treatment.

The worldwide market potential for anti-acne skin preparations alone was estimated to be USD 3300 million in 2013 (GMR Data 2013). The growing market and the willingness of people to take up treatments that have not been clinically proven to be effective means that research into the use and marketing of novel treatments, such as light therapies, is important. If light therapies prove effective, they could offset the cost of acne-related treatments. If, however, light therapies are ineffective, their use should be stopped.

Hence, establishing the evidence to support treatment of acne with light of different wavelengths is critical. The plans for this review were published as a protocol 'Light therapies for acne' (Car 2009).

Objectives

To explore the effects of light treatment of different wavelengths for acne.

Methods

Criteria for considering studies for this review

Types of studies

Randomised controlled trials (RCTs), which were of two types: those which compared two groups of participants where one group was randomised to receive treatment and the other served as the control group; and those which applied treatment randomly to one part of a participant's body compared with another part which served as the control (such as split-face studies).

We did not include cross-over trials because an intervention for acne may have had a lasting effect that could have carried over to subsequent periods of the trial.

Types of participants

Anyone with a diagnosis of mild, moderate, or severe acne vulgaris defined by any classification system.

Types of interventions

We searched for any therapy based on the healing properties of light for the treatment of acne vulgaris. We also accepted therapies that combined light with other treatments to boost the effect of the light. We focused on a comparison between the effectiveness of treatment with light of different properties - coherence, wavelength, and intensity.

Types of outcome measures

Primary outcomes
  1. Participant's global assessment of improvement. This was recorded using a Likert or Likert-like scale (for instance, selecting from the following categories the extent of change of their acne after treatment: acne has worsened a lot; worsened a little; stayed the same; improved a little; or improved a lot) or other scales.

  2. Investigator-assessed change in lesion count.

    1. The change or percentage change from baseline in the number of:

      1. inflamed lesions (ILs) (papules or pustules or both);

      2. non-inflamed lesions (NILs) (blackheads or whiteheads or both); or

      3. nodules and cysts (for nodulocystic acne only

    2. If individual lesion counts were not available, then the change or percentage change from baseline in the number of:

      1. ILs and NILs; or

      2. combined count of all lesion types.

  3. Investigator-assessed severe adverse effects. If blistering or scarring of the skin followed treatment with light therapy then, if possible, we reported on the severity of the adverse effect and whether it resolved in the short-term or was permanent.

Secondary outcomes
  1. Investigator-assessed change in acne severity. The change in acne severity from baseline, using a published grading scale (like the Leeds grading system, which involves counting lesions and weighting them according to severity to give a combined grade) or a severity index determined by the lesion count.

  2. Investigator's global assessment of improvement recorded using a Likert or Likert-like scale or other scales.

  3. Changes in quality of life assessed using a recognised tool.

Other adverse outcomes

We recorded the incidence and, when possible, severity of all other adverse events reported in the included studies. We used the system organ classes (SOCs) defined in MedDRA (MedDRA 2010), version 15.1. MedDRA® ('the Medical Dictionary for Regulatory Activities, terminology is the international medical terminology developed under the auspices of the International Conference on Harmonization of Technical Requirements for Registration of Pharmaceuticals for Human Use (ICH). MedDRA® trademark is owned by the International Federation of Pharmaceutical Manufacturers & Associations (IFPMA) on behalf of ICH').

Timing of outcome assessment

We considered short-term (two to four weeks after final treatment), medium-term (five to eight weeks after final treatment), and long-term (longer than eight weeks after final treatment) follow-up periods. The long-term data were the primary endpoint, but we were also interested in short-term data, indicating early improvement, which may have encouraged participants to continue with treatment.

Exclusion criteria
  1. Studies which were not RCTs.

  2. Studies not focused on the healing properties of light in the management of acne.

  3. Studies on light therapies for acne scars.

Search methods for identification of studies

We aimed to identify all relevant RCTs regardless of language or publication status (published, unpublished, in press, or in progress).

Electronic searches

We searched the following databases up to 29 September 2015:

  1. the Cochrane Skin Specialised Register using the following terms: acne and (laser* or sunlight or phototherap* or photolysis or photochemotherapy or “ ultraviolet therap*” or “photosensitizing agent*” or “light therap*” or “photodynamic therap*” or “photosensitising agent*”);

  2. the Cochrane Central Register of Controlled Trials (CENTRAL; the Cochrane Library 2015, Issue 8) using the search strategy in Appendix 1;

  3. MEDLINE via Ovid (from 1946) using the strategy in Appendix 2;

  4. Embase via Ovid (from 1974) using the strategy in Appendix 3; and

  5. LILACS (Latin American and Caribbean Health Science Information database, from 1982) using the strategy in Appendix 4.

We searched the following databases up to 28 September 2015:

  1. ISI Web of Science using the strategy in Appendix 5; and

  2. Dissertation Abstracts International (1861) using the strategy in Appendix 6.

Trials registers

We searched the following trials registers up to 28 September 2015:

  1. The metaRegister of Controlled trials (isrctn.com/) using the strategy in Appendix 7.

  2. The U.S. National Institutes of Health Ongoing Trials Register (clinicaltrials.gov) using the strategy in Appendix 8.

  3. The Australian and New Zealand Clinical Trials Registry (anzctr.org.au) using the strategy in Appendix 9.

  4. The World Health Organization International Clinical Trials Registry Platform (who.int/ictrp/en/) using the strategy in Appendix 10.

  5. The EU Clinical Trials Register (clinicaltrialsregister.eu/) using the strategy in Appendix 11.

This review fully incorporates the results of searches conducted up to September 2015. A search update conducted in July 2016 identified a further 15 reports of trials, which we have added to ‘Studies awaiting classification’ and will incorporate into the review at the next update. See Characteristics of studies awaiting classification.

Searching other resources

Grey literature

We attempted to find unpublished studies by searching the following grey literature:

  1. Google Scholar using the strategy in Appendix 12 up to 7 October 2015; and

  2. OpenGrey using the strategy in Appendix 13 up to 29 September 2015.

We also used Internet search engines such as Google.

We consulted trial authors of included and excluded trials published in the last 15 years and other experts in the field of optical therapies for acne, in order to identify further unpublished RCTs.

Reference lists

We checked the bibliographies of published studies and reviews for further references to relevant trials.

Adverse effects

We did not perform a separate search for adverse effects of the target intervention. We recorded adverse effects reported in the included trials and discussed the implications of those adverse outcomes.

Data collection and analysis

We followed the protocol for this review (Car 2009). When this was not possible, we clearly stated and further clarified it in the Differences between protocol and review section.

Selection of studies

Two review authors (JB and RA, PP or MC) screened the titles and abstracts of studies identified by the searches. If studies did not address the study of a light therapy for acne, we excluded them. If any of the review authors felt that a paper could have been relevant, we retrieved the full text, and each author independently checked that it met the pre-defined selection criteria. We resolved differences of opinion by discussion with the review team.

Data extraction and management

Two review authors (JB and RA or MC) independently recorded data using a specially designed data extraction form. When data were available only in graph or figure format, two review authors (JB and RA or MC) extracted them independently. A third team member (JC or LG) resolved any differences of opinion. One author (JB) inserted the data into Review Manager (RevMan) (RevMan 2014). Two review authors (MC and LG, RA or PP) cross-checked the data for accuracy.

We defined treatment success as anything above the first category of improvement on a Likert scale or more than 50% improvement from baseline on a continuous scale for participant's global assessment of improvement (primary outcome 1) and secondary outcomes 1, 2, and 3. When individual patient data were not available, we extracted summary data as they were reported. Effects of interventions on investigator-assessed change in lesion count (primary outcome 2) were recorded as the actual or percentage change from baseline.

In addition we reported on the following:

  1. the baseline and comparisons of the participants for age, sex, duration, location, and severity of acne;

  2. light source identity, dose, duration of treatment, and adequacy of instructions if self-administered;

  3. whether outcome measures were described and their assessment was standardised;

  4. whether previous acne treatment was discontinued in a timely manner prior to the trial;

  5. whether concomitant acne treatment was permitted and if so, whether standardised; and

  6. the use and appropriateness of statistical analyses, where data were not reported appropriately in the original publication.

Assessment of risk of bias in included studies

Two review authors (JB and RA or MC) used Cochrane's tool for assessing risk of bias, described in section 8.5 of the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011a), to independently assess the methodological quality of each included study. We assessed the following as 'low risk of bias', 'high risk of bias', or 'unclear risk of bias':

  1. how the randomisation sequence was generated;

  2. whether allocation was adequately concealed;

  3. whether participants, clinicians, or outcome assessors were blinded as appropriate, who was blinded and not blinded (participants, clinicians, outcome assessors) if this was appropriate;

  4. incomplete outcome data and how it was addressed;

  5. possible selective outcome reporting; and

  6. possible other bias.

We compared the assessments and discussed and resolved any disagreements in the gradings between the review authors. We also contacted the corresponding researchers for clarification or additional data when necessary.

Measures of treatment effect

We expressed the results as risk ratio (RR) and 95% confidence intervals (CIs) for dichotomous outcomes. When the relative risk was unreliable due to the lack of events occurring in control groups or body sites, we provided event rates instead of RR and calculated risk differences (RD) with 95% CI. We clarified this in the Effects of interventions section, under 'Primary outcome 3'. Although there were no cases where standardised mean differences were needed, we would have computed them if cases existed where comparable measures on different scales had been used across trials. We used only mean differences where appropriate (Deeks 2011). We expressed the results as 'number needed to treat for an additional beneficial outcome' (NNTB) and 'number needed to treat for an additional harmful outcome' (NNTH) for dichotomous outcomes where appropriate, following guidance in section 12.5.2 of the Cochrane Handbook for Systematic Reviews of Interventions (Schünemann 2011a).

Unit of analysis issues

Where there were multiple intervention groups within a trial, we made pair-wise comparisons of light therapies with different wavelengths versus no treatment, placebo, and conventional treatment. When the level of clinical and methodological heterogeneity was acceptable, we considered pooling studies that had a split-face or split-back design with studies that had a parallel-group design in a meta-analysis using the inverse variance method, described in the Cochrane Handbook for Systematic Reviews of Interventions section 9.4.3 (Deeks 2011). However, we did not pool studies with different designs due to the nature of the results, as there was considerable methodological and clinical heterogeneity, which is outlined in the Effects of interventions section.

Dealing with missing data

If participant drop-out led to missing data, we conducted an intention-to-treat (ITT) analysis. We contacted trial authors or sponsors of studies that were less than 15 years old to provide missing statistics, such as standard deviations. For dichotomous outcomes, we regarded participants with missing outcome data as treatment failures (to be conservative) and included these in the analysis as an imputed value. For continuous outcomes, we imputed missing outcomes by carrying forward the last recorded value for participants with missing outcome data (Higgins 2011b).

Assessment of heterogeneity

We followed updated guidance in sections 9.4.1 and 9.5.1 of the Cochrane Handbook for Systematic Reviews of Interventions (Deeks 2011) on the appropriateness of meta-analysis. To determine whether it would be clinically meaningful to quantitatively combine results of different studies, we considered differences in interventions (wavelengths, doses, active substances used in PDT, number of light sessions, and frequency of application) together with differences in comparator interventions (no treatment, placebo, other light interventions, and various topical treatments and their various combinations). For comparisons where no substantial clinical diversity existed with regard to the above, we assessed statistical heterogeneity using the I² statistic (Higgins 2003) and synthesised data using meta-analysis techniques when appropriate (i.e. when I² statistic was lower than 50%) following guidance in section 9.5.2 of the Cochrane Handbook for Systematic Reviews of Interventions (Deeks 2011).

Assessment of reporting biases

We planned to test publication bias by the use of a funnel plot when adequate data were available for similar light therapies, following guidance in section 10.4 of the Cochrane Handbook for Systematic Reviews of Interventions (Sterne 2011). However, we were unable to implement this method in the current review and test publication bias by the use of a funnel plot due to the nature of our results.

Data synthesis

For studies with acceptable levels of clinical and methodological heterogeneity, we performed a meta-analysis to calculate a weighted treatment effect across trials, using a random-effects model. Where it was not possible to perform a meta-analysis due to substantial clinical and methodological heterogeneity, we narratively synthesised the results, following guidance in section 11.7.2 of the Cochrane Handbook for Systematic Reviews of Interventions (Schünemann 2011b).

Subgroup analysis and investigation of heterogeneity

If substantial statistical heterogeneity (I² statistic greater than 50%) existed between studies for the primary outcome, we looked for the reasons for this, such as differences in disease severity, exposure, and duration of treatment. We planned to undertake further subgroup analysis if sufficient information was given. The groups were to include those with different severity or onset of acne and the age of participants (child or adult). However, subgroup analyses were not performed in the current review due to the nature of the results of the meta-analyses (the I² statistic was lower than 50% for primary outcomes).

Sensitivity analysis

We intended to undertake sensitivity analyses to determine the effects of excluding the poorer quality trials and those with an unclear or high risk of bias as defined in theCochrane Handbook of Systematic Reviews of Interventions (Deeks 2011).

Adverse outcomes

We described:

  1. whether the methods used to record adverse events were appropriate; and

  2. whether reporting of adverse outcomes was adequate.

Other

Where necessary, we contacted the trial authors for clarification.

We created 'Summary of findings' tables using GRADEpro Guideline Developement Tool (GRADEpro GDT 2015).

Results

Description of studies

Please see the Characteristics of included studies tables, Characteristics of excluded studies tables, Characteristics of studies awaiting classification tables, and Characteristics of ongoing studies tables in this review.

Results of the search

The 'Study flow diagram' summarises the results of our incorporated searches up to September 2015 (see Figure 1). We identified 862 records through searching the Cochrane Skin Specialised Register, CENTRAL, MEDLINE, Embase and LILACS. We identified a further 907 records through searching ISI Web of Science and Dissertation Abstracts International. We identified 51 records through other searches. (Please see 'Clinical trials registers and 'Grey literature searches' section below for details.)

Figure 1.

Study flow diagram.

Our searches retrieved a total of 1820 records. We removed 1018 duplicates leaving 802 records. We excluded 648 records based on the titles and abstracts. We obtained full text copies of the remaining 154 records when appropriate. After assessing full texts, we excluded 25 records (corresponding to 24 studies) for reasons outlined in the Characteristics of excluded studies tables.

We included a total of 98 records in a narrative synthesis (corresponding to 71 studies). We were unable to obtain enough information to include or exclude 28 records (corresponding to 23 studies), which we listed in the Characteristics of studies awaiting classification tables. A further three studies are ongoing (EU 2014-005235-13; NCT02217228; NCT02431494).

We included three studies in a quantitative meta-analysis (NCT00594425; NCT00933543; Pariser 2013).

We only included final results of the clinical trials registers and grey literature searches in the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) chart for reasons of clarity (Figure 1; Moher 2009).

Our final searches in July 2016 identified 13 additional studies (14 references): Demina 2015; Du 2015; Elgendy 2015; Ganceviciene 2015; Kwon 2016; Lekakh 2015; Moftah 2016; NCT02647528; Nestor 2016; Park 2015; Sadick 2016; Voravutinon 2016; Wang 2016. We have added a further report as a secondary reference to a previously identified study (Pariser 2013). We will incorporate the additional studies into the next update of this review.

Clinical trials registers and grey literature searches

Clinical trials registers and Open Grey returned a total of 377 records. Of these, 33 identifiers were relevant for the review. We matched 12 identifiers to 11 included studies identified through searches of other databases (Bissonnette 2010; Darne 2011; Haedersdal 2008; Hörfelt 2006; Karsai 2010; McGill 2008; Orringer 2007; Orringer 2010; Pariser 2013 (two identifiers); Uebelhoer 2007; Wiegell 2006b), while one identifier was matched to two separate studies, one included (Paithankar 2015) and one excluded (Owczarek 2014). We matched two identifiers to one study awaiting classification (Shaheen 2011). We were unable to match 18 identifiers with any of the studies identified through searches of other databases. They corresponded to 17 studies, as one study (NCT00237978) was registered in two different registers. We excluded one of these studies after contacting the study authors for clarification (NCT00613444). We obtained full results for three studies (NCT00594425; NCT00673933; NCT00933543) and results of one study were available in the register (NCT00706433), so we included them in our analysis. Nine are among studies awaiting classification (NCT00237978 (two identifiers); NCT00814918 ; NCT01245946; NCT01472900; NCT01584674; NCT01689935; ISRCTN73616060; ISRCTN78675673; ISRCTN95939628). Three studies are ongoing (EU 2014-005235-13; NCT02217228; NCT02431494).

A search of Google Scholar retrieved 963 records, and after screening, we found nine records of potentially relevant studies not identified through searches of the other databases.

We identified nine additional records through other sources (including authors' suggestions, reference lists of papers, and a Google search).

We have described our attempts to contact the authors of individual studies in the 'Notes' sections of the Characteristics of included studies tables, Characteristics of studies awaiting classification tables, Characteristics of ongoing studies tables, or Characteristics of excluded studies tables.

Included studies

We included 71 studies, with a total of 4211 included participants, of which 40 were studies of light therapies, excluding comparisons with photodynamic therapy (PDT) and randomised a total of 2485 participants, and 31 were studies of PDT (including comparisons with light therapies) which included a total of 1726 participants. Please see the Characteristics of included studies tables for details.

Design

All included studies were RCTs. Most had a parallel-group design (40 studies), or a split-face design (28 studies), two had a split-back design (NCT00673933; Pollock 2004), and one had a split-face and split-back design (Barolet 2010).

Eleven of the 40 studies above had a parallel-group design, but within each group, a different intervention was administered to each side of the face or other body part; six studies with such a design randomised both groups and face sides (Bissonnette 2010; Oh 2009; Orringer 2004; Seaton 2003; Yeung 2007; Yilmaz 2011); two studies randomised groups, but not face sides (Liu 2014; Yin 2010); three other studies randomised participants to groups, but it was unclear whether within those groups, treatments were also randomly applied to one part of a participant's body compared with another part that served as control (Genina 2004; Hongcharu 2000; Sami 2008).

Most studies reported, or study authors later provided information that ethical approval was obtained, but this was unclear in 22 studies (Baugh 2005; Bernstein 2007; Bowes 2003; Cheng 2008; Elman 2003; Fadel 2009; Genina 2004; Gold 2011; Hongcharu 2000; Jih 2006; Kim 2009; Ling 2010; Liu 2014; NCT00706433; Ou 2014; Papageorgieu 2000; Sadick 2010a; Taub 2007; Tzung 2004; Zhang 2009a; Zhang 2013a; Zhang 2013b).

The majority of studies reported, or later provided information regarding sponsorship and conflict of interest, but this remained unclear for 20 studies (Bernstein 2007; Borhan 2014; Bowes 2003; Chen 2015; Cheng 2008; de Arruda 2009; Elman 2003; Hong 2013; Ling 2010; Liu 2011; McGill 2008; Na 2011; Ou 2014; Papageorgieu 2000; Pollock 2004; Sami 2008; Tzung 2004; Zhang 2009a; Zhang 2013a; Zhang 2013b). The authors of 20 studies declared no conflict of interest and no commercial sponsors (Anyachukwu 2014; Chang 2007; Choi 2010; Fadel 2009; Ianosi 2013; Jung 2009; Jung 2012; Karsai 2010; Kim 2009; Lee 2010; Leheta 2009; Liu 2014; Mei 2013; Na 2007; Oh 2009; Song 2014; Wiegell 2006a; Wiegell 2006b; Yilmaz 2011; Yin 2010). In 25 studies, the authors declared some sort of conflict of interest or were industry sponsored (Ash 2015; Barolet 2010; Baugh 2005; Bissonnette 2010; Darne 2011; Genina 2004; Gold 2005; Gold 2011; Haedersdal 2008; Hongcharu 2000; Hörfelt 2006; Jih 2006; NCT00594425; NCT00673933; NCT00706433; NCT00933543; Orringer 2004; Orringer 2007; Paithankar 2015; Pariser 2013; Seaton 2003; Taub 2007; Uebelhoer 2007; Wang 2006; Yeung 2007). In five studies, the authors declared that they had no conflicts of interest, but it was unclear who provided the device or the sham device (Kwon 2013) or whether there was commercial sponsorship (Moneib 2014; Ragab 2014; Sadick 2010a; Sadick 2010b). One study had non-commercial sponsors but it was unclear whether the authors had some sort of conflict of interest (Orringer 2010).

Only 18 studies clearly performed power calculations (Ash 2015; Barolet 2010; Bissonnette 2010; Darne 2011; Gold 2005; Hörfelt 2006; Karsai 2010; Ling 2010; NCT00594425; NCT00933543; Orringer 2004; Orringer 2007; Orringer 2010; Pariser 2013; Sadick 2010b; Seaton 2003; Wiegell 2006b; Yeung 2007).

Sample sizes

Individual sample sizes varied from 7 to 738, with an average sample size of 59 participants and median size of 31 participants. Studies of light-only therapies, excluding comparisons with PDT, had an average sample size of 62 and median size of 36.5 participants. Studies of PDT (including comparisons with light therapies) had an average sample size of 56 and median size of 25 participants.

Twelve studies randomised more than 100 participants (Ianosi 2013; Ling 2010; Liu 2014; NCT00594425; NCT00706433; NCT00933543; Papageorgieu 2000; Pariser 2013; Yin 2010; Zhang 2009a; Zhang 2013a, Zhang 2013b); five studies randomised 60 to 90 participants (Cheng 2008; de Arruda 2009; Karsai 2010; Ou 2014; Sadick 2010b).

Setting

Most studies were performed in a single centre or it was unclear whether they were single or multicenter. Only 13 studies were clearly multicenter (Gold 2005; Hörfelt 2006; Kwon 2013; Ling 2010; NCT00594425; NCT00673933; NCT00706433; NCT00933543; Paithankar 2015; Pariser 2013; Sadick 2010b; Tzung 2004; Uebelhoer 2007).

Twenty-seven studies were performed in Asia, 21 in North America, 14 in Europe, seven in Africa, and one in South America (de Arruda 2009). No studies were conducted in Australia. One multicenter study, Sadick 2010b, was conducted in North America and Asia.

Study authors reported several means of recruitment. The most common way was through outpatient clinics and dermatology departments - reported in 33 studies. Around one third of studies (23) did not describe recruitment methods.

Participants

The lowest age as an inclusion criterion was nine years. The age of included participants ranged from 11 to 59 years. In 46 studies, the mean age of included participants was between 20 and 30 years, and 38 of these studies also reported age ranges of included participants (means of age ranges were 17 to 37 years, medians of age ranges 18 to 37.5 years). Seven studies had a mean age lower than 20 (de Arruda 2009; Elman 2003; Hörfelt 2006; Karsai 2010; NCT00933543; Pariser 2013; Ragab 2014) and three, higher than 30 (Gold 2005; McGill 2008; Wang 2006).

Two studies reported no data on age (Bowes 2003; Na 2011), three reported only the inclusion criterion (Ash 2015; Fadel 2009; Wiegell 2006a), one study reported on median age and inclusion criterion only (Ianosi 2013), six reported only the age range (Genina 2004; Hong 2013; Kwon 2013; Pollock 2004; Seaton 2003; Zhang 2013a), and two reported the age range and inclusion criterion (Haedersdal 2008; Leheta 2009).

Most studies enrolled both male and female participants. One study was female only (Chang 2007), and one was male only (Anyachukwu 2014). Sex of participants was unclear in 10 studies (Bowes 2003; Fadel 2009; Jung 2009; Jung 2012; Leheta 2009; Na 2011; Taub 2007; Tzung 2004; Wiegell 2006aWiegell 2006b).

All studies included participants with clinically evident acne. Most studies included participants with mild to moderate acne (27 studies) or moderate to severe acne (18 studies). Four studies did not report severity of acne assessment when including the participants (Bernstein 2007; Jung 2012; Na 2011; Orringer 2010).

Most studies defined severity by various grading scores (34 studies). Twelve studies defined severity using lesion counts (Gold 2005; Haedersdal 2008; Ianosi 2013; Jih 2006; NCT00673933; Papageorgieu 2000; Sadick 2010b; Uebelhoer 2007; Wiegell 2006a; Wiegell 2006b; Yeung 2007; Yilmaz 2011), and eleven studies used both grading scores and lesion counts (Barolet 2010; Bissonnette 2010; Darne 2011; Hörfelt 2006; NCT00594425; NCT00706433; NCT00933543; Paithankar 2015; Pariser 2013; Seaton 2003; Taub 2007). It was unclear how ten studies performed severity assessment when including participants (Baugh 2005; Bowes 2003; Elman 2003; Fadel 2009; Genina 2004; Kim 2009; Leheta 2009; Na 2007; Tzung 2004; Wang 2006).

Studies included participants with different skin responses to sun exposure, that is, different phototypes. According to the commonly used Fitzpatrick's classification, phototypes range from type I (pale white skin which always burns and never tans) to type VI (deeply pigmented dark brown to black skin which never burns and tans very easily) (Fitzpatrick 1988). Ten studies included participants with Fitzpatrick Skin Types (FPTs) I to III (Barolet 2010; Baugh 2005; Bernstein 2007; Haedersdal 2008; Hörfelt 2006; Karsai 2010; McGill 2008; Paithankar 2015; Sadick 2010a; Yilmaz 2011), and five studies, FPT I to IV (Bissonnette 2010; Gold 2011; Hongcharu 2000; Ianosi 2013; NCT00594425). Eight studies included FPT III to IV (Borhan 2014; Chang 2007; Liu 2011; Oh 2009; Sami 2008; Song 2014; Tzung 2004; Yin 2010), and four studies included participants with FPTs III to V ( Choi 2010; Jung 2012; Kwon 2013; Ragab 2014). Three studies included FPT II-IV (Mei 2013; Taub 2007; Wang 2006), two included FPT V to VI (Anyachukwu 2014; NCT00673933), two included FPT IV to V (Hong 2013; Yeung 2007), one included only FPT III (Lee 2010) and 12 studies included participants with 4 or more different FPTs from I to VI (Ash 2015; Darne 2011; Jih 2006; NCT00706433; NCT00933543; Orringer 2007; Orringer 2010; Pariser 2013; Pollock 2004; Sadick 2010b; Wiegell 2006b). Twenty-four studies did not report FPTs.

Interventions

We observed a substantial heterogeneity in interventions. To present them in a clearer way, we first separated studies of light-only therapies (excluding comparisons with PDT and studies of PDT (including comparisons with light-only therapies)). We then made subgroups according to comparison interventions (such as placebo or no treatment, topical treatments, and other comparisons) and wavelengths used in light interventions. To describe light of different wavelengths, we used corresponding colours ('green light' for wavelengths 495 to 570 nm, 'yellow light' for wavelengths 570 to 590 nm etc.). We additionally grouped PDT studies according to active substances used: methyl aminolevulinate (MAL), aminolevulinic acid (ALA), MAL versus ALA, and other active substances.

Below we have listed light-only studies from 1 to 3 and PDT studies from 4 to 7, as well as their subgroups. If a study had more than one comparison, we listed it for every comparison it included.

1. Light versus placebo or no treatment

a) Green light versus placebo: three studies (Baugh 2005; Bowes 2003; Yilmaz 2011)
b) Yellow light versus placebo or no treatment: two studies (Orringer 2004; Seaton 2003)
c) Infrared light versus no treatment: three studies (Darne 2011; Moneib 2014; Orringer 2007)
d) Blue light versus placebo or no treatment: three studies (Elman 2003; Gold 2011; Tzung 2004)
e) Red light versus no treatment: one study (Na 2007)
f) Blue-red light versus placebo: two studies (Kwon 2013; Papageorgieu 2000)
g) Broad spectrum light versus placebo: one study (Sadick 2010b)
h) Intense pulsed light (IPL) versus no treatment: one study (McGill 2008)

2. Light versus topical treatment

a) Light versus benzoyl peroxide (BPO): three studies; one blue light (de Arruda 2009) and two blue-red light (Chang 2007; Papageorgieu 2000)
b) Light versus clindamycin: two studies (Gold 2005; Lee 2010)
c) Light and other topical treatments: seven studies (Anyachukwu 2014; Ash 2015; Borhan 2014; Ianosi 2013; Karsai 2010; Leheta 2009; Zhang 2009a)

3. Light versus other comparators

a) Comparison of light therapies of different wavelengths: seven studies (Cheng 2008; Choi 2010; Jung 2009; Liu 2011; Liu 2014; Papageorgieu 2000; Sami 2008)
b) Comparison of light therapies of different doses: four studies (Bernstein 2007; Jih 2006; NCT00706433; Uebelhoer 2007
c) Comparison of light therapies of different treatment application intervals: one study (Yilmaz 2011)
d) Light alone versus combined with microdermoabrasion: one study (Wang 2006)
e) Light in combination with carbon lotion (topical carbon suspension) versus no treatment: one study  (Jung 2012)
f) Light in combination with oral therapy versus other comparators: four studies (Ling 2010; Ou 2014; Zhang 2009a; Zhang 2013b)
g) Intense pulsed light (IPL) alone versus IPL in combination with vacuum: one study (Ianosi 2013)

4. MAL-PDT versus other comparators

a) MAL-PDT versus red light alone: five studies (Hörfelt 2006; NCT00594425; NCT00673933; NCT00933543; Pariser 2013)
b) MAL-PDT versus yellow light alone: one study (Haedersdal 2008)
c) MAL-PDT versus placebo or no treatment: one study (Wiegell 2006b)
d) MAL-PDT other: four studies (Bissonnette 2010; Hong 2013; NCT00594425; Yeung 2007)

5. ALA-PDT versus other comparators

a) ALA-PDT versus red light alone: three studies (Chen 2015; Pollock 2004; Zhang 2013a)
b) ALA-PDT versus blue light alone: one study (NCT00706433)
c) ALA-PDT versus blue-red light alone: one study (Liu 2014)
d) ALA-PDT versus IPL alone: four studies (Liu 2014; Mei 2013; Oh 2009; Ragab 2014). (Please note that different filters were used.)
e) ALA-PDT versus green light alone: one study (Sadick 2010a)
f) ALA-PDT versus placebo or no treatment: two studies (Orringer 2010; Pollock 2004)
g) ALA-PDT other: six studies (Barolet 2010; Hongcharu 2000; NCT00706433; Pollock 2004; Taub 2007; Yin 2010)

6. MAL-PDT versus ALA-PDT

a) One study compared these interventions (Wiegell 2006a)

7. Other (non-MAL, non-ALA) PDT versus other comparators

a) Indocyanine green (ICG) PDT: two studies (Genina 2004; Kim 2009)
b) Indole-3-acetic acid (IAA) PDT: one study (Na 2011)
c) Topical liposomal methylene blue (TLMB) PDT: one study (Fadel 2009)
d) Chlorophyll-a (CHA) PDT: one study (Song 2014)
e) Gold microparticles PDT: one study (Paithankar 2015)

Seven studies had a single light treatment session in one of the interventions (Barolet 2010; Genina 2004; Hongcharu 2000; Kim 2009; Orringer 2004; Seaton 2003; Wiegell 2006a).

Most interventions had two to four sessions, two studies had five sessions (Ianosi 2013; McGill 2008), two studies had six sessions (Leheta 2009; Ou 2014), 12 studies had eight sessions (Anyachukwu 2014; de Arruda 2009; Elman 2003; Genina 2004; Gold 2005; Lee 2010; Ling 2010; Liu 2011; Song 2014; Tzung 2004, Zhang 2009a; Zhang 2013b), one study had up to 24 sessions (Cheng 2008), one study had 28 sessions (Ash 2015) and one study had 84 sessions (Papageorgieu 2000). Two self-administered interventions had a total of 56 (Kwon 2013) and 112 sessions (Na 2007).

Four studies included endpoints, such as time to resolution or interventions in which treatments were applied until a certain improvement threshold was reached (Gold 2011; Liu 2014; Sadick 2010b; Sami 2008), so the number of light sessions differed between study arms. Please see the Characteristics of included studies tables for details.

The frequency of application varied from twice a day to once a month.

Outcome assessment
Timing of outcome assessment

The majority of studies (52) conducted short-term assessments, two to four weeks after the final treatment. The most common assessment time point was four weeks after final treatment (42 studies), followed by two weeks after final treatment (16 studies), with some of these studies containing assessments at both time points.

About a third of studies (27) conducted medium-term assessments, five to eight weeks after final treatment (Bernstein 2007; Borhan 2014; Chen 2015; Choi 2010; Elman 2003; Fadel 2009; Jung 2009; Kwon 2013; Lee 2010; Leheta 2009; Mei 2013; NCT00594425; NCT00706433; NCT00933543; Oh 2009; Orringer 2004; Orringer 2007; Orringer 2010; Paithankar 2015; Pariser 2013; Ragab 2014; Sadick 2010a; Seaton 2003; Wang 2006; Wiegell 2006a; Wiegell 2006b; Zhang 2013a). The most common assessment time point was eight weeks after final treatment (18 studies), followed by six weeks after final treatment (12 studies).

About a third of studies (25) conducted assessments longer than eight weeks after final treatment (Bissonnette 2010; Darne 2011; Fadel 2009; Haedersdal 2008; Hongcharu 2000; Hörfelt 2006; Jih 2006; Leheta 2009; McGill 2008; Mei 2013; Moneib 2014; NCT00594425; Oh 2009; Orringer 2004; Orringer 2010; Ou 2014; Paithankar 2015; Sadick 2010a; Seaton 2003; Taub 2007; Uebelhoer 2007; Wang 2006; Wiegell 2006b; Yeung 2007; Yin 2010), but the majority at no longer than three months after final treatment. The most common assessment time point was 12 weeks after final treatment (18 studies).

Please note that we listed studies multiple times if they assessed outcomes at multiple time points corresponding to the short-, medium-, or long-term time points defined by our protocol.

We included four studies which had a final evaluation at last treatment (de Arruda 2009; Ianosi 2013; Na 2007; Papageorgieu 2000) and reported their results at the final assessment. In three studies, the time points of assessments were unclear (Anyachukwu 2014, Borhan 2014; Leheta 2009). Comparison of interventions and the outcomes at time points as defined by our protocol was not possible for studies with time-to-resolution or time to a pre-defined improvement threshold (Gold 2011; Liu 2014; Sadick 2010b; Sami 2008), apart from comparison for primary outcome 3, 'Investigator-assessed severe adverse effects', as well as 'Other adverse effects'.

Primary outcome measures
Primary outcome measure 1: Participant's global assessment of improvement

A total of 23 studies addressed this outcome. Of these 13 used Likert or Likert-like scales (Bernstein 2007; Chang 2007; Choi 2010; Darne 2011; Haedersdal 2008; Lee 2010; Moneib 2014; NCT00706433; Oh 2009; Papageorgieu 2000; Ragab 2014; Wiegell 2006b; Yin 2010). Five used visual analogue scales (VAS) (Hong 2013; Jung 2009; Jung 2012; Kwon 2013; Na 2007). In three studies other methods were used (Baugh 2005; Kim 2009; Orringer 2007), and in two studies, it was unclear which method was used (Liu 2011; Taub 2007).

In an additional split-face study, this outcome was also addressed, but not for separate face sides (Jih 2006).

Primary outcome measure 2: Investigator-assessed change in lesion count

The majority of studies (51) addressed this outcome.

Primary outcome measure 3: Investigator-assessed severe adverse effects and other adverse effects

Please note that methods used for assessment of 'Investigator-assessed severe adverse effects' and 'Other adverse effects' are listed under 'Adverse effects', in the 'Outcomes' sections of the Characteristics of included studies tables.

Five studies did not record or report on adverse effects (Bowes 2003; Cheng 2008; Ling 2010; Orringer 2004; Tzung 2004).

Seventeen studies that reported on adverse effects did not report the method they used to record them (Chang 2007; Elman 2003; Jung 2009; Kwon 2013; Moneib 2014; Na 2007; Na 2011; Orringer 2007; Orringer 2010; Ou 2014; Paithankar 2015; Papageorgieu 2000; Song 2014; Taub 2007; Zhang 2009a; Zhang 2013a; Zhang 2013b).

Secondary outcome measures
Secondary outcome measure 1: Investigator-assessed change in acne severity

A total of 30 studies addressed this outcome. The most commonly used scale was the Leeds revised grading scale (O'Brien 1998), reported in 12 studies (Darne 2011; Fadel 2009; Ianosi 2013; Jung 2009; Leheta 2009; McGill 2008; Orringer 2004; Orringer 2007; Orringer 2010; Seaton 2003; Wiegell 2006a; Wiegell 2006b) and an additional five studies used the same scale referring to it as Cunliffe's grading system (Choi 2010; Hong 2013; Jung 2012; Kim 2009; Song 2014). Five studies, Baugh 2005; Bowes 2003; Hongcharu 2000; Tzung 2004; Yilmaz 2011, used the Michaëlsson grading score (Michaelsson 1977). Two studies, Bernstein 2007; Uebelhoer 2007, used the Allen-Smith scale (Allen 1982). One study used the Korean Acne grading system (Chang 2007), one (Bissonnette 2010) used the Global Acne Grading System, and four used non-standardised grading scales (Gold 2005; Hörfelt 2006; NCT00706433; Taub 2007).

Secondary outcome measure 2: Investigator's global assessment of improvement

A total of 32 studies addressed this outcome. The most commonly used scale was the Investigators' Global Assessment (IGA) suggested by the U.S. Food and Drug Administration (FDA) guidance for developing drugs for the treatment of acne vulgaris (FDA 2005), used in six studies (Borhan 2014; Kwon 2013; NCT00594425; NCT00933543; Paithankar 2015; Pariser 2013); eight studies used various Likert or Likert-like scales (Barolet 2010; Baugh 2005; Bernstein 2007; Gold 2005; Karsai 2010; Sadick 2010a; Uebelhoer 2007; Wiegell 2006b), and 17 studies used various per cent improvement scales (Chen 2015; Cheng 2008; Fadel 2009; Hongcharu 2000; Ianosi 2013; Leheta 2009; Ling 2010; Liu 2011; Mei 2013; Moneib 2014; Oh 2009; Ou 2014; Papageorgieu 2000; Yin 2010; Zhang 2009a; Zhang 2013a; Zhang 2013b). In one study, it was unclear which method was used (Taub 2007).

Secondary outcome measure 3: Changes in quality of life

Only three studies recorded this outcome, two using the Dermatology Life Quality Index (DLQI) (McGill 2008; Karsai 2010) and one using the Cardiff Acne Disability Index (CADI) (Ianosi 2013).

Excluded studies

We excluded a total of 24 studies (25 records). Please see the reasons for exclusion in the Characteristics of excluded studies tables. We excluded 16 studies because they were not RCTs. Three studies started as RCTs but then did not follow the protocol and no longer met our inclusion criteria thereafter (Alexiades-Armenakas 2006; Morton 2005; Tuchin 2003). Three studies were not focusing on direct effects of light therapies for acne (Shin 2012; Yang 2013; Zhan 1997). One RCT had a cross-over design (Owczarek 2014), and one was focusing on acne scars (Yoon 2014). Please see the Characteristics of excluded studies tables and Methods for details.

Studies awaiting classification

Please see the Characteristics of studies awaiting classification tables. We identified 23 studies we were unable to include or exclude. Clinical trials registers recorded four studies as completed, but results were not published (NCT01245946; NCT01472900; NCT01584674; ISRCTN78675673). Four studies were discontinued or terminated (Berson 2006; NCT00237978; NCT00814918; ISRCTN73616060), and one was completed, but the author confirmed that data were not available (ISRCTN95939628). Two studies were pilot studies, and three studies were conference proceedings without enough information provided to include or exclude them (Kim 2012; Lee 2012; Passeron 2011; Song 2012; Troilius 2005), and for an additional study it was unclear to us whether it was a RCT (Faghihi 2011). We were unsuccessful when we attempted to contact the responsible parties and obtain further information and results of these studies. Responsible parties of two studies provided information that the trials had been completed, but they were aiming at publishing the results and therefore couldn't provide the data (Sakamoto 2012; Shaheen 2011). One of the clinical trial records (NCT01689935) could correspond to Sakamoto 2012, but we were unable to confirm this with the study authors. One study was completed, but there was ambiguity regarding the randomisation method, and the raw data we obtained were unclear and we were not able to interpret it (Edwards 2006). For two studies published in Mandarin, we were unable to obtain full texts (Lin 2011; Zhang 2009b). We were unable to obtain the full text of one study in Spanish (Pinto 2011); attempts to contact the study authors were unsuccessful. Similarily, we were unsuccessful in obtaining the full text and additional information on a study we identified in a reference list and through grey literature searches (Nataloni 2003).

Ongoing studies

Please see Characteristics of ongoing studies for details about the three ongoing studies we identified (EU 2014-005235-13; NCT02217228; NCT02431494).

Risk of bias in included studies

Selection bias was unclear for the majority of studies, with about half of studies describing adequate methods of random sequence generation and less than a third of studies describing adequate allocation concealment methods. Performance bias was also unclear in more than half of studies, high in about a quarter, and unclear in the remaining studies. Out of 26 studies which included participant-assessed outcomes, detection bias was low in only two studies, high in 10 studies, and unclear in the remaining studies. Detection bias was low in over half of studies for investigator-assessed outcomes and unclear in most of the rest. Attrition bias was low in over half of studies, high in about a quarter, and unclear in a few studies only. Reporting bias was similar. Other risk of bias was low in about a third of studies. Two thirds of studies had unclear risk because of possible conflicts of interest or sponsorship, or both, were not declared; they were industry-sponsored; or they reported some sort of conflict of interest, and a few studies had a high risk due to other reasons, such as baseline imbalances and concomitant treatment.

Please see Figure 2 for details. Please note that studies which did not include participant-assessed outcomes also have 'Detection bias for patient-assessed outcomes' marked as 'unclear' in Figure 2. It is therefore not possible to distinguish them in Figure 2 alone from studies which included such outcomes, but had 'unclear' risk of bias. In the corresponding 'Risk of bias' tables for the individual studies, we have clearly stated when studies did not include participant-assessed outcomes.

Figure 2.

Risk of bias summary: review authors' judgements about each risk of bias item for each included study.

Please see the Characteristics of included studies tables for details on risk of bias in individual studies.

Allocation

Random sequence generation

We judged the risk of bias as low in 34 studies in which study authors reported or later clarified how they generated the allocation sequence; four using coin toss (Barolet 2010; Baugh 2005; Moneib 2014; Uebelhoer 2007); 13 using computer software (Bissonnette 2010; Darne 2011; Genina 2004; Ianosi 2013; Karsai 2010; Kwon 2013; NCT00594425; NCT00673933; NCT00933543; Ou 2014; Papageorgieu 2000; Seaton 2003; Yin 2010); 10 using 'randomised code' (Ash 2015; Choi 2010; Oh 2009; Orringer 2004; Orringer 2007; Orringer 2010; Pariser 2013; Sadick 2010b; Song 2014; Yeung 2007); and seven using drawing lots (Anyachukwu 2014; Haedersdal 2008; Lee 2010; Liu 2014; McGill 2008; Mei 2013; Wiegell 2006b). We judged the risk of bias as unclear in 37 reports which did not include the method used to generate the allocation sequence.

Allocation concealment

We judged the risk of bias as low in 19 studies. Authors of 15 studies reported, or later clarified, that they used sealed envelopes or boxes to conceal the allocation sequence (Anyachukwu 2014; Ash 2015; Darne 2011; Haedersdal 2008; Lee 2010; Liu 2014; McGill 2008; NCT00594425; NCT00673933; NCT00933543; Oh 2009; Ou 2014; Ragab 2014; Seaton 2003; Uebelhoer 2007). One study used kits with randomised codes (Pariser 2013). One study communicated patient allocation via phone by an independent investigator prior to enrolment of each participant (Ianosi 2013). One study reported that assignments were concealed by securing randomisation codes until all data were entered (Kwon 2013) and 'by blinded sponsor numerical allocation' in another study (Sadick 2010b).

Fifty-one studies did not specifically report the intention or method (or both) of concealing the allocation sequence, so we judged the risk of bias as unclear. We judged the risk of bias as high in one study as the study authors clarified that they did not conceal allocation (Mei 2013).

Blinding

Performance bias

We judged the risk of bias as low in 11 studies. In eight studies, the authors described or later clarified blinding of both participants and personnel that we judged as adequate (Hörfelt 2006; Kwon 2013; Mei 2013; NCT00594425; NCT00673933; NCT00933543; Pariser 2013; Wiegell 2006a). Three split-face trials described blinding of participants that we judged as adequate, with unclear blinding of performing clinicians (Baugh 2005; Bowes 2003; Na 2011), but systematic differences between face sides in the care that was provided or in exposure to factors other than the interventions of interest were unlikely.

We judged the risk of bias as unclear in 40 studies, most of which did not report intention to blind participants or performing clinicians, or both, and did not present evidence that participants or clinicians were blinded. Given the nature of the interventions, it is not likely that participants or performing clinicians were blinded in those studies, but without the necessary information, we were unable to clearly judge the risk based on these assumptions.

We judged the risk of bias as high in 20 studies. In 14 studies, the authors specifically reported or later clarified that they did not blind both participants and performing clinicians (Bissonnette 2010; Darne 2011; Gold 2005; Karsai 2010; Kim 2009; Leheta 2009; Na 2007; Oh 2009; Orringer 2004; Orringer 2007; Orringer 2010; Sadick 2010a; Wang 2006; Wiegell 2006b). One study was an open trial (de Arruda 2009). One study described an adequate blinding of performing clinicians, but inadequate blinding of participants (Lee 2010). Four studies described blinding of participants that we judged as ineffective, with unclear blinding of performing clinicians (Elman 2003; Haedersdal 2008; Papageorgieu 2000; Seaton 2003).

Detection bias
Participant-assessed outcomes

Please note that 45 studies which did not include participant-assessed outcomes (participant's global assessment of improvement and changes in quality of life, or both) have 'Detection bias' for participant-assessed outcomes marked as 'unclear' in Figure 2. It is therefore not possible to distinguish them in Figure 2 alone from 14 studies which included such outcomes and had 'unclear' risk of bias (Bernstein 2007; Chang 2007; Choi 2010; Hong 2013; Ianosi 2013; Jung 2009; Jung 2012; Liu 2011; McGill 2008; Moneib 2014; NCT00706433; Ragab 2014; Taub 2007; Yin 2010). In the corresponding 'Risk of bias' tables for the individual studies, we clearly state when studies did not include participant-assessed outcomes. Two studies described blinding of participants that we judged as adequate (Baugh 2005; Kwon 2013) and the risk of bias as low. We judged the risk of bias as high in 10 studies. In nine studies, the authors specifically reported that they did not blind participants (Darne 2011; Karsai 2010; Kim 2009; Lee 2010; Na 2007; Oh 2009; Orringer 2007; Papageorgieu 2000; Wiegell 2006b). In one study, the authors reported that they unsuccessfully attempted to blind the participants (Haedersdal 2008).

Investigator-assessed outcomes

We judged the risk of bias as low in 41 studies. Authors of 20 studies reported blinding by use of photographs (Barolet 2010; Bernstein 2007; Borhan 2014; Darne 2011; Hong 2013; Hongcharu 2000; Ianosi 2013; Jung 2009; Karsai 2010; Liu 2011; McGill 2008; Na 2007; Na 2011; Oh 2009; Orringer 2004; Orringer 2007; Sadick 2010b; Song 2014; Wang 2006; Yeung 2007); 20 studies reported assessment by blinded investigators who did not participate in treatment and were unaware of the intervention status, or both (Ash 2015; Anyachukwu 2014; Bissonnette 2010; Haedersdal 2008; Kwon 2013; Lee 2010; Leheta 2009; Mei 2013; NCT00594425; NCT00673933; NCT00933543; Orringer 2010; Pariser 2013; Pollock 2004; Sami 2008; Seaton 2003; Tzung 2004; Uebelhoer 2007; Wiegell 2006a; Wiegell 2006b); and one study reported blinding of participants and performing clinicians (i.e. those treating the participants) who did outcome assessment that we judged as adequate (Hörfelt 2006).

We judged the risk of bias as unclear in 27 studies. Seven studies stated that they blinded the assessors, but did not describe the method (Chang 2007; Choi 2010; Elman 2003; Fadel 2009; Gold 2005; NCT00706433; Papageorgieu 2000). Four studies reported that they used photographs for evaluation of outcomes, but it was unclear whether they blinded dermatologists (e.g. not performing the treatment and unaware of the intervention status), so we judged the risk of bias as unclear (Chen 2015; Genina 2004; Moneib 2014; Ragab 2014). In 16 studies, there was no report of intended blinding of outcome assessors, and study authors did not provide evidence that they blinded assessors (Baugh 2005; Bowes 2003; Cheng 2008; Gold 2011; Jih 2006; Jung 2012; Kim 2009; Ling 2010; Liu 2014; Ou 2014; Taub 2007; Yilmaz 2011; Yin 2010; Zhang 2009a; Zhang 2013a; Zhang 2013b).

We judged the risk of bias as high in three studies; two studies were open trials (de Arruda 2009; Sadick 2010a) and one study performed both blinded and unblinded assessment (Paithankar 2015).

Incomplete outcome data

We judged the risk of bias as low in 43 studies which reported outcomes for 80% or more of participants randomised for prespecified time points, with reasons for missing data (if there were any) balanced in numbers across intervention groups and unlikely to be related to true outcome.

We judged the risk of bias as unclear in 11 studies. Ten studies did not report number of withdrawals, losses to follow-up, and final number of evaluable participants (Bowes 2003; Elman 2003; Gold 2011; Liu 2011; Liu 2014; Moneib 2014; Na 2011; Ragab 2014; Sami 2008; Song 2014) and reported in a way that did not permit a clear judgement of bias in one study (Genina 2004).

We judged the risk of bias as high in 17 studies which reported outcomes for less than 80% of participants randomised at some of the predefined time points (Anyachukwu 2014; Baugh 2005; Bissonnette 2010; Darne 2011; Fadel 2009; Gold 2005; Ianosi 2013; McGill 2008; Na 2007; NCT00594425; Orringer 2004; Orringer 2007; Orringer 2010; Papageorgieu 2000; Taub 2007; Wiegell 2006b; Yeung 2007). Three studies imputed missing data using various methods that we judged as appropriate (Orringer 2004; Orringer 2007; Orringer 2010). However, we still judged the risk of bias as high in those studies, as we could not obtain information on when the last observation that was carried forward was recorded and there was a high dropout rate. We believe this introduced uncertainty although study authors handled missing data using imputation.

Selective reporting

We judged the risk of bias as low in 44 studies in which prespecified outcomes and those mentioned in the methods section appeared to have been reported at predefined time points or study authors provided them upon our request.

We judged the risk as unclear in eight studies in which baseline data were not reported, or results were reported in graph or figure format or in a way different from prespecified for some outcomes, or both (Chang 2007; Hörfelt 2006; Kim 2009; Liu 2011; Liu 2014; Ragab 2014; Song 2014; Wiegell 2006b).

We judged the risk of bias as high in 19 studies. Thirteen studies did not report results for prespecified outcomes, or results for prespecified time points, or both (Borhan 2014; Fadel 2009; Gold 2005; Gold 2011; Moneib 2014; Na 2007; NCT00594425; Paithankar 2015; Papageorgieu 2000; Pollock 2004; Sadick 2010a; Taub 2007; Uebelhoer 2007). Three studies reported results in graph or figure format only for most outcomes or in a way different from those prespecified (Kwon 2013; Na 2011; Tzung 2004). Two studies did not clearly prespecify the outcomes in the 'Methods' section (Ash 2015; Anyachukwu 2014). In one study, we were unable to obtain statistical data regarding differences between groups to which participants were initially randomised (Orringer 2004).

Other potential sources of bias

We identified no additional sources of bias and judged the risk of bias as low in 23 studies.

We judged the risk of bias as unclear in 44 studies. In 21 studies, we judged the risk of bias as unclear because possible conflicts of interest or sponsorship, or both, were not declared (Bernstein 2007; Bowes 2003; Chen 2015; de Arruda 2009; Elman 2003; Hong 2013; McGill 2008; Moneib 2014; Na 2011; Papageorgieu 2000; Pollock 2004; Ragab 2014; Sadick 2010a; Sami 2008; Tzung 2004). In six of these studies (Cheng 2008; Ling 2010; Ou 2014; Zhang 2009a; Zhang 2013a; Zhang 2013b), additional bias might also have been introduced as these studies were in Mandarin and only one person performed data extraction. In two studies, the authors declared that they had no conflicts of interest, but it was unclear who provided the device or the sham device (Kwon 2013) and whether there was commercial sponsorship (Sadick 2010b). In 21 studies, we judged the risk as unclear because the study authors declared potential conflict of interest or had a commercial sponsor, or both, and it was unclear whether this affected the results (Barolet 2010; Baugh 2005; Bissonnette 2010; Darne 2011; Genina 2004; Gold 2005; Gold 2011; Haedersdal 2008; Hörfelt 2006; Jih 2006; NCT00594425; NCT00673933; NCT00706433; NCT00933543; Paithankar 2015; Pariser 2013; Seaton 2003; Taub 2007; Uebelhoer 2007; Wang 2006; Yeung 2007).

In four studies, we judged the risk of bias as high due to baseline imbalances and concomitant treatment (Anyachukwu 2014; Ash 2015; Borhan 2014; Liu 2011). For two of these studies (Borhan 2014; Liu 2011), sponsorship was also unclear, and in one it was unclear whether potential conflicts of interest might have affected the results (Ash 2015).

Effects of interventions

See: Summary of findings for the main comparison Light therapies (including photodynamic therapy) compared to placebo, no treatment, topical treatment and other comparators for acne vulgaris; Summary of findings 2 MAL-PDT compared to red light only for acne vulgaris; Summary of findings 3 ALA-PDT compared to blue light only for acne vulgaris

We used GRADEpro GDT (GRADEpro GDT 2015) to create a 'Summary of findings' table (Summary of findings for the main comparison) for our primary outcomes Participant's global assessment of improvement, Investigator-assessed change in lesion count, and Investigator-assessed severe adverse effects.

The aim was to illustrate the nature of the results of this review and different aspects of heterogeneity that we took into account when interpreting the results of the included studies. We judged that pooling the results of most of the studies was inappropriate, due to methodological and clinical heterogeneity, including the following:

  1. differences of included participants (Fitzpatrick skin types and acne severity);

  2. differences in design (parallel groups, split-face or split-back studies, and designs combining them);

  3. differences in interventions (wavelengths, doses, different active substances used in PDT and their pharmacokinetic characteristics, incubation time and whether they were administered under occlusion or not, number of light sessions and frequency of application, pre- and post-treatment care);

  4. differences in comparator interventions (most common being no treatment, placebo, other light interventions, and various topical treatments, but also their various combinations);

  5. differences in outcomes assessed, as well as methods and timing of outcome assessment; and

  6. poor reporting and failure to obtain necessary data.

To make it easier for the reader to follow the effects of interventions of the studies we included, we grouped the studies by our outcome (primary and secondary) and then by comparison, as previously described in the Included studies section (under 'Interventions'). For clarity, we used five additional tables to present the effects of interventions (Table 1, Table 2, Table 3, Table 4; Table 5). We reported effects of interventions using the statistics and methods described in the Methods section. When such reporting was not possible, we reported results the way they were available and clarified our reasons for our inability to report them as planned.

Table 1. Participant's global assessment of improvement
  1. ALA = 5-aminolevulinic acid
    BPO = benzoyl peroxide
    CHA = chlorophyll-a
    FPT = Fitzpatrick's Skin Types: based on different reactions to sun exposure and range from type I ('pale white skin which always burns and never tans') to type VI ('deeply pigmented dark brown to black skin which never burns and tans very easily') (Fitzpatrick 1988)
    GAAS = Global acne assessment scoring scale
    IAA = indole 3-acetic acid
    IGA = Investigator global assessment score
    ILs = inflamed lesions
    IPL = intense pulsed light
    IR = infrared
    ITT = intention-to-treat analysis
    LPDL = long pulsed dye laser
    LOCF = last observation carried forward
    LLT = lower level term
    MAL = methyl-aminolevulinate
    NILs = non-inflamed lesions
    NNTB = number needed to treat for an additional beneficial outcome
    OFI = optical fibre intra-tissue irradiation
    PDL = pulsed-dye laser
    PDT = photodynamic therapy
    PT = preferred term
    RCT = randomised controlled trial
    SD = standard deviation
    SE = standard error
    SPF = Sun protection factor
    TER = total effective rate
    TLMB = topical liposomal methylene blue

    Change from baseline i.e. absolute change is calculated by subtracting baseline count from count assessed at certain time point. Percentage change is calculated by dividing the absolute change with baseline count and then multiplying that value by 100 to get percentages.

    Unless specified differently, results presented as reported in the published papers, without performing independent analysis. Please see Characteristics of included studies for details on withdrawals and drop-outs of participants for each study.

Study Participants Intervention(s) and control(s) Participant's global assessment of improvement
1. Light versus placebo or no treatment
Green light versus placebo
Baugh 200525 (4 M, 21 F) aged 19-41 years (mean 27.8), diagnosed with mild to moderate inflammatory facial acne; FPT I–III532 nm pulsed laser vs sham in a split-face trial, both with skin cooling system; two exposures ⁄ week for 2 weeks. Assessed at 1 and 4 weeks post-treatmentNon-standardised scale (overall treatment satisfaction in intervals of 10 percentage points) was used for evaluation. At 4 weeks 4.8% participants reported 30% to 39% satisfaction, 9.5% reported 50% to 59% satisfaction, 23.8% reported 60% to 69% satisfaction, 47.6% reported 70% to 79% satisfaction, 9.5% reported 80% to 89% satisfaction and 4.8 reported 90% to 100% satisfaction. Further data were not provided.
Infrared light versus no treatment
Darne 201138 (7 M, 31 F), aged 18-47 years (mean 28), with moderate-severe facial acne; FPT I-V1450 nm laser (8-9 J/cm²) in a split-face trial, 3 treatments monthly, assessed monthly for 4 months, then at 3-monthly intervals for 12 months after final treatmentNon-standardised scale ('highly satisfied', 'satisfied', 'neutral' or 'unsatisfied' and 'would recommend to a friend') was used for evaluation. At 4 weeks 6/25 (24%) of participants were 'highly satisfied', 9/25 (36%) were 'satisfied', 6/25 (24%) were 'neutral' and 4/25 (16%) reported the treatment to be 'unsatisfactory'. 21/25 (84%) reported that they would 'recommend the treatment to a friend'.
Moneib 201424 (5 M, 19 F), aged 15-8 years (mean 21.5), with moderate-severe acne; FPT II-VFractional Erbium Glass 1559 nm laser, in a split-face trial, 4 treatments, at 2-week intervals, assessed every 3 months for 1 year after final treatmentNon-standardised scale (0 = no improvement; < 25% = mild improvement; 26% to 50% = moderate improvement; 51% to 75% = good improvement; 76% to 100% = excellent improvement) was used for evaluation. Reported in graph format and for treatment face sides only, and at unclear time point. Our interpretation of the graph was that 5% of participants assessed their improvement to be mild, 5% to be moderate, 20% to be good and 70% to be excellent.
Orringer 200746 (10 M, 36 F) enrolled, 30 completed, mean age 23.9 years (range not reported) with clinically apparent active facial acne; FPT II–VI1320 nm Nd:YAG laser in a split-face trial with cooling; 3 treatments at 3-week intervals; assessed at weeks 7 and 14Non-standardised scale (details not given) was used for evaluation. At final treatment, 29/37 of participants who completed the treatments (78%) "indicated that their acne was at least mildly improved on the treated side of the face as compared with baseline", and 16/37 participants (43%) indicated "moderate or better" improvement. Data for non-treated sides were not given, but 22/37 (59%) of participants reported that "their acne had improved at least mildly when compared with the untreated skin".
Red light versus no treatment
Na 200730 (7 M, 23 F) aged 19-33 years (mean 23.6) with mild-moderate acne; skin types not documented635–670 nm portable red light device in a split-face trial, self-administered to the treatment side twice daily for 8 weeks; assessed at weeks 1, 2, 4 and 8VAS: 0-5, none to very severe was used for evaluation. Score (unclear whether mean or median) decreased from baseline 3.9 to 1.8 at final treatment on the treated and from 3.9 to 2.9 on the control side respectively, with significant difference between the sides (P < 0.005). This outcome was not evaluated after final treatment and no further data were provided.
Blue-red light versus placebo
Papageorgieu 200030, mean age 24.8 years in blue–red light group; 25 participants, mean age 25.6 years in white light control group; randomised from the original 107 recruited (33 M, 74 F, age 14-50 years), all with mild-moderate acne; skin types not stated415 nm plus 660 nm light vs cool white light; treated daily for 12 weeks; assessed every 4 weeks for the 12-week treatment periodNon-standardised scale: 'worse' (≤ -10%), 'unchanged' (-9% to 9%), 'mild improvement' (10% to 39%), 'moderate improvement' (40% to 59%), 'marked improvement' (60% to 89%) or 'clearance' (≥ 90%) was used for evaluation, but reported only in graph format and no details were provided. Not evaluated after final treatment. Our interpretation of the graph was that around 4% of participants reported 'clearance', 70% reported 'marked improvement', 20% 'moderate improvement' and 4% reported 'mild improvement' in the blue-red light group, whilst in the white light group around 70% of participants reported 'unchanged' or 'mild improvement', 20% 'moderate improvement' and 8% 'marked improvement'. Further data were not provided. We dichotomised the data to 27/30 of 'success' outcomes in the blue-red and 7/25 in the white light group. Blue red-light was superior to white light with RR (95% CI) of 3.21 (1.70, 6.09), P = 0.0003, and the NNTB was 2 (95% CI 1 to 3)
Kwon 201335 (11 M, 24 F); aged 20-27 years (mean not given), with mild-moderate acne, FPT III-V; 18 participants in the blue-red light group, 17 in the placebo group420 nm plus 660 nm home use LED device vs home-use sham device; self-treatment twice daily for 4 weeks in a split-face trial; assessed 4 and 8 weeks after final treatmentVAS was used for evaluation (10 = same as before the first treatment; 0 = no acne). Mean VAS score 10 at baseline in both groups decreased to 4.3 in the blue-red light group, and stayed at 10 or above in the placebo group (extracted from graph) at 8 weeks after final treatment. No further data (SDs) were provided in text nor in graph format.
2. Light versus topical treatment
Light versus benzoyl peroxide (BPO)
Chang 200730 women aged 23–32 years (mean 25 ± 7) with mild-moderate acne; FPT III-IVIPL with 530–750 nm filter with cooling gel in a split-face trial, 3 sessions, 3 weeks apart, BPO gel used on both sides of the face. Assessed 3 weeks after final treatmentNon-standardised scale (highly satisfied, satisfied, neutral, or dissatisfied) was used for evaluation. At 3 weeks participants were "uniformly satisfied with their treatment, but IPL treatment did not give any additional benefit". No further data were reported.
Papageorgieu 200030 participants, mean age 24.8 years in blue–red light group and 25 participants, mean age 23.4 years in the BPO group, randomised from the original 107 recruited (33 M, 74 F, age 14-50 years) all with mild-moderate acne; skin types not stated415 nm plus 660 nm light vs 5% BPO, parallel groups, treated daily; assessed every 4 weeks for the 12-week treatment periodNon-standardised scale: 'worse' (≤ -10%), 'unchanged' (-9% to 9%), 'mild improvement' (10% to 39%), 'moderate improvement' (40% to 59%), 'marked improvement' (60% to 89%) or 'clearance' (≥ 90%) was used for evaluation, but reported only in graph format and no details were provided. Not evaluated after final treatment. Our interpretation of the graph was that around 4% of participants reported 'clearance', 70% reported 'marked improvement', 20% 'moderate improvement' and 4% reported 'mild improvement' in the blue-red light group, whilst around 35% of participants showed 'marked improvement', 45% 'moderate improvement', 10% 'mild' improvement' and 10% 'unchanged' in the BPO group. Further data were not provided. We dichotomised the data to 27/30 of 'success' outcomes in the blue-red and 20/25 in the BPO group. The difference was non significant, with RR (95% CI) of 1.13 (0.89, 1.42), P = 0.31
Light versus clindamycin
Lee 20109, with inflammatory acne (other characteristics not given)Full-spectrum light twice a week vs 1% clindamycin twice a day, in a split-face trial, for 4 weeks, evaluation weekly whilst on treatment and 2, 4 and 8 weeks after final treatmentNon-standardised scale ('worse', 'no change', 'fair', 'good' and 'excellent') was used for evaluation. Participants rated the treatment as 'good' or 'excellent' (unclear for which intervention and at what time point). Further data were not reported.
Light and other topical treatments
Ash 2015

41 (M/F not reported, study authors clarified "a 50/50 split"), 26 in the intervention, 15 in control group, aged 16–45 years (mean not reported) with mild-moderate acne (Leeds grade); FPT not given:

"Caucasian, Asian and mixed Afro-Caribbean ethnic groups"

Pre-treatment facial wash/weak chemical peel (containing salicylic acid, glycolic acid, lactic acid) followed by treatment with blue light device and then post treatment facial moisturiser (containing salicylic acid, glycolic acid, lactic acid, menthol, niacin) versus unclear control in a parallel group trial, 28 sessions in total, every other day for 8 weeks. Assessed at 12 weeks (4 weeks after final treatment?)Details on scale used for evaluation not given. Results reported as "the majority of subjects reporting that they were satisfied, very satisfied, or extremely satisfied with treatment" in the treatment group. Results were not reported for the control group. No further data were reported.
3. Light versus other comparators
Comparison of light therapies of different wavelengths
Choi 201020 (1 M, 19 F, aged 20-37 years, mean age 26); all with acne (Cunliffe severity grade 2-4), FPT types III-V585 nm PDL vs 530-750 nm IPL, 4 treatments at 2-week intervals, in a split-face trial, assessed 4 and 8 weeks after last treatmentNon-standardised rating scale (from 0-10, neutral to highly satisfied) was used for evaluation. No statistically significant difference in improvement of scores between the two treatments (P > 0.05) was found. They increased from baseline 0 for both to 3.3 for IPL and 3.7 for PDL at 4 weeks after treatment and then to 4.7 for IPL and 5.2 for PDL at eight weeks after treatment. Further data were not reported.
Jung 200918 enrolled, 16 completed (5 M, 11 F, aged 20-31 years, mean age 26); with mild-moderate acne (Cunliffe severity grade 2-5), skin types not given585 nm PDL vs combined 585/1064 nm PDL, in a split-face trial, 3 treatments at 2-week intervals, assessed at 8 and 12 weeks after initial treatmentVAS (0-10, worst imaginable acne state to disease free) was used for evaluation; please note that opposite VAS was used in Jung 2012. Mean scores on the PDL sides and on the 585/1,064-nm laser sides increased from 3.3 and 3.7 at baseline to 6.63 (P = 0.002) and 6.60 (P = 0.001) at 8 weeks respectively. At 12 weeks, they declined to 6.12 at both sides. Further data were not reported.
Liu 201120 (6 M/14 F) completed the study, number of randomised participants not reported, 10 completed in the blue-light, 10 in the red-light group, aged 19–28 years (mean 23.6 years) with mild-moderate acne (Global Acne Grading System); FPT III-IVBlue (405 ± 10 nm) vs red (630 ± 10 nm) LED portable device treatments, about 20 cycles of illumination and the corresponding light doses received in each session were 7.2 J/cm² and 11.52 J/cm², in a parallel-group trial, 8 sessions in total, twice weekly for 4 weeks; assessed at 4 weeks after final treatment and at each treatment session"Subjective evaluation was based on the observations of face skin and communications between the patient and researcher (for the follow-ups)." Further details on scale used for evaluation not given. Results reported as "A few participants reported that fresh new acne lesions came out, while the total number of lesions decreased slightly." No further data were reported.
Papageorgieu 200030 participants, mean age 24.8 years in blue–red light group and 27 participants, mean age 23.4 years in the blue-light group, randomised from the original 107 recruited (33 M, 74 F, age 14–50 years) all with mild-moderate acne; skin types not stated415 nm plus 660 nm light vs 415 nm light, parallel groups, treated daily for 12 weeks; assessed every 4 weeks for the 12-week treatment periodNon-standardised scale: 'worse' (≤ -10%), 'unchanged' (-9% to 9%), 'mild improvement' (10% to 39%), 'moderate improvement' (40% to 59%), 'marked improvement' (60% to 89%) or 'clearance' (≥ 90%) was used for evaluation, but reported only in graph format and no details were provided. Not evaluated after final treatment. Our interpretation of the graph was that around 4% of participants reported 'clearance', 70% reported 'marked improvement', 20% 'moderate improvement' and 4% reported 'mild improvement' in the blue-red light group, whilst in the blue-light group around 4% of participants experienced 'clearance', 50% of participants 'marked improvement', 30% 'moderate improvement', 8% 'mild' improvement' and 8% 'unchanged'. Further data were not provided. We dichotomised the data to 27/30 of 'success' outcomes in the blue-red and 23/27 in the blue-light group. The difference was non significant, with RR (95% CI) of 1.06 (0.87, 1.29), P = 0.59.
Comparison of light therapies of different doses
Bernstein 20077 enrolled, 6 completed (1 M, 4 F, aged 23-41 years, mean age 29), all with active papular acne, FPT I-IIIComparison of two 1450 nm laser treatments; single-pass, high-energy (13–14 J/cm²) vs double-pass, low energy (8–11 J/cm²); 4 treatments at monthly intervals, assessed 1 month following each treatment and 2 months after final treatmentNon-standardised rating scale (0= worsening, 1= no change, 2= mild improvement , 3= moderate improvement, 4= marked improvement) was used for evaluation. At 8 weeks average score on the single-pass side was 2.3 (range 1-4) and on the double-pass side 2.3 (range 2-4).
Jih 200620 (10 M, 10 F) age 18-39 years (mean 23) with active inflammatory facial acne; FPT II–VI1450 nm diode laser in a split-face trial using anaesthetic cream and 14 J/cm² in one group and 16 J/cm² in the second, with 3 treatments given at 3–4 week intervals, assessed at 1, 3, 6 and 12 months after final treatmentNon-standardised rating scale (0 = worsening, 1 = no change, 2 = mild improvement , 3 = moderate improvement, 4 = marked improvement) was used for evaluation. The majority of participants reported moderate to marked improvement, 85.3% at the 1-month, 67.7% at the 3-month, 60.0% at the 6-month and 82.1% at the 12-month assessments. No separate data for different doses given.
NCT00706433266 (128 M, 138 F), 68 in the ALA 1000 s group, 65 in the ALA 500 s group, 67 in the vehicle 1000 s group and 66 in the vehicle 500 s group, mean age 20.1 years, inclusion criterion 12 > years, with moderate and severe acne (IGA score 3 and 4, with at least 20 ILs); FPT I-VI20% ALA (45 min incubation) plus blue 1000 s light vs 20% ALA (45 min incubation) plus 500 s blue light vs vehicle (45 min incubation) plus blue 1000 s light vs vehicle (45 min incubation) plus 500 s blue light; in a parallel-group trial; up to 4 treatments at 3 weeks intervals, assessed 3 and 6 weeks after the final treatmentNon-standardised scale ('subject satisfaction score'; excellent-very satisfied; good-moderately satisfied; fair-slightly satisfied; poor-not satisfied at all) was used for evaluation. At 6 weeks after final treatment 20/67 participants in the vehicle 1000 s and 23/66 in the vehicle 500 s group assessed their improvement as 'good'; 23/67 in the vehicle 1000 s and 26/66 in the vehicle 500 s group assessed their improvement as 'excellent'. We dichotomised the data to 43/67 of 'success' outcomes in the vehicle 1000 s and 49/66 in the vehicle 500 s group. The difference between vehicle 1000 s blue light and vehicle 500 s blue light groups was non significant, with RR (95% CI) of 0.86 (0.69, 1.09), P = 0.21.
Light in combination with carbon lotion versus no treatment
Jung 201222 (4 M, 18 F), 20 completed (2 M, 18 F, aged 19-34 years, mean age 25.4), FPT III-IV, acne severity not givenCarbon lotion plus quasi-long pulse and Q-switched 1064 nm Nd:YAG laser vs non-treated control, in a split-face trial, 3 treatments over 4 weeks, evaluation every 2 weeks whilst on treatment and then every 4 weeksVAS (0-10, disease free to initial visit acne status) was used for evaluation; please note that opposite VAS was used in Jung 2009. At 4 weeks after final treatment participants assessed significantly greater improvement on the laser-treated compared to the untreated side (P < 0.05). VAS score mean (SDs not given) decreased from initial 10 at both sides to 5.9 (P < 0.001) on the laser-treated and to 9.2 (P = 0.007) on the untreated side.
4. MAL-PDT versus other comparators
MAL-PDT versus orange light alone
Haedersdal 200815 (5 M, 10 F) age 18-31 years (median 18), with at least 12 facial inflammatory acne lesions; FPT I–IIISplit-face design with non purpuric LPDL 595 nm full-face treatment and MAL cream applied to randomised side of the face for 3 h before laser exposure, with dynamic cooling device; three treatments at 2-week intervals; assessed 4 and 12 weeks after final treatmentNon-standardised numerical scale (0-10, no satisfaction to best imaginable satisfaction) was used for evaluation. Median (25-75 percentiles) score (range) was significantly higher for MAL-LPDL treatment than for LPDL treatment alone at both 4 weeks after final treatment (P = 0.031); 7 (4.75 to 8) vs 6 (3.75 to 8), and at 12 weeks after final treatment (P = 0.034); 8 (6.25 to 9) vs 7.5 (5 to 8.75).
MAL-PDT versus placebo or no treatment
Wiegell 2006b36 participants: 21 in treatment group aged 23 ± 5 years (9 M, 10 F analysed) and 15 in control group aged 24 ± 5 years (3 M, 9 F analysed), with > 12 inflammatory acne lesions; FPT II–VComparison of MAL plus 630 nm with no treatment in a parallel-group trial; two treatments, 2 weeks apart, assessed every 4 weeks for 12 weeks after treatmentNon-standardised grading scale (0-4; acne worse, no change, slight improvement, moderate improvement, marked improvement) was used for evaluation. Results were reported in graph format and no details were provided. Our interpretation of the graph was that at 4, 8 and 12 weeks after final treatment median improvement scores were 3, 2 and 3 in the MAL-PDT group and 1.5, 1 and 1 in the control group respectively.
MAL-PDT other
Hong 201322 (2 M, 20 F), age 19-35 years (mean not given), "at least grade 2 (Cunliffe acne grading system)", FPT IV-VMAL plus 630 nm light vs MAL plus 530-750 nm light in a split-face trial, 3 treatments in total, 2-week intervals, assessed at 4 weeks after treatmentVAS scale (10-0, 10 = same as before the first treatment; 0 = no acne) was used for evaluation. Mean VAS score decreased from baseline 10 on both sides to 5.0 at the red light side, and 4.9 at the IPL side at 4 weeks after final treatment, with no significant difference between the 2 sides. Further data were not provided.
5. ALA-PDT versus other comparators
ALA-PDT versus blue light alone
NCT00706433266 (128 M, 138 F), 68 in the ALA 1000 s group, 65 in the ALA 500 s group, 67 in the vehicle 1000 s group and 66 in the vehicle 500 s group, mean age 20.1 years, inclusion criterion 12 > years, with moderate and severe acne (IGA score 3 and 4, with at least 20 ILs); FPT I-VI20% ALA (45 min incubation) plus blue 1000 s light vs 20% ALA (45 min incubation) plus 500 s blue light vs vehicle (45 min incubation) plus blue 1000 s light vs vehicle (45 min incubation) plus 500 s blue light; in a parallel-group trial; up to 4 treatments at 3-week intervals, assessed 3 and 6 weeks after the final treatmentNon-standardised scale ('subject satisfaction score'; excellent-very satisfied; good-moderately satisfied; fair-slightly satisfied; poor-not satisfied at all) was used for evaluation. At 6 weeks after final treatment 18/68 participants in ALA 1000 s, 28/65 in the ALA 500 s, 20/67 in the vehicle 1000 s and 23/66 in the vehicle 500 s group assessed their improvement as 'good'; 23/68 participants in ALA 1000 s, 11/65 in the ALA 500 s, 23/67 in the vehicle 1000 s and 26/66 in the vehicle 500 s group assessed their improvement as 'excellent'. We dichotomised the data to 41/68 of 'success' outcomes in ALA 1000 s, 39/65 in the ALA 500 s, 43/67 in the vehicle 1000 s and 49/66 in the vehicle 500 s group. The difference between ALA 1000 s and vehicle 1000 s groups was non significant, with RR (95% CI) of 0.94 (0.72, 1.22), P = 0.64, and it was non significant between ALA 500 s and vehicle 500 s groups, with RR (95% CI) 0.81 (0.63, 1.03), P = 0.09.
ALA-PDT versus IPL alone
Oh 200920 (4 M, 16F), aged 18-30 years, 9 in the short incubation group (3M, 6F, mean age ± SD 23 ± 4.12 years) and 10 in the long incubation group (1 M, 9 F and 23 ± 5.53 years), with moderate and severe acne (Evaluator Global Severity Score 3 and 4); FPT II-IV20% ALA plus 590 nm IPL; 2 parallel groups: short incubation (30 min) vs long incubation (3 h), one half of the face within each treated with IPL alone; 3 treatments at 4-week intervals, assessed 4 weeks after each treatment and 8 and 12 weeks after the third treatmentNon-standardised (significant improvement (> 75%), moderate improvement (50% to 75%), mild improvement (25% to 50%), no improvement (0% to 25%), worse (< 0%) relative to baseline) was used for evaluation. At 12 weeks 6/9 (66.7%) participants assessed their improvement as mild and 3/9 (33.3%) as moderate in the short incubation group; 4/11 (36.4%) as mild, 6/11 (54.5%) as moderate and 1/11 (9.1%) as significant in the long incubation group. We dichotomised the data to 3/9 of 'success' outcomes in the short incubation and 7/11 in the long incubation group. The difference was non significant, with RR (95% CI) 0.52 (0.19, 1.46), P = 0.22.
Ragab 201425 (1 M, 24 F), aged 14-39 years, 15 in the ALA-IPL group (mean 19.7) and 10 in the IPL alone group (mean age 19.0), "with mild-moderate facial acne"; FPT III-V20% ALA plus 560? nm IPL versus 560 nm IPL alone; in a parallel-group trial; two treatments at two weeks intervals, assessed 2 and 8 weeks after final treatmentNon-standardised scale (marked improvement = 3; moderate improvement = 2; no change = 1; acne worsened = 0) was used for evaluation. At 8 weeks 5/15 (33%) participants assessed their improvement as moderate and 10/15 (67%) as marked in the ALA-IPL group, whereas 3/10 (30%) of participants assessed their improvement as marked, 4/10 (40%) as mild and 1/10 (10%) as "slight" (a non pre-specified category) in the IPL alone group. 2/10 (20%) of participants in the IPL alone group assessed that there was no change. We dichotomised the data to 10/15 'success' outcomes in the ALA-PDT group and 3/10 in the IPL alone group. The difference was non significant, with RR (95% CI) 2.22 (0.81, 6.11), P = 0.12.
ALA-PDT other
NCT00706433266 (128 M, 138F), 68 in the ALA 1000 s group, 65 in the ALA 500 s group, 67 in the vehicle 1000 s group and 66 in the vehicle 500 s group, mean age 20.1 years, inclusion criterion 12 > years, with moderate and severe acne (IGA score 3 and 4, with at least 20 ILs); FPT I-VI20% ALA (45 min incubation) plus blue 1000 s light vs 20% ALA (45 min incubation) plus 500 s blue light vs vehicle (45 min incubation) plus blue 1000 s light vs vehicle (45 min incubation) plus 500 s blue light; in a parallel-group trial; up to 4 treatments at 3-week intervals, assessed 3 and 6 weeks after the final treatmentNon-standardised scale ('subject satisfaction score'; excellent - very satisfied; good - moderately satisfied; fair - slightly satisfied; poor - not satisfied at all) was used for evaluation. We dichotomised the data to 41/68 of 'success' outcomes in ALA 1000 s and 39/65 in the ALA 500 s group. The difference between ALA 1000 s and ALA 500 s groups was non significant, with RR (95% CI) 1.00 (0.76, 1.33), P = 0.97.
Taub 200722 recruited, 19 participated, mean ± SD age 26.5 ± 9.1 years, 7 M, 12 F, with moderate-severe acne and > 10 inflammatory acne lesions; FPT not givenComparison of PDT with different light sources for activation: ALA activated by IPL (600–850 nm), or a combination of IPL (580–980 nm) and bipolar radiofrequency energies, or blue light (417 nm) in a parallel-group trial; 3 treatments at 2-week intervals; follow up at 1 and 3 months after final treatmentThe method used for evaluation was unclear. At 1 month after treatments differences among groups were not statistically significant (P = 0.3210); median percentage improvement score (96.9% CI) was 58.75 (5-70) in the IPL group, 20 (0-80) in the IPL-RF group and 15 (0-87.5) in the blue-light group. At three months data were only reported for IPL and blue light only groups 72.3 (range 42.5) versus 15 (range 27.5), so the analysis was not possible.
Yin 2010180 (83 M, 97 F), aged 18-38, mean 25.8, with moderate-severe facial acne (Pillsbury), FPT III-IV, 45 participants in each group633 ± 3 nm (red light) plus different ALA concentrations (5%, 10%, 15% and 20%) vs red light alone, 4 treatments every 10 days, 4 parallel groups, each treated with a different concentration on the right side and placebo agent on the left side; assessments at 2, 4, 12 and 24 weeks after last treatmentNon-standardised scale ('marked improvement', 'moderate improvement', 'no charge' or 'acne worse') was used for evaluation. At 24 weeks after treatment a majority of the participants assessed that their acne had improved on both ALA-PDT and control cheeks. In the 20% ALA group 44/45 of participants (98%, 1 drop-out due to adverse effects) experienced a 'marked improvement' in their acne at ALA-PDT sites, 42/45 (95%) in the 15% ALA and 36/40 (90%) in 10% ALA groups. Other data were not reported in text, but in graph format only. Our interpretation of the graph was that 30 participants in the 5% ALA group (67%) reported 'marked improvement', 3/45 (87%) of participants in the 15% ALA, 5/45 (11%) in the 10% ALA, and 9/45 (20%) in the 5% ALA group reported 'moderate improvement'. One participant, 1/45 (2%) in the 10% ALA group, as well as 3/45 (7%) in the 5% ALA group reported 'no change', whereas 3/45 (7%) of participants in both 10% and 5% ALA reported 'acne worse'. We dichotomised the data to 44/45 'success' outcomes in the 20% ALA group, 42/45 in the 15% ALA group, 36/45 in the 10% ALA group and 30/45 in the 5% ALA group. 20% ALA was not superior to 15% ALA with RR (95% CI) of 1.05 (0.96, 1.15) and P = 0.3. However, 20% ALA was more effective than 10% ALA with RR (95% CI) of 1.22 (1.05, 1.42) and P = 0.01, and more effective than 5% ALA with RR (95% CI) of 1.47 (1.19, 1.81) and P = 0.0004. The NNTB were 6 (95% CI 3 to 19) and 4 (95% CI 2 to 6) for the comparison of 20% ALA with 10% and 5% ALA respectively. However, there is no calculable NNTB for the comparison of 20% to 15% ALA since the 95% CI for the risk difference contains zero (i.e. no effect), and this corresponds to an infinite upper 'limit' for the 95% CI for the NNTB, which indicates that there is no true boundary on how large the NNTB could be for this comparison.
7. Other (non-MAL, non-ALA) PDT versus other comparators
ICG-PDT
Kim 200916 (7 M, 9 F, aged 16-34 years, mean age 25 ± 3.09) with mild-moderate acne, skin types not given, 9 in single, 7 in multiple treatment group2 groups randomised: single treatment vs multiple treatments (once weekly over 3 weeks); right cheek of each patient indocyanine green plus 805 nm light, left cheek light only and forehead "spontaneous resolution" control, evaluated 2 and 4 weeks after final treatment, multiple group also at final treatment–100 to +100 scale scoring was used for evaluation, no details were reported. At both 2 and 4 weeks after final treatment difference between PDT and light-only side was statistically significant only in the multiple treatment group (P < 0.05 at all assessment time points). Further data were not reported. Our interpretation of the graph was that at 4 weeks after final treatment mean VAS score was 20 for both PDT and light-only side in the single treatment group; whereas in the multiple treatment group 50 on the light-only side and 60 on the PDT side. SDs not presented in the graph.
Table 2. Investigator-assessed change in lesion count, studies of light-only therapies (excluding comparisons with photodynamic therapy)
  1. ALA = 5-aminolevulinic acid
    BPO = benzoyl peroxide
    CHA = chlorophyll-a
    FPT = Fitzpatrick's Skin Types: based on different reactions to sun exposure and range from type I ('pale white skin which always burns and never tans') to type VI ('deeply pigmented dark brown to black skin which never burns and tans very easily') (Fitzpatrick 1988)
    GAAS = Global acne assessment scoring scale
    IAA = indole 3-acetic acid
    IGA = Investigator global assessment score
    ILs = inflamed lesions
    IPL = intense pulsed light
    IR = infrared
    ITT = intention-to-treat analysis
    LPDL = long pulsed dye laser
    LOCF = last observation carried forward
    LLT = lower level term
    MAL = methyl-aminolevulinate
    NILs = non-inflamed lesions
    NNTB = number needed to treat for an additional beneficial outcome
    OFI = optical fibre intra-tissue irradiation
    PDL = pulsed-dye laser
    PDT = photodynamic therapy
    PT = preferred term
    RCT = randomised controlled trial
    SD = standard deviation
    SE = standard error
    SPF = Sun protection factor
    TER = total effective rate
    TLMB = topical liposomal methylene blue

    Change from baseline i.e. absolute change is calculated by subtracting baseline count from count assessed at certain time point. Percentage change is calculated by dividing the absolute change with baseline count and then multiplying that value by 100 to get percentages.

    Unless specified differently, results presented as reported in the published papers, without performing independent analysis. Please see Characteristics of included studies for details on withdrawals and drop-outs of participants for each study.

Study Participants Intervention(s) and control(s) Investigator-assessed change in lesion counts
1. Light versus placebo or no treatment
Yellow light versus placebo or no treatment
Seaton 200341, 31 randomised to treatment, 10 to control group; with mild-moderate acne, other characteristics not given585 nm pulsed dye laser vs sham laser, parallel-group trial, single treatment, assessed at 2, 4, 8 and 12 weeks after treatmentSignificantly greater improvement from baseline in ILs and total lesion counts (P = 0.024 and 0.023 respectively) in laser-treated group than in placebo group at 12 weeks, whereas the difference in improvement in NILs was non significant (P = 0.14). ILs median (interquartile range) improvement from baseline in the treatment group was 49% (30% to 75%) versus 10% (-8% to 49%) in the placebo group, NILs 40% (0% to 75%) versus -13% (-42% to 23%), total lesion 53% (19% to 64%) versus 9% (-16% to 38%).
Orringer 200440 (24 M, 16 F) enrolled, 26 completed, mean age 20.7 years (range not reported), with facial acne Leeds score > 2; FPT not given ("28 whites, 7 Asians, 2 blacks, 3 unknown")585 nm PDL in a split-face trial, single treatment and 2 treatment groups (2 weeks apart), serially assessed for 12 weeks after final treatmentNon significant differences in changes in means of papules (P = 0.08), pustules (P = 0.12), comedones (P = 0.63) and cysts (P > 0.99) at 12 weeks between treated and untreated face sides. Changes in means (95% CIs) of papules, pustules, comedones and cysts at 12 weeks -4.2 (-7.7 to -0.6), 0 (−1.4 to 1.4), 2.9 (−4.0 to 9.7) and 0 (-0.5 to 0.5) on the treated sides respectively; and −2.2 (−5.3 to 0.9), -1 (-2 to -0.01), 1.6 (-5.2 to 8.4) and 0 (-0.6 to 0.6) on the untreated sides respectively. LOCF method was used for analysis (n = 38). Our analyses using LOCF data (n = 38) confirmed no significant differences in means between treated and untreated face sides at 12 weeks, MD (95% CIs) for investigator-assessed change in ILs (papules) was -2.00 (-6.60, 2.60), P = 0.39, for investigator-assessed change in ILs (pustules) 1.00, P = 0.24, and for investigator-assessed change in NILs 1.30 (-8.00, 10.60), P = 0.78 and for investigator-assessed change in cysts 0.00 (-0.76, 0.76), P = 1.00. Please note that we based all the calculations from the values provided in the table reported, and we double and triple checked the values using both RevMan and R statistical software, but some of our P values did not match up with the ones presented by the study authors.
Infrared light versus no treatment
Darne 201138 (7 M, 31 F), aged 18-47 years (mean 28), with moderate-severe facial acne; FPT I-V1450 nm laser (8-9 J/cm²) in a split-face trial, 3 treatments monthly, assessed monthly for 4 months, then at 3-monthly intervals for 12 months after final treatmentSimilar reduction in ILs at 1 and 12 months on both sides; treated sides median 0 (95% CI -4 to 2) and untreated sides median 0 (95% CI -3.7 to 0).
Orringer 200746 (10 M, 36 F) enrolled, 30 completed, mean age 23.9 years (range not reported) with clinically apparent active facial acne; FPT II–VI1320 nm Nd:YAG laser in a split-face trial with cooling; 3 treatments at 3-week intervals; assessed at weeks 7 and 14No significant differences in changes in papules (P = 0.62), pustules (P = 0.39), open (P = 0.09), nor closed comedones (P = 0.20) between the treated and untreated sides at week 14. Difference in changes in cyst counts was significant (P = 0.04). Mean (SE) changes in papules, pustules, open comedones, closed comedones and cysts reported at week 14: -1.57 (0.95), -2.54 (1.45), -1.08 (1.91), -8.19 (3.48) and 0 (0.08) on the treated sides respectively; and -1.03 (1.31), -1.86 (1.16), 1.84 (1.85), -1.24 (7.45) and 0.43 (0.17) on the untreated sides respectively. LOCF method was used for analysis (n = 37, 9 participants withdrew prior to any clinical endpoint evaluation, and were not included in the analysis). Our analyses using LOCF data (n = 37) confirmed no significant differences in means between treated and untreated face sides at week 14 (i.e. 8 weeks after final treatment), MD (95% CIs) for investigator-assessed change in ILs (papules) was -0.54 (-3.71, 2.63), P = 0.74, for investigator-assessed change in ILs (pustules) -0.73 (-4.37, 2.91), P = 0.69, for investigator-assessed change in NILs (open comedones) -2.92 (-8.13, 2.29), P = 0.27, for investigator-assessed change in NILs (closed comedones) -6.95 (-23.07, 9.17), P = 0.40. The difference in means for investigator-assessed change in cysts was significant (MD -0.43, 95% CI -0.80 to -0.06), P = 0.02. Please note that we based all the calculations from the values provided in the table reported, and we double and triple checked the values using both RevMan and R statistical software, but some of our P values did not match up with the ones presented by the study authors.
Moneib 201424 (5 M, 19 F), age 15-38 years (mean 21.5), with moderate-severe acne; FPT II-VFractional Erbium Glass 1559 nm laser, in a split-face trial, 4 treatments at 2-week intervals; assessed every 3 months for 1 year after final treatmentFollow-up time point unclear. At treated sides mean papules counts (SD) reduced from baseline 15.42 (14.38) to 0.88 (3.35), mean pustules count from baseline 2.58 (3.32) to 0.46 (1.38), open comedones from 4.25 (7.59) to 1.25 (3.07), closed comedones from 1.75 (3.45) to 0.33 (1.01) and nodules from baseline 1.00 (1.87) to 0.08 (0.41) at "follow up". At control sides mean papules counts (SD) changed from baseline 12.83 (10.89) to 14.08 (12.93), mean pustules count from baseline 3.17 (5.21) to 4.21 (7.40), open comedones from baseline 2.58 (3.37) to 2.88 (3.54), closed comedones from baseline 1.79 (3.75) to 1.21 (2.50) and nodules from baseline 0.92 (1.61) to 1.79 (2.00) at "follow up".
Blue light versus placebo or no treatment
Elman 200323 (11 M, 12 F), mean age 18.8 years (range not given) with mild-severe papulopustular acne; skin types not documented405–420 nm laser with skin cooling in a split-face trial, twice weekly for 4 weeks, assessed at each treatment and at 2, 4, and 8 weeks after treatmentILs percentage change median reduction of 30% at final treatment on untreated sides, other data not available. ILs percentage change median reduction at 2, 4 and 8 weeks post treatment 59%, 61% and 53% respectively on treated sides (P = 0.01 at 8 weeks compared to untreated sides, using McNemar test; other statistical data not provided)
Red light versus no treatment
Na 200730 (7 M, 23 F) aged 19–33 years (mean 23.6) with mild-moderate acne; skin types not documented635–670 nm portable red light device in a split-face trial, self-administered to the treatment side twice daily for 8 weeks; assessed at weeks 1, 2, 4 and 8, and then for 8 weeks after final treatmentAt week 8, NILs percentage change -59% on treatment sides versus 3% increase on control sides (P < 0.005), ILs percentage change -66% on treatment side vs 74% increase in ILs on control sides (P < 0.005). Further data not given. At 4 weeks after final treatment 10/25 (40%) of followed-up participants were reported to have "showed an increase in acne lesions", and at 8 weeks 21/22 (95%) were reported to "have complained of acne exacerbation compared with their status during treatment period". Further data were not provided.
Blue-red light versus placebo
Papageorgieu 200030 participants, mean age 24.8 years in blue–red light group; 25 participants, mean age 25.6 years in white light control group; randomised from the original 107 recruited (33 M, 74 F, age 14-50 years), all with mild-moderate acne; skin types not stated415 nm plus 660 nm light vs cool white light; treated daily for 12 weeks; assessed every 4 weeks for the 12-week treatment periodBlue-red light superior at all time points, differences in mean percentage improvements (95% CI) 50.3 (40.1 to 60.5) for ILs and 66.5 (56.0 to 77.0) for comedones at week 12 (final treatment).
Kwon 201335 participants (11 M, 24 F); aged 20-27 years (mean not given), with mild-moderate acne, FPT III-V; 18 participants in the blue-red light group, 17 in the placebo group420 plus 660 nm home use LED device vs home use sham device; self-treatment twice daily for 4 weeks in a split-face trial; assessed 4 and 8 weeks after final treatmentMean IL counts reduced from baseline 22.8 to 5.3 (by 76.7%, P < 0.01) and mean NILs counts reduced from baseline 51.2 to 23.5 (by 53.3%, P < 0.01) at eight weeks after final treatment in the blue-red light group. Mean reduction of ILs and NILs counts in the placebo group was not statistically significant at eight weeks after final treatment (both P > 0.05). Results reported as percentage improvements in graph format (means and SDs not presented).
2. Light versus topical treatment
Light versus BPO
de Arruda 200960 (34 M, 26 F, mean age 17.3, range not given), all with Brasilian Group of Acne grade II or III, skin types described as mixed Brazilians (11), caucasian (47) and unknown (2).407 nm-420 nm (blue light) twice weekly for 4 weeks vs 5% BPO, self-administered, twice daily; parallel groups, assessed 4 weeks after initial treatment and 2weeks after end treatmentNo statistically significant difference in decrease of means of ILs (P = 0.500) and NILs (P = 0.177) between the blue light and 5% BPO group. In the blue-light group ILs means (SD) reduced from baseline 27.87 (18.08) to 23.33 (15.10) at 4 weeks. NILs means (SD) reduced from baseline 111.6 (45.03) to 85.92 (57.78) at 4 weeks. In the BPO group ILs means (SD) reduced from baseline 35.37 (22.16) to 19.14 (17.95) at 4 weeks. NILs means (SD) reduced from baseline 128.67 (90.8) to 93.50 (69.74) at 4 weeks. We calculated that at 4 weeks the mean difference (95% CI) in changes in NILs was 9.49 (-10.84, 29.82); however, the mean difference in changes in ILs was 0 (and since the P value the study authors presented was 0.5, then there are infinitely many possibilities for the standard error, hence the lack of a 95% CI provided for ILs).
Papageorgieu 200030 participants, mean age 24 ± 8 years in blue–red light group and 25 participants, mean age 26 ± 7 years in the BPO group, randomised from the original 107 recruited (33 M, 74 F, age 14-50 years) all with mild-moderate acne; skin types not stated415 nm plus 660 nm light vs 5% BPO, parallel groups, treated daily; assessed every 4 weeks for the 12-week treatment periodBlue–red light superior to BPO at week 12 (P = 0.006). Difference in mean percentage improvements (95% CI) at week 12 was 17.6 (7.5 to 27.6) for IL counts and 0.9 (-9.4 to 11.3) for comedones.
Chang 200730 women aged 23-32 years (mean 25 ± 7); with mild-moderate acne; FPT III-IVIPL with 530–750 nm filter with cooling gel in a split-face trial, 3 sessions, 3 weeks apart, BPO gel used on both sides of the face. Assessed 3 weeks after final treatmentNo significant difference between IPL-treated and untreated sides of the face for changes in mean papule and pustule counts (-3.2 vs -3.1; P > 0.05). Further data not reported
Light versus clindamycin
Gold 200534 (25 completed the trial, 3 M and 22 F) aged 13–55 years (mean 31 ± 0) with mild-moderate acne; skin types described: caucasian (16), African-American (7), American-Indian (1), Chinese (1); 13 participants in clindamycin group and 12 in blue light group417 nm (blue light) twice weekly for 4 weeks vs self-administered topical clindamycin 1%, twice daily, parallel groups, assessed at 4 weeks after final treatmentNILs & ILs count "averages" (ranges) in the blue-light group were 29.4 (9 to 120) and 22.6 (16 to 34) at baseline and 21.4 (8 to 40) and 11.1 (0 to 24) 4 weeks after final treatment respectively. NILs & ILs count "averages" (ranges) in the clindamycin group were 29 (9 to 95) and 17.4 (12 to 32) at baseline and 12 (4 to 38) and 10.4 (4 to 19) 4 weeks after final treatment respectively.
Lee 20109, with inflammatory acne (other characteristics not given)Full-spectrum light twice a week vs 1% clindamycin twice a day, in a split-face trial, for 4 weeks, evaluation weekly whilst on treatment and 2, 4 and 8 weeks after final treatmentReduction in IL counts by 76.8% at light and 25.5% at clindamycin-treated side (time point and other data not given)
Light and other topical treatments
Karsai 201089 randomised, 80 evaluated (38 M, 42 F, 13.3-43.8 years, mean ± SD age 19.7 ± 5.9 years), with mild-moderate acne (Investigator's Static Global Assessment -ISGA score 2-4), FPT I-IIIClindamycin 1% BPO 5% hydrating gel (C ⁄ BPO) alone, once daily "throughout the observation period" vs in combination with two 585 nm PDL treatments. Parallel groups, assessed at 2 and 4 weeks after initial treatmentIn the C/BPO group there was a 36.3% reduction in number of ILs and 9.2% reduction in total lesion count 4 weeks after initial treatment. In the C/BPO plus light group there was a 36.9% reduction in number of ILs and 9.0% reduction in total lesion count. Means and SD reported in graph format. Our interpretation of the graph was that ILs (SD) in the C/BPO group reduced from baseline 37.5 (20) to 25 (15), and in the C/BPO plus light group from 50 (30) to 30 (25) at 4 weeks after initial treatment. Total lesions reduced from baseline 127.5 (70) to 115 (70) in the C/BPO group, and from 175 (105) to 150 (100) in the C/BPO plus light group at 4 weeks after initial treatment. We judged further analyses would be biased due to lack of precise data, so we did not perform them.
Anyachukwu 201440 (all M), 20 randomised to the light group, 20 to the placebo group, mean age 22 ± 4 years (range not reported), Global Acne Grading System (GAGS) > 19, FPTs not given905 nm light combined with "self-management topical agents" ("antibiotic cream", "medicated soap", "talcum powder" or "personal hygiene"), 8 light treatments, twice weekly over 4 weeks, in a parallel-group trial, control group treated with placebo-non radiating light probe combined with "self-management topical agents", details of topical treatment not given, unclear frequency of application; assessed within treatment and 3 days after final treatmentMean percentage change from baseline in combined number of lesions (SD) was 54.98 (16.297) in the laser group and 17.97 (16.472) in the control group 3 days after final treatment. Mean percentage changes from baseline in combined number of lesions at 3 days after final treatment were 70.37, 61.90, 71.43, 71.43 in the laser combined with "antibiotic cream", "medicated soap", "talcum powder" and "personal hygiene" subgroups respectively. Mean percentage change from baseline in combined number of lesions at 3 days after final treatment were 38.71, 45.00, 10.34 and 12.50 in the placebo plus "antibiotic cream", "medicated soap", "talcum powder" and "personal hygiene" subgroups respectively. Further data were not given.
Ash 201541 (M/F not reported, study authors clarified "a 50/50 split"), 26 in the intervention, 15 in control group, aged 16–45 years (mean not reported) with mild-moderate acne (Leeds grade); FPT not given: "Caucasian, Asian and mixed Afro-Caribbean ethnic groups"Pre-treatment facial wash/weak chemical peel (containing salicylic acid, glycolic acid, lactic acid) followed by treatment with blue-light device and then post-treatment facial moisturiser (containing salicylic acid, glycolic acid, lactic acid, menthol, niacin) versus unclear control in a parallel-group trial, 28 sessions in total, every other day for 8 weeks. Assessed at 12 weeks (4 weeks after final treatment?)At 12 weeks (4 weeks after final treatment?) mean lesion counts reduced by 50.08% (P = 0.002) in the treatment group. In the control group, mean lesion counts increased by 2.45% (P = 0.0029). Further data not given
Borhan 201440 (8 M, 12 F in the light group, 9 M, 11 F in the control group), mean age 21.3 ± 2.0 in the intervention and 21.05 ± 2.18 in the control group (range 18-25 years), with mild-moderate acne vulgaris (Burton scale), FPT III-IV595 nm light plus "traditional topical antibiotic medication" versus "traditional topical antibiotic medication" alone in a parallel-group trial, 3 light treatments in total, at 4-week intervals, details of topical treatment not given, unclear frequency of application; assessed at week 4, 8 and 12 (final evaluation 4 weeks after final treatment)At week 12 combined number of lesions, reported as "acnes number", (SD) changed from baseline 25.7 (5.88) to 8.75 (2.91) in the laser + topical antibiotics group, and from baseline 25.75 (6.71) to 17.7 (5.14) in the topical antibiotics-alone group ( P = 0.0001).
    
3. Light versus other comparators
Comparison of light therapies of different wavelengths
Papageorgieu 200030 participants, mean age 24.8 years in blue–red light group and 27 participants, mean age 23.4 years in the blue-light group, randomised from the original 107 recruited (33 M, 74 F, age 14–50 years) all with mild-moderate acne; skin types not stated415 nm plus 660 nm light vs 415 nm light, parallel groups, treated daily for 12 weeks; assessed every 4 weeks for the 12-week treatment periodThere was no significant difference between the treatments in ILs at week 12 (P = 0.1), nor in comedone count (P value not given). Difference in mean percentage improvements (95% CI) at week 12 was 13.1 (3.0 to 23.1) for ILs counts and 12.9 (2.5 to 23.2) for comedones.
Liu 201120 (6M/14F) completed the study, number of randomised participants not reported, 10 completed in the blue light, 10 in the red-light group, aged 19–28 years (mean 23.6 years) with mild-moderate acne (Global Acne Grading System); FPT III-IVBlue (405 ± 10 nm) vs red (630 ± 10 nm) LED portable device treatments, about 20 cycles of illumination and the corresponding light doses received in each session were 7.2 J/cm² and 11.52 J/cm², in a parallel-group trial, 8 sessions in total, twice weekly for 4 weeks; assessed at 4 weeks after final treatment and at each treatment sessionIn the blue-light group, the mean ILs count (papules and pustules) dropped from baseline 19.2 to 5.5 (by 71.4%) at final treatment and in the red-light group from baseline 8.2 to 6.6 at final treatment (by 19.5%). SDs and further data not given
Choi 201020 (1 M, 19 F, age 20-37, mean age 26); all with acne (Cunliffe severity grade 2-4), FPT III-V585 nm PDL vs 530-750 nm IPL, 4 treatments at 2-week intervals, in a split-face trial, assessed 4 and 8 weeks after last treatment4 weeks after final treatment greater reductions on PDL sides versus IPL treatment sides for NILs (47% versus 33% reduction), but lower for ILs (62% versus 66%). 8 weeks after final treatment significantly greater improvements on PDL sides versus IPL treatment sides for both ILs (86% versus 35% reductions) and NILs (59% versus 43% reduction). Individual participant data reported at baseline and 8 weeks (n = 17). We calculated means (SD): IPL at baseline: ILs 6.17 (3.67) NILs 15 (8.51), at 8 weeks ILs 2.23 (2.19) NILs 6.52 (4.15) PDL at baseline: ILs 6.76 (4.08) NILs 14.64 (8.65), at 8 weeks ILs 0.82(1.13) NILs 5.41(3.93). Mean differences (95% CI) between the 2 treatments at 8 weeks using t-distribution were 2.00 (-0.85, 4.85), P = 0.178, t = 1.431 for changes in ILs and 0.77 (-3.65, 5.19), P = 0.735, t = 0.355 for changes in NILs. MDs (95% CI) between the two treatments at 8 weeks using normal distribution were 2.00 (-0.74, 4.74), P = 0.15 for changes in ILs and 0.77 (-3.49, 5.03), P = 0.72 for changes in NILs.
Jung 200918 enrolled, 16 completed (5 M, 11 F, aged 20-31 years, mean age 26); with mild-moderate acne (Cunliffe severity grade 2-5), skin types not given585 nm PDL vs combined 585/1064 nm PDL, in a split-face trial, 3 treatments at 2-week intervals, assessed at 8 and 12 weeks after initial treatmentILs and NILs reduced by 86% and 69% respectively on the PDL sides and by 89% and 64% on the 585/1,064 nm laser sides respectively at final evaluation (P values reported as < 0.05 "compared with baseline"). No significant difference in the effect of the two interventions (P values and further data not provided)
Comparison of light therapies of different doses
Bernstein 20077 enrolled, 6 completed (1 M, 6 F, aged 23-41 years, mean age 29), all with active papular acne, FPT I-IIIComparison of two 1450 nm laser treatments; single-pass, high-energy (13–14 J/cm²) vs double-pass, low-energy (8–11 J/cm²); 4 treatments at monthly intervals, assessed 1 month following each treatment and 2 months after final treatmentILs counts means (SD) dropped from 19.5 (11.9) to 4.2 (4.7) on the single-pass face side, and from 16.2 (6.0) to 5.2 (4.5) on the double-pass face side. Individual participant data reported (n = 6). We calculated mean difference (95% CI) of -4.33, 95% CI -13.4 to 4.74, P = 0.372, t = -1.063 using t-distribution and MD (95% CI) of -4.33 (-12.31, 3.65), P = 0.29 using normal distribution
Jih 200620 (10 M, 10 F), aged 18-39 years (mean 23) with active inflammatory facial acne; FPT II–VI1450 nm diode laser in a split-face trial using anaesthetic cream and 14 J/cm² in one group and 16 J/cm² in the second, with three treatments given at 3–4 week intervals, assessed at 1, 3, 6 and 12 months after final treatmentBaseline mean IL counts 16.1 for the 14/J cm² and 16.8 for the 16 J/cm² side (SDs not reported). At 1, 3, 6 and 12 month follow-up percentage reductions were 75.1%, 88.6%, 81.6% and 76.1% on the 14 J/cm² and 70.6%, 81.5%, 84.1% and 70.5% on the 16 J/cm² face side respectively (P < 0.001). There was no significant difference in reduction between the different light intensities. Sponsors provided detailed data and our analyses confirmed that. The mean differences (95% CI) in changes in ILs and percentage changes in ILs calculated using t-distribution were -2.40 (-6.46, 1.66), P = 0.26, t = -1.203 and -3.40 (-14.21, 7.41), P = 0.54, t = - 0.641 respectively at 1 month; -3.20 (-7.43, 1.03), P = 0.15, t = 1.541 and -7.05 (-16.05, 1.95), P = 0.13, t = -1.596 respectively at 3 months; -2.00 (-5.87, 1.87), P = 0.32, t = -1.053 and 2.49 (-6.37, 11.35), P = 0.59, t = 0.572 respectively at 6 months; and -2.40 (-7.13, 2.33), P = 0.33, t = -1.034 and -5.59 (-26.07, 14.89), P = 0.60, t = -0.556 respectively at 12 months. The MDs (95% CI) in changes in ILs and percentage changes in ILs calculated using normal distribution were -2.40 (-6.31, 1.51), P = 0.23 and -3.40 (-13.80, 7.00), P = 0.52 respectively at 1 month; -3.20 (-7.27, 0.87), P = 0.12 and -7.05 (-15.71, 1.61), P = 0.11 respectively at 3 months; -2.00 (-5.72, 1.72), P = 0.29 and 2.49 (-6.04, 11.02), P = 0.57 respectively at 6 months; and -2.40 (-6.95, 2.15), P = 0.30 and -5.59 (-25.30, 14.12), P = 0.58 respectively at 12 months
Uebelhoer 200711 (2 M, 9 F, age 19-39 years, mean age 26), 9 completed, all with ≥ 10 inflammatory papules on each side of the face and Allen-Smith grade ≥ 3 and ≤ 5; skin types not given1450 nm laser single-pass treatment consisting of stacked double pulses vs a double-pass treatment of single pulses; in a split-face trial, treated every 3 weeks for a total of 3 treatments, assessed before each follow-up treatment, and at 3 months after the final treatmentStatistically significant reduction of mean acne lesion counts on both the single-pass side and double-pass side of 57.6% (P = 0.02) and 49.8% (P = 0.02), respectively. Further details not given
NCT00706433266 (128 M, 138F), 68 in the ALA 1000 s group, 65 in the ALA 500 s group, 67 in the vehicle 1000 s group and 66 in the vehicle 500 s group, mean age 20.1 years, inclusion criterion 12 > years, with moderate and severe acne (IGA score 3 and 4, with at least 20 ILs); FPT I-V20% ALA (45 min incubation) plus blue 1000 s light vs 20% ALA (45 min incubation) plus 500 s blue light vs vehicle (45 min incubation) plus blue 1000 s light vs vehicle (45 min incubation) plus 500 s blue light; in a parallel-group trial; up to 4 treatments at 3-week intervals, assessed 3 and 6 weeks after the final treatmentAt 3 weeks after final treatment investigator-assessed median change in ILs (SD) was -19.0 (22.8) in the vehicle 1000 s and -14.5 (24.0) in the vehicle 500 s group; investigator-assessed median percentage change in ILs (SD) was -41.7 (38.82) in the vehicle 1000 s and -37.0 (40.23) in the vehicle 500 s group. At 6 weeks after final treatment investigator-assessed median change in ILs (SD) was -21.0 (23.63) in the vehicle 1000 s and -17.0 (26.71) in the vehicle 500 s group; investigator-assessed median percentage change in ILs (SD) was -48.4 (32.81) in the vehicle 1000 s and -45.2 (50.15) in the vehicle 500 s group. Statistical tests to determine whether any changes were significant could not be performed due to the study authors’ use of median changes rather than the typical mean changes required for significance testing in order to make appropriate comparisons with other included studies. Furthermore, it is not clearly stated whether the study authors implemented an ITT analysis or a LOCF approach to handling missing data.
Light alone versus combined with microdermoabrasion
Wang 200620 (7 M, 13 F) age 19–59 years (mean 34 ± 3) with active inflammatory facial acne; FPT II–IV1450 nm diode laser plus microdermoabrasion in a split-face design with light treatment on the control side of the face with topical anaesthetic to whole face; 4 treatments, 3 weeks apart; assessed at 6 and 12 weeks after the final treatmentMicrodermabrasion plus light treatment decreased the mean acne lesion count by 52.8% by 6 weeks and 54.4% by 12 weeks (P <0.02 compared with baseline counts). Light treatment alone reduced the counts by 53.5% by 6 weeks and 61.1% by 12 weeks (P <0.05 compared with baseline counts). No statistically significant difference between the two treatments at any point
Light in combination with carbon lotion versus no treatment
Jung 201222 (4 M, 18F), 20 completed (2 M, 18F, aged 19-34 years, mean age 25.4), FPT III-IV, acne severity not givenCarbon lotion plus quasi-long pulse and Q-switched 1064 nm Nd:YAG laser vs non treated control, in a split-face trial, 3 treatments over 4 weeks, evaluation every 2 weeks whilst on treatment and then every 4 weeksDifference in means of both ILs and NILs statistically significant between treated and untreated side (P < 0.001), but clear data for non treated side not given. Both ILs and NILs reduced to 58.6% (P < 0.001) and to 52.4% (P < 0.001) respectively on the laser-treated side
Table 3. Investigator-assessed change in lesion count, studies of photodynamic therapy (including comparisons with light-only therapies)
  1. ALA = 5-aminolevulinic acid
    BPO = benzoyl peroxide
    CHA = chlorophyll-a
    FPT = Fitzpatrick's Skin Types: based on different reactions to sun exposure and range from type I ('pale white skin which always burns and never tans') to type VI ('deeply pigmented dark brown to black skin which never burns and tans very easily') (Fitzpatrick 1988)
    GAAS = Global acne assessment scoring scale
    IAA = indole 3-acetic acid
    IGA = Investigator global assessment score
    ILs = inflamed lesions
    IPL = intense pulsed light
    IR = infrared
    ITT = intention-to-treat analysis
    LPDL = long pulsed dye laser
    LOCF = last observation carried forward
    LLT = lower level term
    MAL = methyl-aminolevulinate
    NILs = non-inflamed lesions
    NNTB = number needed to treat for an additional beneficial outcome
    OFI = optical fibre intra-tissue irradiation
    PDL = pulsed-dye laser
    PDT = photodynamic therapy
    PT = preferred term
    RCT = randomised controlled trial
    SD = standard deviation
    SE = standard error
    SPF = Sun protection factor
    TER = total effective rate
    TLMB = topical liposomal methylene blue

    Change from baseline i.e. absolute change is calculated by subtracting baseline count from count assessed at certain time point. Percentage change is calculated by dividing the absolute change with baseline count and then multiplying that value by 100 to get percentages.

    Unless specified differently, results presented as reported in the published papers, without performing independent analysis. Please see Characteristics of included studies for details on withdrawals and drop-outs of participants for each study.

Study Participants Intervention(s) and control(s) Investigator-assessed change in lesion counts
4. MAL-PDT versus other comparators
MAL-PDT versus red light alone
Pariser 2013153 participants (87 M/66 F), 100 in the 80 mg/g MAL-PDT group, 53 in the placebo group, aged 12-35 years (mean 18.6), with severe facial acne vulgaris, IGA score 4, 25-75 ILs and 20-100 NILs on the face, FPT I-VI80 mg/g MAL-PDT under occlusion followed by illumination with 632 nm 37 J/cm²red light vs placebo cream plus 632 nm 37 J/cm² light in a parallel-group trial, 4 treatments at 2-week intervals, assessed at 6 weeks after final treatment15 withdrawals from the MAL-PDT group, 4 withdrawals and 1 lost to follow-up from the placebo group. ITT analysis was performed. Our analyses for the individual study showed that at 6 weeks after final treatment 80 mg/g MAL-PDT was superior to placebo cream plus red light in change in ILs (MD -7.80, 95% CI -14.39 to -1.21), in percentage change in ILs (MD -21.10, 95% CI -37.69 to -4.51), but was not superior in change in NILs (MD -1.10, 95% CI -8.11 to 5.91), nor in percentage change in NILs (MD -3.70, 95% CI -19.30 to 11.90). Please note that the results of this study were combined with those of NCT00933543 and NCT00594425 for the same comparison.
NCT00933543107 participants (48 M/59 F), 54 in the 80 mg/g MAL-PDT group, 53 in the placebo group, aged 11-35 years (mean 17.2), with moderate-severe facial acne vulgaris, IGA score 3-4, 20-100 ILs and 30-120 NILs on the face, FPT I-VI80 mg/g MAL-PDT (without occlusive dressing) followed by illumination with 632 nm 37 J/cm² red light vs placebo cream plus 632 nm 37 J/cm² light (without occlusive dressing) in a parallel-group trial, 4 treatments at 2-week intervals, assessed at 6 weeks after final treatment3 withdrawals in MAL-PDT group, 6 withdrawals and 1 lost to follow-up in placebo group. ITT analysis was performed. Our analyses for the individual study showed that at 6 weeks after final treatment 80 mg/g MAL-PDT was not superior to placebo cream plus red light in change in ILs (MD -0.20, 95% CI -8.19 to 7.79), in percentage change in ILs (MD -5.60, 95% CI -21.50 to 10.30), in change in NILs (MD 2.80, 95% CI -7.13 to 12.73), nor in percentage change in NILs (MD -1.60, 95% CI -18.83 to 15.63). Please note that the results of this study were combined with those of Pariser 2013 and NCT00594425 for the same comparison.
NCT00594425150 participants (59 M/91 F), 50 in the 40 mg/g MAL-PDT group, 48 in the 80 mg/g MAL-PDT group, 52 in the placebo group, aged 15-40 years (mean 21.3), with moderate-severe acne, IGA score 3-4, 20-100 ILs and up to 200 NILs on the face, FPT I-IV80 mg/mL MAL under occlusion (1.5h) plus 632 nm 37 J/cm² light vs 40 mg/mL MAL under occlusion (1.5 h) plus 632 nm 37 J/cm² light vs placebo cream plus 632 nm 37 J/cm² light in a parallel-group trial, 4 treatments at 2-week intervals, assessed at 2, 3, 6, 12 and 24 weeks after final treatment

43 participants completed in the 40 mg/g group, 34 in the 80 mg/g group and 42 participants completed in the placebo-cream group, ITT analysis was performed. Our analyses showed that at 6 weeks after final treatment 40 mg/g MAL-PDT was not superior to placebo cream plus red light in change in ILs (MD -3.00, 95% CI -7.76 to 1.76), P = 0.22, in percentage change in ILs (MD -7.90, 95% CI -22.33 to 6.53), P = 0.28, in change in NILs (MD -7.50, 95% CI -16.07 to 1.07), P = 0.09, while there was a borderline superiority in percentage change in NILs (MD -25.80, 95% CI -51.69 to 0.09), P = 0.05.

Our analyses for the individual study showed that at 6 weeks after final treatment 80 mg/g MAL-PDT was not superior to placebo cream plus red light in change in ILs (MD -0.80, 95% CI -5.61 to 4.01), in percentage change in ILs (MD -4.80, 95% CI -19.85 to 10.25), in change in NILs (MD -6.90, 95% CI -15.38 to 1.58), while there was a borderline superiority in percentage change in NILs (MD -27.50, 95% CI -52.76 to -2.24). Please note that the results of this study were combined with those of Pariser 2013 and NCT00933543 for the same comparison.

NCT0067393320 (11 M, 9 F), age 14-40 years (mean 26 years) with moderate-severe acne, FPT V-VI80 mg/mL MAL plus 653 nm light vs placebo cream plus 653 nm light, in a split-back trial, 2 sessions, 2 weeks apart, assessed 4 weeks after final treatmentBaseline ILs means (full range) were 5.9 (5 to 11) in the PDT group and 6.0 (5 to 10) in the placebo group. Mean change from baseline in number of ILs ± SD at 4 weeks was -3.70 ± 2.43 in MAL-PDT group and -3.90 ± 2.07 in the placebo plus red light group. Baseline NILs means (full range) were 6.5 (1 to 21) in the PDT group and 5.4 (2 to 17) in the placebo group. Mean change from baseline in number of NILs ± SD at 4 weeks was -2.95 ± 4.84 in MAL-PDT group and -2.50 ± 2.65 in the placebo plus red light group. Using t-distribution, we calculated that at 4 weeks after final treatment the mean difference (95% CI) in changes in lesion counts on back sides treated with MAL-PDT and those treated with placebo cream plus red light was non significant for ILs 0.20 (-1.24, 1.64), P = 0.79, t = 0.280, as well as for NILs -0.45 (-2.95, 2.05), P = 0.73, t = -0.365. 17 participants completed the study, results reported for 20, ITT analysis performed. MD (95% CI) in changes in lesion counts on back sides treated with MAL-PDT and those treated with placebo cream plus red light was non significant for ILs 0.20 (-1.20, 1.60), P = 0.78, as well as for NILs -0.45 (-2.87, 1.97), P = 0.72 in the analyses using normal distribution.
Hörfelt 200630 (25 M, 5 F), 27 completed, aged 15-28 years (mean 18) with moderate-severe inflammatory facial acne (Leeds score 5–10); FPT types I–III635 nm light plus MAL vs placebo cream and light in a split-face trial, two treatments, 2 weeks apart, assessed at 4 and 10 weeks after treatmentMAL–PDT significantly more effective than light alone for IL: median percentage reduction 63% (95% CI 50% to 71%) versus 28% (95% CI 19% to 47%) at 4 weeks (P = 0.0004), and 54% (95% CI 35% to 64%) versus 20% (95% CI 8% to 50%) at 10 weeks (P = 0.0006). No statistically significant difference in treating NILs observed between two interventions (open comedones P = 0.6875, closed comedones P = 1.00). Intention-to-treat analysis (last observation carried forward method) results reported (n = 30). Study authors provided further data on changes and percentage changes in ILs (ITT population). Changes in means (SD) in ILs were 9.4 (7.4) at 4 weeks and 8.2 (7.4) at 10 weeks after final treatment in the MAL-PDT group and 6.8 (7.8) at 4 weeks and 5.7 (8.7) at 10 weeks respectively in the placebo cream plus light group. Percentage changes in means (SD) in ILs were 53.6% (29.1) at 4 weeks and 45.7 (34.5) at 10 weeks after final treatment in the MAL-PDT group and 29.7% (30.7) at 4 weeks and 26.6% (38.6%) at 10 weeks respectively in the placebo cream plus light group. We calculated that MAL-PDT was not superior to placebo cream plus light in change in ILs at 4 weeks nor at 12 weeks, with mean differences (95% CI) of -2.60 (-6.45, 1.25), P = 0.19 and -2.50 (-6.59, 1.59), P = 0.23 respectively. Howewer, it was superior in percentage change in ILs at 4 weeks and percentage change in ILs at 10 weeks, with mean differences (95% CI) of -23.90 (-39.04, -8.76), P = 0.002 and -19.10 (-37.63, -0.57), P = 0.04 respectively.
MAL-PDT versus yellow light alone
Haedersdal 200815 (5 M, 10 F) age 18-31 years (median 18), with at least 12 facial inflammatory acne lesions; FPT I–IIISplit-face design with non-purpuric LPDL 595 nm full-face treatment and MAL cream applied to randomised side of the face for 3 h before laser exposure, with dynamic cooling device; 3 treatments at 2-week intervals; assessed 4 and 12 weeks after final treatmentMedian percentage reduction in IL counts was significantly greater with MAL–LPDL than with LPDL at 4 weeks (70% versus 50%, P = 0.03) and 12 weeks (80% versus 67%, P = 0.004). Median percentage reduction in NILs lesions was significantly greater on the MAL–LPDL side at 4 weeks (P = 0.035) but difference between the treatments (53% versus 42%) did not achieve statistical significance at final follow-up (P = 0.158). Median IL counts (25% to 75% percentiles) at baseline, 4 and 12 weeks were 21.0 (16-36), 7 (4.75-15) and 3.5 (2-9.5) on the MAL-LPDL side, and 22 (14-36), 10 (6.5-16) and 7 (2-9.5) on the LPDL side respectively. Median NIL counts (25% to 75% percentiles) at baseline, 4 and 12 weeks were 33 (26-41), 23 (17-40) and 15 (9-21) on the MAL-LPDL side, and 32 (25-41), 26 (17-33) and 20 (12-27) on the LPDL side respectively
MAL-PDT versus placebo or no treatment
Wiegell 2006b36 participants: 21 in treatment group age 23 ± 5 years (9 M, 10 F analysed) and 15 in control group age 24 ± 5 years (3 M, 9 F analysed), with > 12 inflammatory acne lesions; FPT II–VComparison of MAL plus 630 nm with no treatment in a parallel-group trial; two treatments, 2 weeks apart, assessed every 4 weeks for 12 weeks after treatmentA significantly greater median reduction in ILs in the treatment group at 8 weeks (P = 0.023) and 12 weeks (P = 0.0023) at 12 weeks. Median ILs change from baseline (range) at 12 weeks was 24 (-4 to 55) in the MAL-PDT group and 0 (-39 to 19) in the control group. Median ILs count (range) at baseline, 4 , 8 and 12 weeks were 46 (13 to 99), 24 (9 to 68), 22 (8 to 83) and 14 (4 to 44) in the MAL-PDT group and 32 (13 to 99), 32 (8 to 128), 42 (9 to 109) and 40 (13 to 80) in the control group. Non significant difference in median change in NILs between the MAL-PDT and control group (P = 0.90) at 12 weeks. Median NILs change from baseline (range) at 12 weeks was 6 (-15 to 18) in the MAL-PDT group and 2 (-14 to 35) in the control group. Median NILs count (range) at baseline, 4, 8 and 12 weeks were 17 (2 to 73), 22 (0 to 56), 24 (6 to 59) and 24 (9 to 74) in the MAL-PDT group and 24 (2 to 64), 19 (0 to 76), 21 (2 to 81) and 31 (5 to 59) in the control group.
MAL-PDT other
Bissonnette 201044 participants, 33 completed (M/F not stated), aged 18-40 years (mean 24.4), 22 randomised to each group,10 ≥ ILs on each side of the face and a Global Acne Severity score 3 ≥, FPT I-IV80 mg/mL MAL plus 630 nm 25 J/cm² light vs 80 mg/mL MAL plus 630 nm 37 J/cm² light in a parallel-group trial, split-face randomisation within each group to occlusion or no occlusion, 4 treatments at 2-week intervals, assessed at 4 and 12 weeks after final treatmentILs means (95% CIs) changed from baseline 16.7 (11.8 to 21.5), 16.6 (12.6 to 20.5), 14.9 (12.3 to 17.1) and 15.7 (13.17 to 18.8) on the non-occluded 25 J/cm², occluded 25 J/cm², non-occluded 37 J/cm² and occluded 37 J/cm² face sides, respectively to 11.0 (8.7 to 13.4), 9.4 (6.3 to 12.4), 8.6 (5.2 to 11.9) and 8.9 (5.5 to 11.8) respectively at 12 weeks after final treatment. NILs means (95% CIs) changed from baseline 10.8 (7.0 to 14.6), 11.3 (7.9 to 14.7), 14.6 (7.8 to 21.4) and 15.1 (8.9 to 21.3) on the non-occluded 25 J/cm², occluded 25 J/cm², non-occluded 37 J/cm² and occluded 37 J/cm² face sides, respectively to 8.6 (5.7 to 11.5), 7.5 (4.9 to 10.1), 12.7 (5.8 to 19.6) and 12.2 (5.8 to 18.6) respectively at 12 weeks after final treatment. The number of ILs was significantly lower than baseline on all face sides but the non-occluded 25 J/cm² (based on non-overlapping 95% CI). No statistically significant difference in mean reduction of ILs between face sides with and without occlusion, for both 25 J/cm² and 37 J/cm². No statistically significant difference in NILs mean change from baseline between the treatments at 12 weeks follow-up. ITT analysis (LOCF method) results reported
Hong 201322 (2 M, 20 F), aged 19-35 years (mean not given), "at least grade 2 (Cunliffe acne grading system)", FPT IV-VMAL plus 630 nm light vs MAL plus 530-750 nm light in a split-face trial, 3 treatments in total, 2-week intervals, assessed at 4 weeks after treatmentAt 4 weeks after treatment, there was no statistically significant difference between red-light and IPL-treated sides in mean percentage reduction of ILs (69.5% versus 72.0% respectively) and NILs (43.4% versus 46.3% respectively). Further data not provided
NCT00594425150 participants (59 M/91 F), 50 in the 40 mg/g MAL-PDT group, 48 in the 80 mg/g MAL-PDT group, 52 in the placebo group, aged 15-40 years (mean 21.3), with moderate-severe acne, IGA score 3-4, 20-100 ILs and up to 200 NILs on the face, FPT I-IV80 mg/mL MAL under occlusion (1.5 h) plus 632 nm 37 J/cm² light vs 40 mg/mL MAL under occlusion (1.5 h) plus 632 nm 37 J/cm² light vs placebo cream plus 632 nm 37 J/cm² light in a parallel-group trial, 4 treatments at 2-week intervals, assessed at 2, 3, 6, 12 and 24 weeks after final treatment37 participants completed in the 80 mg/g group, and 43 completed in the 40 mg/g group, ITT analysis was performed. Our analyses showed that at 6 weeks after final treatment 80 mg/g MAL-PDT was not superior to 40 mg/g MAL-PDT in change in ILs (MD 2.20, 95% CI -2.57 to 6.97), P = 0.37, in percentage change in ILs (MD 3.10 95% CI -11.8 to 17.38), P = 0.67, in change in NILs (MD 0.6, CI 95% -6.36 to 7.56), P = 0.87, nor in percentage change in NILs (MD -1.7, 95% CI -20.67 to 17.27), P = 0.94
Yeung 200730 participants (8 M, 15 F ) aged 18-41 years (mean 25) with moderate facial acne; FPT IV–VAll participants used topical adapalene 0.1% gel at night and were randomised to 2 split-face treatment groups: 530–750 nm light with contact cooling gel plus MAL vs IPL only; and IPL with contact cooling gel vs topical adapalene-only control; 4 treatments with intervals of 3 weeks, assessed after each treatment and at 4 and 12 weeks post-treatmentOnly the control face side showed a statistically significant mean reduction (P =0.01) in IL counts. At 4 weeks and 12 weeks IL counts means (SE) were reported to be reduced by 52.7% (52.5) and 64.5% (54.8) on the MAL–PDT face sides; 22.1% (55.3) and 22.9% (52.2) on the light-only face sides; and 72.4% (19.9) and 88% (12.5) on the control face sides. A significant reduction in comedones on the MAL–PDT (P = 0.05) and light-only (P = 0.01) face sides at 12 weeks compared with the control face sides. At 4 weeks and 12 weeks NIL counts means (SE) reduced by 51.6 (26.1) and 38 (53.5) on the MAL–PDT face sides; 15.5 (42.3) and 43.6 (26.5) on the light-only face sides. 4 weeks after final treatment NIL counts means (SE) reduced by 13.8% (34) on the control face sides, but increased by 15.1% (SE) 12 weeks after final treatment. We performed analyses based on t-distribution and found that MAL-PDT was not superior to IPL alone in percentage change in ILs at both 4 weeks and at 12 weeks, with mean differences (95% CI) of -30.60 (-70.37, 9.17), P = 0.141, t = -1.567 and -41.60 (-81.90, -1.30), P = 0.052, t = -2.103 respectively. However, we found a transient superior effect on percentage change in NILs at 4 weeks, which was lost at 10 weeks, with mean differences (95% CI) of -36.10 (-60.18, -12.02), P = 0.006, t = -3.054 and 5.60 (-29.13, 40.33), P = 0.754, t = 0.328 respectively. We found no difference in effect between adapalene and MAL-PDT in percentage change in ILs at both 4 weeks and at 12 weeks, with mean differences (95% CI) of 19.70 (-15.32, 54.72), P = 0,283, t = 1.170 and 23.50 (-11.68, 58.68), P = 0.205, t = 1.390 respectively. However, MAL-PDT also had a transient superior effect to adapalene on percentage change in NILs at 4 weeks, which was lost at 10 weeks, with mean differences (95% CI) of -37.80 (-63.97, -11.63), P = 0.01, t = -3.005 and -53.10 (-119.64, 13.44), P = 0.133, t = -1.660 respectively. Results of our analyses based on normal distribution were not substantially different and also showed that MAL-PDT was not superior to IPL alone in percentage change in ILs at both 4 weeks and at 12 weeks, with mean differences (95% CI) of -30.60 (-68.86, 7.66), P = 0.133 and -41.60 (-80.38, -2.82), P = 0.0401 respectively. We also found a transient superior effect on percentage change in NILs at 4 weeks, which was lost at 10 weeks, with mean differences (95% CI) of -36.10 (-59.27, -12.93), P = 0.014 and 5.60 (-27.82, 39.02), P = 0.683 respectively. We also found no difference in effect between adapalene and MAL-PDT in percentage change in ILs at both 4 weeks and at 12 weeks, with mean differences (95% CI) of 19.70 (-13.30, 52.70), P = 0.240 and 23.50 (-9.65, 56.65), P = 0.162 respectively. We also found a transient superior effect of MAL-PDT as compared to adapalene in percentage change in NILs at 4 weeks, which was lost at 10 weeks, with mean differences (95% CI) of -37.80 (-62.46, -13.14), P = 0.007 and -53.10 (-115.80, 9.60), P = 0.120 respectively
5. ALA-PDT versus other comparators
ALA-PDT versus red light alone
Pollock 200410 (9 M, 1 F) age 16–40 years (mean 26.9) with mild-moderate acne of the back, Leeds grades 2-4; FPT I-VFour equal 30 cm² areas on the back: 635 nm light plus ALA vs light alone; ALA alone; untreated control; treated weekly for 3 weeks, assessed at each treatment and 3 weeks after final treatmentStatistically significant reduction from baseline in ILs counts from second treatment (P < 0.005) at the ALA–PDT site but not the other sites: reduction in acne was 69% at 21 days' follow-up. Further data reported in graph format, mean ILs at baseline 8.3 and 11.6 at light alone and ALA-PDT areas respectively decreased to 6.1 and 3.6 respectively at 3 weeks' follow-up. Other data not given
ALA-PDT versus blue light alone
NCT00706433266 (128 M, 138F), 68 in the ALA 1000 s group, 65 in the ALA 500 s group, 67 in the vehicle 1000 s group and 66 in the vehicle 500 s group, mean age 20.1 years, inclusion criterion 12 > years, with moderate and severe acne (IGA score 3 and 4, with at least 20 ILs); FPT I-V20% ALA (45 min incubation) plus blue 1000 s light vs 20% ALA (45 min incubation) plus 500 s blue light vs vehicle (45 min incubation) plus blue 1000 s light vs vehicle (45 min incubation) plus 500 s blue light; in a parallel-group trial; up to 4 treatments at 3-week intervals, assessed 3 and 6 weeks after the final treatmentAt 3 weeks after final treatment investigator-assessed median change in ILs (SD) was -18.0 (26.3) in ALA 1000 s, -14.0 (26.8) in the ALA 500 s, -19.0 (22.8) in the vehicle 1000 s and -14.5 (24.0) in the vehicle 500 s group; investigator-assessed median percentage change in ILs (SD) was -37.5 (38.79) in ALA 1000 s, -29.2 (46.68) in the ALA 500 s, -41.7 (38.82) in the vehicle 1000 s and -37.0 (40.23) in the vehicle 500 s group. At 6 weeks after final treatment investigator-assessed median change in ILs (SD) was -18.5 (30.15) in ALA 1000 s, -13.0 (28.74) in the ALA 500 s, -21.0 (23.63) in the vehicle 1000 s and -17.0 (26.71) in the vehicle 500 s group; investigator-assessed median percentage change in ILs (SD) was -34.4 (37.8) in ALA 1000 s, -29.0 (42.57) in the ALA 500 s, -48.4 (32.81) in the vehicle 1000 s and -45.2 (50.15) in the vehicle 500 s group. Statistical tests to determine whether any changes were significant could not be performed due to the study authors’ use of median changes rather than the typical mean changes required for significance testing in order to make appropriate comparisons with other included studies. Furthermore, it is not clearly stated whether the study authors implemented an ITT analysis or a LOCF approach to handling missing data.
ALA-PDT versus IPL alone
Oh 200920 (4 M, 16 F) , aged 18-30 years, 9 in the short incubation group (3 M, 6 F, mean age ± SD 23 ± 4.12 years) and 10 in the long incubation group (1 M, 9 F and 23 ± 5.53 years), with moderate and severe acne (Evaluator Global Severity Score 3 and 4); FPT II-IV20% ALA plus 590 nm IPL; 2 parallel groups: short incubation (30 min) vs long incubation (3 h), one half of the face within each treated with IPL alone; 3 treatments at 4-week intervals, assessed 4 weeks after each treatment and 8 and 12 weeks after the third treatmentMean reduction of ILs 84.4% in the long-incubation-time group, 72.6% in the short-incubation-time group and 65.9% on the face sides treated with IPL alone at 4 weeks (P < 0.001 in all cases). Mean reduction of ILs 89.5% in the long-incubation-time group, 83.0% in the short-incubation-time group and 74.0% on the face sides treated with IPL alone at 12 weeks (P < 0.001 in all cases). Mean reduction significantly greater in the long-incubation sides versus the IPL-alone sides (P = 0.01). The difference was not statistically significant between short-incubation and placebo-treated sides (P = 0.21). Further data not given
Mei 201341 (24 M, 17 F), mean age 24 years, 21 in the ALA-IPL PDT group, 20 in the placebo cream-IPL group, II–IV Pillsbury grade acne; FPT II-IV10% ALA plus 420–950 nm light versus placebo cream plus 420–950 nm light in a parallel-group trial, 4 treatments in total, weekly, assessed 4, 8 and 12 weeks after treatmentILs counts (% mean ± SE) reduced by 76.3 ± 3.7, 81.5 ± 4.6 and 83.6 ± 4.1 at 4, 8 and 12 weeks after final treatment respectively in the ALA-IPL group and by 64.9 ± 4.1, 68.3 ± 4.4 and 69.8 ± 4.6 respectively in the IPL-only group. Mean NILs counts (% mean ± SE) reduced by 44.9 ± 5.2, 49.9 ± 6.6 and 57.5 ± 6.8 at 4, 8 and 12 weeks after final treatment respectively in the ALA-IPL group and by 29.3 ± 5.6, 30.7 ± 6.7 and 30.7 ± 6.7 in the IPL only group respectively. Our analyses based on t-distribution showed that ALA-PDT was superior to light alone in percentage changes in ILs, with mean differences (95% CI) of 13.80 (1.34, 26.26), P = 0.04, t = 2.240 and in percentage changes in NILs, with MDs (95% CIs) of 24.10 (4.65, 43.55), P = 0.02, t = 2.506. Analyses based on normal distribution showed similar results; ALA-PDT was superior to light alone in percentage changes in ILs, with mean differences (95% CI) of 13.80 (1.72, 25.88), P = 0.03 and in percentage changes in NILs, with MDs (95% CIs) of 24.10 (5.25, 42.95), P = 0.01
Ragab 201425 (1 M, 24 F), age 14-39 years, 15 in the ALA-IPL group (mean 19.7) and 10 in the IPL alone group (mean age 19.0), "with mild-moderate facial acne" ; FPT III-V20% ALA plus 560? nm IPL versus 560 nm IPL alone; in a parallel-group trial; 2 treatments at 2-week intervals, assessed 2 and 8 weeks after final treatment

Mean ILs counts decreased from baseline 15.7 to 7.7 and 5.4 at 2 and 8 weeks respectively in the ALA-IPL group; and from baseline 9.6 to 5.2 and 4.4 at 2 and 8 weeks respectively in the IPL-alone group. Mean NILs (comedones) counts decreased from baseline 50.9 to 36.9 and 31.3 at 2 and 8 weeks respectively in the ALA-IPL group; and from baseline 41.8 to 23.8 and 24.4 at 2 and 8 weeks respectively in the IP- alone group. Mean combined lesion counts decreased from baseline 66.6 to 35.7 at 8 weeks in the ALA-IPL group; and from baseline 51.4 to 28.8 at 8 weeks in the IPL-alone group. SDs were not reported.

Mean percentage reductions from baseline at 8 weeks in ALA-IPL group compared with IPL alone were reported to be 73.4 versus 18.9% (P = 0.012) for ILs, 33.6 versus 29.8% (P = 0.739) for NILs (comedones) and 45.6 versus 27.8% (P = 0.202) for combined lesion counts respectively.

ALA-PDT versus no treatment
Orringer 201099 screened, 44 enrolled (14 M, 30 F), age range 15-50, mean age 25, all with clinically evident facial acne, all FPT included20% ALA plus PDL compared with no treatment in a split-face trial, 3 treatments at 2-week intervals, evaluated every 2 weeks for a total of 16 weeks.No statistically significant differences reported between treated and untreated control skin in papules (P = 0.62), pustules (P = 0.85), cysts (P = 0.49), closed (P = 0.21) and open comedones (P = 0.27) at week 16. Transient statistically significant decrease from baseline in mean papule counts on treated sides when compared with untreated sides (P = 0.01) at week 10. No statistically significant difference between treated and untreated control sides in all other lesion counts at week 10. At week 12 mean changes from baseline (95% CIs) in papules, pustules, cysts, closed and open comedones were -1.79 (-5.98 to 2.39), -2.72 (-6.65 to 1.20), 0.38 (-0.20 to 0.96), -6.97 (-13.30 to -0.63) and -4.79 (-11.62 to 2.04) on the treated sides respectively, and -0.97 (-4.32 to 2.39), -2.62 (-6.25 to 1.01), 0.24 (-0.33 to 0.82), -4.07 (-9.12 to 0.98) and -6.79 (-13.88 to 0.29) on the untreated sides respectively. Our analyses using LOCF data (n = 44) confirmed transient statistically significant decrease from baseline in investigator-assessed change in ILs (papules) on treated sides when compared with untreated sides at week 10 of the study (i.e. 4 weeks after final treatment), with MD -4.50 (95% CI -8.28 to -0.72), P = 0.02. We found no significant differences in means between treated and untreated face sides for investigator-assessed change in ILs (pustules) -0.60 (-5.09, 3.89), P = 0.79, for investigator-assessed change in NILs (open comedones) -0.37 (-7.76, 7.02), P = 0.92, for investigator-assessed change in NILs (closed comedones) -3.90 (-12.05, 4.25), P = 0.35, and for cysts 0.03 (-0.53, 0.59), P = 0.92. Our analyses also confirmed no significant differences in means between treated and untreated face sides at week 16 (i.e. 10 weeks after final treatment), MD (95% CIs) for investigator-assessed change in ILs (papules) was -0.82 (-6.03, 4.39), P = 0.76, for investigator-assessed change in ILs (pustules) -0.10 (-5.29, 5.09), P = 0.97, for investigator-assessed change in NILs (open comedones) 2.00 (-7.51, 11.51), P = 0.68, for investigator-assessed change in NILs (closed comedones) -2.90 (-10.78, 4.98), P = 0.47, and for cysts 0.14 (-0.66, 0.94), P = 0.73. Please note that we based all the calculations from the values provided in the table reported, and we double and triple checked the values using both RevMan and R statistical software, but some of our P values did not match up with the ones presented by the study authors.
Pollock 200410 (9 M, 1 F) age 16–40 years (mean 26.9) with mild-moderate acne of the back, Leeds grades 2-4; FPT I-VFour equal 30 cm² areas on the back: 635 nm light plus ALA vs light alone; ALA alone; untreated control; treated weekly for 3 weeks, assessed at each treatment and 3 weeks after final treatmentStatistically significant reduction from baseline in ILs counts from second treatment (P < 0.005) at the ALA–PDT site but not the other sites: reduction in acne was 69% at 21 days' follow-up. Further data reported in graph format, mean ILs at baseline 11.6 and 10.1 at ALA-PDT and no treatment control areas respectively decreased to 3.6 and 6.3 respectively at 3 weeks' follow-up. Other data not given
ALA-PDT other
Barolet 201010 (7M, 3F, aged 13-54, mean age 26.2), with mild-moderate acne, with ≥10 acne lesions, FPT I-III970 nm IR pre-treatment plus ALA and 630 nm PDT vs ALA-PDT alone, 1 treatment in a split-face or split-back design, evaluated after 4 weeksSignificantly greater improvement in IL medians on the IR pre-treated versus control side 4 weeks after treatment (P < 0.0001). Median percentage reduction (95% CI for mean?) in ILs was 73% (51% to 81%) on the IR pre-treated side versus 38% (8% to 55%) on the control side. Further data not provided, 95% CI reported for means, but means were not given
NCT00706433266 (128 M, 138 F), 68 in the ALA 1000 s group, 65 in the ALA 500 s group, 67 in the vehicle 1000 s group and 66 in the vehicle 500 s group, mean age 20.1 years, inclusion criterion 12 > years, with moderate and severe acne (IGA score 3 and 4, with at least 20 ILs); FPT I-V20% ALA (45 min incubation) plus blue 1000 s light vs 20% ALA (45 min incubation) plus 500 s blue light vs vehicle (45 min incubation) plus blue 1000 s light vs vehicle (45 min incubation) plus 500 s blue light; in a parallel-group trial; up to 4 treatments at 3-week intervals, assessed 3 and 6 weeks after the final treatmentAt 3 weeks after final treatment investigator-assessed median change in ILs (SD) was -18.0 (26.3) in ALA 1000 s, and -14.0 (26.8) in the ALA 500 s, -19.0 (22.8) group; investigator-assessed median percentage change in ILs (SD) was -37.5 (38.79) in ALA 1000 s and -29.2 (46.68) in the ALA 500 s group. At 6 weeks after final treatment investigator-assessed median change in ILs (SD) was -18.5 (30.15) in ALA 1000 s and -13.0 (28.74) in the ALA 500 s group; investigator-assessed median percentage change in ILs (SD) was -34.4 (37.8) in ALA 1000 s and -29.0 (42.57) in the ALA 500 s group. Statistical tests to determine whether any changes were significant could not be performed due to the study authors’ use of median changes rather than the typical mean changes required for significance testing in order to make appropriate comparisons with other included studies. Furthermore, it is not clearly stated whether the study authors implemented an ITT analysis or a LOCF approach to handling missing data.
Pollock 200410 (9 M, 1 F) age 16–40 years (mean 26.9) with mild-moderate acne of the back, Leeds grades 2-4; FPT I-VFour equal 30 cm² areas on the back: 635 nm light plus ALA vs light alone; ALA alone; untreated control; treated weekly for 3 weeks, assessed at each treatment and 3 weeks after final treatmentStatistically significant reduction from baseline in ILs counts from second treatment (P < 0.005) at the ALA–PDT site but not the other sites: reduction in acne was 69% at 21 days' follow up. Further data reported in graph format, mean ILs at baseline 6.6 and 11.6 at light-alone, ALA-alone, ALA-PDT and no-treatment control areas respectively decreased to 4.6 and 3.6 respectively at 3 weeks follow-up. Other data not given
Taub 200722 recruited, 19 participated, mean ± SD age 26.5 ± 9.1 years, 7 M, 12 F, with moderate-severe acne and > 10 inflammatory acne lesions; FPT not givenComparison of PDT with different light sources for activation: ALA activated by IPL (600–850 nm), or a combination of IPL (580–980 nm) and bipolar radiofrequency energies, or blue light (417 nm) in a parallel-group trial; 3 treatments at 2-week intervals; follow-up at 1 and 3 months after final treatmentReductions in counts in all 3 groups, highest in the IPL activation group and lowest in the blue-light group, but the difference was not statistically significant (P values not given). Median lesion count percentage reductions (96.9% CI) at 1 month after treatment were 76.8 (12.5 to 86.4) in the IPL group, 47 (8.3 to 82.2) in the IPL-RF group and 52.8 (-88.9 to 66.7) in the blue-light group. At 3 months after treatment, median lesion count percentage reduction (range, defined as "difference between the upper and lower ends of 96.9% CI, indicated when <5 data points are available") was 73.2 (72.4) in the IPL group, 41.6 (167.5%) in the IPL-RF group and -88.9 (123.3) in the blue-light group
Yin 2010180 ( 83 M, 97 F), aged 18-38 years, mean 25.8, with moderate-severe facial acne (Pillsbury), FPT III-IV633 ± 3 nm (red light) plus different ALA concentrations (5%, 10%, 15% and 20%) vs red light alone, 4 treatments every 10 days, 4 parallel groups, each treated with a different concentration on the right side and placebo agent on the left side; assessments at 2, 4, 12 and 24 weeks after last treatmentGreater reduction in both IL and NIL counts at sides treated by ALA-PDT of all concentrations compared with the controls treated by red light alone at 2 weeks (P < 0.001), 4 weeks (P < 0.05), 12 weeks (P < 0.001) and 24 weeks (P < 0.001). Combined data from all follow-up visits, the higher-concentration ALA treatment groups showed more improvement than the lower-concentration groups (P<0.01). Means (SD) reported in graph format only. Our interpretation of the graph was that ILs reduced from baseline 21 (5), 20.5 (5.5), 19 (5), 21 (5) and 20 (4) in the 20% ALA group, 15% ALA group, 10% ALA group, 5% ALA group and control face sides respectively to 1 (0.5), 1.3 (0.5), 3.3 (1), 4 (1) and 5 (1) in the 20% ALA group, 15% ALA group, 10% ALA group, 5% ALA group and control face sides respectively. NILs reduced from baseline 12.9 (4.5), 13 (3.5), 13 (4), 12.5 (3.5) and 11.5 (4) in the 20% ALA group, 15% ALA group, 10% ALA group, 5% ALA group and control face sides respectively to 1.4 (1), 1.4 (0.5),1.5 (0.5), 2.5 (0.5) and 5.5 (1.5) in the 20% ALA group, 15% ALA group, 10% ALA group, 5% ALA group and control face sides respectively at 24 weeks after final treatment. We judged further analyses would be biased due to lack of precise data, so we did not perform them.
6. MAL-PDT versus ALA-PDT
Wiegell 2006a15 participants > 18 years but age range not given, with > 12 inflammatory acne lesions; FPT not statedComparison of MAL and ALA creams: 620 nm light with split-face design; one full-face PDT treatment with MAL on one side and ALA on the other, assessed at 6 and 12 weeks after treatmentNo significant differences in reductions of ILs between ALA- and MAL-treated sides at 6 weeks' (P = 0.061) and 12 weeks' (P = 0.08) follow-up. Baseline differences in ILs counts (P = 0.0049). Median ILs counts (inter-quartile range) at baseline, 6 and 12 weeks after treatment were 19 (13 to 27), 8 (6 to 14) and 8 (3 to 11) on the MAL-treated sides and 16 (11 to 22), 5 (3 to 11) and 5 (3 to11) on the ALA-treated sides respectively. No significant differences in reductions of NIL between ALA- and MAL-treated sides at 6 weeks' (P = 0.18) and 12 weeks' (P = 0.052) follow-up. Median NILs counts (inter-quartile range) at baseline, 6 and 12 weeks after treatment were 14 (6 to 22), 21 (17 to 31) and 17 (9 to 29) on the MAL-treated sides and 17 (7 to 21), 18 (13 to 29) and 20 (17 to 38) on the ALA-treated sides respectively.
7. Other (non-MAL, non-ALA) PDT versus other comparators
Indocyanine green-PDT
Genina 200412 (5 M, 7 F) aged 17-27 years (mean age not given) with light-severe acne on the face or back; FPT not given803 nm low-intensity diode laser ± indocyanine green (ICG), single (8 participants) and multiple (4 participants) treatment groups, multiple treatment group had 2 treatments weekly for 4 weeks, assessed 1 week and 1 month after treatmentIL counts improved by 23% at 4 weeks for the single treatment groups and by 7% for control at ICG plus light sites; 80% improvement at 4 weeks for the multiple treatment group versus no improvement for control. More improvement was seen in participants with severe acne.
Kim 200916 (7 M, 9 F) aged 16-34 years, mean age 25 ± 3.09, with mild-moderate acne, skin types not given, 9 in single, 7 in multiple treatment group, FPT not given2 groups randomised: single treatment vs multiple (once-weekly over 3 weeks); right cheek of each patient ICG plus 805 nm light, left cheek light only and forehead "spontaneous resolution" control, evaluated 2 and 4 weeks after final treatment, multiple group also at final treatmentSignificant improvement only in mean number of closed comedones at PDL-treated side at all assessment periods, and at light-only side at 4 weeks post-treatment when compared to "spontaneous resolution" control (P < 0.05 in all cases). ILs improved at all sites, but non significantly (other data not given). Not reported whether there were differences between the two groups. Further data not given and part of the results reported in graph format. Our interpretation of the graph was that mean counts of closed comedones reduced from baseline 15 to 9 on the PDT sides and from 16 to 14 on the light-only sides respectively at final evaluation in the single treatment group, and from baseline 12 to 8 on the PDT sides and from 13 to 10 on the light-only sides in the multiple treatment group respectively.
Indole 3-acetic acid (IAA)-PDT
Na 201114 participants with inflammatory acne, sex, age, acne severity and FPT not given520 nm green light plus 0.015% IAA vs placebo cream plus green light, split-face trial, 3 treatments at 2-week intervals, assessed 0, 2, 4 and 6 weeks of treatmentImprovement in ILs count was observed on both sides. Difference between treatment and control group statistically significant from week 4 after final treatment (P < 0.05). Further data not given and reported in graph format. Our interpretation of the graph was that mean? ILs counts reduced from baseline 16.5 to 15.2 on the control sides, and from 16.3 to 14 on the treatment sides
Topical liposomal methylene blue (TLMB)-PDT
Fadel 200920 (M/F not stated), age not stated (> 18 years), with mild-moderate acne, FPT not givenTLMB plus 650 nm light vs no treatment in a split-face trial, 2 treatments in total, weekly, assessed every 2 weeks for 3 months after treatment.At 4 weeks IL counts decreased by 83.3% and NILs by 63.6% on the treated sides. Results for control sides not reported in narrative form. At 12 weeks reduction was also significant for ILs (P < 0.01) and NILs (P < 0.01). Further data not given
Chlorophyll-a (CHA)-PDT
Song 201424 (14 M, 10 F), mean age 23.4 ± 3.5 years; range 18-32 years, "acne on both sides of the face", Cunliffe grades 2-4, FPT III-IV430 plus 660 nm light combined with CHA vs 430 plus 660 nm light alone in a split-face trial, 8 treatments in total, twice weekly, final assessment 2 weeks after last treatment2 weeks after final treatment papule counts reduced from baseline 13.0 to 5.1 on the CHA plus light sides and from baseline 13.1 to 8.6 on the light-only sides (P = 0.030, SDs not given); pustule counts reduced from baseline 3.8 to 1.3 on the CHA-plus-light sides and from baseline 4.2 to 3.0 on the light-only sides (P < 0.001, precise P value not given, SDs not given); open comedone counts reduced from baseline 9.0 to 4.2 on the CHA-plus-light sides and from baseline 9.1 to 6.7 on the light-only sides (P = 0.011, SDs not given); closed comedones counts reduced from baseline 18.4 to 8.5 on the CHA-plus-light sides and from baseline 18.4 to 13.3 on the light-only sides (P = 0.014, SDs not given); nodules & cyst counts reduced from baseline 0.6 to 0.1 on the CHA-plus-light sides and from baseline 0.55 to 0.3 on the light-only sides (P value not given, data extracted from figure). Further data were not given
Gold microparticle PDT versus other comparators
Paithankar 201551 (14 M, 37 F), mean age 21.4 years, age range 16-26 years, IGA scores 3–4 with at least 25 total papules and pustules on face, FPT I-IIIGold microparticle suspension plus light (details not given) vs microparticle suspension vehicle (without light-absorbing particles) plus light (details not given) in a parallel-group trial, 3 treatments in total, weekly, assessed at 6, 10 and 14 weeks after final treatmentAt 6 weeks after final treatment, the mean percentage change in inflammatory lesion count was −44.0% and −14.0% for the active treatment and sham arms, respectively. At 10 weeks after final treatment, the mean percentage change in inflammatory lesion count was −49.0% and −21.7% for the active treatment and sham arms, respectively (P = 0.015). At 14 weeks after final treatment changes were −53% and −30% for the active treatment and sham arms, respectively (P = 0.04). Other data were not given
Table 4. Adverse effects
  1. 1We reported the adverse events as provided by the study authors or sponsors (we did not perform coding ourselves).

    2'Investigator-assessed severe adverse effects' are presented in bolded text.

    ALA = 5-aminolevulinic acid
    BPO = benzoyl peroxide
    FPT = Fitzpatrick's Skin Types: based on different reactions to sun exposure and range from type I ('pale white skin which always burns and never tans') to type VI ('deeply pigmented dark brown to black skin which never burns and tans very easily') (Fitzpatrick 1988)
    GAAS = Global Acne Assessment Scoring
    ILs = inflamed lesions
    IPL = intense pulsed light
    IR = Infrared
    ITT = Intention-to-treat analysis
    LLT = Lower Level Term
    MAL = methyl-aminolevulinate
    NILs = non-inflamed lesions
    OFI = optical fibre intra-tissue irradiation
    PDL = pulsed-dye laser
    PDT = photodynamic therapy
    PT = Preferred term
    RCT = randomised controlled trial
    SD = standard deviation
    SOC* = System Organ Class
    SPF = sun protection factor

    *MedDRA®, the Medical Dictionary for Regulatory Activities, terminology is the international medical terminology developed under the auspices of the International Conference on Harmonization of Technical Requirements for Registration of Pharmaceuticals for Human Use (ICH). MedDRA® trademark is owned by the International Federation of Pharmaceutical Manufacturers & Associations (IFPMA) on behalf of ICH.

Study SOC skin and subcutaneous tissue disorders SOC general disorders and application site conditions Other SOCs
1. Light versus placebo or no treatment
Green light versus placebo
Baugh 2005None reportedNone reportedNone reported
Yilmaz 2011None reportedNone reportedNone reported
Yellow light versus placebo or no treatment
Seaton 2003In the yellow-light group: 2/31 (6.4%) pain of skin, 1/31 (3.2%) purpura, 1/31 (3.2%) pruritus, 2/31 (6.4%) dry skin. In the placebo group: 1/10 (10%) pruritus, 2/10 (20%) dry skinNone reportedIn the yellow-light group: SOC Eye disorders: 1/31 (3.2%) lacrimation increased
Infrared light (IR) versus no treatment
Darne 2011None reportedApplication site erythema in "most" of 38 participantsNone reported
Moneib 20142/24 (83%) dry skin (in the nasolabial fold), 2/24 (83%) pustular rash (LLT pustular skin eruption). Unclear whether treated or untreated face sidesApplication site erythema, and "decreased oiliness" in "all" of 24 participants. Unclear whether treated or untreated face sidesNone reported
Orringer 2007IR sides: 2/46 (4.4%) post-inflammatory pigmentation change (LLT post-inflammatory hyperpigmentation)IR sides: application site discomfort: 34/46 (74%) moderate, 12/46 (24%) substantial causing 2 withdrawals; 2/46 (4.4%) application site vesicle (LLT application site blister)2IR sides: SOC Psychiatric disorders: 1/46 (2.2%) panic attack, caused one withdrawal
Blue light versus placebo or no treatment
Elman 2003None reportedNone reportedNone reported
Red light versus no treatment
Na 2007None reportedNone reportedRed-light sides: SOC Nervous systems 1/30 (3.3%) burning sensation
Blue-red light versus placebo
Papageorgieu 2000In the blue-red light group: 2/30 (6.6%) acne (LLT acne exacerbation), 1/30 (3.3%) dry skin and pruritus, 2/30 (6.6%) rash (LLT facial rash). In the placebo group: 2/25 (8%) acne (LLT acne exacerbation), 2/25 (8%) dry skin and pruritusNone reportedIn the blue-red light group: SOC Nervous system disorders: 1/30 (3.3%) headache
Kwon 20132/18 (11%) dry skin 1/18 (6%) erythema and skin exfoliation (unclear in which group)None reportedNone reported
Broad spectrum light versus placebo
Sadick 2010bNone reportedNone reportedNone reported
IPL versus no treatment 
McGill 20081/10 (10%) acne (LLT acne exacerbation), on both IPL and control sides, reported as "A further patient experienced an acne flare-up following the first treatment. However, this was bilateral and so was felt to be unrelated to the IPL treatment."IPL sides: 1/10 (10%) application site vesicle (LLT application site blister); "One patient developed minor blistering after the fifth treatment, which resolved without scarring. This occurred in areas where double passing treatment was carried out, and were most likely due to the second pass taking place too quickly after the first."None reported
2. Light versus topical treatment
Light versus benzoyl peroxide (BPO)
de Arruda 2009In the BPO group: 28/30 (93.3%) "some level of erythema, desquamation, dryness or burning". In the blue-light group: 7/30 (23.3%) skin exfoliation and dry skin, reported as "all of mild intensity".None reportedNone reported
Chang 2007Post-inflammatory pigmentation change (LLT post-inflammatory hyperpigmentation) 3/30 (10%) (unclear on which face-sides)None reportedNone reported
Papageorgieu 2000In the blue-red light group: 2/30 (6.6%) acne (LLT acne exacerbation), 1/30 (3.3%) dry skin and pruritus, 2/30 (6.6%) rash (LLT facial rash). In the BPO group: 2/25 (8%) acne (LLT acne exacerbation), 8/25 (32%) dry skin and pruritus, 2/25 (8%) rash (LLT facial rash)None reportedIn the blue-red light group: SOC Nervous system disorders: 1/30 (3.3%) headache
Light versus clindamycin
Gold 2005None reportedNone reportedNone reported
Lee 2010No "significant" adverse effectsNo "significant" adverse effectsNo "significant" adverse effects
Light and other topical treatments
Ash 2015None reportedNone reportedNone reported
Ianosi 201311/180 (6%) scab, reported as "Eleven patients with dark III and IV phototypes presented with hematic crusts", unclear in which group, 34/60 (57%) in the vacum-IPL group and 4/60 (7%) reported as "sebum secretion increase"Application site erythema in light treatment groups, reported as "persistent erythema during a period of 24 h was noted in almost all patients", lasting for 72 h in 10/60 (17%) in the vacum-IPL group and 3/60 (5%) in the IPL-only group. 12/60 (20%) application site ecchymosis in the vacum-IPL groupNone reported
Karsai 2010In the C/BPO plus laser group: 1/51 (2%) purpura; reported as "one case of mild purpura lasting 3 days (incidence 2%)"None reportedNone reported
Zhang 2009aNone reported

In the blue and red light in combination with antibiotics group:

120/508 (23.6%) application site discomfort (reported as "participants from the blue and red light in combination with antibiotics found the red light ‘too intense’"; exact effects not specified); 2/508 (0,4%) withdrew due to application site discomfort

None reported
3. Light versus other comparators
Comparison of light therapies of different wavelengths
Choi 2010None reportedMild application site oedema, mild application site erythema (unclear on which face sides, exact numbers not given)None reported
Jung 2009Pain of skin, dry skin, skin exfoliation; reported as "one patient reported mild dryness that disappeared after a few days. All participants tolerated pain well" (unclear whether PDL or combined 585/1,064 nm laser-treated sides)Application site erythema, reported as "All patients reported mild erythema" (unclear whether PDL or combined 585/1,064 nm laser-treated sides).None reported
Liu 2011Dry skin. Reported as "A few patients stated certain dryness of skin after exposure to light sources for 20-min session… The result (shown in Fig. 5) demonstrated that there was no obvious change in skin color." Further details not givenNone reportedNone reported
Liu 2014None reportedIn the IPL group: 28/50 (56%) application site erythema. In the LED group: 3/50 (6%) application site erythemaSOC Nervous system disorders: 28/50 (56%) in the IPL group and 3/50 (6%) in the LED group paraesthesia (reported as "slight stinging sensations immediately after the procedures that disappeared within approximately 2 h")
Papageorgieu 2000In the blue-red light group: 2/30 (6.6%) acne (LLT acne exacerbation), 1/30 (3.3%) dry skin and pruritus, 2/30 (6.6%) rash (LLT facial rash). In the blue light-only group: 3/27 (11.1%) acne (LLT acne exacerbation), 3/27 (11.1%) dry skin and pruritus, 1/27 (3.7%) rash (LLT facial rash)None reportedIn the blue-red light group: SOC Nervous system disorders: 1/30 (3.3%) headache
Sami 2008In the PDL group: 3/15 (20%) post-inflammatory pigmentation change (LLLT post-inflammatory hyperpigmentation), "mild purpura" (participants' numbers unclear)In the PDL group: application site discomfort (participants' numbers unclear). In the IPL group: application site erythema (participants' numbers unclear). In the LED group application site warmth (participants' numbers unclear).In the IPL group SOC Nervous system disorders: paraesthesia (participants' numbers unclear)
Comparison of light therapies of different doses
Bernstein 2007None reportedApplication site oedema 12/30 (41.6%) on the single-pass side and 7/30 (23%) on the double-pass side; application site erythema 30/30 (100%) on both sidesNone reported
Jih 2006None reportedApplication site oedema, application site erythema, pain of skin ("most common side effects", numbers not given)None reported
Uebelhoer 2007None reported11/11 (100%) application site oedema and 11/11 (100%) application site erythema on both sides. On the single-pass side: 1/11 (9%) application site discolouration (LLT Application site hyperpigmentation).1/11 (9%) application site vesicle (LLT application site blister) reported as "We also experienced a cryogen failure that resulted in a single blister that resolved completely with proper wound care"None reported
Light alone versus combined with microdermoabrasion
Wang 2006Pain of skin (numbers reported unclearly), 1/20 (5%) post-inflammatory pigmentation change (post inflammatory hypopigmentation) on the laser plus microdermoabrasion sideApplication site oedema, application site erythema, application site papules (numbers reported unclearly)None reported
Light in combination with carbon lotion versus no treatment
Jung 2012Mild pain of skin (numbers reported unclearly), 15/22 (75%) dry skin, skin exfoliation reported as "mild dryness and desquamation of the treated side"22/22 (100%) application site erythemaNone reported
Light in combination with oral therapy versus other comparators
Ou 2014In the Yinhua decoction with electric light synergy group: 1/43 who completed (2.3%) dry skin; in the Yinhua decoction in combination with red and blue light treatment group 7/40 who completed (17.5%) dry skin and pruritus. Number of participants randomised to each group unclearNone reported

In the Yinhua decoction in combination with red and blue light treatment group: SOC Gastrointestinal disorders 1/40 who completed (2.5%) LLT

Diarrhoea (reported as: "after having yinhua concoction – side effects subsided after participant changed to having yinhua concoction after meals"). None reported in the intervention group

Zhang 2013bNone reportedIn the red-blue combined with jinhua xiaocuo pills and chloramphenicol tincture group 2/60 (3.3%) subjects reported "mild facial erythema, itching and scaling". No adverse effects were reported in the jinhua xiaocuo pills and chloramphenicol tincture alone groupNone reported
4. MAL-PDT versus other comparators
MAL-PDT versus red light alone
NCT00594425 1None reportedIn the 40 mg/g MAL plus red light group: 3/50 (6%) application site discolouration, 1/50 (2%) application site dryness, 40/50 (80%) application site erythema, 3/50 (6%) application site exfoliation, 31/50 (62%) application site irritation, 32/50 (64%) application site pain, 3/50 (6%) application site paraesthesia, 13/50 (26%) application site pruritus, 3/50 (6%) application site scab, 2/50 (4%) application site warmth. In the 80 mg/g MAL plus red light group: 7/48 (16%) application site discolouration, 3/48 (6%) application site dryness, 35/48 (73%) application site erythema, 3/48 (6%) application site exfoliation, 26/48 (54%) application site irritation, 31/48 (65%) application site pain, 9/48 (19%) application site paraesthesia, 10/48 (21%) application site pruritus, 3/48 (6%) application site scab, 0/48 (0%) application site warmth. In the placebo cream plus red light group: 3/52 (6%) application site discolouration, 0/52 (0%) application site dryness, 17/52 (33%) application site erythema, 0/52 (0%) application site exfoliation, 4/52 (8%) application site irritation, 5/52 application site pain, 4/52 (8%) application site paraesthesia, 5/52 (10%) application site pruritus, 0/52 (0%) application site scab, 4/52 (8%) application site warmth. We only included treatment-related adverse effects in this table. Frequency threshold above which adverse effects were reported was 2%. Sponsors confirmed that there were no reports of application site blisters.In the 40 mg/g MAL plus red light group: SOC Nervous system disorders 2/50 (4%) headache. In the 80 mg/g MAL plus red light group: SOC Nervous system disorders: 2/48 (4%) headache. In the placebo cream plus red light group: SOC Nervous system disorders 0/52 (0%) headache
NCT00933543 1In the 80 mg/g MAL plus red light group: 26/54 (48%) erythema, 14/54 (26%) pruritus, 12/54 (23%) skin burning sensation, 4/54 (7.4%) skin irritation. In the placebo cream plus red light group: 9/53 (17%) erythema, 8/53 (16%) pruritus, 4/53 (8%) skin burning sensation, 0/54 (0%) skin irritationIn 80 mg/g MAL plus red light group: 4/54 (7%) facial pain, 2/54 (4%) "feeling hot", 27/54 (50%) pain. In placebo cream plus red light group: 4/53 (8%) facial pain, 2/53 (4%) 'feeling hot', 6/53 (11%) pain. Sponsors confirmed that there were no reports of application site blistersIn 80 mg/g MAL plus red light group: SOC Infections and infestations: 10/54 (19%) nasopharyngitis, SOC Nervous disorders: 9/54 (17%) paraesthesia. In placebo cream plus red light group: SOC Infections and Infestations: 5/53 (9%) nasopharyngitis, SOC Nervous disorders: 1/53 (2%) paraesthesia
Pariser 2013 1In the 80 mg/g MAL plus red light group: 17/100 (17%) pain of skin, 15/100 (15%) skin burning sensation, 8/100 (8%) pruritus, 4/100 (4%) erythema, 2/100 (2%) rash, 2/100 (2%) scab, 2/100 (2%) skin hyperpigmentation. In the placebo cream plus red light group: 0/53 (0%) pain of skin, 0/53 (0%) skin burning sensation, 1/53 (2%) pruritus, 0/53 (0%) erythema, 1/53 (2%) rash, 0/53 (0%) scab, 0/53 (0%) skin hyperpigmentation. Data provided by sponsors for adverse events experienced by more than one participant in each treatment groupIn the 80 mg/g MAL plus red light group: 1/100 (1%) application site blister. In the placebo cream plus red light group: None reportedIn 80 mg/g MAL plus red light group: SOC Musculoskeletal and connective tissue disorders: 2/100 (2%) back pain. In placebo cream plus red light group: None reported. We only included treatment-related adverse effects in this table
NCT00673933 1On the MAL-PDT area: 4/20 (20%) erythema, 1/20 (5%) pain of skin, 5/20 (25%) pruritus, 4/20 (20%) skin burning sensation, 7/20 (35%) skin warmth. On the placebo cream plus red light area: 0/20 (0%) erythema, 0/20 (0%) pain of skin, 1/20 (5%) pruritus, 4/20 (20%) skin burning sensation, 5/20 (25%) skin warmth. Frequency threshold above which adverse effects were reported was 5%None reported. Sponsors confirmed that there were no reports of application site blistersOn the MAL-PDT area: SOC Nervous system disorders: 1/20 (5%) paraesthesia. SOC Vascular disorders: 1/20 (5%) hematoma. On the placebo cream plus red light area: SOC Nervous system disorders: 3/20 (15%) paraesthesia. SOC Vascular disorders: 1/20 (5%) haematoma
Hörfelt 20069/30 (30%) pain of skin (unclear on which face side)8/30 (27%) application site erythema, 5/30 (17%) application site oedema (unclear on which face side); 1/30 application site blister (3%) on the MAL-PDT sideNone reported
MAL-PDT versus yellow light alone
Haedersdal 2008On the MAL-LPDL side 15/15 (100%) pain of skin, 1/15 (6.6%) MAL-LPDL scab (LLT crust). On the LPDL only side: 4/15 (26.6%) pain of skin15/15 (100%) MAL-LPDL, 12/15 (80%) LPDL application site erythema; 15/15 (100%) MAL-LPDL, 4/15 (26.6%) LPDL application site oedema; 12/15 (80%) MAL-LPDL, 5/15 (33.3%) LPDL application site pustules/rash pustular (LLT pustular skin eruption)None reported
(4c) MAL-PDT versus placebo or no treatment
Wiegell 2006bPain of skin, pustular rash (LLT pustular skin eruption) ("in almost all patients", number of participants unclear); scab (LLT crust) "a third of patients", skin exfoliation ("in some patients"; numbers and groups unclearApplication site erythema, application site oedema (number of participants unclear)SOC Social conditions: "Approximately half of the patients did not go to school or work for between 1 day and 1 week after treatment due to their appearance" (we were unable to find an appropriate MedDRA PT)
MAL-PDT other
Bissonnette 2010 1None reportedPlease note that total unit of analyses numbers were reported as n = 154 in the 25 J/cm² group, and n = 169 in the 37 J/cm² group. 22 participants randomised to each group initially (2 withdrew after an adverse event from the 25 J/cm² group (first because of a pustular eruption on the face following MAL-PDT; second due to pain during light exposure). We were unable to obtain further explanations by the study authors. On the non-occluded 25 J/cm² side: 15/154 (9.7%) erythema, 39/154 pain (25.3%), 2/154 pruritus (1.3%), 1/154 (0.6%) scab, 1/154 (0.6%) pustular eruption and 3/154 (2%) paraesthesia. On the occluded 25 J/cm² face sides: 1/154 (0.6%) dryness, 5/154 (3.2%) erythema (3.2%), 60/154 pain, 2/154 pruritus (1.3%), 1/154 (0.6%) scab, 1/154 (0.6%) pustular eruption, 4/154 (2.6%) paraesthesia and 1/154 (0.6%) desquamation. On the non-occluded 37 J/cm² group there were 14/169 (8.2%) reports of erythema, 48/169 (28.4%) pain and 6/169 (3.5%) paraesthesia. On the occluded 37 J/cm² face sides 7/169 (4.1%) reports of erythema and 59/169 (35%) pain, 6/169 (3.5%) paraesthesia and 1/169 blister (0.6%). "Other adverse events" 18/154 (11.7%) in the 25 J/cm² group, 28/169 (16.6%) in the 37 Jcm² group. Regarding application site blisters, sponsors provided information that there was "1 report from 44 Visonac treated patients" 1/44 (2%)None reported
Hong 2013On the MAL plus 630 nm light side: 1/22 (4.5%) erythema, On the MAL plus 530-750 nm light side: 1/22 (4.5%) post-inflammatory pigmentation change (LLT post inflammatory hyperpigmentation)22/22 (100%) application site pain (both sides)None reported
Yeung 20071/11 (9%) scab and skin hyperpigmentation in the MAL PDT group, 2/12 (16%) scab and skin hyperpigmentation in IPL-only group; dermatitis acneiform (LLT rash acneiform) "in some patients", details not provided. Unclear reportingApplication site stinging, application site oedema, and application site erythema, caused 4? withdrawals ('MAL-PDT side', details not given)SOC Nervous system disorders: 4? withdrawals because of paraesthesia (LLTs skin burning sensation). Unclear reporting
5. ALA-PDT versus other comparators
ALA-PDT versus red light alone
Chen 2015None reportedIn the ALA-PDT group: 7/25 (28%) combination of application site erythema, application site oedema, application site pain and application site paraesthesia ("burning sensations"); 3/25 (12%) application site discolouration (LLT application site hyperpigmentation); 2/25 (8%) application site pustules/rash pustular (LLT pustular skin eruption; reported as "Two patients developed small pimples and were diagnosed with acute acneform lesions, which were topically treated successfully with mupirocin ointment". In the control group 2/25 (8%) application site erythema and application site dryness (reported as "two patients experienced flushing and dryness of the face")None reported
Pollock 2004On the ALA-PDT site: 10/10 (100%) urticaria (LLT erythema urticarial, reported as "urticated erythema"), 10/10 (100%) post-inflammatory pigmentation change (post-inflammatory hyperpigmentation), resolved within 1 month in 9/10 (90%), and within 3 months in 1/10 (10%), FPT VOn the ALA-PDT site: 1/10 (10%) application site discomfortOn the ALA-PDT site: SOC Nervous system disorders: 10/10 (100%) paraesthesia (LLTs skin tingling, tingling sensation, skin burning sensation), 4/10 (40%) perifollicular eruption (we were unable to find an appropriate MedDRA PT)
Zhang 2013aNone reportedIn the ALA-PDT group: unclear reporting, 6/63 (9.5%) (LLT application site blister)?, together with varying degrees of application site oedema, application site erythema, application site burning sensation?. Reported as:"Three days after treatment, 6 participants experienced varying degrees of erythema, burning heat sensation, swelling, water blisters. These side effects disappeared after cold compress within 5 to 10 days in 5 of these participants. The side effects in the other participant disappeared 3 days after taking metacortandracin and undergoing cold compress." Not reported for the red-light-only groupNone reported
ALA-PDT versus blue light alone 
NCT00706433 1None reportedReported as "Injury, poisoning and procedural complications". In the ALA 1000 s group: "Stinging/Burning" 17/68 (25.00%), "Dry skin" 7/68 (10.29%), "Erythema" 13/68 (19.12%), "Itching of face" 14/68 (20.59%), "Scabbing" 4/68 (5.88%), "Peeling of skin" 6/68 (8.82%), "Tightness of skin" 4/68 (5.88%), "Facial pain" 4/68 (5.88%) In the ALA 500 s group: "Stinging/Burning" 17/65 (26.15%), "Dry skin" 4/65 (6.15%), ‘Erythema’ 5/65 (7.69%), "Itching of face" 20/65 (30.77%), "Peeling of skin" 5/65 (7.69%), "Tightness of skin’ 6/65 (9.23%), ‘Facial pain’ 6/65 (9.23%). In the vehicle 1000 s group: ‘Stinging/Burning’ 6/67 (8.96%), ‘Dry skin’3/67 (4.48%), ‘Erythema’ 1/67 (1.49%), ‘Itching of face’ 6/67 (8.96%), ‘Peeling of skin" 1/67 (1.49%), "Tightness of skin" 2/67 (2.99%). In the vehicle 500 s group:"Stinging/Burning" 5/66 (7.58%), "Dry skin" 4/66 (6.06%), "Erythema" 1/66 (1.52%), "Itching of face" 6/66 (9.09%), "Peeling of skin" 2/66 (3.03%), "Tightness of skin" 1/66 (1.52%)’In the ALA 1000 s group: Gastrointestinal disorders:1/68 (1.47%) vomiting; Infections and infestations: 5/68 (7.35%) Upper respiratory tract infection; Nervous system disorders: Headache 5/68 (7.35%); Respiratory, thoracic and mediastinal disorders: Nasopharyngitis 6/68 (8.82%). In the ALA 500 s group Gastrointestinal disorders: 1/65 (1.54%) vomiting; Infections and infestations: 2/65 (3.08%) Upper respiratory tract infection; Nervous system disorders: Headache 4/65 (6.15%); Respiratory, thoracic and mediastinal disorders: Nasopharyngitis 7/65 (10.77%). In the vehicle 1000 s group: Gastrointestinal disorders 1/67 (1.49%) vomiting; Nervous system disorders: Headache 3/67 (4.48%); Respiratory, thoracic and mediastinal disorders: Nasopharyngitis 4/67 (5.97%), Sinus congestion 1/67 (1.49%) In the vehicle 500 s group: 3/66 (4.55%) vomiting; Infections and infestations: 5/66 (7.58%) Upper respiratory tract infection; Nervous system disorders: Headache 2/66 (3.03%); Respiratory, thoracic and mediastinal disorders: Nasopharyngitis 4/66 (6.06%), Sinus congestion 4/66 (6.06%)
ALA-PDT versus blue-red light alone
Liu 2014In the ALA-PDT group: 2/50 (4%) post-inflammatory pigmentation change (LLT post-inflammatory hyperpigmentation); 10/50 (20%) "brightening of skin tone and improvements of skin texture after treatment" (we were unable to find an appropriate MedDRA PT)In the ALA-PDT group 46/50 (92%) combination of application site pain, application site erythema and application site oedema. In the LED group: 3/50 (6%) application site erythemaSOC Nervous system disorders: 3/50 (6%) in the LED group paraesthesia (reported as "slight stinging sensations immediately after the procedures that disappeared within approximately 2 h")
ALA-PDT versus IPL alone
Oh 2009Short incubation ALA-PDT side: 1/9 (11.1%) dermatitis acneiform (LLT acneiform eruption or rash acneiform)Short incubation ALA-PDT side: 1/9 (11.1%) application site discolouration (LLT application site hyperpigmentation). Application site erythema and application site oedema (unclear reporting)None reported
Liu 2014In the ALA-PDT group: 2/50 (4%) post-inflammatory pigmentation change (LLT post-inflammatory hyperpigmentation); 10/50 (20%) "brightening of skin tone and improvements of skin texture after treatment" (we were unable to find an appropriate MedDRA PT)In the IPL group: 28/50 (56%) application site erythema.SOC Nervous system disorders: 28/50 (56%) in the IPL group paraesthesia (reported as "slight stinging sensations immediately after the procedures that disappeared within approximately 2 h")
Mei 2013In the ALA-IPL group: 3/21 (14%) erythema, 3/21 (14%) dermatitis acneiform (LLT acneiform eruption)In the ALA-IPL group: 21/21 (100%) application site painNone reported
Ragab 2014In the ALA-IPL group: 15/15 (100%) pain of skin, of which 4/15 mild, 8/15 moderate and 3/15 severe. In the IPL-alone group: 10/10 (100%) pain of skin, of which 8/10 mild and 2/10 moderateIn the ALA-IPL group: 4/15 (27%) application site discolouration (LLT application site hyperpigmentation); 10/15 (67%) application site exfoliation, of which 5/15 mild and 5/15 moderate; 14/15 (93%) application site erythema, of which 6/15 mild, 6/15 moderate and 2/15 severe. In the IPL-alone group: 1/10 (10%) application site discolouration (LLT application site hyperpigmentation); 2/10 (20%) application site exfoliation, of which 2/2 moderate; 8/10 (80%) application site erythema, of which 7/10 mild and 1/10 moderateNone reported
ALA-PDT versus green light alone
Sadick 2010aNone reportedNone reportedNone reported
ALA-PDT versus placebo or no treatment
Orringer 2010On the ALA-PDT side: 2/44 participants (4.5%) skin desquamation, 2/44 (4.5%) post inflammatory pigmentation change (LLT post inflammatory hyperpigmentation). Both of these participants withdrew from trial1/44 patient (2.3%) application site vesicle (LLT application site blister). Resolved without permanent consequencesNone reported
ALA-PDT other
Barolet 2010Scab (exact numbers of participants not given). 2/10 (20%) participants had acneiform folliculitis (we were unable to find an adequate term in MedDRA). Not clear whether this refers to IR-LED treatment or PDTApplication site erythema. Not clear whether this refers to IR-LED treatment or PDTSOC Nervous system disorders: paraesthesia (LLTs skin burning sensation). Not clear whether this refers to IR-LED treatment or PDT
Hongcharu 2000Pain of skin, pruritus, skin burning sensation, post-inflammatory pigmentation change (post-inflammatory hyperpigmentation), lasting more than 20 weeks in 55% of multiple treatment group participants; numbers and group those participants were assigned to reported unclearly. Acne (LLT exacerbation of acne; reported as "acute eruption of inflammatory acneiform lesions") in all participantsApplication site erythema, application site oedema, numbers reported unclearly. 1/11? (9%) application site vesicle (LLT application site blister) reported as: "one subject in the single PDT group developed severe blistering in the PDT site after vigorous aerobic exercise while wearing a tight outfit the day after treatment".Transient purpura in 10% of multiple treatment participants (following "superficial but very prominent exfoliation"). Numbers reported unclearly
NCT00706433 1SeeALA-PDT versus blue light alone aboveSee ALA-PDT versus blue light alone above.See ALA-PDT versus blue light alone above
Taub 2007In the IPL group: 1 severe erythema and skin exfoliation, 1 alopecia; in the IPL-RF group: 1 severe erythema and skin exfoliation, 1 acne (LLT exacerbation of acne) 1 contusion (LLT bruise). In ALA-PDT plus blue light 1 acne (LLT exacerbation of acne). Numbers of participants per group were not statedIn the IPL-RF group: 1 application site vesicle (LLT application site blister). Numbers of participants per group were not statedNone reported
Yin 2010In 5%, 10%, 15% and 20% ALA groups 1/45 (2%), 2/45 (4%), 2/45 (4%), 5/45 (11%) respectively combination of mild dry skin and skin exfoliation. In the 20% ALA-PDT group 5/45 (%) marked dry skin and skin exfoliationIn the 20% ALA group: 30/45 (67%) application site discomfort, 3/45 (7%) severe application site oedema and application site erythema, scab (exact numbers of participants not given), 1/45 (2%) combination of application site erythema, application site oedema and application site vesicle (LLT application site blister);"treated with systemic glucocorticoids and resolution took place in 2 weeks, with no persistent clinical sequelae or permanent scarring". 2/45 (4%), 5/45 (11%), 7/45 (16%), 10/45 (22%) application site discolouration (LLT application site hyperpigmentation) in 5%, 10%, 15% and 20% ALA-PDT groups respectivelySOC Nervous system disorders: paraesthesia (LLTs skin burning sensation): "occurred in almost all the patients, but seldom led to considerable pain and normally disappeared within 5 min".
6. MAL-PDT versus ALA-PDT
Wiegell 2006aPain of skin, reported as "The two treatments were equally painful during illumination"; ALA sides: 6/19 (31.5%), scab (LLT crust) reported as "yellow crusting", treated with antibiotics to avoid infectionApplication site oedema, application site erythema, pustular rash (LLT pustular skin eruption), reported as "After illumination edema and severe inflammation were seen in the treatment area. In the following days, a pustular eruption and epithelial exfoliation occurred." Participants' numbers not given, 12/15 (80%) of adverse effects more prominent on the ALA side as compared to MAL. 3/15 (20%) no differences between the sides in adverse effectsSOC Social conditions: "Approximately half of the patients did not go to school or work the following days due to their appearance" (we were unable to find an appropriate MedDRA PT)
7. Other (non MAL, non ALA) PDT versus other comparators
Indocyanine green-PDT
Genina 2004None reportedNone reportedNone reported
Kim 2009None reported2/16 (12.5%) application site oedema and application site erythema (1 in single and 1 in multiple treatment 'group') 1/7 (14.3%) (in the multiple treatment 'group') application site discolouration (LLT application site hyperpigmentation) and scab (LLT crust). Please see 'Notes' in the Characteristics of Included StudiesNone reported
Indole 3-Acetic Acid-PDT
Na 2011None reportedNone reportedNone reported
Topical liposomal methylene blue-PDT
Fadel 2009Pain of skin 7/20 (35%)10/20 (50%) application site erythema, 3/20 (15%) skin erythema, 3/20 (15%) application site discolouration (LLT application site hyperpigmentation), reported to be 'transient'.None reported
Chlorophyll-a (CHA)-PDT 
Song 2014None reportedNone reportedNone reported
Gold microparticle PDT versus other comparators
Paithankar 2015None reportedApplication site pain (reported as: "Treatment was well tolerated, with a mean pain score of 3.5 in the active treatment group.", further information not given)None reported
Table 5. Secondary outcomes other than adverse effects
  1. ALA = 5-aminolevulinic acid
    BPO = benzoyl peroxide
    CHA = chlorophyll-a
    FPT = Fitzpatrick's Skin Types: based on different reactions to sun exposure and range from type I ('pale white skin which always burns and never tans') to type VI ('deeply pigmented dark brown to black skin which never burns and tans very easily') (Fitzpatrick 1988)
    GAAS = Global acne assessment scoring scale
    IAA = indole 3-acetic acid
    IGA = Investigator global assessment score
    ILs = inflamed lesions
    IPL = intense pulsed light
    IR = infrared
    ITT = intention-to-treat analysis
    LPDL = long pulsed dye laser
    LOCF = last observation carried forward
    LLT = lower level term
    MAL = methyl-aminolevulinate
    NILs = non-inflamed lesions
    NNTB = number needed to treat for an additional beneficial outcome
    OFI = optical fibre intra-tissue irradiation
    PDL = pulsed-dye laser
    PDT = photodynamic therapy
    PT = preferred term
    RCT = randomised controlled trial
    SD = standard deviation
    SE = standard error
    SPF = Sun protection factor
    TER = total effective rate
    TLMB = topical liposomal methylene blue

    Unless specified differently, results presented as reported in the published papers, without performing independent analysis. Please see Characteristics of included studies for details on withdrawals and drop-outs of participants for each study.

    Change from baseline i.e. absolute change is calculated by subtracting baseline count from count assessed at certain time point. Percentage change is calculated by dividing the absolute change with baseline count and then multiplying that value by 100 to get percentages.

Study Participants Intervention(s) and control(s) Secondary outcomes other than adverse effects
1. Light versus placebo or no treatment
Green light versus placebo
Baugh 200525 (4 M, 21 F) aged 19-41 years (mean 27.8), diagnosed with mild-moderate inflammatory facial acne; FPT I–III532 nm pulsed laser vs sham in a split-face trial, both with skin cooling system; 2 exposures a week for 2 weeks. Assessed at 1 and 4 weeks post treatmentAt week 4 mean Michaelsson acne severity score decreased from baseline 42.9 to 34.1 (by 21%) on the treated side and increased from baseline 41.2 to 51.4 (by 25%) on the control side (P = 0. 089, SDs not given). At 4 weeks investigators assessed that 14.3% of participants had 50% to 59% improvement, 14.3% had 60% to 69% improvement, 57.1% had 70% to 79% and 14.3% had 80% to 89% improvement. Results for control sides not given
Bowes 200311 (M ⁄ F proportion not given) with mild-moderate acne; skin types not given532 nm pulsed laser vs sham in a split-face trial, both with skin cooling system; 2 treatments weekly for 2 weeks; assessed at 1 week and 1 month after final treatmentAt 4 weeks Michaelsson acne severity score decreased by 35.9% on the treated and increased by 1.8% on the untreated side (SDs not given)
Yilmaz 201144; 38 completed, 20 participants in the once-weekly group (12 M, 8 F) and 18 in twice weekly group (12 M, 6 F); mean ages (± standard deviation) of the participants were 21.0 ± 3.5 and 20.7 ± 2.7 in each group respectively; all with ≥ 4 inflammatory acne lesions, FPT I-III532 nm KTP laser, 2 randomised groups, application once weekly for 4 weeks vs twice weekly for 2 weeks. Within each group 1 side of the face randomised to assigned treatment and the other to no treatment; evaluated at 0, 1 and 4 weeks after final treatmentBoth sides improved, but decrease in Michaelsson severity score was significantly greater on the treated side - 31% versus 6% (P = 0.005) in once-weekly group and by 40% versus 13% in twice-weekly group (P < 0.001). Means and SDs were not given, further data not given
Yellow light versus placebo or no treatment
Seaton 200341, 31 randomised to treatment, 10 to control group; with mild-moderate acne, other characteristics not given585 nm PDL vs sham laser, parallel-group trial, single treatment, assessed at 2, 4, 8 and 12 weeks after treatmentMedian (inter quartile range) improvements in Leeds score were 1.9 (1.8) in the treated group and 0.1 (1.4) in the placebo group (P = 0.007)
Orringer 200440 (24 M, 16 F) enrolled, 26 completed, mean age 20.7 years (range not reported), with facial acne Leeds score > 2; FPT not given ("28 whites, 7 Asians, 2 blacks, 3 unknown")585 nm PDL in a split-face trial, single treatment and 2 treatment (2 weeks apart) groups, serially assessed for 12 weeks after final treatmentChanges in means (SE) Leeds scores were not statistically significant at week 4 (P = 0.56) nor at week 12 (P > 0.99) for both treated and untreated sides. Changes in means (SE) were 0.07 (0.17) on the treated and 0.01 (0.10) on the untreated side at week 4 and 0.04 (0.15) and 0.04 (0.09) at week 12 on each side respectively. ITT analysis (LOCF method) reported
Infrared light versus no treatment
Darne 201138 (7 M, 31 F), aged 18-47 years (mean 28), with moderate-severe facial acne; FPT I-V1450 nm laser (8-9 J/cm²) in a split-face trial, 3 treatments monthly, assessed monthly for 4 months, then at 3-monthly intervals for 12 months after final treatmentSimilar reduction in Leeds grade on both treated and untreated sides at 1 and at 12 months after final treatment with median difference between sides 0 (95% CI -1 to 0) and 0 (95% CI -1 to 0.7) respectively
Orringer 200746 (10 M, 36 F) enrolled, 30 completed, mean age 23.9 years (range not reported) with clinically apparent active facial acne; FPT II–VI1320 nm Nd:YAG laser in a split-face trial with cooling; 3 treatments at 3-week intervals; assessed at weeks 7 and 14Modified Leeds acne severity scale was used. At week 7, both sides graded as slightly worse, by 0.07 (0.23) units for the treated side and by 0.18 (0.22) units for the untreated side (P = 0.46) for 37 participants who completed, and had similar baseline scores with means (SE of the mean) of 2.97 (0.26) and 2.99 (0.26) for treated and untreated sides, respectively. At week 14 both sides graded as slightly improved, but not statistically significant; by 0.20 (0.21) and 0.23 (0.18) units for treated and untreated sides, respectively (P = 0.85) for 32 participants who completed, and had similar baseline means (SE) of 2.88 (0.29) and 2.85 (0.28) for treated and untreated sides, respectively
Moneib 201424 (5 M, 19 F), age 15-38 years (mean 21.5), with moderate-severe acne; FPT II-VFractional Erbium Glass 1559 nm laser, in a split-face trial, 4 treatments at 2-week intervals, assessed every 3 months for 1 year after final treatmentNon-standardised scale (0 = no improvement; < 25% = mild improvement; 26% to 50% = moderate improvement; 51% to 75% = good improvement; 76% to 100% = excellent improvement) was used for evaluation. Reported in graph format and for treatment face sides only, and at unclear time point. Our interpretation of the graph was that investigators assessed 5% participants had moderate, 25% good and 70% excellent improvement
Blue light versus placebo or no treatment
Tzung 200431 (28 completed: 10 M, 18 F) age15–32 years (mean 20.79) with mild-moderate acne; all Taiwanese; FPT III–IV420 nm light in a split-face trial, twice-weekly for 4 consecutive weeks, assessed after each treatment and at 1 month after final treatmentMichaelsson modified grade percentage improvement light compared to control was reported as 52% and 12% respectively at 8 weeks, P = 0.009. Unclear whether mean or median
Blue-red light versus placebo
Papageorgieu 200030 participants, mean age 24.8 years in blue–red light group; 25 participants, mean age 25.6 years in white light control group; randomised from the original 107 recruited (33 M, 74 F, aged 14-50 years), all with mild-moderate acne; skin types not stated415 nm plus 660 nm light vs cool white light; treated daily for 12 weeks; assessed every 4 weeks for the 12-week treatment periodNon-standardised scale: 'worse' (≤ -10%), 'unchanged' (-9% to 9%), 'mild improvement' (10% to 39%), 'moderate improvement' (40% to 59%), 'marked improvement' (60% to 89%) or 'clearance' (≥ 90%) was used for evaluation, but reported only in graph format and no details were provided. Not evaluated after final treatment. Our interpretation of the graph was that in the blue-red light group 4% of participants were reported to have their acne as 'unchanged', 4% as 'mild improvement', 25% as 'moderate improvement', 55% as 'marked improvement' and 6% as 'clearance'. In the white light group 38% of participants were reported as 'unchanged', 38% as 'mild improvement', 15% as 'moderate improvement' and 9% as 'marked improvement'. We dichotomised the data to 26/30 'success' outcomes in the blue-red group and 6/25 in the white light group. Blue red-light was superior to white light with RR (95% CI) of 3.61 (1.77, 7.36), P = 0.0004 and the 'number needed to treat for an additional beneficial outcome' (NNTB) was 2 (95% CI 1 to 3)
Kwon 201335 participants (11 M, 24 F); aged 20-27 years (mean not given), with mild-moderate acne, FPT III-V; 18 participants in the blue-red light group, 17 in the placebo group420 nm plus 660 nm home use LED device vs home use sham device; self-treatment twice daily for 4 weeks in a split-face trial; assessed 4 and 8 weeks after final treatmentNo difference in the distribution of IGA-score between 2 groups at baseline (P > 0.05). At 8 weeks after final treatment 14/18 participants (77.8%) in the blue-red light group and 2/17 (11.8%) in the placebo group had grade 0 (clear) or grade 1 (almost clear) and the difference in distribution of participants was statistically significant (P < 0.01)
Intense pulsed light (IPL) versus no treatment
McGill 200810 (3 M, 7 F), 7 completed, 5 evaluated, aged 18-47 years (mean 30), with mild-moderate facial acne; FPT I-IIIPL, ‘upper’ and ‘lower’ halves of face sides treated with different filters; 550-1100 nm filter ("585 filter"), and the "Dual band" filter (blue light), versus no treatment? (unclear intervention on control half-sides), in a split-face trial, 5 treatments at 2-week intervals, assessed at 1, 3 and 6 months after final treatment

Leeds grade reduced from baseline 3.1 ± 1.7 on the 585 half-sides (n = 10) to 1.6 ± 1.1 at 1 month (n = 8), 1.9 ± 1.4 at 3 months (n = 7) and 2.2 ± 1.8 at 6 months (n = 5); from baseline 2.4 ± 1.8 on the 585 control half-sides to 1.9 ± 1.9 at 1 month, 1.3 ± 1.3 at 3 months and 1.6 ± 1.5 at 6 months; from baseline 3.1 ± 1.7 on the blue-light half-sides to 1.9 ± 1.1 at 1 month, 1.9 ± 1.2 at 3 months and 2.2 ± 1.8 at 6 months; and from baseline 2.5 ± 1.8 on the blue-light control half-sides to 2.0 ± 1.8 at 1 month, 1.6 ± 1.0 at 3 months and 1.8 ± 1.3 at 6 months. At 6 months after final treatment, our calculations using t-distribution showed that there were no significant differences in changes in Leeds grade between 585 half sides and control sides (MD 0.60, 95% CI -1.88 to 3.08), P = 0.64, nor between blue-light and control sides (MD 0.40, 95% CI -1.95 to 2.75), P = 0.74.

Mean (± SD) pretreatment Dermatology Life Quality Index (DLQI) scores were 11 ± 5 (range 3 to 19). At 1 month DLQI score had decreased to 6 ± 5 (range 0 to 12), at 3 months to 5 ± 2 (range 2 to 7) and at 6 months it increased to 7 ± 4 (range 4 to 12). Not reported for separate face half-sides

2. Light versus topical treatment
Light versus benzoyl peroxide (BPO)
Papageorgieu 200030 participants, mean age 24 ± 8 years in blue–red light group and 25 participants, mean age 26 ± 7 years in the BPO group, randomised from the original 107 recruited (33 M, 74 F, age 14–50 years) all with mild-moderate acne; skin types not stated415 nm plus 660 nm light vs 5% BPO parallel groups, treated daily; assessed every 4 weeks for the 12-week treatment periodNon-standardised scale: 'worse' (≤ -10%), 'unchanged' (-9% to 9%), 'mild improvement' (10% to 39%), 'moderate improvement' (40% to 59%), 'marked improvement' (60% to 89%) or 'clearance' (≥ 90%) was used for evaluation, but reported only in graph format and no details were provided. Not evaluated after final treatment. Our interpretation of the graph was that in the blue-red light group 4% of participants were reported to have their acne as 'unchanged', 4% as 'mild improvement', 25% as 'moderate improvement', 55% as 'marked improvement' and 6% as 'clearance'. In the BPO group 10% of participants were reported as 'unchanged', 25% as 'mild improvement', 30% as 'moderate improvement', 30% as 'marked improvement' and 4% as 'clearance'. We dichotomised the data to 26/30 'success' outcomes in the blue-red group and 16/25 in the BPO group.The difference was non significant, with RR (95% CI) of 1.35 (0.98, 1.88), P = 0.07
Light versus clindamycin
Gold 200534 (25 completed the trial, 3 M and 22 F) aged 13-55 years (mean 31 ± 0) with mild-moderate acne; skin types described: caucasian (16), African-American (7), American-Indian (1), Chinese (1); 13 participants in clindamycin group and 12 in blue light group417 nm (blue light) twice weekly for 4 weeks vs self-administered topical clindamycin 1% twice daily, parallel groups, assessed at 4 weeks after final treatmentInvestigator-assessed change in acne severity and global assessment of improvement reported as similar for both groups (figures not given in paper)
Light and other topical treatments
Borhan 201440 (8 M, 12 F in the light group, 9 M, 11 F in the control group), mean age 21.3 ± 2.0 in the intervention and 21.05 ± 2.18 in the control group (range 18-25 years), with mild-moderate acne vulgaris (Burton scale), FPT III-IV595 nm light plus "traditional topical antibiotic medication" versus "traditional topical antibiotic medication" alone in a parallel-group trial, 3 light treatments in total, at 4-week intervals, details of topical treatment not given, unclear frequency of application; assessed at week 4, 8 and 12 (final evaluation 4 weeks after final treatment)At week 12 investigators assessed that 19/20 participants (95%) had marked and 1/20 (5%) had moderate improvement in the laser combined with topical antibiotics group. In the topical antibiotics-alone group 19/20 participants (95%) had mild improvement and 1/20 (5%) had moderate improvement
Ianosi 2013180 participants (56 M, 124 F), aged 24 years (median), 60 in each group, with mild-moderate acne, FPT I-IV500-1200 nm light plus vacuum vs IPL alone 400–700 nm and 870–1200 nm vs anti-acne micellar solution, light applied once a week for 5 weeks, micellar solution unclear, final assessment at final treatmentGreater reduction in Leeds score in light-treatment groups compared to micellar-solution group reported in graph format and no further data provided. Significantly greater effect on quality of life (using Cardiff Acne Disability Index) in vacuum plus IPL group compared to micellar solution group (P = 0.004). Further data not given
Karsai 201089 randomised, 80 evaluated (38 M, 42 F, aged 13.3-43.8 years, mean ± SD age 19.7 ± 5.9 years), with mild-moderate acne (Investigator's Static Global Assessment -ISGA score 2-4), FPT I-IIIClindamycin 1%–BPO 5% hydrating gel (C ⁄ BPO) alone, once daily "throughout the observation period" vs in combination with 2 585 nm PDL treatments. Parallel groups, assessed at 2 and 4 weeks after initial treatmentSimilar reduction in investigator's global assessment of improvement in both groups (27.1% versus 24.6%), measured by Investigator's Static Global Assessment Score (ISGA). ISGA score means (SD) in the C/BPO group were 3.17 (0.76) at baseline and 2.31 (0.54) 4 weeks after initial treatment. They were similar in the C/BPO with light group, 3.37 (0.60) at baseline and 2.54 (0.72) 4 weeks after initial treatment. Dermatology Life Quality Index (DLQI) was used for evaluation of life quality (0-1 = no effect at all on patient's life, 2-5 = small effect on patient's life, 6-10 = moderate effect on patient's life, 11-20 = very large effect on patient's life, 21-30 = extremely large effect on patient's life). Significant DLQI points reduction of 2.31 points (54.5%) in the C/BPO only group and 3.06 points (42.5%) in the C/BPO with light group, with no significant difference in reduction between the groups. Means and SD reported in graph format. Our interpretation of the graph was that mean (SD) in the C/BPO group reduced from baseline 4.3 (3.5) to 2 (2) at 4 weeks after initial treatment, and in the C/BPO from baseline 7.1 (6) to 4 (4) at 4 weeks after initial treatment
Leheta 200975 screened, 45 randomised, aged 18-30 years (mean not reported). 13 (6 M, 7 F, mean age ± SD 24.2 ± 4.6 years) completed the study in the PDL group, 13 (8 M, 5 F, 23.2 ± 4.2 years) in the tretinoin and BPO group, 15 (7 M, 8 F, 24.8 ± 3.8 years) in the chemical peeling group; all with mild-moderate acne, FPT II-IV585 nm PDL, 6 treatments at 2-week intervals vs daily self-administered topical 5% BPO and 0.1% tretinoin (treatment duration not specified) vs chemical peeling with 25% trichloroacetic acid, 6 treatments at 2-week intervals + monthly during the follow-up period. Parallel groups, assessed at the end of the treatment period (3 months)Leeds score means (SD) in the PDL group were 1.673 (0.926) at baseline and 0.557 (0.573) 3 months after initial treatment. In the T/BPO group 2.019 (1.012) at baseline and 0.648 (0.469) 3 months after initial treatment. In the TCAA group 2.083 (0.948) at baseline and 0.680 (0.497) 3 months after initial treatment.
Investigator's global assessment of improvement was evaluated using "degree of clinical improvement": marked response (> 75% improvement), moderate response (51% to 75% improvement), mild response (25% to 50% improvement), minimal response (< 25% improvement), no change, or worsening. In the PDL group 6 (46.2%) participants had been assessed to have marked and 7 (53.8%) moderate improvement; in the T/BPO 5 (38.5%) participants had marked improvement participants and 8 (61.5%) had moderate improvement; in the TCAA 6 (40%) participants had marked and 9 (60%) participants moderate improvement. We dichotomised the data to 13/15 'success' outcomes in the PDL group, 13/15 in T/BPO group and 15/15 in the TCAA group. PDL was not superior to T/BPO with RR (95% CI) of 1.00 (0.76, 1.32), P = 1.00, nor to TCAA, RR (95% CI) of 0.87 (0.69, 1.09), P = 0.24
Zhang 2009a738 randomised, 508 (247 M/261 F) in the intervention group, and 230 (112 M/118 F) in the control group, aged 12–53 years (mean not reported), with mild-severe acne (Pillsbury grades I-IV); FPT not given415 ± 5 nm blue and 633 ± 6 red light in combination with clindamycin gel, azithromycin, antisterone or cimetidine versus clindamycin gel, azithromycin, antisterone or cimetidine alone, in a parallel-group trial, 8 light treatments in total, twice weekly, clindamycin gel twice per day and azithromycin 0.5 g/day (on days without light therapy when in the group with light treatments), assessed at 4 weeks after final treatment

Non-standardised method used for evaluation, based on percentage change in combined lesion counts. Percentage change in lesion count = (lesion count before treatment – lesion count after treatment)/lesion count before treatment × 100%; scale based on lesion count percentage change: ≥ 90% improvement = 'full recovery'; 60% to 89% = 'good improvement'; 30% to 59% = 'effective improvement'; ≤ 29% = 'no effect'; Total effective rate (TER) = (number of fully recovered + good improvement)/total number of participants x 100%. At 4 weeks after final treatment TERs were 65.6% in the treatment group and 54.4% in the control group respectively, with a significant difference between the groups (P value reported as < 0.05).

In the intervention group 142 participants were reported to have ‘fully recovered’, 190 had ‘good improvement’ and 151 had ‘effective improvement’. In the control group 44 participants ‘fully recovered’, 81 had ‘good improvement’ and 87 had ‘effective improvement’.

We dichotomised the data following our protocol and ITT approach to 332/508 ‘success’ outcomes in the intervention and 125/230 ‘success’ outcomes in the control group. Antibiotic treatment in combination with blue-red light was superior to antibiotic treatment alone with RR (95% CI) of 1.20 (1.05, 1.38), P = 0.006 . The NNTB was 10 (95% CI 6 to 20)

3. Light other comparators
Comparison of light therapies of different wavelengths
Cheng 200836 participants (29 M, 7 F) in the blue-light group, 28 participants (19 M, 9 F) in the blue-red light group, aged 14-36 years (mean 22.6 years), all with mild-moderate acne, FPT not reported400-410 nm light versus 400-410 nm plus 660 nm light, 2 treatments a week, duration depending on Pillsbury grade; 4 weeks for grade I in a parallel-group trial, up to 12 weeks for Pillsbury III, evaluated at 1 and 4 weeks after treatmentInvestigators assessed improvement using the following scale based on lesion count percentage change: ≥ 90% improvement = 'full recovery'; 70% to 89% = 'good improvement'; 30% to 69% = 'effective improvement'; ≤ 30% = 'no effect'. In the blue-light group in 7/36 (19.4%) participants there was no improvement, in 3/36 (8.3%) participants the improvement was good and 26/36 (72.2%) participants have 'fully recovered'. In the blue-red light group there was no improvement in 3/28 (10.7%) participants, in 10/28 (35.7%) participants the improvement was good and 15/28 (53.5%) have 'fully recovered'. We dichotomised the data to 15/28 'success' outcomes in the blue-red group and 26/36 in the blue light alone group. The difference was non significant with RR (95% CI) of 0.74 (0.50, 1.11) and P = 0.14
Choi 201020 (1 M, 19 F, age 20-37, mean age 26); all with acne (Cunliffe severity grade 2-4), FPT III-V585 nm PDL vs 530-750 nm IPL, 4 treatments at 2-week intervals, in a split-face trial, assessed 4 and 8 weeks after last treatmentNo statistically significant difference in improvement of Cunliffe scores between the two treatments (P > 0.05); decrease from baseline 2.5 for both to 1.2 for IPL and 1.3 for PDL at 4 weeks and to 1.2 for IPL and to 1.0 for PDL at 8 weeks after treatment
Jung 200918 enrolled, 16 completed (5 M, 11 F, aged 20-31 years, mean age 26); with mild-moderate acne (Cunliffe severity grade 2-5), skin types not given585 nm PDL vs combined 585/1064 nm PDL, in a split-face trial, 3 treatments at 2-week intervals, assessed at 8 and 12 weeks after initial treatmentBaseline mean Cunliffe grades of 2.43 on the PDL sides and 2.19 on the 585/1,064-nm laser sides decreased to 0.77 (P < 0.001) and 0.91 (P = 0.001) at the final visit respectively. Further data not given
Liu 201120 (6 M/14 F) completed the study, number of randomised participants not reported, 10 completed in the blue light, 10 in the red-light group, aged 19–28 years (mean 23.6 years) with mild-moderate acne (Global Acne Grading System); FPT III-IVBlue (405 ± 10 nm) vs red (630 ± 10 nm) LED portable device treatments, about 20 cycles of illumination and the corresponding light doses received in each session were 7.2 J/cm² and 11.52 J/cm², in a parallel-group trial, 8 sessions in total, twice weekly for 4 weeks; assessed at 4 weeks after final treatment and at each treatment sessionNon standardised scale used for investigator's global assessment of improvement (‘reduction ≥ 90% = ‘full recovery’; 60% to 89% reduction = ‘significant improvement’, 40% to 59% reduction = ‘moderate improvement’, 20% to 39% reduction = ‘mild improvement’, and ≤ 19% reduction = ‘non- improvement or aggravation’). In the blue-light group 2 participants ‘fully recovered’, 5 had ‘significant improvement’, 1 ‘moderate improvement’, 1 ‘mild improvement’, and 1 ‘non-improvement or aggravation’. In the red-light group there were 4 participants with ‘significant improvement’, 1 ‘moderate improvement’, 1 ‘mild improvement’, and 4 ‘non-improvement or aggravation’. We dichotomised the data following our protocol to 8/10 ‘success’ outcomes in the blue light and 5/10 in the red-light group. The difference was non significant with RR (95% CI) of 1.60 (0.80, 3.20), P = 0.18
Papageorgieu 200030 participants, mean age 24.8 years in blue–red light group and 27 participants, mean age 23.4 years in the blue-light group, randomised from the original 107 recruited (33 M, 74 F, aged 14-50 years) all with mild-moderate acne; skin types not stated415 nm plus 660 nm light vs 415 nm light, parallel groups, treated daily for 12 weeks; assessed every 4 weeks for the 12-week treatment periodNon-standardised scale: 'worse' (≤ -10%), 'unchanged' (-9% to 9%), 'mild improvement' (10% to 39%), 'moderate improvement' (40% to 59%), 'marked improvement' (60% to 89%) or 'clearance' (≥ 90%) was used for evaluation, but reported only in graph format and no details were provided. Not evaluated after final treatment. Our interpretation of the graph was that in the blue-red light group 4% of participants were reported to have their acne as 'unchanged', 4% as 'mild improvement', 25% as 'moderate improvement', 55% as 'marked improvement' and 6% as 'clearance'. In the blue-light group 25% of participants were reported to have their acne as 'unchanged', 4% as 'mild improvement', 30% as 'moderate improvement', 35% as 'marked improvement' and 4% as 'clearance'. We dichotomised the data to 26/30 'success' outcomes in the blue-red group and 19/27 in the blue light alone group. The difference was non significant, with RR (95% CI) of 1.23 (0.93, 1.63), P = 0.15
Comparison of light therapies of different doses
Bernstein 20077 enrolled, 6 completed (1 M, 4 F, aged 23-41 years, mean age 29), all with active papular acne, FPT I-IIIComparison of two 1450 nm laser treatments; single-pass, high-energy (13–14 J/cm²) vs double-pass, low-energy (8–11 J/cm²); 4 treatments at monthly intervals, assessed 1 month following each treatment and 2 months after final treatmentAllen-Smith acne severity score mean (SD) dropped from 3.1 (1.1) to 1 (1.1) on the single-pass face side and from 3.2 (0.7) to 1 (1.1) on the double-pass face side. Single-pass mean (SD) investigator-assessed improvement score mean (SD) was 1.6 (1.1) on the single-pass side of the face and 2.4 (0.9) on the double-pass side of the face
Uebelhoer 200711 (2 M, 9 F, aged 19-39 years, mean age 26), 9 completed, all with ≥ 10 inflammatory papules on each side of the face and Allen-Smith grade ≥ 3 and ≤ 5; skin types not given1450 nm laser single-pass treatment consisting of stacked double pulses vs a double-pass treatment of single pulses; in a split-face trial, treated every 3 weeks for a total of 3 treatments, assessed before each follow-up treatment, and at 3 months after the final treatmentDecrease in acne severity in 8/9 subjects (89%); the mean acne severity scores decreased to 2.1 (range 0 to 5) on the single-pass sides and 2.2 (range 1 to 5) on the double-pass sides from 3.3 (range to 3–5) at baseline. One subject's grade increased from 3 to 5. Data not reported at any time point for Investigator's global assessment of improvement
NCT00706433266 (128 M, 138 F), 68 in the ALA 1000 s group, 65 in the ALA 500 s group, 67 in the vehicle 1000 s group and 66 in the vehicle 500 s group, mean age 20.1 years, inclusion criterion 12 > years, with moderate and severe acne (IGA score 3 and 4, with at least 20 ILs); FPT I-V20% ALA (45 min incubation) plus blue 1000 s light vs 20% ALA (45 min incubation) plus 500 s blue light vs vehicle (45 min incubation) plus blue 1000 s light vs vehicle (45 min incubation) plus 500 s blue light; in a parallel-group trial; up to 4 treatments at 3-week intervals, assessed 3 and 6 weeks after the final treatmentInvestigator Global Assessment (IGA) was used for evaluation (0; clear skin with no ILs or NILs; almost clear; rare NILs with no more than a few small ILs; Mild; > Grade 1; some NILs with some ILs (papules/pustules only; no nodules); Moderate; > Grade 2; up to many NILs and a moderate number of ILs but no more than one small nodule; Severe; > Grade 3; up to many NILs and ILs, but no more than a few nodules); success was defined as a 2 point or more improvement on the IGA scale since baseline. At 3 weeks after final treatment there were 15/67 of 'success' outcomes in the vehicle 1000 s and 11/66 in the vehicle 500 s group. The difference between vehicle 1000 s and vehicle 500 s groups was non significant, with RR (95% CI) of 1.34 (0.67, 2.70), P = 0.43. At 6 weeks after final treatment there were 16/67 of 'success' outcomes in the vehicle 1000 s and 16/66 in the vehicle 500 s group. The difference between vehicle 1000 s and vehicle 500 s groups was non significant, with RR (95% CI) 0.99 (0.54, 1.80), P = 0.96
Comparison of light therapies of different treatment application intervals
Yilmaz 201144; 38 completed, 20 participants in the once-weekly group (12 M, 8 F) and 18 in twice-weekly group (12 M, 6 F); mean ages (± standard deviation) of the participants were 21.0 ± 3.5 years and 20.7 ± 2.7 years in each group respectively; all with ≥ 4 inflammatory acne lesions, FPT I-III532 nm KTP laser, 2 randomised groups, application once weekly for 4 weeks vs twice weekly for 2 weeks. Within each group 1 side of the face randomised to assigned treatment and the other to no treatment; evaluated at 0, 1 and 4 weeks after final treatmentAt 4 weeks there was no statistically significant difference in decrease of acne severity between the treated sides among the 2 groups. Michaelson acne severity scores of treated sides of the face dropped by 41% in once-weekly treatment group and by 40% in twice-weekly group. Differences in Michaelson acne severity score means (SD) of the treated face sides at baseline and at 4 weeks were -5.9 (7.9) in the once-weekly group and -9.3 (7.5) in the twice-weekly group
Light in combination with carbon lotion versus no treatment
Jung 201222 (4 M, 18 F), 20 completed (2 M, 18 F, aged 19-34 years, mean age 25.4), FPT III-IV, acne severity not givenCarbon lotion plus quasi-long pulse and Q-switched 1064 nm Nd:YAG laser vs non treated control, in a split-face trial, 3 treatments over 4 weeks, evaluation every 2 weeks whilst on treatment and then every 4 weeksCunliffe severity grade decreased significantly from 3.2 to 1.7 (P < 0.001) on the laser-treated side and from 2.7 to 2.6 (P < 0.05) on the non-treated side. The difference between the 2 treatments was significant (P = 0.04)
Light in combination with oral therapy versus other comparators
Ling 2010120 (68 M, 52 F), aged 12-32 years, means given for individual groups 21-22 years); 30 in each group, moderate-severe acne according to Pillsbury classification, FPT not reported415 nm plus 630 nm light in combination with sulfotanshinone vs sulfotanshinone alone vs 415 nm plus 630 nm light in combination with sulfotanshinone and prednisolone vs sulfotashinone and prednisolone; blue-red light applied twice weekly, sulfotanshinone 4 times daily and prednisolone 3 times daily in a parallel-group trial, assessed 4 weeks after treatmentInvestigators assessed improvement using the following scale based on lesion count percentage change: ≥ 95% improvement = 'full recovery'; 60% to 95% = 'good improvement'; 20% to 59% = 'effective improvement'; ≤ 20% = 'no effect'. In the blue-red light plus sulfotanshinone group 19/30 (63.3%) participants fully recovered, 7/30 (23.3%) had good improvement, in 3/30 (10%) the treatment was effective and 1/30 (3.33%) there was no effect. In the sulfotanshinone-alone group 9/30 (30%) participants fully recovered, 10/30 (33.33%) had good improvement, in 7/30 (23.3%) the treatment was effective and 4/30 (13.33%) there was no effect. In the blue -red light plus sulfotanshinone plus prednisolone group 8/30 (26.6%) participants fully recovered, 8/30 (26.6%) had good improvement, in 7/30 (23.3%) the treatment was effective and 7/30 (23.3%) that there was no effect. In the sulfotanshinone plus prednisolone group 6/30 (20%) participants fully recovered, 7/50 (23.3%) had good improvement, in 8/30 (26.6%) the treatment was effective and 9/30 (30%) that there was no effect. We dichotomised the data to 26/30 'success' outcomes in the blue-red light plus sulfotanshinone group,19/30 in the sulfotanshinone alone group, 16/30 in the blue-red light plus sulfotanshinone plus prednisolone group and 13/30 in the sulfotanshinone plus prednisolone group. Blue and red light plus sulfotanshinone was superior to sulfotanshinone alone with RR (95% CI) with 1.37 (1.01, 1.86), P = 0.04; to blue and red light plus sulfotanshinone plus prednisolone with RR (95% CI) of 1.63 (1.13, 2.34), P = 0.009; and to sulfotanshinone plus prednisolone with RR (95% CI) of 2.00 (1.30, 3.08), P = 0.002. The NNTB were 3 (95% CI 1 to 9) and 3 (95% CI 1 to 5) for the latter two comparisons with blue-red light plus sulfotanshinone respectively. However, there is no calculable NNTB for the comparison of blue-red light plus sulfotanshinone to sulfotanshinone alone since the 95% CI for the risk difference contains zero (i.e. no effect), and this corresponds to an infinite upper 'limit' for the 95% CI for the NNTB, which indicates that there is no true boundary on how large the NNTB could be for this comparison: this is also seen in the marginal effect seen with the RR
Ou 201490 randomised; number of participants per group not reported (M/F not reported, 43 in the intervention, 40 in the control group), 83 completed (13 M/70 F), aged 18–38 years (mean 25.1), with moderate acne (grade II-III according to the Chinese Acne Treatment Guidelines); FPT not givenYinhua decoction (term as presented in the English translation of the abstract provided by the journal where full text was published in Mandarin) with electric light synergy versus Yinhua decoction in combination with red and blue light treatment, in a parallel-group trial, 6 treatments in total, applied every 2 weeks, assessed at 12 weeks after final treatment

Non-standardised method used for evaluation, based on percentage change in combined lesion counts. Percentage change in lesion count = (lesion count before treatment – lesion count after treatment)/lesion count before treatment × 100%; Fully recovered: percentage change in lesion count ≥ 90%; Good improvement: percentage change in lesion count 60% to 89%; Effective: percentage change in lesion count 30% to 59%; No effect: percentage change in lesion count ≤ 29%; Total effective rate (TER) = (number of fully recovered + good improvement)/total number of participants x 100%. At 12 weeks after final treatment the study authors reported TERs of 70% in the treatment group and of 37.5% in the control group respectively, with a reported significant difference between the groups (P = 0.002).

In the intervention group 6 participants ‘fully recovered’, 24 had ‘good improvement’ and 10 had ‘effective improvement’. In the control group no participants ‘fully recovered’, 15 had ‘good improvement’ and 20 had ‘effective improvement’. 43 participants completed the trial in the intervention group and 40 completed in the control group. We dichotomised the data to 30/43 (69.7% of those who completed) 'success' outcomes in the intervention arm, and 15/40 (37.5% of those who completed) in the control arm. Numbers of randomised participants in each group were not reported, and so we were unable to use ITT approach. YD plus "electric light synergy" were superior to YD in combination with blue-red light with RR (95% CI) of 1.86 (1.19, 2.91), P=0.006. The NNTB was 4 (95% CI 2 to 10)

Zhang 2009a738 randomised, 508 (247 M/261 F) in the intervention group, and 230 (112 M/118 F) in the control group, aged 12–53 years (mean not reported), with mild-severe acne (Pillsbury grades I-IV); FPT not given415 ± 5 nm blue and 633 ± 6 red light in combination with clindamycin gel, azithromycin, antisterone or cimetidine versus clindamycin gel, azithromycin, antisterone or cimetidine alone, in a parallel-group trial, 8 light treatments in total, twice weekly, clindamycin gel twice per day and azithromycin 0.5g/day (on days without light therapy when in the group with light treatments), assessed at 4 weeks after final treatmentPlease see results under Light and other topical treatments as this study could be placed under both comparisons. We were unable to perform subgroup analyses
Zhang 2013b120 (59 M/61 F), 60 in each group, aged 14–40 years (mean 22.1 in the intervention, 23.6 in the control group), with mild-moderate acne (Pillsbury grades I-III); FPT not given

415 ± 5 nm (blue) and 633 ± 3 nm (red)

light combined with jinhua xiaocuo (term as presented in the English translation of the abstract provided by the journal where full text was published in Mandarin) pills and chloramphenicol tincture versus Jinhua xiaocuo pills and chloramphenicol tincture alone, in a parallel-group trial, 8 treatments, applied twice weekly; Jinhua xiaocuo pills 4 g orally 3 times/day, Chloramphenicol tincture 10 mg/mL (applied once in the day once at night), assessed 4 weeks after final treatment

Non-standardised method used for evaluation, based on percentage change in combined lesion counts. Percentage change in lesion count = (lesion count before treatment – lesion count after treatment)/lesion count before treatment × 100%; Fully recovered: percentage change in lesion count ≥ 90%; Good improvement: percentage change in lesion count 60% to 89%; Effective: percentage change in lesion count 30% to 59%; No effect: percentage change in lesion count ≤ 29%; Total effective rate (TER) = (number of fully recovered + good improvement)/total number of participants x 100%). At 4 weeks after final treatment TERs were reported to be 91.7% in the treatment group and 65% in the control group respectively, with a reported significant difference between the groups (P value reported as < 0.05). In the intervention group 25 participants ‘fully recovered’, 30 had ‘good improvement’ and 5 had ‘effective improvement’. In the control group 17 participants ‘fully recovered’, 22 had ‘good improvement’ and 11 had ‘effective improvement’. We dichotomised the data following our protocol to 55/60 ‘success’ outcomes in the intervention and 39/60 ‘success’ outcomes in the control group. Jinhua xiaocuo pills and chloramphenicol tincture in combination with blue-red light were superior to jinhua xiaocuo pills and chloramphenicol tincture alone with RR (95% CI) of 1.41 (1.15, 1.72), P = 0.0008. The NNTB was 4 (95% CI 3 to 9)
IPL alone versus IPL in combination with vacuum
Ianosi 2013180 participants (56 M, 124 F), aged 24 years (median), 60 in each group, with mild-moderate acne, FPT I-IV500-1200 nm light plus vacuum vs IPL alone 400–700 nm and 870–1200 nm vs anti-acne micellar solution, light applied once a week for 5 weeks, micellar solution unclear, final assessment at final treatmentChanges in lesion counts reported as scores 1 = insignificant result (lesion count reduction 0% to 25%) to 4 = very good result (lesion count reduction 76% to 100%). No significant differences found between treatments at final assessment in reduction score of papules and pustules (P reported as 'NS'). Significantlly greater reduction score of comedones in vacuum plus IPL group (P < 0.001). Greater reduction in Leeds score in IPL-only group reported in graph format and no further data provided. Significantly greater effect on quality of life (using Cardiff Acne Disability Index) in vacuum plus IPL group (P = 0.004). Further data not given
4. MAL-PDT versus other comparators
MAL-PDT versus red light alone
Pariser 2013153 participants (87 M/66 F), 100 in the 80 mg/g MAL-PDT group, 53 in the placebo group, aged 12-35 years (mean 18.6), with severe facial acne vulgaris, IGA score 4, 25-75 ILs and 20-100 NILs on the face, FPT I-VI80 mg/g MAL-PDT under occlusion followed by illumination with 632 nm 37J/cm² red light vs placebo cream plus 632 nm 37 J/cm² light in a parallel-group trial, 4 treatments at 2-week intervals, assessed at 6 weeks after final treatment15 withdrawals from the MAL-PDT group, 4 withdrawals and 1 lost to follow-up from the placebo group. ITT analysis was performed. At 6 weeks after final treatment 'success' outcomes as defined by the IGA score were found in 44/100 participants in the 80 mg/g group and 14/53 in the placebo cream group. Our analyses showed borderline superiority of 80 mg/g MAL-PDT to placebo cream activated by red light, with RR 1.67 (95% CI 1.01 to 2.75). Please note that the results of this study were combined with those of NCT00933543 and NCT00594425 for the same comparison
NCT00933543107 participants (48 M/59 F), 54 in the 80 mg/g MAL-PDT group, 53 in the placebo group, aged 11-35 years (mean 17.2), with moderate-severe facial acne vulgaris, IGA score 3-4, 20-100 ILs and 30-120 NILs on the face, FPT I-VI80 mg/g MAL-PDT (without occlusive dressing) followed by illumination with 632 nm 37 J/cm²red light vs placebo cream plus 632 nm 37 J/cm² light (without occlusive dressing) in a parallel-group trial, 4 treatments at 2-week intervals, assessed at 6 weeks after final treatment3 withdrawals in MAL-PDT group, 6 withdrawals and 1 lost to follow-up in placebo group. ITT analysis was performed. At 6 weeks after final treatment 'success' outcomes as defined by the IGA score were found in 5/54 participants in the 80 mg/g group and 1/53 in the placebo-cream group. Our analyses showed that 80 mg/g MAL-PDT was not superior to placebo cream activated by red light, with RR 4.91 (95% CI 0.59 to 40.61). Please note that the results of this study were combined with those of Pariser 2013 and NCT00594425 for the same comparison
NCT00594425150 participants (59 M/91 F), 50 in the 40 mg/g MAL-PDT group, 48 in the 80 mg/g MAL-PDT group, 52 in the placebo group, aged 15-40 years (mean 21.3), with moderate-severe acne, IGA score 3-4, 20-100 ILs and up-200 NILs on the face, FPT I-IV80 mg/mL MAL under occlusion (1.5 h) plus 632 nm 37 J/cm² light vs 40 mg/mL MAL under occlusion (1.5 h) plus 632 nm 37 J/cm² light vs placebo cream plus 632 nm 37 J/cm² light in a parallel-group trial, 4 treatments at 2-week intervals, assessed at 2, 3, 6, 12 and 24 weeks after final treatment43 participants completed in the 40 mg/g group, 34 completed in the 80 mg/g group and 42 completed in the placebo-cream group, ITT analysis was performed (LOCF method). At 6 weeks after final treatment 'success' outcomes as defined by the IGA score were found in 6/50 participants in the 40 mg/g group and 4/52 in the placebo-cream group. Our analyses showed that 40 mg/g MAL-PDT was not superior to placebo cream activated by red light, with RR 1.56 (95% CI 0.47 to 5.20), P = 0.47. At 6 weeks after final treatment 'success' outcomes as defined by the IGA score were found in 6/48 participants in the 80 mg/g group and 4/52 in the placebo cream group. Our analyses showed that 80 mg/g MAL-PDT was not superior to placebo cream activated by red light, with RR 1.63 (95% CI 0.49 to 5.41). Please note that the results of this study were combined with those of Pariser 2013 and NCT00933543 for the same comparison.
Hörfelt 200630 (25 M, 5 F), 27 completed, aged 15-28 years (mean 18) with moderate-severe inflammatory facial acne (Leeds score 5–10); FPT types I–III635 nm light plus MAL vs placebo cream and light in a split-face trial, 2 treatments, 2 weeks apart, assessed at 4 and 10 weeks after treatmentAt 12 weeks investigator-assessed change in acne severity (global severity assessment clear or almost clear) observed in 9/30 participants (30%) for the MAL-PDT side and in 3/30 participants (10%) on the light-only side. Significantly greater improvement on the MAL-PDT side than on placebo-PDT side (P = 0.0143). 12 (40%) participants improved in more than one category on the MAL-PDT side versus 7 (23%) on the placebo-PDT side. We dichotomised the data to 12/30 'success' outcomes on the MAL-PDT sides and 7/30 on the placebo-PDT sides. The difference was non significant, with RR (95% CI) of 1.71 (0.78, 3.75), P = 0.18
MAL-PDT versus placebo or no treatment
Wiegell 2006b36 participants: 21 in treatment group age 23 ± 5 years (9 M, 10 F analysed) and 15 in control group age 24 ± 5 years (3 M, 9 F analysed), with > 12 inflammatory acne lesions; FPT II–VComparison of MAL plus 630 nm with no treatment in a parallel-group trial; 2 treatments, 2 weeks apart, assessed every 4 weeks for 12 weeks after treatmentNo significant difference was observed in reduction in Leeds grade between the two groups (P = 0.24). Median score (range) at 12 weeks was 1 (0 to 5) in the MAL-PDT and 2 (0 to 8) in the control group. In the MAL-PDT group median improvement score was 2 at 4 weeks, 2 at 8 weeks and 3 at 12 weeks. In the control group median improvement score was 1 at 4 weeks, 0 at 8 weeks and 1 at 12 weeks after treatment (results were reported in graph form, our interpretation given). No further data were provided
MAL-PDT other
NCT00594425150 participants (59 M/91 F), 50 in the 40 mg/g MAL-PDT group, 48 in the 80 mg/g MAL-PDT group, 52 in the placebo group, aged 15-40 years (mean 21.3), with moderate to severe acne, IGA score 3-4, 20-100 ILs and up-200 NILs on the face, FPT I-IV80 mg/mL MAL under occlusion (1.5 h) plus 632 nm 37 J/cm² light vs 40 mg/mL MAL under occlusion (1.5 h) plus 632 nm 37 J/cm² light vs placebo cream plus 632 nm 37 J/cm² light in a parallel-group trial, 4 treatments at 2 weeks intervals, assessed at 2, 3, 6, 12 and 24 weeks after final treatment37 participants completed in the 80 mg/g group, and 43 completed in the 40 mg/g group, ITT analysis was performed (LOCF method). At 6 weeks after final treatment 'success' outcomes as defined by the IGA score were found in 6/48 participants in the 80 mg/g group and 6/50 in the 40 mg/g group. Our analyses showed that 80 mg/g MAL-PDT was not superior to 40 mg/g MAL-PDT, with RR 1.04 (95% CI 0.36 to 3.01), P = 0.94
Bissonnette 201044 participants, 33 completed (M/F not stated), aged 18-40 years (mean 24.4), 22 randomised to each group,10 ≥ ILs on each side of the face and a Global Acne Severity score 3 ≥, FPT I-IV80 mg/mL MAL plus 630 nm 25 J/cm² light vs 80 mg/mL MAL plus 630 nm 37 J/cm² light in a parallel-group trial, split-face randomisation within each group to occlusion or no occlusion, 4 treatments at 2-week intervals, assessed at 4 and 12 weeks after final treatmentAt 4 weeks after treatment Global Acne Severity score 0 or 1 ('success') was found on 1/16 (6.3%) of face sides with occlusion and on 0/16 (0%) face sides without occlusion in the 25 J/cm² group; and on 0/17 (0%) of face sides with occlusion and on 1/17 (5.9%) of the face sides without occlusion in the 37 J/cm² group. At 12 weeks ('success') was found on 0/20 (0%) of face sides with occlusion and on 0/20 (0%) face sides without occlusion in the 25 J/cm² group; and on 1/20 (5.6%) of face sides with occlusion and 2/20 (11.1%) of the face sides without occlusion in the 37 J/cm² group. Detailed data provided by the study authors. ITT analysis results reported (LOCF method). At 12 weeks the difference for comparison 37 J/cm² treatment with occlusion versus 37 J/cm² treatment without occlusion was non significant, with RR (95% CIs) 0.50 (0.05, 5.12)
Hong 201322 (2 M, 20 F), aged 19-35 years (mean not given), 'at least grade 2 (Cunliffe acne grading system)', FPT IV-VMAL plus 630 nm light vs MAL plus 530-750 nm light in a split-face trial, 3 treatments in total, 2-week intervals, assessed at 4 weeks after treatmentAt 4 weeks after treatment there was no significant difference in the improvement in acne Cunliffe grade between the red-light side (1.9) and IPL side (2.0). Baseline means extracted from graph as 3.6 on the red-light side and 3.75 on the IPL side. Further data were not provided
5. ALA-PDT versus other comparators
ALA-PDT versus red light alone
Chen 201550, 47 completed (25 M/22 F), 24/25 in the intervention, 23/25 in control group, aged 18–33 years (mean 23.6 in the intervention, 24.1 in the control group), with mild-severe acne (non-standardised scale); FPT not given20% ALA (90 min under plastic film occlusion) plus 633 ± 10 nm red light for 20 min versus 633 ± 10 nm red light for 20 min alone in a parallel-group trial, 3 treatments in total, weekly, assessed at 2, 4 and 6 weeks after final treatmentNon-standardised method used for evaluation, TER (‘Reduction rate was calculated as follows: Reduction rate (%) = (numbers of comedones before treatment - numbers of comedones after treatment)/number of comedones before treatment x 100. Skin lesions with ≥ 90% improvement were classified as cured, skin lesions with 60%-89% improvement were classified as excellent effect, skin lesions with 30%-59% improvement were classified as fair effect and skin lesions with < 30% improvement or exacerbations were classified as no effect. TER was computed as follows: TER (%) = (number of cured cases + excellent effect cases)/total number of cases x 100). TERs in the treatment group were 54.2% at 2 weeks, 75.0% at 4 weeks and 83.3% at 6 weeks, whereas those in the control group were 26.1% at 2 weeks, 43.5% at 4 weeks and 56.5% at 6 weeks. P-values reported for differences between the 2 groups were P = 0.050 at 2 weeks, P = 0.028 at 4 weeks and P = 0.045 at 6 weeks. In the ALA-PDT group 3, 11 and 15 participants were reported to be ‘cured’ at 2, 4 and 6 weeks after final treatment respectively; 10, 7 and 5 had ‘excellent effect’ at 2, 4 and 6 weeks after final treatment respectively. In the red-light group 1, 4 and 6 participants were ‘cured’ at 2, 4 and 6 weeks after final treatment respectively; 5, 6 and 7 had ‘excellent effect’ at 2, 4 and 6 weeks after final treatment respectively; 1 participant dropped out from the ALA-PDT group, and 2 from the red-light-only group, and we treated them as treatment failures as per our protocol. We dichotomised the data following our protocol to 13/25 ‘success’ outcomes at 2 weeks, 18/25 at 4 weeks and 20/25 at 6 weeks in the intervention group, whereas in the control group there were 6/25 ‘success’ outcomes at 2 weeks, 10/25 at 4 weeks and 13/25 at 6 weeks. ALA-PDT was not superior to red light alone with RR (95% CI) of 1.54 (1.01, 2.35), P = 0.05 at 6 weeks. We combined the results of this study with those of Zhang 2013a for assessments at 2 and 4 weeks
Zhang 2013a116 (47 M/59 F) randomised, 63 in the intervention, 53 in control group, aged 16–47 years (mean 24 years in the intervention, 23 years in the control group), with moderate-severe acne (Pillsbury grade II-IV); FPT not given

Unclear % of

5-ALA plus

630 ± 5 nm

red light versus

630±5 nm red light alone, in a parallel-group trial, 3 treatments in total, weekly, assessed at 2, 4 and 8 weeks after final treatment

Non-standardised method used for evaluation, based on percentage change in combined lesion counts. Percentage change in lesion count = (lesion count before treatment – lesion count after treatment)/lesion count before treatment × 100%; Fully recovered: percentage change in lesion count ≥ 90%; Good improvement: percentage change in lesion count 60% to 89%; Effective: percentage change in lesion count 20% to 59%; No effect: percentage change in lesion count ≤ 19%; total percentage effectiveness = (number of fully recovered + good improvement)/total number of participants x100%). TERs were 44.4%, 58,7% and 79.4%, in the treatment group at 2, 4 and 8 weeks after final treatment respectively. TERs in the control group were 13.2%, 28.3% and 41.5% at 2, 4 and 8 weeks after final treatment respectively. In the intervention group 5, 12 and 24 participants ‘fully recovered’ at 2, 4 and 8 weeks after final treatment respectively; 23, 25 and 26 had ‘good improvement’ at 2, 4 and 8 weeks after final treatment respectively; and 33, 26, 13 had ‘effective improvement’ at 2, 4 and 8 weeks after final treatment respectively. In the control group no participants ‘fully recovered’ at 2, 4 nor at 8 weeks after final treatment; 7, 15 and 22 had ‘good improvement’ at 2, 4 and 8 weeks after final treatment respectively; and 3, 21 and 19 had ‘effective improvement’ at 2, 4 and 8 weeks after final treatment respectively. We dichotomised the data following our protocol to 28/63, 37/63, and 50/63 ‘success’ outcomes in the intervention group at 2, 4 and 8 weeks after final treatment respectively; and 7/53, 15/53, and 22/53 ‘success’ outcomes in the control group at 2, 4 and 8 weeks after final treatment respectively. ALA-PDT was superior to red light alone with RR (95% CI) of 1.91 (1.36, 2.70), P = 0.0002 at 8 weeks. The NNTB was 3 (95% CI 2 to 5) at 8 weeks. We combined the results of this study with those of Chen 2015 for assessments at 2 and 4 weeks
ALA-PDT versus blue light alone
NCT00706433266 (128 M, 138 F), 68 in the ALA 1000 s group, 65 in the ALA 500 s group, 67 in the vehicle 1000 s group and 66 in the vehicle 500 s group, mean age 20.1 years, inclusion criterion 12 > years, with moderate and severe acne (IGA score 3 and 4, with at least 20 ILs); FPT I-V20% ALA (45 min incubation) plus blue 1000 s light vs 20% ALA (45 min incubation) plus 500 s blue light vs vehicle (45 min incubation) plus blue 1000 s light vs vehicle (45 min incubation) plus 500 s blue light; in a parallel-group trial; up to 4 treatments at 3 weeks intervals, assessed 3 and 6 weeks after the final treatmentIGA was used for evaluation (0; clear skin with no ILs or NILs; almost clear; rare NILs with no more than a few small ILs; mild; > Grade 1; some NILs with some ILs (papules/pustules only; no nodules); moderate; > grade 2; up to many NILs and a moderate number of ILs but no more than one small nodule; severe; > Grade 3; up to many NILs and ILs, but no more than a few nodules); success was defined as a 2 point or more improvement on the IGA scale since baseline. At 3 weeks after final treatment there were 13/68 of 'success' outcomes in ALA 1000 s, 11/65 in the ALA 500 s, 15/67 in the vehicle 1000 s and 11/66 in the vehicle 500 s group. The difference between ALA 1000 s and vehicle 1000 s groups was non significant, with RR (95% CI) 0.85 (0.44, 1.65), P = 0.64, and it was non significant between ALA 500 s and vehicle 500 s groups, with RR (95% CI) 1.02 (0.47, 2.18), P = 0.97. At 6 weeks after final treatment there were 15/68 of 'success' outcomes in ALA 1000 s, 11/65 in the ALA 500 s, 16/67 in the vehicle 1000 s and 16/66 in the vehicle 500 s group. The difference between ALA 1000 s and vehicle 1000 s groups was non significant, with RR (95% CI) 0.92 (0.50, 1.71), P = 0.80, and it was non significant between ALA 500 s and vehicle 500 s groups, with RR (95% CI) 0.70 (0.35, 1.39), P = 0.31
ALA-PDT versus IPL alone
Oh 200920 (4 M, 16 F) , aged 18-30 years, 9 in the short-incubation group (3 M, 6 F, mean age ± SD 23 ± 4.12 years) and 10 in the long-incubation group (1 M, 9 F and 23 ± 5.53 years), with moderate and severe acne (Evaluator Global Severity Score 3 and 4); FPT II-IV20% ALA plus 590 nm IPL; 2 parallel groups: short incubation (30 min) vs long incubation (3 h), half of the face within each treated with IPL alone; 3 treatments at 4 weeks intervals, assessed 4 weeks after each treatment and 8 and 12 weeks after the third treatmentAt 12 weeks investigators assessed improvement as mild in 3/9 participants (33.3%) and as moderate in 6/9 participants (66.7%) in the short incubation group; as mild in 2/11 participants (18.2%), as moderate in 5/11 participants (45.4%) and as significant in 4/11 participants (36.4%) in the long incubation group. We dichotomised the data to 6/9 'success' outcomes in short incubation group and 9/11 in the long incubation group. The difference was non significant, with RR (95% CI) of 0.44 (0.06, 3.51), P = 0.44
Mei 201341 (24 M, 17 F), mean age 24 years, 21 in the ALA-IPL PDT group, 20 in the placebo cream-IPL group, II–IV Pillsbury grade acne; FPT II-IV.10% ALA plus 420–950 nm light versus placebo cream plus 420–950 nm light in a parallel-group trial, 4 treatments in total, weekly, assessed 4, 8 and 12 weeks after treatment.At 12 weeks after final treatment investigators assessed an improvement of 75% to 100% in all lesions in 13/21, of 50% to 75% in 5/21 and of 25% to 50% in 2/21 participants and no improvements 1/21 participants in the ALA-IPL group. In the control group an improvement of 75% to 100% in all lesions was achieved in 3/20, of 50% to 75% in 9/20, of 25% to 50% in 6/20 and no improvements in 2/20 participants. We dichotomised the data to 18/21 'success' outcomes in the ALA-IPL group and 12/20 in IPL-alone group. The difference was non significant, with RR (95% CI) of 1.43 (0.96, 2.13), P = 0.08
ALA-PDT versus green light alone
Sadick 2010a10 randomised (M/F not reported), 8 (2 M,6 F) completed, all > 18 years, mean age and age range not reported, all with moderate-severe acne IGA 3-4, FPT I-III20% ALA plus KTP 532 nm laser compared with KTP 532 nm laser alone in a split-face trial, 3 treatments spaced at 3-4 week intervals, evaluated after each treatment and at 2, 6 and 12 weeks after final treatmentIGA was used for evaluation (Grade 0 = clear skin, no inflammatory lesions; grade 1 = almost clear, rare non-inflammatory lesions, few small inflammatory lesions; grade 2 = mild severity, some non-inflammatory lesions, some inflammatory lesions (papules, pustules, no nodular lesions); grade 3 = moderate severity, many non-inflammatory and moderate inflammatory lesions, no more than one nodular lesion; grade 4 = severe, many non-inflammatory and inflammatory lesions, nodular lesions are present). On the ALA-PDT sides IGA score (mean ± SE) reduced from baseline 3.50 ± 0.19 to 2.29 ± 0.29 (35% improvement) after first treatment and to 2.13 ± 0.40 (39% improvement) after second treatment. On the light-only sides IGA score (mean ± SE) reduced from baseline 3.63 ± 0.18 to 2.42 ± 0.30 (33% improvement) after first treatment and to 2.38 ± 0.33 (34% improvement) after second treatment. Further details and results of evaluations after final treatment were not given (reported as "Similar results were recorded after the third treatment session that was evaluated at week 12")
ALA-PDT versus placebo or no treatment
Orringer 201099 screened, 44 enrolled (14 M, 30 F) aged 15-50 years, mean 25, all with clinically evident facial acne, all FPT included20% ALA plus PDL compared with placebo in a split-face trial, 3 treatments spaced at 2-week intervals, evaluated every 2 weeks for a total of 16 weeksStatistically significant improvement (P = 0.01) in mean Leeds score on treated skin versus untreated skin at week 16. Mean change in score from baseline (95% CI) was -1.07 (-1.69 to -0.45) on the treated sides and -0.52 (-1.07 to 0.04) on the control sides
ALA-PDT other
Barolet 201010 (7 M, 3 F), aged 13-54 years, mean age 26.2, with mild-moderate acne, with ≥ 10 acne lesions, FPT I-III970 nm IR pre-treatment plus ALA and 630 nm PDT vs ALA-PDT alone, one treatment in a split-face or split-back design, evaluated after 4 weeks4 weeks after treatment greater improvement in Global Severity Assessment Score medians on the IR pre-treated (1, 95% CIs 0.74 to 1.34) versus control side (2, 95% CIs 1.17 to 1.72). 95% CI reported for means, but means were not given
Hongcharu 200022 participants, aged 18-44 years; 11 in single-treatment group, mean age 30 years, 9 M, 2 F; 11 in multiple-treatment group, mean age 27 years, 8 M, 3 F; all with mild-moderate acne of the back; FPT I–IV4 areas on the back of each participant: 550–700 nm light source used. ALA–light; ALA alone; light alone; untreated control. Single and multiple treatment groups, assessed at 1, 2, 3, 10 and 20 weeksChange from baseline in Michaelsson acne severity score was significantly better in ALA-PDT than other three areas at 3, 10 and 20 weeks after single treatment (P values not given) and at all visits after multiple treatment (P < 0.05). ALA–PDT and multiple ALA treatment sites showed more improvement than single treatment (P < 0.001 and P = 0.007, respectively).
Investigator's global assessment of improvement scores also significantly better for the ALA-PDT areas than other 3 areas where some improvement was also observed in both single and multiple treatment groups. These comparisons, as well as comparison between single and multiple treatment groups were reported in an unclear way
NCT00706433266 (128 M, 138 F), 68 in the ALA 1000 s group, 65 in the ALA 500 s group, 67 in the vehicle 1000 s group and 66 in the vehicle 500 s group, mean age 20.1 years, inclusion criterion 12 > years, with moderate and severe acne (IGA score 3 and 4, with at least 20 ILs); FPT I-V20% ALA (45 min incubation) plus blue 1000 s light vs 20% ALA (45 min incubation) plus 500 s blue light vs vehicle (45 min incubation) plus blue 1000 s light vs vehicle (45 min incubation) plus 500 s blue light; in a parallel-group trial; up to 4 treatments at 3 weeks intervals, assessed 3 and 6 weeks after the final treatmentIGA was used for evaluation (0; clear skin with no ILs or NILs; almost clear; rare NILs with no more than a few small ILs; mild; > grade 1; some NILs with some ILs (papules/pustules only; no nodules); moderate; > grade 2; up to many NILs and a moderate number of ILs but no more than one small nodule; severe; > grade 3; up to many NILs and ILs, but no more than a few nodules); success was defined as a 2 point or more improvement on the IGA scale since baseline. At 3 weeks after final treatment there were 13/68 of 'success' outcomes in ALA 1000 s and 11/65 in the ALA 500 s group. The difference between ALA 1000 s and ALA 500 s groups was non significant, with RR (95% CI) 1.13 (0.55, 2.34), P = 0.33. At 6 weeks after final treatment there were 15/68 of 'success' outcomes in ALA 1000 s and 11/65 in the ALA 500 s group. The difference between ALA 1000 s and ALA 500 s groups was non significant, with RR (95% CI) 1.30 (0.65, 2.62), P = 0.74
Taub 200722 recruited, 19 participated, mean ± SD aged 26.5 ± 9.1 years, 7 M, 12 F, with moderate-severe acne and > 10 inflammatory acne lesions; FPT not givenComparison of PDT with different light sources for activation: ALA activated by IPL (600–850 nm), or a combination of IPL (580–980 nm) and bipolar radiofrequency energies, or blue light (417 nm) in a parallel-group trial; 3 treatments at 2-week intervals; follow up at 1 and 3 months after final treatmentMedian acne grade score (96.9% CI) at baseline, and 1 month after treatment were 2.75 (2.5-4.0) and 1.5 (1.0-2.5) in the IPL group, 2.5 (2.0-4.0) and 2.25 (1.5-3.5) in the IPL-RF group and 3.25 (2.5-3.5) and 1.50 (1.0-3.5) in the blue-light group. At 3 months after treatment median acne grade score (range) was 1.75 (1.5) in the IPL group, 1.5 (2) in the IPL-RF group and 2.00 (1) in the blue-light group. Investigator-assessed improvement was highest with IPL activation and lowest with blue light, and the differences between groups reached borderline statistical significance at 3 months (P = 0.0498). At 1 month after treatment median percentage improvement score (96.9% CI) was 56.25 (27.5-85.0) in the IPL group, 23.75 (2.5-85.0) in the IPL-RF group and 20 (0-62.5) in the blue-light group. At 3 months after treatment median percentage improvement score (range) was 72.5 (42.5) in the IPL group, 50 (47.5) in the IPL-RF group and 25 (40) in the blue-light group
Yin 2010180 (83 M, 97 F), aged 18-38 years, mean 25.8, with moderate-severe facial acne (Pillsbury), FPT III-IV, 45 participants in each group633 ± 3 nm (red light) plus different ALA concentrations (5%, 10%, 15% and 20%) vs red light alone, 4 treatments every 10 days, 4 parallel groups, each treated with a different concentration on the right side and placebo agent on the left side; assessments at 2, 4, 12 and 24 weeks last after treatmentAssessed by a grading scale that was defined as -3 for > 50% exacerbation, -2 for 25% to 50% exacerbation, -1 for 1% to 25% exacerbation, 0 if unchanged, 1 for 1% to 25% improvement, 2 for 25% to 50% improvement, 3 for 50% to 75% improvement, 4 for 75% to 99% improvement, and 5 for 100% improvement, compared with baseline. Significant difference among the different ALA concentration groups (P values not given), with a clear positive correlation between global improvement score and ALA concentration (P < 0.05). Further data were expressed in graph format. Our interpretation of the graph was that there were mean improvements (SE) of 3.9 (0.2), 4 (0.5), 3.9 (0.5), 3 (1) and 1.9 (1.5) in the 20% ALA group, 15% ALA group, 10% ALA group, 5% ALA group and control face sides respectively at 24 weeks after last treatment
6. MAL-PDT versus ALA-PDT
Wiegell 2006a15 participants > 18 years but age range not given, with > 12 inflammatory acne lesions; FPT not statedComparison of MAL and ALA creams: 620 nm light with split-face design; one full-face PDT treatment with MAL on one side and ALA on the other, assessed at 6 and 12 weeks after treatmentThe median of Leeds revised acne global severity grade reduced from 2 before treatment to 1 at 12-week follow-up in both the MAL-PDT- and ALA-PDT-treated sides of the face. Thre were no significant differences between the two treatments (P = 0.250)
7. Other (non-MAL, non-ALA) PDT versus other comparators
Indocyanine green-PDT
Kim 200916 (7 M, 9 F, aged 16-34 years, mean age 25 ± 3.09) with mild-moderate acne, skin types not given, 9 in single, 7 in multiple treatment group2 groups randomised: single treatment vs multiple (once weekly over 3 weeks); right cheek of each participant indocyanine green plus 805 nm light, left cheek light only and forehead "spontaneous resolution" control, evaluated 2 and 4 weeks after final treatment, multiple group also at final treatmentSignificant improvement in Cunliffe acne severity score in both groups at 2 and 4 weeks after final treatment (P < 0.05). Not reported whether there were differences between the 2 groups. Part of the results reported in graph format. Our interpretation of the graph was that Cunliffe grade reduced from baseline 3.8 to 2.5 on the single-treatment sides, and from baseline 3.5 to 2.1 on the multiple-treatment sides respectively at final evaluation. Results not reported for sides treated only with light
Topical liposomal methylene blue-PDT
Fadel 200920 (M/F not stated), age not stated (> 18 years), with mild-moderate acne, FPT not givenTopical liposomal methylene blue plus 650 nm light vs no treatment in a split-face trial, 2 treatments in total, weekly, assessed every 2 weeks for 3 months after treatmentAt 12 weeks median Leeds severity grade on the treated side was 1 (range 0-2) and on the untreated side 3 (range 2-4). No baseline data given. At 12 weeks 7/13 (54%) participants had marked, 4/13 (31%) participants had moderate and 2/13 (15%) participants had slight improvement. "Approximately the same improvements" after 4 weeks and 8 weeks. Control areas reported to have no change or worsening of acne with no details provided
Chlorophyll-a (CHA)-PDT
Song 201424 (14 M, 10 F), mean age 23.4 ± 3.5 years; range 18-32 years, "acne on both sides of the face", Cunliffe grades 2-4, FPT III-IV430 plus 660 nm light combined with chlorophyll-a (CHA) vs 430 plus 660 nm light alone in a split-face trial, 8 treatments in total, twice weekly, final assessment 2 weeks after last treatment2 weeks after final treatment Cunliffe grade reduced from baseline 3.1 to 1.8 on the CHA plus light sides and from baseline 3.1 to 2.2 on the light-only sides (P = 0.027). Further data were not given
Gold microparticle PDT versus other comparators
Paithankar 201551 (14 M, 37 F), mean age 21.4 years, age range 16-26 years, IGA scores 3–4 with at least 25 total papules and pustules on face, FPT I-IIIGold microparticle suspension plus light (details not given) vs microparticle suspension vehicle (without light-absorbing particles) plus light (details not given) in a parallel-group trial, 3 treatments in total, weekly, assessed at 6, 10 and 14 weeks after final treatmentAt 10 weeks after final treatment, "40% of subjects in the treatment arm, whereas none in the sham arm, showed Investigator’s Global Assessment (IGA) score reduction in two or higher". Further data were not given

We tried to present our analyses in numerical order but this was not always possible due to the nature of the many comparisons and our desire to present the outcomes for a particular comparison together.

We identified three studies of 80 mg/g MAL-PDT in combination with red light compared with red light alone (NCT00594425; NCT00933543; Pariser 2013) where no substantial statistical heterogeneity existed for primary outcomes (I² statistic was 39% for change in inflamed lesions (ILs), 19% for percentage change in ILs, 11% for change in non-inflamed lesions (NILs), and 35% for percentage change in NILs), and although there was some clinical heterogeneity, we synthesised data using meta-analysis techniques. We have presented the results in Summary of findings 2. We have not performed subgroup analyses because the I² statistic was lower than 50%, the threshold defined in our protocol.

In the following section, we provide details on why pooling data was not possible for each outcome and intervention subgroup, together with a narrative synthesis of the effects of interventions for individual studies where appropriate.

Primary outcome 1: Participant's global assessment of improvement

We have presented the details of participants, interventions, and the effects of interventions for this outcome in Table 1.

1. Participant's global assessment of improvement: 1. Light versus placebo or no treatment
1.1.a. Green light versus placebo

One split-face study, Baugh 2005, of four treatments included 18 participants (FPT I to III, with mild to moderate acne). A non-standardised scale (overall treatment satisfaction in intervals of 10 percentage points) was used for evaluation. At four weeks, 4.8% of participants reported 30% to 39% satisfaction, 9.5% reported 50% to 59% satisfaction, 23.8% reported 60% to 69% satisfaction, 47.6% reported 70% to 79% satisfaction, 9.5% reported 80% to 89% satisfaction, and 4.8% reported 90% to 100% satisfaction. Further data were not provided.

1. Participant's global assessment of improvement: 1. Light versus placebo or no treatment
1.1.c. Infrared light versus no treatment

One split-face study, Darne 2011, of three treatments randomised 38 participants (FPT I to V, with moderate to severe or mild but treatment-resistant acne). A non-standardised scale ('highly satisfied', 'satisfied', 'neutral' or 'unsatisfied' and 'would recommend to a friend') was used for evaluation. At four weeks, 6/25 (24%) of participants were 'highly satisfied', 9/25 (36%) were 'satisfied', 6/25 (24%) were 'neutral', and 4/25 (16%) reported the treatment to be 'unsatisfactory'. A total of 21/25 (84%) reported that they would "recommend the treatment to a friend".

A split-face study of four treatments, Moneib 2014, randomised 24 participants (FPT II to V, with moderate to severe acne). A non-standardised scale (0 = no improvement; less than 25% = mild improvement; 26% to 50% = moderate improvement; 51% to 75% = good improvement; 76% to 100% = excellent improvement) was used for evaluation. Results were reported at an unclear time point, in graph format, and for treated face sides only. Our interpretation of the graph was that 5% of participants assessed their improvement to be mild, 5% to be moderate, 20% to be good, and 70% to be excellent.

Another split-face study of three treatments, Orringer 2007, randomised 46 participants (FPT II to VI, with clinically active facial acne). A non-standardised scale (details not given) was used for evaluation. At final treatment, 29/37 of participants who completed the treatments (78%) "indicated that their acne was at least mildly improved on the treated side of the face as compared with baseline", and 16/37 participants (43%) indicated "moderate or better" improvement. Data for non-treated sides were not given, but 22/37 (59%) of participants reported that "their acne had improved at least mildly when compared with the untreated skin".

1. Participant's global assessment of improvement: 1. Light versus placebo or no treatment
1.1.e. Red light versus no treatment  

One split-face study of 122 self-administered treatments (twice daily for eight weeks) randomised 30 participants (FPT not reported, with mild to moderate acne) (Na 2007). Visual analogue scale (VAS) (0 to 5, none to very severe) was used for evaluation. Score (unclear whether mean or median) decreased from baseline 3.9 to 1.8 at final treatment on the treated and from 3.9 to 2.9 on the control side, respectively, with significant difference between the sides (P < 0.005). The study did not evaluate this outcome after final treatment, and no further data were provided.

1. Participant's global assessment of improvement: 1. Light versus placebo or no treatment
1.1.f. Blue-red light versus placebo

Two parallel-group studies, Kwon 2013; Papageorgieu 2000, included this comparison for this outcome but we were unable to pool data due to substantial methodological heterogeneity (84 versus 56 treatments, different scales and timings of outcome assessment).

Kwon 2013, with 56 treatments, randomised 18 patients to the blue-red light group and 17 to the placebo group (FPT III to V, with mild to moderate acne). A VAS scale was used for evaluation (10 = same as before the first treatment; 0 = no acne). Mean VAS score 10 at baseline in both groups decreased to 4.3 in the blue-red light group and stayed at 10 or above in the placebo group (extracted from graph) at eight weeks after final treatment. No further data (standard deviations (SDs)) were provided in text or in graph format.

Papageorgieu 2000, with 84 treatments, randomised 30 participants to the blue-red light group and 25 to the white light group (FPTs not reported, all with mild to moderate acne). A non-standardised scale ('worse', -10% or less; 'unchanged', -9% to 9%; 'mild improvement', 10% to 39%; 'moderate improvement', 40% to 59%; 'marked improvement', 60% to 89%; or 'clearance', 90% or above) was used for evaluation. At final treatment the assessments were "in favour of blue-red light", but reported only in graph format, and no details were provided. Final evaluation was performed at final treatment. We extracted the data from the graph and dichotomised them to 27/30 of 'success' outcomes in the blue-red and 7/25 in the white light group. Blue-red light was superior to white light with RR 3.21, 95% CI 1.70 to 6.09, P = 0.0003 (Analysis 1.1), and the 'number needed to treat for an additional beneficial outcome' (NNTB) was 2 (95% CI 1 to 3).

1. Participant's global assessment of improvement: 2. Light versus topical treatment
1.2.a Light versus benzoyl peroxide (BPO)

One split-face, Chang 2007, and one parallel-group study, Papageorgieu 2000, included this outcome for this comparison, so we did not perform quantitative synthesis. Light interventions had different light sources, numbers, and frequency of sessions. Timing of outcome assessment was also different.

Chang 2007 compared a combination of BPO and three sessions of 530 nm to 750 nm light with BPO alone and included 30 women (FPT III to IV, with mild to moderate acne). A non-standardised scale (highly satisfied, satisfied, neutral, or dissatisfied) was used for evaluation. At three weeks participants were "uniformly satisfied with their treatment, but intense pulsed light (IPL) treatment did not give any additional benefit". No further data were reported.

Papageorgieu 2000 randomised 30 participants to the blue-red light group and 25 to the BPO group (FPTs not reported, all with mild to moderate acne). A non-standardised scale was used for evaluation (please see above) and reported in graph format only. We extracted the data from the graph and dichotomised them to 27/30 of 'success' outcomes in the blue-red and 20/25 in the BPO group. The difference was non significant, RR 1.13, 95% CI 0.89 to 1.42), P = 0.31 (Analysis 2.1).

1. Participant's global assessment of improvement: 2. Light versus topical treatment
1.2.b. Light versus clindamycin

One split-face study (Lee 2010) compared eight full-spectrum light treatments to 1% clindamycin twice daily over four weeks and randomised nine participants (FPT III, with moderate to severe acne). A non-standardised scale ('worse', 'no change', 'fair', 'good', and 'excellent') was used for evaluation. Participants rated the treatment as 'good' or 'excellent' (it is unclear for which intervention and at what time point). Further data were not reported.

1. Participant's global assessment of improvement: 2. Light versus topical treatment
1.2.c. Light and other topical treatments

One parallel-group study (Ash 2015 ), randomised 26 participants to the blue-light group (28 sessions in total) and 15 to the control group with unclear (probably topical treatment) intervention (FPTs I-V, all with mild to moderate acne). A non-standardised scale was used for evaluation. Results reported as "the majority of subjects reporting that they were satisfied, very satisfied, or extremely satisfied with treatment" in the blue-light group, and not reported for the control group. No further data were reported nor supplied upon request.

1. Participant's global assessment of improvement: 3. Light versus other comparators
1.3.a. Comparison of light therapies of different wavelengths

Two split-face studies (Choi 2010; Jung 2009) and two parallel-group studies (Liu 2011; Papageorgieu 2000) included this comparison for this outcome, but we were unable to pool data due to substantial methodological heterogeneity (different wavelengths used as comparators, different number of sessions, and different evaluation scales).

Choi 2010 compared three sessions of 585 nm pulsed-dye lasers (PDL) with combined 585/1064 nm PDL and included 20 participants (FPT III to V, with mild to moderate acne). A non-standardised rating scale (from 0 to 10, neutral to highly satisfied) was used for evaluation. No statistically significant difference in improvement of scores between the two treatments (P > 0.05) was found. They increased from baseline 0 for both to 3.3 for IPL and 3.7 for PDL at four weeks after treatment and then to 4.7 for IPL and 5.2 for PDL at eight weeks after treatment. Further data were not reported.

Jung 2009 compared three sessions of 585 nm PDL with combined 585/1064 nm PDL and included 18 participants (FPT not reported, with mild to moderate acne). A VAS (0 to 10, worst imaginable acne state to disease free) was used for evaluation; please note that the opposite VAS was used in Jung 2012. Mean scores on the PDL sides and on the 585/1064 nm-laser sides increased from 3.3 and 3.7 at baseline to 6.63 (P = 0.002) and 6.60 (P = 0.001) at eight weeks respectively. At 12 weeks, they declined to 6.12 at both sides. Further data were not reported.

Liu 2011 included results for 20 participants (FPTs III-IV, all with mild to moderate acne) who completed the trial of eight sessions of blue light in one group (405 ± 10 nm, power of 30 mW/cm²) and red light (630 ± 10 nm, power of 48 mW/cm²) in the other group. A non-standardised scale was used for evaluation. Results were reported as, "A few patients reported that fresh new acne lesions came out, while the total number of lesions decreased slightly". Further data were not reported.

Papageorgieu 2000 randomised 30 participants to the blue-red light group and 27 to the blue-light group (FPTs not reported, all with mild to moderate acne). A non-standardised scale was used for evaluation (please see above) and reported in graph format only. We extracted the data from the graph and dichotomised them to 27/30 of 'success' outcomes in the blue-red and 23/27 in the blue-light group. The difference was non significant, RR 1.06, 95% CI 0.87 to 1.29, P = 0.59 (Analysis 3.1).

1. Participant's global assessment of improvement: 3. Light versus other comparators
1.3.b. Comparison of light therapies of different doses

Two split-face trials (Bernstein 2007; Jih 2006) compared different numbers of sessions, passes and doses of 1450 nm lasers, in participants with different FPT and different timings of outcome assessment, so we did not perform a meta-analysis.

Bernstein 2007 compared four sessions of 1450 nm laser treatments; single-pass, high-energy (13 to 14 J/cm²) versus double-pass, low-energy (8 to 11 J/cm²) and included seven participants (all with active papular acne, FPT I to III). A non-standardised rating scale (0 = worsening, 1 = no change, 2 = mild improvement , 3 = moderate improvement, 4 = marked improvement) was used for evaluation. At eight weeks, the average score on the single-pass side was 2.3 (range 1 to 4) and on the double-pass side 2.3 (range 2 to 4).

Jih 2006 also compared three sessions of 14 J/cm² and 16 J/cm² 1450 nm laser and included 20 participants (all with active inflammatory facial acne, FPT II to VI). A non-standardised rating scale (0 = worsening, 1 = no change, 2 = mild improvement , 3 = moderate improvement, 4 = marked improvement) was used for evaluation. The majority of participants reported moderate to marked improvement, 85.3% at the one-month, 67.7% at the three-month, 60.0% at the six-month, and 82.1% at the 12-month assessments. No separate data for different doses were reported.

One parallel-group trial (NCT00706433) compared four interventions:

  1. 20% ALA (45 min incubation) plus 1000 s of blue light;

  2. 20% ALA (45 min incubation) plus 500 s of blue light;

  3. vehicle (45 min incubation) plus 1000 s of blue light; and

  4. vehicle (45 min incubation) plus 500 s of blue light.

The study included a total of 266 participants (FPT I-VI, with moderate to severe acne, IGA score 3 and 4, with at least 20 ILs); 67 in the vehicle-1000 s group and 66 in the vehicle-500 s group. A non-standardised scale ('subject satisfaction score'; excellent = very satisfied; good = moderately satisfied; fair = slightly satisfied; poor = not satisfied at all) was used for evaluation. At six weeks after final treatment 20/67 participants in the vehicle-1000 s and 23/66 in the vehicle-500 s group assessed their improvement as 'good'; 23/67 participants in the vehicle-1000 s and 26/66 in the vehicle-500 s group assessed their improvement as 'excellent'. We dichotomised the data to 43/67 of 'success' outcomes in the vehicle-1000 s and 49/66 in the vehicle-500 s group. The difference between vehicle-1000 s blue light and vehicle-500 s blue light groups was non significant, with RR 0.86, 95% CI 0.69 to 1.09, P = 0.21 (Analysis 4.1).

1. Participant's global assessment of improvement: 3. Light versus other comparators
1.3.e. Light in combination with carbon lotion versus no treatment

One split-face study (Jung 2012) compared three sessions of quasi-long pulse and Q-switched 1064 nm Nd:YAG laser plus carbon lotion with non-treated control and included 22 participants (FPT III to V, unclear severity). A VAS (0 to 10, disease-free to initial visit acne status) was used for evaluation (please note that the opposite VAS was used in Jung 2009). At four weeks after final treatment, participants assessed significantly greater improvement on the laser-treated side compared with the untreated side (P < 0.05). The VAS score mean (SDs not given) decreased from an initial 10 at both sides to 5.9 (P < 0.001) on the laser-treated side and to 9.2 (P = 0.007) on the untreated side.

1. Participant's global assessment of improvement: 4. MAL-PDT versus other comparators
1.4.b. MAL-PDT versus yellow light alone

One split-face study (Haedersdal 2008) compared three sessions of 595 nm long-pulsed dye laser (LPDL) plus methyl aminolevulinate (MAL) with LPDL only and included 15 participants (FPT I to III, with at least 12 facial ILs). A non-standardised numerical scale (0 to 10, no satisfaction to best imaginable satisfaction) was used for evaluation. Median (25 to 75 percentiles) score (range) was significantly higher for MAL-LPDL treatment than for LPDL treatment alone at both four weeks after final treatment (P = 0.031); 7 (4.75 to 8) versus 6 (3.75 to 8), and at 12 weeks after final treatment (P = 0.034); 8 (6.25 to 9) versus 7.5 (5 to 8.75).

1. Participant's global assessment of improvement: 4. MAL-PDT versus other comparators
1.4.c. MAL-PDT versus placebo or no treatment

A parallel-group study (Wiegell 2006b) of two treatments of 630 nm plus 160 mg/g MAL included 21 participants in the treatment group and 15 in the control group (FPT II to V, with at least 12 facial ILs). A non-standardised grading scale (0 to 4; acne worse, no change, slight improvement, moderate improvement, marked improvement) was used for evaluation. Results were reported in graph format, and no details were provided. Our interpretation of the graph was that at 4, 8, and 12 weeks after final treatment, median improvement scores were 3, 2, and 3 in the MAL-PDT group and 1.5, 1, and 1 in the control group respectively.

1. Participant's global assessment of improvement: 4. MAL-PDT versus other comparators
1.4.d. MAL-PDT other

One split-face study (Hong 2013) compared three sessions of 160 mg/g MAL plus red light with three sessions of MAL plus IPL and included 22 participants (FPT IV to V). The VAS scale (10 to 0, 10 = same as before the first treatment; 0 = no acne) was used for evaluation. Mean VAS score decreased from baseline 10 on both sides to 5.0 at the red light side, and 4.9 at the IPL side at four weeks after final treatment, with no significant difference between the two sides. Further data were not provided.

1. Participant's global assessment of improvement: 5. ALA-PDT versus other comparators
1.5.b. ALA-PDT versus blue light alone

One parallel-group trial (NCT00706433) compared four interventions:

  1. 20% ALA (45 min incubation) plus 1000 s of blue light;

  2. 20% ALA (45 min incubation) plus 500 s of blue light;

  3. vehicle (45 min incubation) plus 1000 s of blue light; and

  4. vehicle (45 min incubation) plus 500 s of blue light.

The study included a total of 266 participants (FPT I-VI, with moderate to severe acne, IGA score 3 and 4, with at least 20 ILs); 68 in the ALA-1000 s group, 65 in the ALA-500 s group, 67 in the vehicle-1000 s group and 66 in the vehicle-500 s group. A non-standardised scale ('subject satisfaction score'; excellent = very satisfied; good = moderately satisfied; fair = slightly satisfied; poor = not satisfied at all) was used for evaluation. At six weeks after final treatment 18/68 participants in ALA-1000 s, 28/65 in the ALA-500 s, 20/67 in the vehicle-1000 s and 23/66 in the vehicle-500 s group assessed their improvement as 'good'; 23/68 participants in ALA-1000 s, 11/65 in the ALA-500 s, 23/67 in the vehicle-1000 s and 26/66 in the vehicle-500 s group assessed their improvement as 'excellent'. We dichotomised the data to 41/68 of 'success' outcomes in ALA-1000 s, 39/65 in the ALA-500 s, 43/67 in the vehicle-1000 s and 49/66 in the vehicle-500 s group. The difference between ALA-1000 s and vehicle-1000 s groups was non significant, with RR 0.94, 95% CI 0.72 to 1.22, P = 0.64 (Analysis 5.1), and it was non significant between ALA-500 s and vehicle-500 s groups, with RR 0.81, 95% CI 0.63 to 1.03, P = 0.09 (Analysis 5.1). The difference between ALA-PDT and vehicle plus blue light was non significant when we combined results for the 1000 s and 500 s subgroups using a random-effects model, with RR 0.87, 95% CI 0.72 to 1.04, P = 0.12 (Analysis 5.1). See Summary of findings 3 where we rated the evidence for this outcome as low quality.

1. Participant's global assessment of improvement: 5. ALA-PDT versus other comparators
1.5.d ALA-PDT versus IPL alone

One split-face study (Oh 2009) and one parallel-group study (Ragab 2014) included this comparison for this outcome. We were unable to combine their results due to methodological (different outcome assessment methods) and clinical differences (including numbers of treatment, application intervals, wavelengths used, incubation times).

Oh 2009 compared three sessions of 20% aminolevulinic acid (ALA) plus IPL (one face side randomised to either 30 minutes' or three hours' incubation) with IPL-only and included 20 participants (FPT III to IV, with moderate to severe acne). A non-standardised scale (significant improvement (over 75%), moderate improvement (50% to 75%), mild improvement (25% to 50%), no improvement (0% to 25%), worse (less than 0%) relative to baseline) was used for evaluation. We dichotomised the data to 3/9 of 'success' outcomes in the short-incubation and 7/11 in the long-incubation group. The difference was non significant, with RR 0.52, 95% CI 0.19 to 1.46, P = 0.22 (Analysis 6.1). Results were not reported for IPL-only sides.

Ragab 2014 (FPT III to V, with mild to moderate facial acne) compared two treatments of 20% ALA-PDT plus IPL (15 participants randomised) with IPL alone (10 participants randomised). A non-standardised scale (marked improvement = 3; moderate improvement = 2; no change = 1; acne worsened = 0) was used for evaluation. We dichotomised the data at eight weeks to 10/15 'success' outcomes in the ALA-PDT group and 3/10 in the IPL alone group. The difference was non significant, with RR 2.22, 95% CI 0.81 to 6.11, P = 0.12. (Analysis 7.1).

1. Participant's global assessment of improvement: 5. ALA-PDT versus other comparators
1.5.g. ALA-PDT other

Three parallel-group studies (NCT00706433; Taub 2007; Yin 2010) included this comparison for this outcome but we were unable to pool data due to substantial methodological heterogeneity (different number of treatments, different ALA concentrations, different light wavelengths used for activation). Methods (scales and timings of outcome assessment) were unclear in one study, and we were unable to obtain additional data and clarification.

One parallel-group trial (NCT00706433) compared four interventions:

  1. 20% ALA (45 min incubation) plus 1000 s of blue light;

  2. 20% ALA (45 min incubation) plus 500 s of blue light;

  3. vehicle (45 min incubation) plus 1000 s of blue light; and

  4. vehicle (45 min incubation) plus 500 s of blue light.

The study included a total of 266 participants (FPT I-VI, with moderate to severe acne, IGA score 3 and 4, with at least 20 ILs); 68 in the ALA 1000 s group, 65 in the ALA 500 s group, 67 in the vehicle 1000 s group and 66 in the vehicle 500 s group. A non-standardised scale was used for evaluation (please see above). We dichotomised the data to 41/68 of 'success' outcomes in ALA-1000 s and 39/65 in the ALA-500 s group. The difference between ALA-1000 s and ALA-500 s groups was non significant, with RR 1.00, 95% CI 0.76 to 1.33, P = 0.97 (Analysis 8.1).

Taub 2007 compared three ALA-PDT treatments with different light sources for activation: IPL (600 nm to 850 nm) versus a combination of IPL (580 nm to 980 nm) and bipolar radiofrequency (RF) energies versus blue light (417 nm) and included 19 participants (FPT II to IV, with > 10 facial ILs, moderate to severe acne). The method used for evaluation was unclear. One month after the treatments, differences among the groups were not statistically significant (P = 0.3210); the median percentage improvement score was 58.75 (96.9% CI 5 to 70) in the IPL group, 20 (96.9% CI 0 to 80) in the IPL-RF group, and 15 (96.9% CI 0 to 87.5) in the blue-light group. At three months, data were only reported for IPL and blue-light-only groups 72.3 (range 42.5) versus 15 (range 27.5), so analysis was not possible.

Yin 2010 compared four red light ALA-PDT treatments with different ALA concentrations: 5% versus 10%, versus 15% versus 20% and included 180 participants (FPT III to IV, with moderate to severe acne). A non-standardised scale ('marked improvement', 'moderate improvement', 'no charge', or 'acne worse') was used for evaluation. At 24 weeks after treatment, a majority of the participants assessed that their acne had improved on both the ALA-PDT-treated and control cheeks. We dichotomised the data to 44/45 'success' outcomes in the 20% ALA group, 42/45 in the 15% ALA group, 36/45 in the 10% ALA group, and 30/45 in the 5% ALA group. 20% ALA was not superior to 15% ALA with RR 1.05, 95% CI 0.96 to 1.15 and P = 0.3 (Analysis 9.1). However, 20% ALA was more effective than 10% ALA with RR 1.22, 95% CI 1.05 to 1.42 and P = 0.01 (Analysis 10.1) and more effective than 5% ALA with RR 1.47, 95% CI 1.19 to 1.81 and P = 0.0004 (Analysis 11.1). The NNTB were 6 (95% CI 3 to 19) and 4 (95% CI 2 to 6) for the comparison of 20% ALA with 10% and 5% ALA, respectively. However, there was no calculable NNTB for the comparison of 20% to 15% ALA since the 95% CI for the risk difference contained zero (i.e. no effect), and this corresponded to an infinite upper 'limit' for the 95% CI for the NNTB, which indicated that there was no true boundary on how large the NNTB could be for this comparison.

1. Participant's global assessment of improvement: 6. MAL-PDT versus ALA-PDT

No studies reported results for this outcome for this comparison.

1. Participant's global assessment of improvement: 7. Other (non-MAL, non-ALA) PDT versus other comparators
1.7.a. ICG-PDT versus other comparators

One parallel-group study (Kim 2009) of a single treatment of topical application of indocyanine green (ICG) dye applied to the right cheek compared with three treatments of indocyanine green plus 805 nm light (right cheek), 805 nm light alone (left cheek), and 'spontaneous resolution' control (forehead) included 16 participants (FPTs not reported, with mild to moderate acne). A VAS score on a scale from –100 to +100 was used for evaluation; no details were reported. At both two and four weeks after final treatment, the difference between the PDT and light-only sides was statistically significant only in the multiple treatment group (P < 0.05 at all assessment time points). Further data were not reported. Our interpretation of the graph was that at four weeks after final treatment, mean VAS score was 20 for both the PDT and the light-only side in the single treatment group whereas in the multiple treatment group, mean VAS score was 50 on the light-only side and 60 on the PDT side. SDs were not presented in the graph format.

Primary outcome 2: Investigator-assessed change in lesion count: the change or percentage change from baseline in number of lesions

We have presented the details of participants, interventions, and the effects of interventions for this outcome in Table 2 for studies of light only therapies (excluding comparisons with PDT) and in Table 3 for studies of PDT (including comparisons with light-only therapies). Please note that we calculated change from baseline (absolute change) by subtracting baseline count from count assessed at a certain time point. We calculated percentage change by dividing the absolute change with baseline count and then multiplying that value by 100 to get percentages.

2. Investigator-assessed change in lesion count: 1. Light versus placebo or no treatment
2.1.b. Yellow light versus placebo or no treatment

Two studies included this comparison for this outcome, one parallel-group trial with a total of 41 participants (FPT not reported, with mild to moderate acne), which compared a single light treatment with placebo (Seaton 2003), and another split-face trial with a total of 40 participants (FPT not reported, Leeds severity greater than 2) which compared single or two light treatments with no treatment (Orringer 2004). Orringer 2004 initially randomised participants into single-treatment and two-treatment groups, with split-face design within each group. The study authors reported only combined group data and were unable to provide separate data for the groups. Seaton 2003 reported medians of lesion counts, and we were unable to obtain original data, so we were unable to combine the results in a meta-analysis. Results of the two studies were inconsistent.

Seaton 2003 found significantly greater improvement from baseline in ILs and total lesion counts in the laser-treated group than in the placebo group at 12 weeks: ILs median (interquartile range) improvement from baseline in the treatment group was 49% (30% to 75%) versus 10% (-8% to 49%) in the placebo group P = 0.024, and total lesions 53% (19% to 64%) versus 9% (-16% to 38%) in the placebo group P = 0.023. NILs median (interquartile range) improvement from baseline in the treatment group was 40% (0% to 75%) versus -13% (-42% to 23%) in the placebo group, with non significant difference between the groups (P = 0.14).

However Orringer 2004 reported non significant differences in changes in means of papules, pustules, comedones, and cysts at 12 weeks between the treated and untreated sides of the face. Our analyses using last observation carried forward (LOCF) data (n = 38) confirmed no significant differences in means between the treated and untreated sides of the face at 12 weeks: investigator-assessed change in ILs (papules) was MD -2.00, 95% CI -6.60 to 2.60, P = 0.39 (Analysis 12.1); investigator-assessed change in ILs (pustules) MD 1.00, 95% CI -0.66 to 2.66, P = 0.24 (Analysis 12.1); investigator-assessed change in NILs, MD 1.30, 95% CI -8.00 to 10.60, P = 0.78 (Analysis 12.1); and investigator-assessed change in cysts MD 0.00, 95% CI -0.76 to 0.76, P = 1.00 (Analysis 12.1).

2. Investigator-assessed change in lesion count: 1. Light versus placebo or no treatment
2.1.c. Infrared light versus no treatment

Two split-face trials (Darne 2011; Orringer 2007) of three treatments included a total of 84 participants (FPT I to VI, with mild to moderate acne). Meta-analysis was not possible for this outcome because of timings and methods of outcome assessment and because the report of one of them included only medians of lesion counts. However, both studies had consistent results.

Darne 2011 randomised 38 participants (FPT I to V, with moderate to severe or mild but treatment-resistant acne) and found similar reduction in ILs at one and 12 months on both sides; the treated sides' median was 0 (95% CI -4 to 2) and untreated sides' median was 0 (95% CI -3.7 to 0).

Orringer 2007 randomised 46 participants (FPT II to VI, with clinically active facial acne) and reported no significant differences in changes in papules, pustules, and open or closed comedones between the treated and untreated sides at week 14. Difference in changes in cyst counts was reported to be significant. Our analyses using LOCF data (n = 37, 9 participants withdrew prior to any clinical endpoint evaluation, and were not included in the analysis) confirmed no significant differences in means between treated and untreated face sides at week 14 (i.e. eight weeks after final treatment): investigator-assessed change in ILs (papules) was MD -0.54, 95% CI -3.71 to 2.63, P = 0.74 (Analysis 13.1); investigator-assessed change in ILs (pustules) MD -0.73, 95% CI -4.37 to 2.91, P = 0.69 (Analysis 13.1); investigator-assessed change in NILs (open comedones) MD -2.92, 95% CI -8.13 to 2.29, P = 0.27 (Analysis 13.1); investigator-assessed change in NILs (closed comedones) MD -6.95, 95% CI -23.07 to 9.17, P = 0.40 (Analysis 13.1). The difference in means for investigator-assessed change in cysts was significant, favouring infrared light (MD -0.43, 95% CI -0.80 to -0.06, P = 0.02) (Analysis 13.2).

Another smaller split-face trial (Moneib 2014) of four treatments, randomised 24 participants (FPT II to V, with moderate to severe acne), but the time point of reported assessment was unclear. Results were inconsistent with Darne 2011 and Orringer 2007 (above). On the treated sides, mean papule counts (SD) reduced from a baseline of 15.42 (14.38) to 0.88 (3.35), mean pustule counts from a baseline of 2.58 (3.32) to 0.46 (1.38), open comedones from a baseline of 4.25 (7.59) to 1.25 (3.07), closed comedones from a baseline of 1.75 (3.45) to 0.33 (1.01), and nodules from a baseline of 1.00 (1.87) to 0.08 (0.41) at 'follow-up'. On the control sides mean papule counts (SD) changed from baseline 12.83 (10.89) to 14.08 (12.93), mean pustule counts from a baseline of 3.17 (5.21) to 4.21 (7.40), open comedones from a baseline of 2.58 (3.37) to 2.88 (3.54), closed comedones from a baseline of 1.79 (3.75) to 1.21 (2.50), and nodules from a baseline of 0.92 (1.61) to 1.79 (2.00) at 'follow-up'.

2. Investigator-assessed change in lesion count: 1. Light versus placebo or no treatment
2.1.d. Blue light versus no treatment

One split-face study (Elman 2003) of eight treatments included 23 participants with mild to severe acne and unclear FPT. ILs percentage change median reduction at two, four, and eight weeks post-treatment were 59%, 61%, and 53%, respectively on treated sides (P = 0.01 at eight weeks compared with untreated sides, using McNemar test; other statistical data not provided). ILs percentage change median reduction was 30% at final treatment on untreated sides; other data were not available.

2. Investigator-assessed change in lesion count: 1. Light versus placebo or no treatment
2.1.e. Red light versus no treatment

One split-face study (Na 2007) of 122 self-administered treatments (twice daily for eight weeks) included 30 participants (FPT not reported, with mild to moderate acne). At week eight, NILs percentage change -59% on treatment sides versus 3% increase on control sides (P < 0.005), ILs percentage change was -66% on treatment side versus 74% increase in ILs on control sides (P < 0.005). Further data were not given. At four weeks after final treatment 10/25 (40%) of followed-up participants were reported to have "showed an increase in acne lesions", and at eight weeks 21/22 (95%) were reported to "have complained of acne exacerbation compared with their status during treatment period". Further data were not provided.

2. Investigator-assessed change in lesion count: 1. Light versus placebo or no treatment
2.1.f. Blue-red light versus placebo

Two parallel-group studies (Kwon 2013; Papageorgieu 2000) included this comparison for this outcome but we were unable to pool data due to substantial methodological heterogeneity (84 versus 56 treatments, different timings of outcome assessment). We were also unable to obtain additional data and clarifications.

Kwon 2013, with 56 treatments, randomised 18 participants to the blue-red light group and 17 to the placebo group (FPT III-V, with mild to moderate acne). Mean IL counts reduced from baseline 22.8 to 5.3 (by 76.7%, P < 0.01) and mean NIL counts reduced from baseline 51.2 to 23.5 (by 53.3%, P < 0.01) at eight weeks after final treatment in the blue-red light group. Mean reduction of IL and NIL counts in the placebo group was not statistically significant at eight weeks after final treatment (both P > 0.05). Results were reported as percentage improvements in graph format (means and SDs not presented).

Papageorgieu 2000, with 84 treatments, randomised 30 participants to the blue-red light group and 25 to the white light group (FPTs not reported, all with mild to moderate acne). Blue-red light was reported to be superior at all time points, differences in mean percentage improvements 50.3 (95% CI 40.1 to 60.5) for ILs and 66.5 (95% CI 56.0 to 77.0) for comedones at week 12 (final treatment).

2. Investigator-assessed change in lesion count: 2. Light versus topical treatment
2.2.a. Light versus benzoyl peroxide (BPO)

Two parallel-group trials included comparison of blue (de Arruda 2009) and blue-red light (Papageorgieu 2000) with 5% BPO. A total of 115 participants were included (FPTs not reported, with mild to moderate acne). We did not carry out meta-analysis due to differences in light wavelengths (blue versus blue-red light), number of light treatment sessions (eight versus 84), number of daily applications of BPO (single versus twice daily), different outcomes recorded and timing of their assessment. We did not combine them with results of a split-face study (Chang 2007) which compared a combination of BPO and three sessions of 530–750 nm light with BPO alone and included 30 women (FPT III-IV, with mild to moderate acne). The results of these studies were inconsistent.

de Arruda 2009, with eight treatments, randomised 60 participants (unclear FPT, Brazilian group of Acne Grade II-III) to two groups and found no statistically significant difference in decrease of means of ILs (P = 0.500) and NILs (P = 0.177) between the blue light and 5% BPO group. We calculated that at four weeks the MD in changes in NILs was 9.49, 95% CI -10.84 to 29.82; however, the MD in changes in ILs was 0 (and since the P value the study authors presented was 0.5, there are infinitely many possibilities for the standard error (SE), hence, the lack of a 95% CI provided for ILs).

Papageorgieu 2000, 84 treatments in total, randomised 30 participants to the blue-red light group and 25 to the BPO group (FPTs not reported, all with mild to moderate acne). Blue–red light was reported to be superior to BPO at week 12 (P = 0.006). Difference in mean percentage improvements at week 12 was 17.6 (95% CI 7.5 to 27.6) for IL counts and 0.9 (95% CI -9.4 to 11.3) for comedones.

Chang 2007 compared a combination of BPO and three sessions of 530–750 nm light with BPO alone and included 30 women (FPT III-IV, with mild to moderate acne and found no significant difference between IPL-treated and untreated sides of the face for changes in mean papule and pustule counts (-3.2 versus -3.1; P > 0.05). Further data were not reported.

2. Investigator-assessed change in lesion count: 2. Light versus topical treatment
2.2.b. Light versus clindamycin

One parallel-group trial (Gold 2005) compared eight sessions of 417 nm blue light with self-administered topical clindamycin and included 34 participants (FPT not reported, with mild to moderate acne). This study found that NILs & ILs counts' 'averages' (ranges) in the blue-light group were 29.4 (9 to 120) and 22.6 (16 to 34) at baseline and 21.4 (8 to 40) and 11.1 (0 to 24) four weeks after final treatment respectively. NILs & ILs counts' 'averages' (ranges) in the clindamycin group were 29 (9 to 95) and 17.4 (12 to 32) at baseline and 12 (4 to 38) and 10.4 (4 to 19) 4 weeks after final treatment respectively.

One split-face trial (Lee 2010) compared eight treatments of full-spectrum light with 1% clindamycin twice daily and included nine participants (FPT III, with moderate to severe acne).

We were unable to combine the results of these two trials quantitatively due to clinical and methodological differences and unclear reporting of timings of outcome assessment in one of the studies (Lee 2010).

2. Investigator-assessed change in lesion count: 2. Light versus topical treatment
2.2.c. Light and other topical treatments

Four parallel-group studies included this comparison, but they all had different topical treatments or combinations of topical treatments comparisons, so we did not perform a meta-analysis.

Karsai 2010 compared clindamycin 1%–benzoyl peroxide 5% hydrating gel (C/BPO) alone with C/BPO in combination with two 585 nm PDL treatments and included 89 participants (FPT I-III, with mild to moderate acne). C/BPO was applied daily over four weeks. In the C/BPO group, there was a 36.3% reduction in the number of ILs and 9.2% reduction in total lesion count four weeks after initial treatment (at final treatment). In the C/BPO plus light group, there was a 36.9% reduction in number of ILs and 9.0% reduction in total lesion count. Means and SD were reported in graph format. Our interpretation of the graph was that ILs (SD) in the C/BPO group reduced from baseline 37.5 (20) to 25 (15), and in the C/BPO plus light group from 50 (30) to 30 (25) at four weeks after initial treatment. Total lesions reduced from baseline 127.5 (70) to 115 (70) in the C/BPO group, and from 175 (105) to 150 (100) in the C/BPO plus light group. We judged further analyses would be biased due to lack of precise data, so we did not perform them.

There were three studies where details of topical treatments that were used were not specified or the control intervention was unclear.

Anyachukwu 2014 randomised 40 men (FPT unclear, Global Acne Grading System > 19) either to eight treatments of 905 nm light combined with 'self-management topical agents' (including 'antibiotic cream', 'medicated soap', 'talcum powder' or 'personal hygiene'), or to the control group, who were treated with placebo-non radiating light probe combined with 'self-management topical agents'. Mean percentage change from baseline in combined number of lesions (SD) was 54.98 (16.297) in the laser group and 17.97 (16.472) in the control group three days after final treatment. Mean percentage changes from baseline in combined number of lesions at three days after final treatment were 70.37, 61.90, 71.43, 71.43 in the laser combined with 'antibiotic cream', 'medicated soap', 'talcum powder' and 'personal hygiene' subgroups respectively. Mean percentage change from baseline in combined number of lesions at three days after final treatment were 38.71, 45.00, 10.34 and 12.50 in the placebo plus 'antibiotic cream', 'medicated soap', 'talcum powder' and 'personal hygiene' subgroups respectively. Further data were not provided.

Ash 2015 randomised 26 participants to the blue-light group (28 sessions in total) and 15 to the control group with an unclear (probably topical treatment) intervention (FPTs not reported I-V, all with mild to moderate acne). At 12 weeks (four weeks after final treatment) mean lesion counts reduced by 50.08% (P = 0.002) in the treatment group and increased by 2.45% in the control group (P = 0.0029). Further data not given nor supplied upon request.

The other study (Borhan 2014) compared three treatments of 595 nm light plus "traditional topical antibiotic medication" with "traditional topical antibiotic medication" alone. A total of 40 participants were randomised (FPT III-IV, with mild to moderate acne). At week 12 the combined number of lesions, reported as "acnes number" (SD) changed from a baseline of 25.7 (5.88) to 8.75 (2.91) in the laser combined with topical antibiotics group, and from a baseline of 25.75 (6.71) to 17.7 (5.14) in the topical antibiotics-alone group (P = 0.0001).

2. Investigator-assessed change in lesion count: 3. Light versus other comparators
2.3.a. Comparison of light therapies of different wavelengths

Four trials included different comparisons; blue and red light (Liu 2011); blue and blue-red light (Papageorgieu 2000); 585 nm pulsed dye laser (PDL) with four 530-750 nm IPL (Choi 2010) and 585 nm PDL with combined 585/1064nm PDL (Jung 2009), so we did not perform quantitative synthesis.

Papageorgieu 2000 (parallel-group trial) had 84 treatments in total and randomised 30 participants to the blue-red light group and 27 to the blue-light group (FPTs not reported, all with mild to moderate acne). There was no significant difference between the treatments in ILs at week 12 (P = 0.1), nor in comedone count (P value not given). Difference in mean percentage improvements at week 12 was 13.1 (95% CI 3.0 to 23.1) for IL counts and 12.9 (95% CI 2.5 to 23.2) for comedones.

Liu 2011 (parallel-group study) included results for 20 participants (FPTs III-IV, all with mild to moderate acne) who completed the trial of eight sessions of blue light in one group (405 ± 10 nm, power of 30 mW/cm²) and red light (630 ± 10 nm, power of 48 mW/cm²) in the other group. In the blue-light group, the mean ILs count dropped from baseline 19.2 to 5.5 (by 71.4%) at final treatment and in the red-light group from baseline 8.2 to 6.6 at final treatment (by 19.5%). SDs and further details were not given.

Choi 2010 (split-face trial) compared four sessions of 585 nm PDL with four 530-750 nm IPL sessions and included 20 participants (FPT III-V, with mild to moderate acne). Individual participant data were given in the paper (n = 17). Our analyses based on t-distributions showed that at eight weeks PDL was not superior to IPL in changes in ILs (MD 2.00, 95% CI -0.85 to 4.85, P = 0.178, t = 1.431 Analysis 14.1) nor in changes in NILs (MD 0.77, 95% CI -3.65 to 5.19, P = 0.735, t = 0.355 Analysis 14.1). Results of the analyses using t-distribution did not substantially differ from the ones in which we used normal distribution (Analysis 14.2).

Jung 2009 (split-face trial) compared three sessions of 585 nm PDL with combined 585/1064nm PDL and included 18 participants (FPT not reported, with mild to moderate acne). ILs and NILs reduced by 86% and 69% respectively on the PDL sides and by 89% and 64% on the 585/1,064-nm laser sides respectively at final evaluation (P values reported as < 0.05 "compared with baseline"). There was no significant difference in the effect of the two interventions (P values and further data not provided).

2. Investigator-assessed change in lesion count: 3. Light versus other comparators
2.3.b. Comparison of light therapies of different doses

Three split-face trials (Bernstein 2007; Jih 2006; Uebelhoer 2007) compared different numbers of sessions, passes and doses of 1450 nm lasers, in participants with different FPT and different timings of outcome assessment, so we did not perform a meta-analysis.

Bernstein 2007 compared four sessions of two 1450 nm laser doses: single-pass, high-energy (13 to 14 J/cm²) and double-pass, low-energy (8 to 11 J/cm²) and included 30 participants (FPT I-III, with mild to moderate acne). Individual participant data were given in the paper (n = 6). We found no significant difference at eight weeks, with MD -4.33, 95% CI -13.4 to 4.74, P = 0.372, t = -1.063 (Analysis 15.1). Results of the analyses using t-distribution did not substantially differ from the ones in which we used normal distribution (MD -4.33, 95% CI -12.31 to 3.65) (Analysis 15.2).

Jih 2006 compared three sessions of different (infrared) light intensities of 1450 nm diode laser: 14 J/cm² and 16 J/cm² and included 20 participants (FST II-VI, with at least 20 ILs). Sponsors provided detailed data and our analyses confirmed no significant difference in reduction between the different light intensities. The MDs in changes in ILs and percentage changes in ILs see (Analysis 16.1) were: MD -2.40, 95% CI -6.46 to 1.66, P = 0.26, t = -1.203 and MD -3.40, 95% CI -14.21 to 7.41, P = 0.54, t = - 0.641 respectively at one month; MD -7.05, 95% CI -16.05 to 1.95, P = 0.13, t = -1.596 and MD -3.20, 95% CI -7.43 to 1.03, P = 0.15, t = 1.541 respectively at three months; MD -2.00, 95% CI -5.87 to 1.87, P = 0.32, t = -1.053 and MD 2.49, 95% CI -6.37 to 11.35, P = 0.59, t = 0.572 respectively at six months; and MD -2.40, 95% CI -7.13 to 2.33, P = 0.33, t = -1.034 and MD -5.59, 95% CI -26.07 to 14.89, P = 0.60, t = -0.556 respectively at 12 months. Results of the analyses using t-distribution did not substantially differ from the ones in which we used normal distribution (Analysis 16.2).

Uebelhoer 2007 compared three sessions of single-pass with double-pass of 1450 nm laser treatment and included 11 participants (FPT not given, with at least 10 ILs on each side of the face). There was a statistically significant reduction of mean acne lesion counts on both the single-pass side and double-pass side of 57.6% (P = 0.02) and 49.8% (P = 0.02), respectively. Further details were not given.

One parallel-group trial (NCT00706433) compared four interventions:

  1. 20% ALA (45 min incubation) plus 1000 s of blue light;

  2. 20% ALA (45 min incubation) plus 500 s of blue light;

  3. vehicle (45 min incubation) plus 1000 s of blue light; and

  4. vehicle (45 min incubation) plus 500 s of blue light.

The study included a total of 266 participants (FPT I-VI, with moderate to severe acne, IGA score 3 and 4, with at least 20 ILs); 67 in the vehicle 1000 s group and 66 in the vehicle 500 s group. At three weeks after final treatment investigator-assessed median change in ILs (SD) was -19.0 (22.8) in the vehicle 1000 s and -14.5 (24.0) in the vehicle 500 s group; investigator-assessed median percentage change in ILs (SD) was -41.7 (38.82) in the vehicle 1000 s and -37.0 (40.23) in the vehicle 500 s group. At six weeks after final treatment investigator-assessed median change in ILs (SD) was -21.0 (23.63) in the vehicle 1000 s and -17.0 (26.71) in the vehicle 500 s group; investigator-assessed median percentage change in ILs (SD) was -48.4 (32.81) in the vehicle 1000 s and -45.2 (50.15) in the vehicle 500 s group. We could not perform statistical tests to determine whether any changes were significant due to the study authors’ use of median changes rather than the typical mean changes required for significance testing in order to make appropriate comparisons with other included studies. Furthermore, it is not clearly stated whether the study authors implemented an ITT analysis or a LOCF approach to handling missing data.

2. Investigator-assessed change in lesion count: 3. Light versus other comparators
2.3.d. Light alone versus combined with microdermoabrasion

One split-face trial (Wang 2006) compared four sessions of 1450 nm diode laser plus microdermoabrasion with 1450 nm diode laser therapy alone. The trial included 20 participants (FPT II-IV, with moderate to severe acne). Microdermoabrasion plus light treatment decreased the mean acne lesion count by 52.8% by six weeks and 54.4% by 12 weeks (P < 0.02 compared with baseline counts). Light treatment alone reduced the counts by 53.5% by six weeks and 61.1% by 12 weeks (P < 0.05 compared with baseline counts). There was no statistically significant difference between the two treatments at any point.

2. Investigator-assessed change in lesion count: 3. Light versus other comparators
2.3.e Light in combination with carbon lotion versus no treatment

One split-face trial (Jung 2012) compared three sessions of quasi-long pulse and Q-switched 1064 nm Nd:YAG laser plus carbon lotion with non-treated control and included 22 participants (FPT III-V, with unclear severity of acne). The difference in means of both ILs and NILs was statistically significant between treated and untreated sides (P < 0.001), but clear data for non treated sides were not given. Both ILs and NILs reduced to 58.6% (P < 0.001) and to 52.4% (P < 0.001), respectively on the laser-treated side.

2. Investigator-assessed change in lesion count: 4. MAL-PDT versus other comparators

We have presented the details of participants, studies of PDT (including comparisons with light-only therapies), and the effects of interventions for this outcome in Table 3.

2. Investigator-assessed change in lesion count: 4. MAL-PDT versus other comparators
2.4.a. MAL-PDT versus red light alone

We combined results of three parallel-group studies (NCT00594425; NCT00933543; Pariser 2013) comparing four sessions of red light plus MAL with placebo cream and red light, with a final evaluation at six weeks after the last treatment. We combined and compared two groups from these studies: 80 mg/g MAL-PDT groups (a total of 202 participants) and placebo cream groups (a total of 158 participants). The participants had FPT I-VI and moderate to severe acne. NCT00594425 had an additional group of 50 participants treated with 40 mg/g MAL-PDT whom we did not include in the meta-analysis (see below). The statistical heterogeneity across studies was not substantial, that is, the I² statistic fitted the criteria we stated in our protocol (I² statistic had to be lower than 50%). I² was 39% for change in ILs, 19% for percentage change in ILs, 11% for change in NILs and 35% for percentage change in NILs. Therefore we judged it was appropriate to combine the results. However, there was some clinical heterogeneity across studies to take into account. We have narratively summarised it here, please check Characteristics of included studies tables of each study for details.

Pariser 2013 included only people with severe acne, whilst NCT00594425 and NCT00933543 included people with both severe and moderate acne (the sponsor later provided information that less than 20% of the included participants had severe acne in those two trials). Pariser 2013 and NCT00933543 included all skin types, whilst NCT00594425 included only skin types I-IV. Occlusive dressing was used during incubation in Pariser 2013 and NCT00594425, but was not used in NCT00933543. Sponsors later clarified that investigators used the Aktilite lamp in NCT00594425, and the Nedax lamp in Pariser 2013 and NCT00933543. Both lamps produce a wavelength of 632 nm, but the illumination field is four times larger with the Nedax lamp. The angle between the LED panel and the face is also different (90° for the Aktilite and 60° for the Nedax lamp).

Meta-analysis of these three studies (n = 360), using a random-effects model, showed that MAL-PDT was not superior to red light alone for: change in ILs (MD -2.85, 95% CI -7.51 to 1.81) (Analysis 17.1); percentage change in ILs (MD -10.09, 95% CI -20.25 to 0.06) (Analysis 17.2); change in NILs (MD -2.01, 95% CI -7.07 to 3.05) (Analysis 17.3); nor for percentage change in NILs (MD -8.09, 95% CI -21.51 to 5.32). (Analysis 17.4). See Summary of findings 2 where we rated the evidence as moderate quality for these outcomes. Please note that these studies are not presented in Table 3.

NCT00594425 was a three-arm parallel-group trial, which also randomised 50 participants in the 40 mg/g MAL-PDT group (FPT I-IV, with moderate to severe acne, IGA score 3 to 4, 20 to 100 ILs and up to 200 NILs on the face). Four treatments at two-week intervals were applied; 43/50 participants completed treatment in the 40 mg/g group and 42/52 completed treatment in the placebo (vehicle cream) group. We used the data as provided by the sponsors, who used both ITT and the LOCF method to account for missing data within their analyses. Our analyses showed that at six weeks after final treatment 40 mg/g MAL-PDT was not superior to placebo cream plus red light in change in ILs (MD -3.00, 95% CI -7.76 to 1.76, P = 0.22) (Analysis 18.1), in percentage change in ILs (MD -7.90, 95% CI -22.33 to 6.53, P = 0.28) (Analysis 18.2), and in change in NILs (MD -7.50, 95% CI -16.07 to 1.07, P = 0.09) (Analysis 18.3), while there was a borderline superiority in percentage change in NILs (MD -25.80, 95% CI -51.69 to 0.09, P = 0.05) (Analysis 18.4).

Two more trials included this comparison for these outcomes, but we were unable to combine their results quantitatively because one was a split-back trial (two 8 m² x 8 cm² areas) which included only participants with FPT V-VI (NCT00673933) and the other was a split-face trial, which compared only two sessions of 635 nm light plus 160 mg/g MAL with placebo cream and light (Hörfelt 2006). Both of these studies were assessed at different time points.

NCT00673933 compared two sessions of red light plus 80 mg/g MAL with placebo cream and red light. It included a total of 20 participants (FPT V-VI, with moderate to severe acne). Our analyses based on t-distributions showed that at four weeks after final treatment MAL- PDT was not superior in changing the ILs count (MD 0.20 CI 95% -1.24 to 1.64, P = 0.79, t = 0.280) (Analysis 19.1) nor the NILs count (MD -0.45 CI 95% -2.95 to 2.05, P = 0.73, t = -0.365) (Analysis 19.1). ITT analysis results were given (n = 20). Results of the analyses using t-distribution did not substantially differ from the ones in which we used normal distribution (Analysis 19.2)

Hörfelt 2006 compared two sessions of 635 nm light plus 160 mg/g MAL with placebo cream and light. The trial included 30 participants (FPT I-III, with moderate to severe acne). MAL–PDT was reported to be significantly more effective than light alone for ILs: median percentage reduction 63% (95% CI 50% to 71%) versus 28% (95% CI 19% to 47%) at four weeks (P = 0.0004), and 54% (95% CI 35% to 64%) versus 20% (95% CI 8% to 50%) at 10 weeks (P = 0.0006). No statistically significant difference in treating NILs was observed between two interventions (open comedones P = 0.6875, closed comedones P = 1.00). The study authors used the LOCF method to account for missing data for three participants who dropped out due to adverse effects. The study authors stated that they used both ITT and LOCF, in this way, within their analyses. Study authors provided further data on changes and percentage changes in ILs. We calculated that MAL-PDT was not superior to placebo cream plus light in change in ILs at four weeks nor at 10 weeks, with MD -2.60, 95% CI -6.45 to 1.25, P = 0.19 (Analysis 20.1) and MD -2.50, 95% CI -6.59 to 1.59, P = 0.23 (Analysis 20.1) respectively. However, it was superior in percentage change in ILs at four weeks and percentage change in ILs at 10 weeks, with MD -23.90, 95% CI -39.04 to -8.76, P = 0.002 (Analysis 20.2) and MD -19.10, 95% CI -37.63 to -0.57, P = 0.04 (Analysis 20.2), respectively.

2. Investigator-assessed change in lesion count: 4. MAL-PDT versus other comparators
2.4.b. MAL-PDT versus yellow light alone

One split-face study (Haedersdal 2008) compared three sessions of 595 nm LPDL plus 160 mg/g MAL with LPDL only and included 15 participants (FPT I to III, with at least 12 facial ILs). Median percentage reduction in IL counts was significantly greater with MAL–LPDL than with LPDL alone at four weeks (70% versus 50%, P = 0.03) and 12 weeks (80% versus 67%, P = 0.004). Median percentage reduction in NILs lesions was significantly greater on the MAL–LPDL side at four weeks (P = 0.035), but the difference between the treatments (53% versus 42%) did not achieve statistical significance at final follow-up (P = 0.158). Median IL counts (25% to 75% percentiles) at baseline, four and 12 weeks were 21.0 (16 to 36), 7 (4.75 to 15) and 3.5 (2 to 9.5) on the MAL-LPDL side, and 22 (14 to 36), 10 (6.5 to 16) and 7 (2 to 9.5) on the LPDL side respectively. Median NIL counts (25% to 75% percentiles) at baseline, four and 12 weeks were 33 (26 to 41), 23 (17 to 40) and 15 (9 to 21) on the MAL-LPDL side, and 32 (25 to 41), 26 (17 to 33) and 20 (12 to 27) on the LPDL side respectively.

2. Investigator-assessed change in lesion count: 4. MAL-PDT versus other comparators
2.4.c. MAL-PDT versus placebo or no treatment

This was a parallel-group study (Wiegell 2006b) of two treatments of 630 nm plus 160 mg/g MAL which included 21 participants in the treatment group and 15 in the control group (FPT II to V, with at least 12 facial ILs). There was a significantly greater median reduction in ILs in the treatment group at eight weeks (P = 0.023) and 12 weeks (P = 0.0023). Median ILs change from baseline (range) at 12 weeks was 24 (-4 to 55) in the MAL-PDT group and 0 (-39 to 19) in the control group. Median ILs count (range) at baseline, 4, 8 and 12 weeks were 46 (13 to 99), 24 (9 to 68), 22 (8 to 83) and 14 (4 to 44) in the MAL-PDT group and 32 (13 to 99), 32 (8 to 128), 42 (9 to 109) and 40 (13 to 80) in the control group. There was a non-significant difference in median change in NILs between the MAL-PDT and control group (P = 0.90) at 12 weeks. Median NILs change from baseline (range) at 12 weeks was 6 (-15 to 18) in the MAL-PDT group and 2 (-14 to 35) in the control group. Median NILs count (range) at baseline, 4, 8 and 12 weeks were 17 (2 to 73), 22 (0 to 56), 24 (6 to 59) and 24 (9 to 74) in the MAL-PDT group and 24 (2 to 64), 19 (0 to 76), 21 (2 to 81) and 31 (5 to 59) in the control group.

2. Investigator-assessed change in lesion count: 4. MAL-PDT versus other comparators
2.4.d. MAL-PDT other

Due to substantial clinical and methodological heterogeneity of four studies with different interventions and comparators (Bissonnette 2010; Hong 2013; NCT00594425; Yeung 2007) we did not perform quantitative synthesis of their results.

Bissonnette 2010 (parallel-group trial) randomised 44 participants (FPT I to IV, with 10 or more ILs on each face side) to compare 80 mg/g MAL with or without occlusion followed by different red light intensity exposure; participants randomised in four groups with 25 J/cm² or 37 J/cm² and with or without occlusion; four treatments, assessed at four and 12 weeks after the final treatment.

ILs means changed from baseline 16.7 (95% CI 11.8 to 21.5), 16.6 (95% CI 12.6 to 20.5), 14.9 (95% CI 12.3 to 17.1) and 15.7 (95% CI 13.17 to 18.8) on the non-occluded 25 J/cm², occluded 25 J/cm², non-occluded 37 J/cm² and occluded 37 J/cm² face sides, respectively to 11.0 (95% CI 8.7 to 13.4), 9.4 (95% CI 6.3 to 12.4), 8.6 (95% CI 5.2 to 11.9) and 8.9 (95% CI 5.5 to 11.8) respectively at 12 weeks after final treatment.

NILs means changed from baseline 10.8 (95% CI 7.0 to 14.6), 11.3 (95% CI 7.9 to 14.7), 14.6 (95% CI 7.8 to 21.4) and 15.1 (95% CI 8.9 to 21.3) on the non-occluded 25 J/cm², occluded 25 J/cm², non-occluded 37 J/cm² and occluded 37 J/cm² face sides, respectively to 8.6 (95% CI 5.7 to 11.5), 7.5 (95% CI 4.9 to 10.1), 12.7 (95% CI 5.8 to 19.6) and 12.2 (95% CI 5.8 to 18.6) respectively at 12 weeks after final treatment. The number of ILs was significantly lower than baseline on all face sides except the non-occluded 25 J/cm² (based on non-overlapping 95% CI). There was no statistically significant difference in mean reduction of ILs between face sides with and without occlusion, for both 25 J/cm² and 37 J/cm². There was no statistically significant difference in NILs mean change from baseline between the treatments at 12 weeks follow-up, based on overlapping CIs. The study authors stated using both ITT and LOCF within their analyses, please see the 'Risk of bias' table of this study for details.

Hong 2013 (split-face study) compared three sessions of 160 mg/g MAL plus red light with three sessions of MAL plus IPL and included 22 participants (FPT IV to V). At four weeks after treatment, there was no statistically significant difference between red light and IPL treated sides in mean percentage reduction of ILs (69.5% versus 72.0% respectively) and NILs (43.4% versus 46.3% respectively). Further data were not provided.

NCT00594425 (three-arm parallel-group trial) randomised 48 participants to the 80 mg/g MAL-PDT arm and 50 participants to the 40 mg/g MAL-PDT arm (FPT I to IV, with moderate to severe acne, IGA score 3 to 4, 20 to 100 ILs and up to 200 NILs on the face). Four treatments at two-week intervals were applied, 37 participants completed treatment in the 80 mg/g group, and 43 completed treatment in the 40 mg/g group. Our analyses showed that at six weeks after final treatment 80 mg/g MAL-PDT was not superior to 40 mg/g MAL-PDT in change in ILs (MD 2.20, 95% CI -2.57 to 6.97, P = 0.37) (Analysis 21.1), in percentage change in ILs (MD 3.10, 95% CI -11.8 to 17.38, P = 0.67) (Analysis 21.2), in change in NILs (MD 0.6, CI 95% -6.36 to 7.56, P = 0.87) (Analysis 21.3), nor in percentage change in NILs (MD -1.7, 95% CI -20.67 to 17.27, P = 0.94) (Analysis 21.4).

Yeung 2007 30 participants (FPT IV to V, with moderate acne) used topical adapalene 0.1% gel at night and were randomised to two split-face treatment groups: 530 nm to 750 nm light plus 160 mg/g MAL versus IPL light (11 participants completed treatment) or IPL versus adapalene-only control (12 participants completed treatment). Four light treatments were applied. We performed analyses based on t-distribution and found that MAL-PDT was not superior to IPL alone in percentage change in ILs at both four weeks and at 12 weeks, with MD -30.60, 95% CI -70.37 to 9.17, P = 0.141, t = -1.567 (Analysis 22.1) and MD -41.60, 95% CI -81.90 to -1.30, P = 0.052, t = -2.103 (Analysis 22.1) respectively. However, we found a transient superior effect on percentage change in NILs at four weeks, which was lost at 12 weeks, with MD -36.10, 95% CI -60.18 to -12.02, P = 0.006, t = -3.054 (Analysis 22.1) and MD 5.60, 95% CI -29.13 to 40.33, P = 0.754, t = 0.328 (Analysis 22.1) respectively. Results of the analyses using t-distribution did not substantially differ from the ones in which we used normal distribution (Analysis 22.2).

We found no difference in effect between adapalene and MAL-PDT in percentage change in ILs at both four weeks and at 12 weeks, with MD 19.70, 95% CI -15.32 to 54.72, P = 0.283, t = 1.170 (Analysis 23.1) and MD 23.50, 95% CI -11.68, 58.68), P = 0.205, t = 1.390 (Analysis 23.1) respectively. However, MAL-PDT also had a transient superior effect to adapalene on percentage change in NILs at four weeks, which was lost at 12 weeks, with MD -37.80, 95% CI -63.97 to -11.63, P = 0.01, t = -3.005 (Analysis 23.1) and MD -53.10, 95% CI -119.64 to 13.44, P = 0.133, t = -1.660 (Analysis 23.1) respectively. Results of the analyses using t-distribution did not substantially differ from the ones in which we used normal distribution (Analysis 23.2).

2. Investigator-assessed change in lesion count: 5. ALA-PDT versus other comparators
2.5.a. ALA-PDT versus red light alone

One split-back trial (Pollock 2004) compared three sessions of 635 nm light plus 20% ALA with light alone, ALA alone and untreated control. The trial included 10 participants (FPT I to III and V, with mild to moderate acne). There was a statistically significant reduction from baseline in IL counts from the second treatment (P < 0.005) at the ALA–PDT site but not the other sites: reduction in acne was 69% at 21 days' follow-up. Further data was reported in graph format. Mean baseline IL counts were 8.3, and 11.6 respectively at the light-alone and ALA-PDT areas. At three weeks' follow-up IL counts at the light-alone and ALA-PDT areas decreased to 6.1 and 6.3 respectively. Other data were not given.

2. Investigator-assessed change in lesion count: 5. ALA-PDT versus other comparators
2.5.b. ALA-PDT versus blue light alone

One parallel-group trial (NCT00706433) compared four interventions:

  1. 20% ALA (45 min incubation) plus 1000 s of blue light;

  2. 20% ALA (45 min incubation) plus 500 s of blue light;

  3. vehicle (45 min incubation) plus 1000 s of blue light; and

  4. vehicle (45 min incubation) plus 500 s of blue light.

The study included a total of 266 participants (FPT I to VI, with moderate to severe acne, IGA score 3 and 4, with at least 20 ILs); 68 in the ALA 1000 s group, 65 in the ALA 500 s group, 67 in the vehicle 1000 s group and 66 in the vehicle 500 s group. At three weeks after final treatment investigator-assessed median change in ILs (SD) was -18.0 (26.3) in ALA 1000 s, -14.0 (26.8) in the ALA 500 s, -19.0 (22.8) in the vehicle 1000 s and -14.5 (24.0) in the vehicle 500 s group; investigator-assessed median percentage change in ILs (SD) was -37.5 (38.79) in ALA 1000 s, -29.2 (46.68) in the ALA 500 s, -41.7 (38.82) in the vehicle 1000 s and -37.0 (40.23) in the vehicle 500 s group. At six weeks after final treatment investigator-assessed median change in ILs (SD) was -18.5 (30.15) in ALA 1000 s, -13.0 (28.74) in the ALA 500 s, -21.0 (23.63) in the vehicle 1000 s and -17.0 (26.71) in the vehicle 500 s group; investigator-assessed median percentage change in ILs (SD) was -34.4 (37.8) in ALA 1000 s, -29.0 (42.57) in the ALA 500 s, -48.4 (32.81) in the vehicle 1000 s and -45.2 (50.15) in the vehicle 500 s group. Statistical tests to determine whether any changes were significant could not be performed due to the study authors’ use of median changes rather than the typical mean changes required for significance testing in order to make appropriate comparisons with other included studies. Furthermore, it was not clearly stated whether the study authors implemented an ITT analysis or a LOCF approach to handling missing data. See Summary of findings 3 where we rated the evidence as very low quality for this comparison.

2. Investigator-assessed change in lesion count: 5. ALA-PDT versus other comparators
2.5.d. ALA-PDT versus IPL alone

Three trials included this comparison, but one had a split-face design (Oh 2009) and included three treatments with different incubation times in participants with moderate to severe acne, whilst the other two were parallel-group trials, of different ALA doses, numbers of treatments, application intervals and incubation times, and included participants of different acne severity (Mei 2013, Ragab 2014). We did not combine results because of this heterogeneity.

Oh 2009 compared three sessions of 20% ALA plus IPL (one side of the face randomised to either 30 minutes' or three hours' incubation) with IPL only and included 20 participants (FPT III to IV, with moderate to severe acne). Mean reduction of ILs was 84.4% in the long-incubation time group, 72.6% in the short-incubation time group and 65.9% on the sides of the face treated with IPL alone at four weeks (P < 0.001 in all cases). Mean reduction of ILs was 89.5% in the long incubation time group, 83.0% in the short incubation time group and 74.0% for the sides of the face treated with IPL alone at 12 weeks (P < 0.001 in all cases). Mean reduction was significantly greater in the group where the sides of the face were treated for the long incubation time compared to the IPL-alone treated sides (P = 0.01). The difference was not statistically significant between short incubation and placebo-treated sides (P = 0.21). Further data were not given.

Mei 2013 (FPT II to IV, with severe acne) compared four treatments of 10% ALA plus IPL (21 participants randomised) to placebo cream plus IPL (20 participants randomised). Our analyses based on t-distribution showed that ALA-PDT was superior to light alone in percentage changes in ILs, with MD 13.80, 95% CI 1.34 to 26.26, P = 0.04, t = 2.240 (Analysis 24.1) and in percentage changes in NILs, with MD 24.10, 95% CI 4.65 to 43.55, P = 0.02, t = 2.506 (Analysis 24.1). Results of the analyses using t-distribution did not substantially differ from the ones in which we used normal distribution (Analysis 24.2).

Ragab 2014 (FPT III to V, with mild to moderate facial acne) compared two treatments of 20% ALA plus IPL (15 participants randomised) with IPL alone (10 participants randomised). Mean IL counts decreased from a baseline of 15.7 to 7.7 and 5.4 at two and eight weeks respectively in the ALA-IPL group; and from a baseline of 9.6 to 5.2 and 4.4 at two and eight weeks respectively in the IPL alone group. Mean NIL (comedones) counts decreased from a baseline of 50.9 to 36.9 and 31.3 at two and eight weeks respectively in the ALA-IPL group; and from a baseline of 41.8 to 23.8 and 24.4 at two and eight weeks respectively in the IPL alone group. Mean combined lesion counts decreased from a baseline of 66.6 to 35.7 at eight weeks in the ALA-IPL group; and from a baseline of 51.4 to 28.8 at eight weeks in the IPL alone group. SDs were not reported.

2. Investigator-assessed change in lesion count: 5. ALA-PDT versus other comparators
2.5.f. ALA-PDT versus placebo or no treatment

One split-face trial (Orringer 2010) compared three sessions of 20% ALA plus PDL with untreated control. The trial included 44 participants (all FPTs, severity of acne unclear). The study authors reported no statistically significant difference between treated and untreated control skin in papules, pustules, cysts, closed and open comedones at week 16, but there was a transient statistically significant decrease from baseline in mean papule counts on treated sides when compared with untreated sides at week 10. There was no statistically significant difference between treated and untreated control sides in all other lesion counts at week 10. Our analyses using LOCF data (n = 44) confirmed a transient statistically significant decrease from baseline in investigator-assessed change in ILs (papules) on treated sides when compared with untreated sides at week 10 of the study (i.e. four weeks after final treatment) see (Analysis 25.1), with MD -4.50, 95% CI -8.28 to -0.72, P = 0.02. We found no significant differences in means between treated and untreated sides of the face for investigator-assessed change in ILs (pustules) MD -0.60, 95% CI -5.09 to 3.89, P = 0.79, for investigator-assessed change in NILs (open comedones) MD -0.37, 95% CI -7.76 to 7.02, P = 0.92, for investigator-assessed change in NILs (closed comedones) MD -3.90, 95% CI -12.05 to 4.25, P = 0.35, and for cysts MD 0.03, 95% CI -0.53 to 0.59, P = 0.92. Our analyses also confirmed no significant differences in means between treated and untreated sides of the face at week 16 (i.e. 10 weeks after final treatment): investigator-assessed change in ILs (papules) was MD -0.82, 95% CI -6.03 to 4.39, P = 0.76; investigator-assessed change in ILs (pustules) MD -0.10, 95% CI -5.29 to 5.09, P = 0.97; investigator-assessed change in NILs (open comedones) MD 2.00, 95% CI -7.51 to 11.51, P = 0.68; investigator-assessed change in NILs (closed comedones) MD -2.90, 95% CI -10.78 to 4.98, P = 0.47; and cysts MD 0.14, 95% CI -0.66 to 0.94, P = 0.73.

One split-back trial (Pollock 2004) compared three sessions of 635 nm light plus 20% ALA with light alone, ALA alone and untreated control. The trial included 10 participants (FPT I to III and V, with mild to moderate acne). There was a statistically significant reduction from baseline in IL counts from the second treatment (P < 0.005) at the ALA–PDT site but not the other sites: reduction in acne was 69% at 21 days follow up. Further data was reported in graph format. Mean baseline IL counts were 11.6 and 10.1 respectively at the ALA-PDT and untreated control areas. At three weeks' follow-up, IL counts at the ALA-PDT and untreated control areas decreased to 3.6 and 6.3 respectively. Other data were not given.

2. Investigator-assessed change in lesion count: 5. ALA-PDT versus other comparators
2.5.g. ALA-PDT other

Due to substantial clinical and methodological heterogeneity of five studies with different interventions and comparators (Barolet 2010; NCT00706433; Pollock 2004; Taub 2007; Yin 2010) we did not perform quantitative synthesis of their results.

Barolet 2010 (split-face or split-back trial) compared a single treatment of 970 nm IR (radiant infrared) pre-treatment plus 20% ALA and 630 nm PDT with ALA-PDT alone. The trial included 10 participants (FPT I to III, with mild to moderate acne). There was a significantly greater improvement in IL medians on the IR pre-treated versus control side four weeks after treatment (P < 0.0001). Median percentage reduction (95% CI for mean, as reported) in ILs was 73% (95% CI 51% to 81%) on the IR pre-treated side versus 38% (95% CI 8% to 55%) on the control side. Further data were not provided, 95% CI reported for means, but means were not given.

One parallel-group trial (NCT00706433) compared four interventions:

  1. 20% ALA (45 min incubation) plus 1000 s of blue light;

  2. 20% ALA (45 min incubation) plus 500 s of blue light;

  3. vehicle (45 min incubation) plus 1000 s of blue light; and

  4. vehicle (45 min incubation) plus 500 s of blue light.

The study included a total of 266 participants (FPT I to VI, with moderate to severe acne, IGA score 3 and 4, with at least 20 ILs); 68 in the ALA 1000 s group, 65 in the ALA 500 s group. At three weeks after final treatment investigator-assessed median change in ILs (SD) was -18.0 (26.3) in ALA 1000 s and -14.0 (26.8) in the ALA 500 s group; investigator-assessed median percentage change in ILs (SD) was -37.5 (38.79) in ALA 1000 s group and -29.2 (46.68) in the ALA 500 s group. At six weeks after final treatment investigator-assessed median change in ILs (SD) was -18.5 (30.15) in ALA 1000 s, and -13.0 (28.74) in the ALA 500 s group; investigator-assessed median percentage change in ILs (SD) was -34.4 (37.8) in ALA 1000 s and -29.0 (42.57) in the ALA 500 s group. We could not perform statistical tests to determine whether any changes were significant due to the study authors’ use of median changes rather than the typical mean changes required for significance testing in order to make appropriate comparisons with other included studies. Furthermore, it was not clearly stated whether the study authors implemented an ITT analysis or a LOCF approach to handling missing data.

Pollock 2004 (split-back trial) compared three sessions of 635 nm light plus 20% ALA with light alone, ALA alone and untreated control. The trial included 10 participants (FPT I to III and V, with mild to moderate acne). There was a statistically significant reduction from baseline in IL counts from the second treatment (P < 0.005) at the ALA–PDT site but not the other sites: reduction in acne was 69% at 21 days' follow-up. Further data was reported in graph format. Mean baseline IL counts were 6.6 and 11.6 respectively at the ALA-alone and ALA-PDT areas. At three weeks' follow-up IL counts at the ALA alone and ALA-PDT areas decreased to 4.6 and 3.6 respectively. Other data were not given.

Taub 2007 (parallel-group trial) compared three 20% ALA-PDT treatments with different light sources for activation: IPL (600 nm to 850 nm) versus a combination of IPL (580 nm to 980 nm) and bipolar RF energies versus blue light (417 nm) and included 19 participants (FPT II to IV, with more than 10 facial ILs, moderate to severe acne). Reductions in counts were found in all three groups, with the highest in the IPL-activation group and the lowest in the blue-light group, but the difference was not statistically significant (P values not given). Median lesion count percentage reductions at one month after treatment were 76.8 (96.9% CI 12.5 to 86.4) in the IPL group, 47 (96.9% CI 8.3 to 82.2) in the IPL-RF group and 52.8 (96.9% CI -88.9 to 66.7) in the blue-light group. At three months after treatment, median lesion count percentage reduction (range, defined as "difference between the upper and lower ends of 96.9% CI, indicated when <5 data points are available") was 73.2 (72.4) in the IPL group, 41.6 (167.5%) in the IPL-RF group and -88.9 (123.3) in the blue-light group.

Yin 2010 (parallel-group trial) compared four red light ALA-PDT treatments with different ALA concentrations: 20%, 15%, 10% and 5%, and included a total of 180 participants (FPT III to IV, with moderate to severe acne). Each participant was treated with the assigned concentration on the right side and placebo agent on the left side of the face. Greater reduction in both IL and NIL counts was found at sides treated by ALA-PDT of all concentrations compared with the controls treated by red light alone at two weeks (P < 0.001), four weeks (P < 0.05), 12 weeks (P < 0.001) and 24 weeks (P < 0.001). Combined data from all follow-up visits showed more improvement in the higher-concentration ALA treatment groups than the lower-concentration groups (P < 0.01).

Means (SD) were reported in graph format only. Our interpretation of the graph was that ILs reduced from a baseline of 21 (5), 20.5 (5.5), 19 (5), 21 (5) and 20 (4) in the 20% ALA group, 15% ALA group, 10% ALA group, 5% ALA group and control face sides, respectively to 1 (0.5), 1.3 (0.5), 3.3 (1), 4 (1) and 5 (1) in the 20% ALA group, 15% ALA group, 10% ALA group, 5% ALA group and control face sides, respectively. NILs reduced from a baseline of 12.9 (4.5), 13 (3.5), 13 (4), 12.5 (3.5) and 11.5 (4) in the 20% ALA group, 15% ALA group, 10% ALA group, 5% ALA group and control face sides, respectively to 1.4 (1), 1.4 (0.5),1.5 (0.5), 2.5 (0.5) and 5.5 (1.5) in the 20% ALA group, 15% ALA group, 10% ALA group, 5% ALA group and control face sides, respectively at 24 weeks after final treatment. We judged further analyses would be biased due to lack of precise data, so we did not perform them. The study authors reported that at 24 weeks for ILs "a significant statistical difference was found in multiple comparisons between 5%, 10%, 15% and 20% ALA (P < 0.05), except between 15% and 20% ALA (P = 0.148)" and for NILs "a significant statistical difference was found in multiple comparisons between 5%, 10%, 15% and 20% ALA (P < 0.05), except for 5% ALA vs. control (P = 1.734) and 15% vs. 20% ALA (P = 0.327)."

2. Investigator-assessed change in lesion count: 6. MAL-PDT versus ALA-PDT
2.6.a. MAL-PDT versus ALA-PDT

One split-face trial (Wiegell 2006a) compared single 620 nm PDT treatments with different creams: 20% ALA versus 160 mg/g MAL. The trial included 19 participants (FPT not given, with more than 12 ILs). There were no significant differences in reductions of ILs between ALA-treated and MAL-treated sides at six weeks' (P = 0.061) and 12 weeks' (P = 0.08) follow-up. Baseline differences in IL counts (P = 0.0049). Median IL counts (inter-quartile range) at baseline, six and 12 weeks after treatment were 19 (13 to 27), 8 (6 to 14) and 8 (3 to 11) on the MAL-treated sides and 16 (11 to 22), 5 (3 to 11) and 5 (3 to 11) on the ALA treated sides respectively. There were no significant differences in reductions of NILs between ALA-treated and MAL-treated sides at six weeks' (P = 0.18) and 12 weeks' (P = 0.052) follow-up. Median NIL counts (inter-quartile range) at baseline, six and 12 weeks after treatment were 14 (6 to 22), 21 (17 to 31) and 17 (9 to 29) on the MAL-treated sides and 17 (7 to 21), 18 (13 to 29) and 20 (17 to 38) on the ALA-treated sides respectively.

2. Investigator-assessed change in lesion count: 7. Other (non-MAL, non-ALA) PDT versus other comparators
2.7.a. Indocyanine green (ICG)-PDT versus other comparators

Two parallel-group trials (Genina 2004; Kim 2009) included these comparisons, but Genina 2004 evaluated single and multiple treatments whilst Kim 2009 compared a single treatment with three treatments. We were unable to quantitatively combine the results because of different study designs and lack of data.

Genina 2004 compared single and multiple treatments with 803 nm low-intensity diode laser in combination with ICG. An area of each participant's face or back was then assigned to be treated with ICG, and the other area was used as 'control'. Twelve participants were included (FPT not given, with mild to moderate acne). IL counts improved by 23% at four weeks for the single treatment groups and by 7% for control at ICG plus light sites; 80% improvement at four weeks for the multiple treatment group versus no improvement for control. More improvement was seen in participants with severe acne.

Kim 2009 compared a single treatment with three treatments of ICG plus 805 nm light (right cheek), 805 nm light alone (left cheek) and 'spontaneous resolution' control (forehead). The study included 16 participants (FPT, with mild to moderate acne). Participants were evaluated two and four weeks after final treatment. Significant improvement was found only in the mean number of closed comedones on the PDL-treated side at all assessment periods, and on the light-only side at four weeks post-treatment when compared to 'spontaneous resolution' control (P < 0.05 in all cases). ILs improved at all sites, but non significantly (other data not given). The study did not report whether there were differences between the two groups. Further data were not given and part of the results were reported in graph format. Our interpretation of the graph was that mean counts of closed comedones reduced from a baseline of 15 to 9 on the PDT sides and from 16 to 14 on the light-only sides, respectively at final evaluation in the single treatment group, and from a baseline of 12 to 8 on the PDT sides and from 13 to 10 on the light-only sides in the multiple treatment group, respectively.

2. Investigator-assessed change in lesion count: 7. Other (non-MAL, non-ALA) PDT versus other comparators
2.7.b. Indole 3-acetic acid (IAA)-PDT versus other comparators

One split-face trial (Na 2011) compared three sessions of 520 nm green light plus IAA with green light plus placebo cream. The trial included 14 participants (FPT not reported, severity not specified). Improvement in IL counts was observed on both sides. The difference between the treatment and control groups was statistically significant from week four after final treatment (P < 0.05). Further data was not given and was reported only in graph format. Our interpretation of the graph was that mean (we were unsure that this was a measurement of the mean) IL counts reduced from baseline 16.5 to 15.2 on the control sides, and from 16.3 to 14 on the treatment sides.

2. Investigator-assessed change in lesion count: 7. Other (non-MAL, non-ALA) PDT versus other comparators
2.7.c. Topical liposomal methylene blue (TLMB)-PDT versus other comparators

One split-face trial (Fadel 2009) compared two sessions of TLMB plus 650 nm light with no treatment. The trial included 20 participants (FPT not reported, with mild to moderate acne). At four weeks IL counts decreased by 83.3% and NILs by 63.6% on the treated sides. Results for control sides were not reported in narrative form. At 12 weeks the reduction was also significant for ILs (P < 0.01) and NILs (P < 0.01). Further data were not given.

2. Investigator-assessed change in lesion count: 7. Other (non-MAL, non-ALA) PDT versus other comparators
2.7.d. Chlorophyll-a (CHA)-PDT versus other comparators

One split-face trial (Song 2014) compared 430 plus 660 nm light combined with CHA with 430 plus 660 nm light alone and included 24 participants (FPT III to IV, acne of Cunliffe grades 2 to 4). Two weeks after final treatment papule counts reduced from baseline 13.0 to 5.1 on the CHA plus light sides and from baseline 13.1 to 8.6 on the light-only sides (P = 0.030, SDs not given); pustule counts reduced from baseline 3.8 to 1.3 on the CHA plus light sides and from baseline 4.2 to 3.0 on the light-only sides (P < 0.001, precise P value not given, SDs not given ); open comedone counts reduced from baseline 9.0 to 4.2 on the CHA plus light sides and from baseline 9.1 to 6.7 on the light-only sides (P = 0.011, SDs not given ); closed comedone counts reduced from baseline 18.4 to 8.5 on the CHA plus light sides and from baseline 18.4 to 13.3 on the light-only sides (P = 0.014, SDs not given ); nodules & cysts' counts reduced from baseline 0.6 to 0.1 on the CHA plus light sides and from baseline 0.55 to 0.3 on the light-only sides (P value not given, data extracted from figure). Further data were not given.

2. Investigator-assessed change in lesion count: 7. Other (non-MAL, non-ALA) PDT versus other comparators
2.7.e. Gold microparticle PDT versus other comparators

One parallel-group trial (Paithankar 2015) compared three sessions (applied one week apart) of gold microparticle suspension plus light (details not given) with vehicle (without light-absorbing particles) plus light (details not given) control. The trial included 51 participants (FPT I to III, with IGA scores 3 to 4 with at least 25 total papules and pustules on the face). At six weeks after final treatment, the mean percentage change in inflammatory lesion count was −44.0% and −14.0% for the active treatment and sham arms, respectively. At 10 weeks after final treatment, the mean percentage change in inflammatory lesion count was −49.0% and −21.7% for the active treatment and sham arms, respectively (P = 0.015). At 14 weeks after final treatment changes were −53% and −30% for the active treatment and sham arms, respectively (P = 0.04). Other data were not given.

Primary outcome 3: Investigator-assessed severe adverse effects

We have presented the adverse effects of interventions in (Table 4). There is no separate additional table for 'Investigator-assessed severe adverse effects', but this outcome is included in Table 4 together with other adverse effects that were reported.

Adverse effects were reported as defined in MedDRA (MedDRA 2010) and coded into System Organ Classes (SOCs) in only a few studies. To report them uniformly in this review, we coded adverse effects reported in other studies using MedDRA lowest level terms (LLTs) where possible and corresponding SOCs, as prespecified in our protocol.

Most studies of light-only therapies and PDT therapies did not report blistering and there were no reports of scarring. Ten studies (two studies of infrared light, one study on intense pulsed light, two studies of 80 mg/g MAL plus red light, one study of 160 mg/g MAL plus red light, four studies of 20% ALA plus 635 nm light) reported "application site vesicle" (that is, blister; lower level term (LLT): "application site blister") as an adverse effect. Two of them were studies of infrared light (Orringer 2007; Uebelhoer 2007), one was a study of intense pulsed light (McGill 2008), three MAL-PDT activated by red light (Bissonnette 2010; Hörfelt 2006; Pariser 2013) and four of ALA-PDT (Hongcharu 2000; Orringer 2010; Taub 2007; Yin 2010). However none of these studies reported the severity adequately (number and size of blisters). Five of the ten studies that reported blistering as an adverse effect (McGill 2008; Hongcharu 2000; Orringer 2007; Orringer 2010; Taub 2007) reported that there was no long-term scarring.

We have only presented details of effects of interventions for comparisons which included at least one report of 'investigator-assessed severe adverse effect' in this section. Many studies used very different light sources and applied photosensitisers with different vehicles for variable durations which may have influenced penetration into the follicle. In addition time for and between treatments on different sites challenged comparisons as there are many more pilosebaceous follicles on the face compared to the trunk so one might expect different outcomes with such heterogeneity. All of these sources of clinical and methodological heterogeneity led us to refrain from performing a meta-analysis, as substantial bias would, indeed, be incurred, hence jeopardising the validity and reliability of any combined results.

The relative risk was unreliable for comparisons in studies which included a report of blister due to the lack of events occurring in control groups or body sites. We were unable to calculate for the same reason. We provided application site blister rates instead and calculated risk differences (RD) with 95% CI for individual studies that included reports of blisters and the comparison in which we were able to combine three studies quantitatively.

3. Investigator-assessed severe adverse effects: 1. Light versus placebo or no treatment
3.1.c. Infrared light versus no treatment

Two split-face trials of 38 participants (FPT I to V, with moderate to severe or mild but treatment resistant acne; Darne 2011) and 24 participants (FPT II to V, with mild to severe acne; Moneib 2014) reported 0% application site blisters on either the treatment or control sides.

One split-face trial of three treatments and application intervals of three weeks (Orringer 2007) randomised 46 participants (FPT II to VI, with clinically active facial acne). There were two reports of application site vesicle (LLT application site blister) on the treated side 2/46 (4.3%) and no reports on the untreated sides (0%), with RD 0.04, 95% CI -0.03 to 0.11, P = 0.23 (Analysis 13.3).

We did not combine the studies due to different laser characteristics (1450 nm laser (8-9 J/cm²) (Orringer 2007), 1320 nm Nd:YAG laser (Darne 2011) and 1550 nm Fractional Erbium Glass Laser (Moneib 2014)). There were also differences in number of treatments, and time intervals between treatments and different application intervals (four versus three weeks).

3. Investigator-assessed severe adverse effects: 1. Light versus placebo or no treatment
3.1.h. Intense pulsed light (IPL) versus no treatment

One split-face trial (McGill 2008) randomised ten participants (FPT I to II, with mild to moderate facial acne). IPL was applied, with ‘upper’ and ‘lower’ halves of face sides treated with different filters; 550 nm to 1100 nm filter (‘585 filter’), and the ‘dual band’ filter (blue light), whereas the other face half-sides served as control. Intervention on the control face sides was unclear, but it was most likely no-treatment control. Five treatments were applied at two weeks intervals. There was a report of application side blister (LLT application site blister) on the IPL sides, 1/10 (10%), reported as, "One patient developed minor blistering after the fifth treatment, which resolved without scarring. This occurred in areas where double passing treatment was carried out, and were most likely due to the second pass taking place too quickly after the first." We calculated RD 0.10, 95% CI -0.14 to 0.34, P = 0.41 (Analysis 26.1).

3. Investigator-assessed severe adverse effects: 2. Light versus topical treatment

There were no results for this outcome for this comparison.

3. Investigator-assessed severe adverse effects: 3. Light versus other comparators
3.3.b. Comparison of light therapies of different doses

One split-face trial (Uebelhoer 2007) compared three sessions of single-pass with double-pass of 1450 nm infrared laser treatment and included 11 participants (FPT not given, with at least 10 ILs on each side of the face). There was a report of application site vesicle (LLT application site blister) on the single-pass side, 1/11 (9%), reported as, "We also experienced a cryogen failure that resulted in a single blister that resolved completely with proper wound care". We calculated RD 0.09, 95% CI -0.13 to 0.31, P = 0.42 (Analysis 27.1).

We were unable to quantitatively combine this study with other studies of infrared light (such as Darne 2011, Orringer 2007 and Moneib 2014) due to substantial clinical heterogeneity in interventions and their comparators.

3. Investigator-assessed severe adverse effects: 4. MAL-PDT versus other comparators
3.4.a. MAL-PDT versus red light alone

We combined results of three parallel-group studies (NCT00594425; NCT00933543; Pariser 2013) comparing four sessions of red light plus 80 mg/g MAL with placebo cream and red light. NCT00594425 had a group of 50 participants treated with 40 mg/g MAL-PDT whom we did not include in the meta-analysis. We have presented the results in Analysis 17.5 and Summary of findings 2. We took into account different aspects of methodological, clinical and statistical heterogeneity of the combined studies when considering meta-analysis, described previously under primary outcome 2 for this comparison. Application site blister rates in the red light-only groups were 0/158 (0%) and in the MAL-PDT groups were 1/202 (0.5 %), RD 0.00, 95% CI -0.02 to 0.02, P = 0.73.

We also considered combining results of a split-back study (n = 20) on 80 mg/g MAL-PDT (NCT00673933) for this outcome, but we did not include it as only two sessions were applied, only participants of FPT V and VI were included, and the treatment was applied on the back, where there are fewer pilosebaceous follicles than on the face.

An additional split-face trial (Hörfelt 2006) (n = 30) was not included in the meta-analysis because 160 mg /g MAL was used with 635 nm light and it also included only two sessions. Sponsors provided information that there was one report of application site blister on the MAL-PDT treated sides in that study; 1/30 (30%). We found RD 0.03, 95% CI -0.05 to 0.12, P = 0.4958 (Analysis 20.3).

3. Investigator-assessed severe adverse effects: 4. MAL-PDT versus other comparators
3.4.d. MAL-PDT other

A parallel-group trial (Bissonnette 2010) which randomised 44 participants (FPT I to IV, with 10 ILs or more on each face side) to compare 80 mg/g MAL with or without occlusion followed by different red light intensity exposure. Participants were randomised in four groups with 25 J/cm² or 37 J/cm², with or without occlusion on different sides of the face. It included one report of application site blister, 1/22 (4.5%) on the occluded 37 J/cm² face sides; and 0/22 (0%) on the non-occluded 37 J/cm² face sides, 0/22 (0%) on the occluded and 0/22 (0%) on the non-occluded 22 J/cm² sides respectively. For 37 J/cm² with and 37 J/cm² without occlusion face sides, we calculated RD 0.05, 95% CI -0.07 to 0.16, P = 0.45 (Analysis 28.1).

3. Investigator-assessed severe adverse effects: 5. ALA-PDT versus other comparators
3.5.f. ALA-PDT versus placebo or no treatment

One split-face trial (Orringer 2010) compared three sessions of 20% ALA plus PDL with untreated control. The trial included 44 participants (all FPTs, severity of acne unclear). There was one report of application site vesicle (LLT application site blister); 1/44 participants (2.3%). It resolved without permanent consequences. We calculated RD 0.02, 95% CI -0.04 to 0.08, P = 0.46 (Analysis 25.2).

3. Investigator-assessed severe adverse effects: 5. ALA-PDT versus other comparators
3.5.g. ALA-PDT other

One study (Hongcharu 2000) included 22 participants (FPT I to IV, with mild to moderate acne) and randomised 11 of them to the single treatment group, and the other 11 to the multiple treatment group. Four areas on the back of each participant were treated with 20% ALA–plus 550 nm to 700 nm light; or 20% ALA alone; or 550 nm to 700 nm light alone and the fourth area served as an untreated control. There was one report 1/11 (9%) of application site vesicle (LLT application site blister) in the single treatment group on the ALA-PDT site, "...after vigorous aerobic exercise while wearing a tight outfit [a] day after treatment. This area healed without scarring in three weeks". We calculated RD 0.09, 95% CI -0.13 to 0.31, P = 0.42 (Analysis 29.1).

One parallel-group trial (Taub 2007) compared three 20% ALA-PDT treatments with different light sources for activation: IPL (600 nm to 850 nm) versus a combination of IPL (580 nm to 980 nm) and bipolar RF energies versus blue light (417 nm) and included 19 participants (FPT II-IV, with > 10 facial ILs, moderate to severe acne). There was one report of application site vesicle (LLT application site blister) in the IPL-RF group, but the numbers of participants per group were not stated, so we were unable to perform further analyses.

One parallel-group trial (Yin 2010) compared four red light (633 nm) ALA-PDT treatments with different ALA concentrations: 5%, 10%, 15% and 20% and included a total of 120 participants (FPT III to IV, with moderate to severe acne). Each participant was treated with the assigned concentration on the right side and placebo agent on the left side of the face. In the 20% ALA group there was one report 1/45 (2%) of a combination of application site erythema, application site oedema and application site vesicle (LLT application site blister); "treated with systemic glucocorticoids and resolution took place in 2 weeks, with no persistent clinical sequelae or permanent scarring". No reports of adverse effects were made for the other concentrations of ALA. We calculated RD 0.02, 95% CI -0.04 to 0.08, P = 0.46 for all three comparisons (Analysis 9.2; Analysis 10.2; Analysis 11.2).

We considered combining the results of the above ALA-PDT studies, together with one more split-back study of 10 participants (Pollock 2004), as well as one parallel-group trial of 20 participants (Oh 2009). However, we judged this was inappropriate due to substantial clinical heterogeneity including different pre- and post-treatment care which was applied, incubation times, occlusion regimens, wavelengths and doses used for activation, numbers of treatment sessions, intervals between them etc.

Secondary outcomes 1: Investigator-assessed change in acne severity; 2: Investigator's global assessment of improvement; and 3: Changes in quality of life

We have presented the details of participants, interventions and the effects of interventions for these outcomes in (Table 5). For studies which had no reports of our primary outcome 2 (Investigator-assessed change in lesion count) and for which we were therefore unable to provide narrative summary in the previous section, we provide it here (Baugh 2005; Bowes 2003; Chen 2015; Cheng 2008; Hongcharu 2000; Ianosi 2013; Leheta 2009; Ling 2010; McGill 2008; Ou 2014; Sadick 2010a; Tzung 2004; Yilmaz 2011; Zhang 2009a; Zhang 2013a; Zhang 2013b). For studies which included both primary outcome 2 and secondary outcomes 1, 2 and 3, please find the full details on secondary outcomes in Table 5. Where appropriate, we also clarified why we did not perform meta-analysis.

Secondary outcomes: 1. Light versus placebo or no treatment
Secondary outcomes 1.a. Green light versus placebo

Three split-face trials (Baugh 2005; Bowes 2003; Yilmaz 2011) of four treatments included a total of 80 participants (FST I to III or not reported, with mild to moderate acne or  more than 4 facial ILs). All three studies used the Michaelsson score (where a decrease in the score signifies a decrease in acne severity, Michaelsson 1977 ) for acne severity evaluation, but meta-analysis of change in acne severity was not possible because necessary data were not reported nor provided upon request. All three studies reported greater decreases on light-treated sides at four weeks after final treatment.

The Michaelsson score decreased from a baseline of 42.9 to 34.1 (by 21%) on the treated side and increased from a baseline of 41.2 to 51.4 (by 25%) on the control side (P = 0. 089, SDs not given) in Baugh 2005, and in Bowes 2003 decreased by 35.9% on the treated side and increased by 1.8% on the untreated side (SDs not given). In Yilmaz 2011 (split-face within a parallel-group design), which also compared single and multiple treatment groups, both sides improved, but decrease in Michaelsson severity score was significantly greater on the treated side - 31% versus 6% (P = 0.005) in the once-weekly group and by 40% versus 13% in twice-weekly group (P < 0.001). Means and SDs were not given; further data were not given.

Secondary outcomes 1.b. Yellow light versus placebo or no treatment

We were unable to pool results of two studies. One parallel-group study (Seaton 2003) reported median (interquartile range) improvements in Leeds grade whilst the other, split-face study (Orringer 2004), reported changes in means with SEs of Leeds scores. We were unable to obtain additional data.

Secondary outcomes 1.c. Infrared light versus no treatment

Two split-face trials (Darne 2011; Orringer 2007) of three treatments included a total of 84 participants (FST I to VI, with mild to moderate acne). Meta-analysis was not possible because of different types of lasers used, different application intervals and timings of outcome assessment, although both used the Leeds score and reported it with 95% confidence interval (Darne 2011) and SE (Orringer 2007). Another split-face trial (Moneib 2014) included the outcome 'Investigator's global assessment of improvement', using a non-standardised scale and reported assessments at an unclear time point.

Secondary outcomes 1.d. Blue light versus no treatment  

One split-face study (Tzung 2004) randomised 31 participants (FPT III to IV, with mild to moderate acne). The Michaelson modified grade percentage improvement in the blue-light group compared to the control group was reported as 52% and 12% respectively at eight weeks, P = 0.009.

Secondary outcomes 1.f. Blue-red light versus placebo

Two parallel-group studies included this comparison; one (Papageorgieu 2000) included 'Investigator-assessed change in acne severity', another (Kwon 2013) included 'Investigator's global assessment of improvement'. We were therefore unable to pool data. One study (Papageorgieu 2000) randomised 30 participants to the blue-red-light group and 25 to the white-light group (FPTs not reported, all with mild to moderate acne). A non-standardised scale was used for evaluation (please see Table 5) and reported in graph format only. We extracted the data from the graph and dichotomised them to 26/30 'success' outcomes in the blue-red group and 6/25 in the white-light group. Blue-red light was superior to white light with RR 3.61, 95% CI 1.77 to 7.36, P = 0.0004 (Analysis 1.1) and the NNTB was 2 (95% CI 1 to 3).

Secondary outcomes 1.h. Intense pulsed light (IPL) versus no treatment

One split-face trial (McGill 2008) randomised ten participants (FPT I to II, with mild to moderate facial acne). IPL was applied, with ‘upper’ and ‘lower’ halves of face sides treated with different filters; 550 nm to 1100 nm filter (‘585 filter’), and the ‘dual band’ filter (blue light), whereas the other half served as control. Intervention on the control face sides was unclear, but it was most likely no-treatment control. Five treatments were applied at two-week intervals, and assessed at one, three and six months after final treatment. Seven participants completed the study, and five were evaluated. At six months after final treatment for the outcome 'Investigator-assessed change in acne severity', our calculations using t-distribution showed that there were no significant differences in changes in the Leeds grade between 585 half sides and control sides (MD 0.60, 95% CI -1.88 to 3.08), P = 0.64 (Analysis 26.2), nor between blue-light and control sides (MD 0.40, 95% CI -1.95 to 2.75), P = 0.74 (Analysis 26.2). Results of the analyses using t-distribution did not substantially differ from the ones in which we used normal distribution (Analysis 26.3).

Our third secondary outcome was 'Changes in quality of life'. Mean (± SD) pretreatment Dermatology Life Quality Index (DLQI) scores were 11 ± 5 (range 3 to 19). At one month DLQI score had decreased to 6 ± 5 (range 0 to 12), at three months to 5 ± 2 (range 2 to 7) and at six months it increased to 7 ± 4 (range 4 to 12). Not reported for separate face half-sides.

Secondary outcomes: 2. Light versus topical treatment.
Secondary outcomes 2.a. Light versus benzoyl peroxide (BPO)

Only one study (Papageorgieu 2000) included this outcome (Investigator's global assessment of improvement) for this comparison, which randomised 30 participants to the blue-red light group and 25 to the BPO group (FPTs not reported, all with mild to moderate acne). A non-standardised scale was used for evaluation (please see above) and reported in graph format only. We extracted the data from the graph and dichotomised them to 26/30 'success' outcomes in the blue-red group and 16/25 in the BPO group. The difference was non significant, with RR 1.35, 95% CI 0.98 to 1.88) P = 0.07 (Analysis 2.1).

Secondary outcomes 2.b. Light versus clindamycin

Only one study (Gold 2005) included the outcomes 'investigator-assessed change in acne severity' and 'global assessment of improvement' for this comparison. It was a parallel-group trial that compared eight sessions of 417 nm blue light with self-administered topical clindamycin (34 participants, FPT not reported, with mild to moderate acne). Investigator-assessed change in acne severity and global assessment of improvement were reported as similar for both groups (figures were not given in the paper).

Secondary outcomes 2.c. Light and other topical treatments

Five parallel-group studies included this comparison, but their interventions included different modalities of light and topical treatments, so we were unable to combine their results (Borhan 2014; Ianosi 2013; Karsai 2010; Leheta 2009; Zhang 2009a ). The outcomes they assessed also differed.

Ianosi 2013 included 180 participants (FPT I to IV, with mild to moderate acne) and randomised 60 participants to 500nm to 1200 nm light plus vacuum group, 60 participants to IPL alone group (400nm to 700 nm and 870 nm to 1200 nm) and 60 participants to anti-acne micellar solution. Light treatments were applied once a week for five weeks, and final evaluation was done at the last treatment. There was a greater reduction in the Leeds score in the light-treatment groups compared to the micellar-solution group, which was reported only in graph format and no further data were provided. There was also a significantly greater effect on quality of life (using the Cardiff Acne Disability Index) in vacuum plus IPL group compared to the micellar solution group (P = 0.004). Further data were not given.

Leheta 2009 (parallel-group study). We dichotomised the data for 'investigator's global assessment of improvement' to 3/15 'success' outcomes in the PDL group, 13/15 in 5% BPO in combination with tretinoin (T/BPO) group and 15/15 in the 0.025% retinoic acid cream combined with trichloroacetic acid peeling (TCAA) group. PDL was not superior to T/BPO with RR 1.00, 95% CI 0.76 to 1.32, P = 1.00 (Analysis 30.1), nor to TCAA, RR 0.87, 95% CI 0.69 to 1.09, P = 0.24 (Analysis 31.1).

Zhang 2009a (parallel-group trial) compared blue and red light in combination with clindamycin gel, azithromycin, antisterone or cimetidine with clindamycin gel, azithromycin, antisterone or cimetidine alone. The trial included 738 participants (FPT not given, with mild to severe acne, Pillsbury grades I to IV). Evaluation was performed four weeks after treatment. Investigators assessed improvement using the following scale based on lesion count percentage change: 90% improvement or above = 'full recovery'; 60% to 89% = 'good improvement'; 30% to 59% = 'effective improvement'; 29% or less = 'no effect'. We dichotomised the data following our protocol and using the ITT approach to present the outcome 'investigator's global assessment of improvement' as 332/508 ‘success’ outcomes in the intervention and 125/230 ‘success’ outcomes in the control group. Antibiotic treatment in combination with blue-red light was superior to antibiotic treatment alone with RR 1.20, 95% CI 1.05 to 1.38, P = 0.006 (Analysis 32.1). The NNTB was 10 (95% CI 6 to 30).

Secondary outcomes: 3. Light versus other comparators
Secondary outcomes 3.a. Comparison of light therapies of different wavelengths

Two parallel-group trials (Cheng 2008; Papageorgieu 2000) included comparison of blue and blue-red light. Meta-analysis was not done because of the differences in the number of sessions (84 versus 8 to 24) and timing of their assessment. Another three trials compared different interventions, namely eight sessions of blue LED with eight sessions of red LED (Liu 2011); four sessions of 585 nm PDL compared with four sessions of 530 to 750 nm IPL (Choi 2010) and three sessions of 585 nm PDL compared with three sessions of combined 585/1064 nm PDL (Jung 2009), so quantitative synthesis was not appropriate.

Cheng 2008 (secondary outcomes only reported) included 64 participants (FPT not reported, with mild to moderate acne), who were randomised to the 400 nm to 410 nm light group or to the 400 nm to 410 nm plus 660 nm light group. Investigators assessed improvement using the following scale based on lesion count percentage change: 90% improvement or above = 'full recovery'; 70% to 89% = 'good improvement'; 30% to 69% = 'effective improvement'; 30% or less = 'no effect'. We dichotomised the data to present the outcome 'investigator's global assessment of improvement' as 15/28 'success' outcomes in the blue-red group and 26/36 in the blue-light-alone group. The difference was non significant with RR 0.74, 95% CI 0.50 to 1.11, P = 0.14 (Analysis 33.1).

Liu 2011 (parallel-group study) compared blue with red light and included results for 20 participants (FPTs III to IV, all with mild to moderate acne), who completed the trial of eight sessions of blue light in one group (405 ± 10 nm, power of 30 mW/cm²) and red light (630 ± 10 nm, power of 48 mW/cm² ) in the other group. Investigators assessed improvement using the following scale based on lesion count percentage change: reduction 90% or above = ‘full recovery’; 60% to 89% reduction= ‘significant improvement’, 40% to 59% reduction = ‘moderate improvement’, 20% to 39% reduction = ‘mild improvement’, and 19% reduction or below = ‘non- improvement or aggravation’. We dichotomised the data to present the outcome 'investigator's global assessment of improvement' as 8/10 ‘success’ outcomes in the blue-light and 5/10 in the red-light group. The difference was non significant with RR 1.60, 95% CI 0.80 to 3.20, P = 0.18 (Analysis 34.1).

Papageorgieu 2000 randomised 30 participants to the blue-red light group and 27 to the blue-light group (FPTs not reported, all with mild to moderate acne). A non-standardised scale was used for evaluation (please see above) and reported in graph format only. We extracted the data from the graph and dichotomised them to present the outcome 'investigator's global assessment of improvement' as 26/30 'success' outcomes in the blue-red group and 19/27 in the blue-light-alone group. The difference was non significant, with RR 1.23, 95% CI 0.93 to 1.63, P = 0.15. (Analysis 3.1).

Secondary outcomes 3.b. Comparison of light therapies of different doses

Two split-face (Bernstein 2007; Uebelhoer 2007) trials compared single and double passes of 1450 nm lasers, but had different numbers of sessions and timings of outcome assessment so we did not quantitatively combine the data.

One parallel-group trial (NCT00706433) compared four interventions:

  1. 20% ALA (45 min incubation) plus 1000 s of blue light;

  2. 20% ALA (45 min incubation) plus 500 s of blue light;

  3. vehicle (45 min incubation) plus 1000 s of blue light; and

  4. vehicle (45 min incubation) plus 500 s of blue light.

The study included a total of 266 participants (FPT I to VI, with moderate to severe acne, IGA score 3 and 4, with at least 20 ILs); 67 in the vehicle 1000 s group and 66 in the vehicle 500 s group. At three weeks after final treatment there were 15/67 'success' outcomes in the vehicle 1000 s and 11/66 in the vehicle 500 s group. The difference between vehicle 1000 s and vehicle 500 s groups for the outcome 'investigator's global assessment of improvement' was non significant, with RR 1.34, 95% CI 0.67 to 2.70, P = 0.43 (Analysis 4.2). At six weeks after final treatment there were 16/67 'success' outcomes in the vehicle 1000 s and 16/66 in the vehicle 500 s group. The difference between vehicle 1000 s and vehicle 500 s groups was non significant, with RR 0.99, 0.54 to 1.80, P = 0.96 (Analysis 4.2).

Secondary outcomes 3.c. Comparison of light therapies of different treatment application intervals

Only one study (Yilmaz 2011) included this comparison for this outcome. This was a parallel-group RCT (split-face within groups) which randomised two groups; application of 532 nm (green) light once weekly for four weeks versus twice weekly for two weeks. Within each group one side of the face was randomised to assigned treatment and the other to no treatment. It included a total of 44 participants (FST I to III, with more than 4 facial ILs). Differences in Michaelson acne severity score means (SD) of the treated face sides at baseline and at four weeks were -5.9 (7.9) in the once-weekly group and -9.3 (7.5) in the twice-weekly group.

Secondary outcomes 3.e. Light in combination with carbon lotion versus no treatment

Only one study (Jung 2012) included this outcome for this comparison. This was a split-face trial that compared three sessions of quasi-long pulse and Q-switched 1064 nm Nd:YAG laser plus carbon lotion with non-treated control and (22 participants, FPT III to V, with unclear severity of acne). The Cunliffe severity grade decreased significantly from 3.2 to 1.7 (P < 0.001) on the laser-treated side and from 2.7 to 2.6 (P < 0.05) on the non-treated side. The difference between the two treatments was significant (P = 0.04).

Secondary outcomes 3.f. Light in combination with oral therapy versus other comparators

Four parallel-group studies included this comparison, but their interventions included different modalities and combinations of light, oral and topical treatments, so we were unable to combine their results (Ling 2010; Ou 2014; Zhang 2009a; Zhang 2013b).

Ling 2010 compared blue and red light plus sulfotanshinone, versus sulfotanshinone alone, versus blue and red light plus sulfotanshinone plus prednisolone, versus sulfotanshinone plus prednisolone. The trial included 30 participants in each of the four groups (FPT not given, with moderate to severe acne). Evaluation was performed four weeks after treatment. Investigators assessed improvement using the following scale based on lesion count percentage change: 95% improvement or above = 'full recovery'; 60% to 95% = 'good improvement'; 20% to 59% = 'effective improvement'; 20% or less = 'no effect'. We dichotomised the data to 26/30 'success' outcomes in the blue-red light plus sulfotanshinone group, 19/30 in the sulfotanshinone-alone group, 16/30 in the blue-red light plus sulfotanshinone plus prednisolone group and 13/30 in the sulfotanshinone plus prednisolone group. Blue and red light plus sulfotanshinone was superior (marginally) to sulfotanshinone alone for the outcome 'investigator's global assessment of improvement' with RR 1.37, 95% CI 1.01 to 1.86, P = 0.04 (Analysis 35.1); to blue and red light plus sulfotanshinone plus prednisolone with RR 1.63, 95% CI 1.13 to 2.34, P = 0.009 (Analysis 36.1); and to sulfotanshinone plus prednisolone with RR 2.00, 95% CI 1.30 to 3.08, P = 0.002 (Analysis 37.1).The NNTBs were 3 (95% CI 1 to 9) and 3 (95% CI 1 to 5) for the latter two comparisons with blue-red light plus sulfotanshinone respectively. However, there is no calculable NNTB for the comparison of blue-red light plus sulfotanshinone to sulfotanshinone alone since the 95% CI for the risk difference contains zero (i.e. no effect), and this corresponds to an infinite upper 'limit' for the 95% CI for the NNTB, which indicates that there is no true boundary on how large the NNTB could be for this comparison: this is also seen in the marginal effect seen with the RR.

Ou 2014 compared Yinhua decoction (YD, term as presented in the English translation of the abstract provided by the journal where full text was published in Mandarin) with 'electric light synergy' versus YD in combination with red and blue light treatment. The trial included 90 participants, and 83 completed the trial (FPT not given, with moderate acne grade II to III Chinese Acne Treatment Guidelines). Evaluation was performed twelve weeks after final treatment. Investigators assessed improvement using the following scale based on lesion count percentage change: 90% improvement or above = 'full recovery'; 60% to 89% = 'good improvement'; 30% to 59% = 'effective improvement'; 29% or under = 'no effect'. We dichotomised the data for the outcome 'investigator's global assessment of improvement' 30/43 (69.7% of those who completed) success outcomes in the intervention arm, and 15/40 (37.5% of those who completed) in the control arm. Numbers of randomised participants in each group were not reported, and so we were unable to use ITT approach. YD plus 'electric light synergy' were superior to YD in combination with blue-red light with RR 1.86, 95% CI 1.19 to 2.91, P = 0.006 (Analysis 38.1). The NNTB was 4 (95% CI 2 to 10).

Zhang 2013b compared blue and red light plus Jinhua xiaocuo (term as presented in the English translation of the abstract provided by the journal where full text was published in Mandarin) pills and chloramphenicol tincture versus Jinhua xiaocuo pills and chloramphenicol tincture alone. The trial included 60 in each group (FPT not given, with mild to moderate acne, Pillsbury grades I to III). Evaluation was performed four weeks after final treatment. Investigators assessed improvement using the following scale based on lesion count percentage change: 90% improvement or above = 'full recovery'; 60% to 89% = 'good improvement'; 30% to 59% = 'effective improvement'; 29% or under = 'no effect'. We dichotomised the data following our protocol to 55/60 ‘success’ outcomes in the intervention and 39/60 ‘success’ outcomes in the control group. Jinhua xiaocuo pills and chloramphenicol tincture in combination with blue-red light were superior to jinhua xiaocuo pills and chloramphenicol tincture alone with RR 1.41, 95% CI 1.15 to 1.72, P = 0.0008 (Analysis 39.1). The NNTB was 4 (95% CI 3 to 9).

Zhang 2009a compared blue and red light in combination with clindamycin gel, azithromycin, antisterone or cimetidine with clindamycin gel, azithromycin, antisterone or cimetidine alone. Please see (Analysis 32.1), further details and the results under Secondary outcomes 2.c. Light and other topical treatments, as this study could be placed under both comparisons.

Secondary outcomes 3.g. IPL alone versus IPL in combination with vacuum

One parallel-group trial (Ianosi 2013) randomised a total of 180 participants (FPT I to IV, with mild to moderate acne) to 500 nm to 1200 nm light plus vacuum group or to an IPL-alone group (400 nm to 700 nm and 870 nm to 1200 nm). Changes in lesion counts were reported as scores 1 = insignificant result (lesion count reduction 0% to 25%) to 4 = very good result (lesion count reduction 76% to 100%). No significant differences were found between the two treatments at final assessment in a reduction of the score of papules and pustules (P reported as 'NS'). There was a significantly greater reduction in the score of comedones in the vacuum plus IPL group (P < 0.001). There was a greater reduction in the Leeds score in the IPL-only group reported in graph format and no further data provided. There was a significantly greater effect on quality of life (using Cardiff Acne Disability Index) in the vacuum plus IPL group (P = 0.004). Further data were not given.

Secondary outcomes: 4. MAL-PDT versus other comparators
Secondary outcomes 4.a. MAL-PDT versus red light alone

We combined results of three parallel-group studies (NCT00594425; NCT00933543; Pariser 2013) comparing four sessions of red light plus MAL with placebo cream and red light, with final evaluation at six weeks after last treatment (please see above for details). Meta-analysis showed that MAL-PDT was superior to red light alone in IGA score improvement ('success' outcome defined by decrease in the IGA score by at least two grades from baseline), with RR 1.74, 95% CI 1.11 to 2.74 (Analysis 17.6), moderate quality evidence (Summary of findings 2). There were 120 reports of treatment 'success' as defined by IGA score decrease per 1000 study population in the red-light-alone group, and 209 per 1000 study population (95% CI 133 to 329) in the MAL-PDT group. The absolute effect was 89 more treatment 'success' outcomes per 1000 (95% CI from 13 more to 209 more). The NNTB was 7 (95% CI 5 to 11). Please see Analysis 17.6 and Summary of findings 2 for details. Please note that these studies are not presented in Table 5.

NCT00594425 (three-arm trial) randomised 50 participants in the 40 mg/g MAL-PDT group (FPT I to V, with moderate to severe acne, IGA score 3 to 4, 20 to 100 ILs and up to 200 NILs on the face). Four treatments at two-week intervals were applied. At six weeks after final treatment 'success' outcomes (defined by decrease in IGA score by at least two grades from baseline) were found in 6/50 participants in the 40 mg/g group and 4/52 in the placebo-cream group. Our analyses showed that 40 mg/g MAL-PDT was not superior to placebo cream activated by red light for the outcome 'investigator's global assessment of improvement', with RR 1.56, 95% CI 0.47 to 5.20, P = 0.47 (Analysis 18.5).

We were unable to combine results from one more split-face trial, which compared only two sessions of 635 nm light plus 160 mg/g MAL with placebo cream and light and also had different assessment time points (Hörfelt 2006). In that study, we dichotomised the data to 12/30 'success' outcomes on the MAL-PDT sides and 7/30 on the placebo-PDT sides. The difference was non significant, with RR 1.71, 95% CI 0.78 to 3.75, P = 0.18 (Analysis 20.4; Table 5).

Secondary outcomes 4.c. MAL-PDT versus placebo or no treatment

Only one study (Wiegell 2006b) included this comparison for this outcome. This was a parallel-group study of two treatments of 630 nm plus 160 mg/g MAL (21 participants in the treatment group and 15 in the control group; FPT II to V, with at least 12 facial ILs). No significant difference was observed in reduction in the Leeds grade between the two groups (P = 0.24).

Secondary outcomes 4.d. MAL-PDT other

Due to substantial clinical and methodological heterogeneity of three studies with different interventions and comparators (Bissonnette 2010; Hong 2013; NCT00594425) we did not perform quantitative synthesis of their results.

NCT00594425 (three-arm parallel-group trial) randomised 48 participants to the 80 mg/g MAL-PDT arm and 50 participants to the 40 mg/g MAL-PDT arm (FPT I to IV, with moderate to severe acne, IGA score 3 to 4, 20 to 100 ILs and up to 200 NILs on the face). Four treatments at two-week intervals were applied, and 37 participants completed treatment in the 80 mg/g group, and 43 completed in the 40 mg/g group. Our analyses showed that at six weeks after final treatment 80 mg/g MAL-PDT was not superior to 40 mg/g MAL-PDT by the 'investigator's assessment of improvement' (a 'success' outcome was defined by a decrease in the IGA score by at least two grades from baseline), (RR 1.04, 95% CI 0.36 to 3.01; n = 98, P = 0.94) (Analysis 21.5).

Bissonnette 2010 (parallel-group trial) randomised 44 participants (FPT I to IV, with 10 or more ILs on each face side) to compare 80 mg/g MAL with or without occlusion followed by different red light intensity exposure; participants were randomised in four groups with 25 J/cm² or 37 J/cm² and with or without occlusion; there were four treatments, assessed at four and 12 weeks after the final treatment. At 12 weeks for the outcome 'investigator's assessment of improvement' the difference in 'success' outcomes (defined by decrease in the IGA score by at least two grades from baseline) was non significant for the comparison 37 J/cm² treatment with occlusion versus 37 J/cm² treatment without occlusion, (RR 0.50, 95% CI 0.05 to 5.12; n = 44) (Analysis 28.2).

Hong 2013 (split-face study) compared three sessions of 160 mg/g MAL plus red light with three sessions of MAL plus IPL and included 22 participants (FPT IV to V). At four weeks after treatment there was no significant difference in the improvement in acne Cunliffe grade between the red light side (1.9) and IPL side (2.0).

Secondary outcomes: 5. ALA-PDT versus other comparators
Secondary outcomes 5.a. ALA-PDT versus red light alone

One parallel-group trial (Chen 2015) compared three red light (633 nm) 20% ALA-PDT treatments with three treatments of red light alone and included a total of 50 participants (FPT not given, with mild to severe acne). A non-standardised method was used for the investigators' evaluation (90% or above improvement = 'cured', 60% to 89% improvement = 'excellent effect', 30% to 59% improvement = 'fair effect', 30% improvement or exacerbations or less = 'no effect'). One participant dropped out from the ALA-PDT group, and two dropped out from the red-light only group, so we treated them as treatment failures as per our protocol. We dichotomised the data following our protocol ('success' defined as anything above the first category of improvement) to 13/25 ‘success’ outcomes at two weeks, 18/25 at four weeks and 20/25 at six weeks in the intervention group, whereas in the control group there were 6/25 ‘success’ outcomes at two weeks, 10/25 at four weeks and 13/25 at six weeks.

Another parallel-group trial (Zhang 2013a) compared three red light ALA-PDT treatments with three treatments of red light alone and included a total of 116 participants (FPT not given, with moderate to severe acne, Pillsbury grade II to IV). Evaluation was performed two, four and eight weeks after final treatment. Investigators assessed improvement using the following scale based on lesion count percentage change: 90% improvement or above = 'full recovery'; 60% to 89% = 'good improvement'; 20% to 59% = 'effective improvement'; 19% or below = 'no effect'. We dichotomised the data following our protocol ('success' defined as anything above the first category of improvement) to 28/63, 37/63, and 50/63 ‘success’ outcomes in the intervention group at two, four and eight weeks after final treatment respectively; and 7/53, 15/53, and 22/53 ‘success’ outcomes in the control group at two, four and eight weeks after final treatment respectively.

We judged it was appropriate to combine the results of the above two parallel-group studies (Chen 2015; Zhang 2013a). We have presented details of the data and results as reported by the authors of these studies in Table 5. Treatments were applied in weekly intervals in both studies. Both studies also had evaluation time points at two and four weeks after last treatment, but final evaluation was done at six weeks after last treatment in Chen 2015, and eight weeks after last treatment in Zhang 2013a. The statistical heterogeneity across studies was not substantial, I² was 0% at both two weeks and four weeks, and fitted the criteria we stated in our protocol (I² had to be lower than 50%). Therefore we judged it was appropriate to combine the results. However, there was some clinical heterogeneity across studies to take into account. We have narratively summarised it here, please check Characteristics of included studies tables of each study for details. While Chen 2015 included all acne severity grades (mild to severe), Zhang 2013a included only moderate to severe acne. FPTs were not reported in either of the studies. Both studies had the same ALA supplier, however it is unclear whether the same ALA percentage was used. Characteristics of red light also differed, but not substantially.

Meta-analysis, using a random-effects model, showed that ALA-PDT was superior to red light alone in improving the 'investigator global assessment of improvement' score at two weeks with RR 2.74, 95% CI 1.59 to 4.71 (Analysis 40.1), as well as at four weeks with RR 1.95, 95% CI 1.36 to 2.79 (Analysis 40.1). The NNTB was 4 (95% CI 3 to 7) at two weeks, as well as at four weeks. However, results Chen 2015 also showed that at six weeks ALA-PDT was no longer superior to red light alone with RR 1.54, 95% CI 1.01 to 2.35, P = 0.05 (Analysis 40.1). Zhang 2013a did not include six weeks as an assessment time point, but found that ALA-PDT was still superior to red light alone at eight weeks after final treatment with RR 1.91, 95% CI 1.36 to 2.70, P = 0.0002 (Analysis 40.1). The NNTB was 3 (95% CI 2 to 5) at eight weeks.

Secondary outcomes 5.b. ALA-PDT versus blue light alone

One parallel-group trial (NCT00706433) compared four interventions:

  1. 20% ALA (45 min incubation) plus 1000 s of blue light;

  2. 20% ALA (45 min incubation) plus 500 s of blue light;

  3. vehicle (45 min incubation) plus 1000 s of blue light; and

  4. vehicle (45 min incubation) plus 500 s of blue light.

The study included a total of 266 participants (FPT I to VI, with moderate to severe acne, IGA score 3 and 4, with at least 20 ILs). The difference in the; 'investigator global assessment (IGA) of improvement' ('success' outcome defined as a 2 point or more improvement on the IGA scale from baseline) between ALA 1000 s and vehicle 1000 s groups was non significant at three weeks after final treatment, with RR 0.85, 95% CI 0.44 to 1.65, P = 0.64, and it was also non significant between ALA 500 s and vehicle 500 s groups, with RR 1.02, 95% CI 0.47 to 2.18, P = 0.97 (Analysis 5.2; Table 5). At six weeks after final treatment, the difference between ALA 1000 s and vehicle 1000 s groups remained non significant, with RR 0.92, 95% CI 0.50 to 1.71), P = 0.80, and it also remained non significant between ALA 500 s and vehicle 500 s groups, with RR 0.70, 95% CI 0.35 to 1.39, P = 0.31 (Analysis 5.3; Table 5). The difference between ALA-PDT and vehicle plus blue light was non significant when we combined results for the 1000 s and 500 s subgroups using a random-effects model, with RR 0.92, 95% CI 0.56 to 1.52, P = 0.74 at 3 weeks and RR 0.81, 95% CI 0.51 to 1.29, P = 0.38 at six weeks after final treatment respectively (Analysis 5.3). See Summary of findings 3 where we rated the evidence as low quality for this outcome.

Secondary outcomes 5.d. ALA-PDT versus IPL alone

Two trials included this comparison, but one had a split-face design (Oh 2009), and included three treatments with different incubation times in participants with moderate to severe acne, whilst the other was a parallel-group trial, of four treatments and included participants with severe acne (Mei 2013). Different scales were used for assessment. We did not combine results because of this heterogeneity and calculated RR with 95% CI for individual studies.

Oh 2009 compared three sessions of 20% ALA plus IPL (one face side randomised to either 30 minutes' or three hours' incubation) with IPL only and included 20 participants (FPT III to IV, with moderate to severe acne). The difference was non significant, (RR 0.81, 95% CI 0.48 to 1.40) (Analysis 6.1). Results were reported for IPL-only sides.

Mei 2013: the investigators assessed there was no significant difference in improvement between the 10% ALA-PDT and IPL-alone group at 12 weeks after final treatment, (RR 1.43, 95% CI 0.96 to 2.13, P = 0.08) (Analysis 24.3).

Secondary outcomes 5.e. ALA-PDT versus green light alone

Only one split-face trial (Sadick 2010a) compared three 20% ALA (30 min incubation) plus 532 nm potassium titanyl phosphate (KTP) laser light with KTP laser alone. The study included a total of 10 participants (FPT I to III, with moderate to severe acne, IGA score 3 and 4). IGA was also used for evaluation (see above). On the ALA-PDT sides IGA score (mean ± standard error) reduced from baseline 3.50 ± 0.19 to 2.29 ± 0.29 (35% improvement) after the first treatment and to 2.13 ± 0.40 (39% improvement) after the second treatment. On the light-only sides IGA score (mean ± standard error) reduced from baseline 3.63 ± 0.18 to 2.42 ± 0.30 (33% improvement) after the first treatment and to 2.38 ± 0.33 (34% improvement) after the second treatment. Further details and results of evaluations after the final treatment were not given (reported as "Similar results were recorded after the third treatment session that was evaluated at week 12").

Secondary outcomes 5.f. ALA-PDT versus placebo or no treatment

Only one study (Orringer 2010) included investigator-assessed change in acne severity for this comparison. This was a split-face trial that compared three sessions of 20% ALA plus PDL with untreated control. The trial included 44 participants (all FPTs, severity of acne unclear). There was a statistically significant difference in decrease (i.e. improvement, P = 0.01) in the mean Leeds score on treated skin versus untreated skin at week 16 (i.e. 10 weeks after final treatment). Mean change in score from baseline was -1.07, (95% CI -1.69 to -0.45) on the treated sides and -0.52 (95% CI -1.07 to 0.04) on the control sides.

Secondary outcomes 5.g. ALA-PDT other

Due to substantial clinical and methodological heterogeneity of five studies with different interventions and comparators (Barolet 2010; Hongcharu 2000; NCT00706433; Taub 2007; Yin 2010), we did not perform quantitative synthesis of their results. Please see Table 5 and Analysis 8.2 for details.

Barolet 2010 (split-face or split-back trial) compared a single treatment of 970 nm IR pre-treatment plus 20% ALA and 630 nm PDT with ALA-PDT alone. The trial included 10 participants (FPT I to III, with mild to moderate acne). At four weeks after treatment there was greater improvement in Global Severity Assessment Score medians on the IR pre-treated (1, 95% CI 0.74 to 1.34) versus control side (2, 95% CI 1.17 to 1.72). Further data were not provided, 95% CI reported for means, but means were not given.

Hongcharu 2000 randomised 22 participants (FPT I to IV, with mild to moderate acne) into single and multiple treatment groups, with four areas on the back of each participant treated with ALA plus 550 nm to 700 nm light, ALA alone, or 550 nm to 700 nm light, or untreated as control. Change from baseline in Michaelsson acne severity score was significantly better in ALA-PDT than the other three areas at 3, 10 and 20 weeks after single treatment (P values not given) and at all visits after multiple treatment (P < 0.05). ALA–PDT and multiple ALA treatment sites showed more improvement than single treatment (P < 0.001 and P = 0.007, respectively). Investigator's global assessment of improvement scores was also significantly better for the ALA-PDT areas than the other three areas where some improvement has also been observed in both single and multiple treatment groups. These comparisons, as well as comparison between single and multiple treatment groups were reported in an unclear way.

One parallel-group trial (NCT00706433) compared four interventions:

  1. 20% ALA (45 min incubation) plus 1000 s of blue light;

  2. 20% ALA (45 min incubation) plus 500 s of blue light;

  3. vehicle (45 min incubation) plus 1000 s of blue light; and

  4. vehicle (45 min incubation) plus 500 s of blue light.

The study included a total of 266 participants (FPT I to VI, with moderate to severe acne, IGA score 3 and 4, with at least 20 ILs); 68 in the ALA 1000 s group, 65 in the ALA 500 s group, 67 in the vehicle 1000 s group and 66 in the vehicle 500 s group. The improvement of the Investigator Global Assessment (IGA) score at three weeks after final treatment between ALA 1000 s and ALA 500 s groups was non significant, (RR 1.13, 95% CI 0.55 to 2.34, n=143, P = 0.33) (Analysis 8.2), and it remained non significant at 6 weeks after final treatment, (RR 1.30, 95% CI 0.65 to 2.62, n=143, P = 0.74) (Analysis 8.2).

Taub 2007 compared three ALA-PDT treatments with different light sources for activation: IPL (600 nm to 850 nm) versus a combination of IPL (580 nm to 980 nm) and bipolar RF energies versus blue light (417 nm) and included 19 participants (FPT II to IV, with more than 10 facial ILs, moderate to severe acne). Investigator-assessed improvement was highest with IPL activation and lowest with blue light, and the differences between groups reached borderline statistical significance at three months (P = 0.0498). At one month after treatment median percentage improvement score was 56.25 (96.9% CI 27.5 to 85.0) in the IPL group, 23.75 (96.9% CI 2.5 to 85.0) in the IPL-RF group and 20 (96.9% CI 0 to 62.5) in the blue-light group. At three months after treatment median percentage improvement score (range) was 72.5 (42.5) in the IPL group, 50 (47.5) in the IPL-RF group and 25 (40) in the blue-light group.

Yin 2010 compared four red light ALA-PDT treatments with different ALA concentrations: 5%, 10%, 15% and 20% and included 180 participants (FPT III to IV, with moderate to severe acne). A non-standardised scale was used for evaluation. At 24 weeks after treatment, a significant difference among the different ALA concentration groups (P values not given) was reported, with a clear positive correlation between global improvement score and ALA concentration (P < 0.05). Further data were expressed in graph format, please see Table 5 for details.

Secondary outcomes: 6. MAL-PDT versus ALA-PDT
Secondary outcomes 6.a. MAL-PDT versus ALA-PDT

Only one study (Wiegell 2006a) included investigator-assessed change in acne severity for this comparison. This was a split-face trial that compared single 620 nm PDT treatments with different creams: 20% ALA versus 160 mg/g MAL. The trial included 19 participants (FPT not given, with more than 12 ILs). Median of the Leeds revised acne global severity grade reduced from 2 before treatment to 1 at 12-week follow-up in both the MAL-PDT and ALA-PDT treated sides of the face. There were no significant differences between the two treatments (P = 0.250).

Secondary outcomes: 7. Other (non MAL, non ALA) PDT versus other comparators
Secondary outcomes 7.a. Indocyanine green (ICG)-PDT versus other comparators

Only one study (Kim 2009) included investigator-assessed change in acne severity for this comparison. This was a parallel-group study of a single treatment with three treatments of ICG plus 805 nm light (right cheek), 805 nm light alone (left cheek) and 'spontaneous resolution' control (forehead). The study included 16 participants (FPT, with mild to moderate acne). There was significant improvement in the Cunliffe acne severity score in both groups at two and four weeks after final treatment (P < 0.05). It was not reported whether there were differences between the two groups.

Secondary outcomes 7.c. Topical liposomal methylene blue (TLMB)-PDT versus other comparators

Only one study (Fadel 2009) included investigator-assessed change in acne severity for this comparison.This was a split-face trial that compared two sessions of TLMB plus 650 nm light with no treatment. The trial included 20 participants (FPT not reported, with mild to moderate acne). At 12 weeks the median Leeds severity grade on the treated side was 1 (range 0 to 2) and on the untreated side 3 (range 2 to 4). No baseline data given. At 12 weeks 7/13 (54%) participants had marked improvement, 4/13 (31%) participants had moderate and 2/13 (15%) participants had slight improvement. "Approximately the same improvements" after four weeks and eight weeks. Study authors reported that control areas had no change or worsening of acne with no details provided.

Secondary outcomes 7.e. Gold microparticle PDT versus other comparators

Only one parallel-group trial (Paithankar 2015) compared three sessions of gold microparticle suspension plus light (details not given) with vehicle (without light-absorbing particles) plus light (details not given) control. The trial included 51 participants (FPT I to III, with IGA scores 3 to 4 with at least 25 total papules and pustules on the face). At 10 weeks after the final treatment, the study authors stated "40% of subjects in the treatment arm, whereas none in the sham arm, showed Investigator’s Global Assessment (IGA) score reduction in two or higher". Further data were not given.

Other adverse effects

Most commonly reported adverse effects were application site erythema, application site oedema and pain of skin. Please see Table 4 for details and other adverse effects and their incidence reported in individual studies.

Adverse effects were reported inadequately in most studies and most studies did not quantify adverse effects in each intervention group or report them separately for the sides of the face or back assigned to different interventions. Six studies did not explicitly report whether participants experienced any adverse effects (Bowes 2003; Cheng 2008; Gold 2011; Ling 2010; Orringer 2004; Tzung 2004) and ten studies reported that they recorded adverse effects but no adverse effects were observed (Ash 2015; Baugh 2005; Elman 2003; Genina 2004; Gold 2005; Lee 2010; Na 2011; Sadick 2010b; Song 2014; Yilmaz 2011).

Discussion

Summary of main results

We included 71 studies with a total of 4211 participants, of which 40 were studies of light-only therapies with a total of 2485 participants, and 31 were studies of photodynamic therapy (PDT) with a total of 1726 participants. Most studies had a parallel-group design, split-face design, or a design that combined split-face and parallel groups. The majority had small sample sizes (median 31, mean 59). Most studies were single centre and did not report on funding sources, or were sponsored by industry if multicentre. Most studies included participants with a mean age of between 20 and 30 years, of both sexes, with mild to moderate acne. Many studies did not report on Fitzpatrick Skin Types (FPTs) and a great proportion of studies which did, included up to three FPTs, typically I to III or III to V. Light interventions differed greatly in wavelengths, doses, active substances used in PDT and comparator interventions (most common being no treatment, placebo, other light interventions and various topical treatments). The number of light sessions of the interventions varied from one to 112, with two to four sessions being the most common. Frequency of application varied from twice a day to once a month.

We have summarised the comparison of light therapies (including PDT) compared to placebo, no treatment, topical treatment and other comparators in Summary of findings for the main comparison for our primary outcomes. Twenty-three studies addressed our first primary outcome, 'participant's global assessment of improvement'. Most of them had small sample sizes (median sample size 24), used non-standardised scales, were poorly reported, and only a few assessed this outcome at times after the final treatment. We decided not to combine the effect estimates from the different interventions for this outcome, but rated the evidence based on the GRADE considerations as very low quality, as we were uncertain whether light therapies improve acne compared to placebo, no treatment, topical treatment and other comparators.

For our second primary outcome, Investigator-assessed change in lesion counts, 51 studies with 2242 participants addressed this outcome. Here too we were unable to combine the effect estimates from the different interventions and rated the quality of the evidence as very low, so we are uncertain whether light therapies improve lesion counts compared to placebo, no treatment, topical treatment or other comparators.

For our third primary outcome, 'investigator-assessed severe adverse effects', adverse effects were reported inadequately in most studies. Six studies did not report whether any adverse effects were experienced by participants. Adverse effects were reported as defined in MedDRA (MedDRA 2010) and coded into System Organ Classes (SOCs) in a few studies only. There were no reports of scarring in any of the studies and no reports of blistering (application site blister) in 56 studies with a total of 3378 participants. Here too we were unable to combine the effect estimates from the different interventions and rated the quality of the evidence as very low, so we are uncertain whether light therapies caused more adverse effects compared to placebo, no treatment, topical treatment and other comparators.

Please see Summary of findings 2, where MAL-PDT (methyl aminolevulinate-photodynamic therapy) activated by red light was compared to red light only for acne vulgaris. Our primary outcome which was ‘participants’ global assessment of improvement was not addressed by these studies. Meta-analysis of results from three studies comparing four treatments of 80 mg/g MAL plus red light with placebo cream and red light in a total of 360 participants with moderate to severe acne showed that at six weeks after final treatment MAL-PDT was not superior in reducing the counts or the percentage change in inflamed or non-inflamed lesions as assessed by the investigator, which was our second primary outcome. We rated this evidence as of moderate quality and so of moderate certainty. The outcome, Investigator-assessed severe adverse effects found a lack of adverse events, such as application site blisters in the red-light-alone group (0/158, 0%), while there was one in the MAL-PDT group (1/202, 0.5%). For our secondary outcome, ‘investigators’ global assessment of improvement’ we combined three studies (n = 360) which gave statistically significantly greater improvement in the MAL-PDT groups (moderate-quality evidence). The number needed to treat for an additional treatment 'success' was 7 (95% CI 5 to 15) which we did not interpret as a clinically significant result.

The largest clinical trial we identified, with 266 participants, compared ALA-PDT (20% aminolevulinic acid (ALA) activated by 500 s and 1000 s blue light) with vehicle plus 500 s and 1000 s blue light, and found no difference for our outcome ‘participants’ global assessment of improvement’ at six weeks after final treatment (Summary of findings 3). Similarly, for the outcome of 'investigator-assessed treatment 'success' at three and at six weeks after final treatment there was no significant difference between the treatments. Both of these were rated as low-quality evidence, meaning we have low certainty in the result and that future studies may alter this evidence. For our outcomes ‘investigator-assessed change' or 'percentage change in inflamed lesions', or 'severe adverse effects’, we assessed the certainty of the evidence as very low.

We were unable to quantitatively combine the data for most comparisons due to great variation in many aspects of the studies, poor reporting and failure to obtain necessary data. We therefore performed a narrative synthesis of the results for most of the studies.

Briefly, studies comparing the effects of other interventions were inconsistent or had small samples and high risk of bias. We performed only narrative synthesis for the results of the remaining trials, due to great variation in many aspects of the studies, poor reporting, and failure to obtain necessary data. Several studies compared yellow light to placebo or no treatment, infrared light to no treatment, gold microparticle suspension to vehicle, and clindamycin/benzoyl peroxide combined with pulsed dye laser to clindamycin/benzoyl peroxide alone. There were also several other studies comparing MAL-PDT to light-only treatment, to adapalene and in combination with long-pulsed dye laser to long-pulsed dye laser alone. None of these showed any clinically significant effects.

Overall completeness and applicability of evidence

The studies we included were performed in different geographical and cultural settings, which might prevent generalisation of the results to some extent because of factors such as differences in exposure to natural sunlight or impact on non-validated scales for participants' assessment of improvement of their acne. More importantly, this implies that participants of various FPTs may have been included although not reported (Fitzpatrick 1988). This challenges the applicability of evidence to all FPTs, and in particular to FPTs V and VI, which are known to have a greater risk of adverse effects compared to other skin types when applying light therapies (Alexis 2013). In studies which reported FPTs they were, unsurprisingly, different among studies from Europe, Asia and North America. Other important factors which should be considered in the context of limited generalisability are participants' sex and age, with possible differences in the underlying subtypes of acne and their response to treatment (Choi 2011; Dreno 2013; Preneau 2012).

Most studies included participants with mild to moderate acne, but some did not report the severity of the acne. This limits generalisation, as the effect of light therapies in those with severe acne is less clear.

Participants with acne refractory to antibiotic treatments have often been included in the comparison of different modalities of light therapies. When light therapies were compared with topical treatment, it was often unclear whether there was initial resistance to topical antibiotics in acne patients included in topical antibiotic arms of trials. Initial resistance might have caused antibiotic treatments to prove less effective in these participants, but this would not necessarily be the case in other participants who did not have a resistance problem.

Many studies had a split-face design. It is unclear whether there are possible systemic effects that light and other therapies used in such studies could have on the side of the face used as the control, even if it is not treated directly.

A variety of interventions regarding different wavelengths, fluences, numbers of sessions, as well as frequency of application have been included in this review. However, there are still a lot of possibilities in combining different modalities which were not performed in the studies we included. There were only a few studies using the conventional treatments documented in guidelines (Nast 2012; Zaenglein 2016) for acne as a control. Only a few studies had systemic therapy as a comparator. Combination of light therapies with topical therapies, and particularly systemic therapy have rarely been explored.   

Our primary endpoint was long term outcomes, but less than half of studies performed assessments later than eight weeks after final treatment. Clinically, if a treatment did not give at least three months' resolution it could arguably be a failure. Only a few studies assessed outcomes at more than three months after final treatment, and longer-term assessments are mostly not covered in this review. Although long-term data were our primary endpoint, we were also interested in short-term data, indicating early improvement which may have encouraged participants to continue with the treatment and we therefore considered follow-ups of two to eight weeks after final treatment, reported in the majority of studies. We also reported results recorded at final treatment for studies which did not include follow-up thereafter. Possibly, some interventions may have an early transitional effect on outcomes which our review did not cover, as we only considered follow-ups after final treatment (or at final treatment for studies which did not include evaluations after final treatment). Timing of outcome assessment should be taken into account when interpreting our results, as effects may be different at different time points, some of which are not covered by our review.

Only three studies addressed changes in quality of life (Ianosi 2013; Karsai 2010; McGill 2008) making it the most under-investigated outcome in our review.

Quality of the evidence

The body of evidence we identified did not allow a robust conclusion on the effectiveness of light therapies for acne. We included 71 studies with a total of 4211 participants. The overall quality of evidence was very low, as presented in Summary of findings for the main comparison. We decided not to combine the effect estimates from the different interventions. Instead we rated the quality of the evidence based on the GRADE considerations for our three primary outcomes, taking into account factors that decrease the quality level of a body of evidence outlined in the Cochrane Handbook for Systematic Reviews of Interventions section 12.2.2 (Schünemann 2011a).

Studies addressing 'participant's global assessment of improvement' (23 studies, 1033 participants included) used non-standardised scales, were poorly reported and only a few assessed this outcome later than at final treatment (Summary of findings for the main comparison). The evidence for the effectiveness of light therapies on changes and percentage changes of lesion counts was somewhat more robust in terms of numbers of studies and included participants (51 studies, 2242 participants included) and consistency of methods used for outcome assessment. Most studies (66 studies with 3945 included participants) assessed adverse effects and we presented their results for our third primary outcome (‘investigator-assessed severe adverse effects'). We downgraded the body of evidence for all of these outcomes for several reasons. Firstly, most of the evidence came from studies with unclear or high overall risk of bias, and for primary outcome 1 detection bias was high or unclear in all but two studies. Secondly, quality was limited by inconsistency in the results of individual studies and heterogeneity across studies due to diversity of populations, interventions, comparators and methods of outcome assessment. Thirdly, only a few studies included comparisons with standard treatments, and rarely included comparisons with placebo or no treatment, and so their results did not answer our review question directly, and were further limited by variation of participants who had been included (in terms of Fitzpatrick skin types, severity of acne etc.). Furthermore, most studies had small sample sizes, with medians of 24, 30 and 30 for primary outcomes 1, 2 and 3 respectively. For comparisons where individual studies had randomised fewer than 30 participants per arm, we used t-distribution for analyses of continuous outcomes to account for the sample size. However, substantial imprecision should be taken into consideration when assessing the quality of evidence, in particular when assessing the quality of the evidence for comparisons where only such small studies were available. We also downgraded the evidence because our searches identified a number of unpublished studies but with no available data, which we believe raises questions of whether those trials suggested no benefit.

Quantitative synthesis of several studies was only possible for the comparison of MAL-PDT with red light. We graded the body of evidence for that comparison as moderate (Summary of findings 2). Studies did not include comparisons with conventional treatments documented in guidelines (Nast 2012; Zaenglein 2016), placebo or no treatment, and we judged this was a reason to downgrade the quality level of evidence on the basis of indirectness. Although the following were not reasons for downgrading the evidence, we did consider clinical heterogeneity across studies, such as differences among included participants (Fitzpatrick skin types and severity of acne), as well as differences in interventions (use of occlusive dressing during incubation and different lamps). The studies had low overall risk of bias, so we did not downgrade the evidence on that basis, but we did consider the possible impact of high attrition and selective reporting bias in one study and the fact that the studies were industry sponsored.

We also graded the evidence from a single study with 266 participants for comparison of ALA-PDT with blue light as low for 'participant's global assessment of improvement' and for 'investigator's global assessment of improvement' (Summary of findings 3). The study did not include comparisons with standard treatments, placebo or no treatment, and we judged this was a reason to downgrade the quality level of the evidence on the basis of indirectness. We also downgraded the evidence for all outcomes by one level because of risk of bias, as the study had unclear risk of bias in most of the domains. We considered the possible impact of non-standardised scales which were used to measure these outcomes, but have not further downgraded the evidence on that basis. We graded the evidence as very low for 'investigator-assessed change in ILs' and 'investigator-assessed percentage change in ILs' (Summary of findings 3). Only medians with standard deviations of changes for these continuous outcomes were reported, and means were not provided upon request, so we were unable to perform further analyses. This was an additional reason to downgrade the quality of evidence by one level, along with the reasons listed for the evidence on the above outcomes. We also graded the quality of evidence as very low for our third primary outcome 'Investigator-assessed severe adverse effects' (Summary of findings 3). There were no reports of application site blisters among adverse effects, however it is possible that some occurred, but it is impossible to separate those as they were reported together with "Oozing/ Vesiculation/Crusting", so we downgraded it by two levels because of risk of bias.

As previously described, the quality of evidence for other interventions was fairly limited since we were unable to quantitatively combine the data. Individual studies we identified did not present conclusive evidence of high quality.

Potential biases in the review process

To avoid bias, we followed the protocol for this study (Car 2009). However, considerable time has passed since the protocol was produced in 2009 and we had to make a few minor changes, mostly related to updates in Cochrane methodology. Please see Differences between protocol and review for details.

We tried to minimise bias in the review process through a comprehensive search for all eligible studies, irrespective of language in which they were published or publication status. Seven out of 12 studies with the largest samples (more than 100 participants) were identified through grey literature searches or were not in English (Ling 2010; NCT00594425; NCT00706433; NCT00933543; Zhang 2009a; Zhang 2013a; Zhang 2013b). We intended to test for publication bias by the use of a funnel plot for similar light therapies, however we were unable to create funnel plots because most studies were too heterogeneous to be combined. Two studies we did combine in meta-analyses were not published (we identified NCT00594425 and NCT00933543 in clinical trials registers only) so we did not construct a funnel plot for these not-yet published works. According to trial register records, the final data collection date for primary outcome measures for these studies was 2008 (NCT00594425), and 2010 (NCT00933543). Some bias was probably introduced because we were unable to obtain reports or full results of 36 studies which may possibly meet our inclusion criteria in the future. Please see 'Characteristics of studies awaiting classification' section for details. We therefore believe that despite the fact that our efforts to identify unpublished studies were successful to some extent, publication bias may have still affected the results our review.

Further skewing of the results in our review might be due to unclear selection and performance bias in most studies, together with unclear to high overall detection bias for participant-reported outcomes. Most studies which had unclear to low overall bias, good methodological quality and larger sample sizes were industry sponsored, or study authors had reported some sort of conflict of interest, so additional bias might have been introduced. Non industry-sponsored studies, on the other hand, were in general of lower methodological quality, had unclear to high overall bias and smaller sample sizes.

At least two review authors independently assessed studies for eligibility and extracted data. English translations were obtained for studies in other languages when that was possible. For one study in Portuguese that we included (de Arruda 2009) two review authors extracted data independently from an English translation. However only one person screened full texts of studies which were originally in Mandarin. Six of these studies were included in the review (Cheng 2008; Ling 2010; Ou 2014; Zhang 2009a; Zhang 2013a; Zhang 2013b) and this sole person extracted the data from them.

Poor reporting in general may have introduced some bias in our assessment of some studies, as well as our failure to obtain the additional data we needed to clarify ambiguities resulting from such poor reporting. As we were unable to obtain Individual Patient Data for most (or almost none of the) studies, we considered chapter 18 (18.4.2) of the Cochrane Handbook for Systematic Reviews of Interventions (Stewart 2009). We believe we have minimised bias by reporting results in the original papers with the additional limited data obtained from the study authors or sponsors, rather than not reporting results of the majority of studies at all. However, the results we presented should be interpreted with the potential bias such reporting has introduced in mind. Unclear reporting issues, if there were any, are given specifically for each study within Characteristics of included studies and Characteristics of excluded studies when appropriate. Some bias was probably introduced because we had to code adverse effects from most studies in MedDRA (MedDRA 2010) ourselves in order to uniformly report them.

Agreements and disagreements with other studies or reviews

An overview of systematic reviews on treatments for acne (Smith 2011) identified three systematic reviews addressing laser and light therapies from 2009 (Hamilton 2009; Riddle 2009; Taylor 2009). We considered several other systematic reviews (Erceg 2013; Haedersdal 2008a; Wat 2014) and a recent narrative review (Pei 2015). Our conclusions are somewhat different from those of previous reviews. This is partly because we included studies published several years after some of the above reviews were done. We also screened out studies of non-RCT design due to our rigorous assessment of studies against criteria in our protocol. Our search was also more comprehensive as we included studies in languages other than English. Additionally, our extensive grey literature search identified several unpublished studies.

The conclusions of the previous reviews (Haedersdal 2008a; Hamilton 2009; Pei 2015) are in line with our conclusions regarding the general direction of evidence for green light, blue light, blue-red light and infrared light. The authors of earlier reviews emphasised the need for larger studies of better quality, in particular those comparing light therapies to standard treatments, or evaluating possible increased benefit of standard therapies in combination with light as compared to standard therapies alone, which is in agreement with our findings.

Our conclusions regarding the efficacy of pulsed-dye lasers PDL (i.e. yellow light) for acne are different to those of a recent systematic review on the efficacy of PDL for inflammatory skin diseases (Erceg 2013). The authors identified two RCTs included in our review (Orringer 2004; Seaton 2003), together with several non-RCT studies, and acknowledged design of such studies as the main limitation to the conclusions in their review. Erceg et al. graded the evidence according to the Oxford Center for Evidence-based Medicine Levels of Evidence (OCEBM 2011). The authors suggested a B level of recommendation (based on 'studies with consistent evidence from systematic reviews of cohort studies, individual cohort studies, including low quality RCTs, systematic reviews of case-control studies, individual case control studies or extrapolation from systematic reviews of RCTs or individual RCTs') and concluded that 'PDL seems to be an effective treatment for acne vulgaris' (Erceg 2013; OCEBM 2011). As the two RCTs identified in our review presented inconsistent results (Orringer 2004; Seaton 2003), and there is a paucity of further RCTs we believe that the grade of recommendation should be D -'a recommendation based on case reports or expert opinions or troubling, inconsistent or inconclusive studies of any level' (OCEBM 2011).

For similar reasons, our conclusions are different to those of a recent systematic review on intense pulsed light (IPL) for treatment of different dermatologic conditions, which included acne vulgaris (Wat 2014). We considered RCTs only, so we screened out many studies Wat et al included. We found that the evidence is still inconclusive, as opposed to 'treatment of acne vulgaris with IPL alone has the potential to achieve significant improvement in clinical severity and patient satisfaction' and 'IPL-PDT is a good treatment option for acne vulgaris' (Wat 2014). Furthermore, we rigorously assessed risk of bias using the Cochrane tool and found the overall risk of bias to be unclear or high in most of the studies. That, together with consideration of sample sizes and heterogeneity (regarding populations, interventions, controls and outcomes) prevented us from reaching firm conclusions. Additionally, we grouped interventions not only according to whether an active substance was used prior to illumination (IPL alone versus IPL-PDT), but also taking into account filters used to narrow the spectrum to selected wavelengths, as these varied across studies. Although the 530 nm to 750 nm filter ('the acne filter') was used most commonly, there were examples where different filters were used in different interventions even within the same study (Taub 2007). We believe filters introduce considerable heterogeneity and it would thus be inappropriate to lose sight of them when reaching conclusions on the effectiveness of IPL.

Our conclusions regarding the effectiveness of photodynamic therapies (PDT) are different to those of reviews on PDT-only studies (Riddle 2009; Sakamoto 2010; Taylor 2009), broader systematic reviews (Haedersdal 2008a; Hamilton 2009) and a recent narrative review (Pei 2015). We included several new studies on PDT. New studies with larger samples and better quality showed that MAL-PDT was not more effective than red light alone. We presented a larger and more conclusive body of evidence for that comparison. Similarly, the largest study on ALA-PDT in our review was identified through grey literature searches, included a total of 266 participants, and showed that ALA-PDT was not more effective than blue light alone. Recent studies on ALA-PDT activated by red light were also included in our review, including one originally in Mandarin, with 116 participants. Furthermore, new evidence has emerged on PDT modalities other than MAL-PDT and ALA-PDT.

We also found that severe adverse effects as defined in our protocol (blistering) were reported in studies on infrared light, IPL, 37J/cm² MAL-PDT with occlusion and ALA-PDT, whereas previous reviews mostly reported on non-severe adverse effects.

Like other Cochrane Reviews on treatments for acne (Arowojolu 2012; Cao 2015; Garner 2012), we found that many of the included studies had methodological, as well as reporting flaws and identified a lack of standardised outcome measures as an important problem. Previous reviews on core outcome measures in acne have highlighted this problem (Barratt 2009; Tan 2008). Lack of studies comparing light therapies with standard acne treatments is in line with general lack of evidence on comparative effectiveness of common acne therapies (Williams 2012).

Authors' conclusions

Implications for practice

Due to limited evidence, we are unable to draw firm conclusions from the results of our review. In particular, the lack of long-term outcomes was a major drawback because if a treatment does not give at least three months’ benefit, it could arguably be considered a treatment failure.

We identified the greatest body of moderate-quality evidence for the comparison of MAL-PDT and red light only. However, current evidence does not support the use of MAL-PDT as a standard therapy for people with moderate to severe acne.

The use of 20% ALA-PDT activated by blue light as a standard therapy for people with moderate to severe acne, was not supported by the evidence (low and very low quality) as this treatment did not show superior effectiveness in comparison with blue light alone. However, the overall evidence suggests that using lower ALA doses (15% and 10%), together with light modalities other than blue light may be of benefit. This is because several studies found that 20% ALA had more adverse effects (including blistering), whereas individual studies also found that, for example, 20% ALA activated by red light was not more effective than 15% ALA activated by red light, and 10% ALA activated by IPL was more effective than IPL alone.

Although the body of evidence on photodynamic therapies other than MAL-PDT and ALA-PDT has increased, it is still inconclusive, and so we could not draw firm conclusions.

We did not identify additional studies on blue light which would suggest recommending blue light as monotherapy with a greater strength of recommendation. Red light alone has shown promising results in several studies, but these were of high overall risk of bias. The new studies we included in our review also suggest greater effectiveness of blue-red light to that of blue light alone or placebo. Green light was more effective than placebo or no treatment, however these studies were very small.

Although the evidence was not conclusive and we were unable to combine it quantitatively, studies with a larger number of participants and of high overall risk of bias showed that infrared light was not more effective than placebo or no treatment and had more side effects, including severe ones.

Some of the studies in Characteristics of studies awaiting classification may alter the conclusions of the review once fully assessed.

Implications for research

Acne is a common, non-life-threatening condition. Assessment of different therapies are amenable to being tested by randomised controlled trials. However we found that the majority of trials were not properly randomised, with an overall unclear to high risk of bias and were poorly reported. It is well recognised that acne trials are often of poor methodological quality and also affected by poor reporting standards (Ingram 2010).

Methodological issues

Development of detailed guidance for clinical studies as well as standardisation of factors that influence the clinical evaluation of light therapies for acne is needed for future production of high-quality evidence. Several studies have adhered to FDA guidance for developing drugs for the treatment of acne vulgaris (FDA 2005), which is arguably the best available source for this purpose to date. However this is not specifically designed for light therapies and there have been marked technological advancements in the field since 2005.

A range of different assessment methods in acne trials often prevent, complicate and prolong collection, interpretation, extraction and synthesis of data. Economic impact and, more importantly, the impact this has on patient care needs to be addressed.

Although consensus and recommendations on a consistent use of investigator-assessed outcome measures would minimise this problem, consensus has still not been reached in the USA (Zaenglein 2016) or in Europe (Nast 2012). Further evaluation of validity, reliability and reproducibility of current outcome measures is needed to come up with the most appropriate ones to agree upon. This should be complemented by exploring relevant information technology and basic medical research advancements in developing innovative techniques for this purpose.

A minority of studies included participant-assessed outcomes. As with investigator-assessed outcomes, a variety of measures with questionable validity and reliability were used, particularly in trials with a split-face design and long follow-up periods. In individual trials participants commonly assessed their outcomes less often than investigators. This lack of monitoring of the participant perspective on treatment effects prevents adequate comparisons with the investigator perspective. Also, participants were not blinded in most trials, although the investigator assessors were. Due to the nature of interventions and adverse effects, blinding of participants and clinicians is challenging. Even when the participants do not evaluate the effects themselves, their awareness of the intervention may lead to systematic differences in the outcomes unrelated to the effects of interventions of interest (due to possible confounding factors, e.g. different care applied to different face sides or sleeping on the untreated side etc.). Attempts to blind the participants (or lack of such intentions) were not clearly reported in most studies, and so it seems that performance bias has often been overlooked in the studies we included. Future development of participant assessment methods need to be addressed and how they correspond to investigator assessment and compliance. Participant assessment should be performed with similar frequency to investigator assessment in future trials.

Only three studies included a quality-of-life assessment. We believe this important participant-assessed outcome should also be consistently incorporated into future trial protocols. Specific acne quality of life (QoL) instruments for adults and children have been developed (Tan 2008), but need further assessment and validation.

In this review we considered short-term (two to four weeks), medium (five to eight weeks) and long-term (longer than eight weeks) follow-up periods. Standardisation of time points for short, medium and long term assessment after final treatment is needed to enable synthesis of trial data. Furthermore, although of primary interest in this review, long-term data were scarce, similar to evidence for other acne treatments (Williams 2012), indicating a need to incorporate those assessment time points in the protocols of future trials. As patients are often treated at a young age, a way should also be sought to address follow-up and possible unwanted effects of light therapies decades after treatment.

Recent initiatives, such as The Cochrane Skin Group Outcomes Research Initiative (CSG-COUSIN, Schmitt 2016) and the Acne Core Outcomes Research Network (ACORN) (ACORN 2013) may accelerate improvement and standardisation of outcome measurement.

Reporting issues

Tools developed to improve reporting of randomised control trials are freely available, but have not been used in a majority of the reports of included studies. Recommendations of the CONSORT Statement (Schulz 2010) and its extension for non-pharmacologic treatment (NPT) interventions should be applied to all future reports. The following specific aspects of light therapies and acne trials should be reported:

  1. Light source identity including wavelength, fluence, pulse duration and spot size

  2. Total number and frequency of treatments as well as duration of single light treatment

  3. Definition of time of year (months) when treatment was administered

  4. Instructions given and compliance monitoring method if self-administered

  5. Whether sun protection advice was given if appropriate

  6. Whether previous acne treatment was stopped and when

  7. Whether concomitant acne treatment was permitted, and if so whether standardised

  8. Baseline measures of the participants for age, sex, Fitzpatrick skin types, duration and location of acne

  9. Initial severity of condition assessment measured by published grading system or preferably by lesion counts. Initial lesion counts should be reported separately for face sides in split-face trials.

  10. How many investigators performed assessment and their educational background or training

Adverse effects should be reported using lowest level terms (LLTs) as defined in the latest version of MedDRA (MedDRA 2010) and in accordance with CONSORT Statement extension on reporting of harms (Ioannidis 2004). Future studies should also adequately code adverse events into SOCs (System Organ Class) to enable adverse effects to be combined properly in future reviews. Adverse effects such as oedema, erythema, discolouration etc. which occur locally on the laser application site should consistently be coded using LLTs, which reflect the information that the reaction occurred locally at the application site (e.g. in SOC general disorders and application site conditions or in both that SOC and SOC skin and subcutaneous tissue disorders, and not solely in SOC skin and subcutaneous tissue disorders), taking into account directions set out in the MedDRA (MedDRA 2010).

Full results of a number of studies presented in conferences or registered in trials' registers were not published and study authors were unable to provide the full data, or reasons for their early termination. We believe that these details should be added to trials registers' records when appropriate or reported in the form of short communications to journals. Establishing a database for full results of acne clinical trials to enable storing data in a timely manner could also be considered.

Many study authors did not respond to our requests or were unable to provide original data when it was appropriate to combine them with results from other studies. Full results tables should be added as online supplementary material in journals when possible. Adequate data on participants' FPTs, sex, age and severity of acne would enable subgroup analyses and aid identification of differences in the treatment response of acne subtypes in future updates of this review. Furthermore, overcoming of reporting flaws together with standardisation of methodological aspects would enable multiple-treatment (network) meta-analyses of different light and other therapies for acne (Caldwell 2005).

Therapies

We have prioritised clinical outcomes in this review. However, further research on the underlying mechanisms of action, (including impact on seborrhoea, effects on sebocytes and sebaceous gland function, antimicrobial and immunomodulatory effects) are required to inform and guide future decisions about the conduct of clinical trials as well as clinical practice in treating acne with light therapies.

Future research must take into account the methodological and reporting issues, as well as whether the following have implications for practice: the possible superior effectiveness of MAL-PDT in those with severe acne; the use of blue light, red light, blue-red light and green light alone; 15% ALA-PDT activated by red or blue-red light; as well as PDT modalities other than MAL- and ALA-PDT, compared to conventional treatments, placebo or no treatment.

In summary, more robust, well planned studies with greater sample sizes comparing the effectiveness of common acne treatments with light therapies and their effect on reducing lesion counts would be welcomed together with prospective trial registration and adherence to the CONSORT guidelines.

Acknowledgements

We wish to thank those trial authors and sponsors who responded to our request for additional information; Elizabeth Doney, Nikolaos Mastellos, Tim Reeves, and Maggie Yin for their support in searches; Dr Quan Yang and Elicia Toon Yuan Ni for their help with screening of studies in Mandarin and data extraction from such studies we subsequently included; Marijan Sember and Dr Adriana Andric for their help with MedDRA; Toby Lasserson, Matthew Grainge and Dr Monika Semwal for their helpful comments and suggestions. We also wish to thank Finola Delamere, Laura Prescott, Hywel Williams and Helen Scott from Cochrane Skin for their continuous support during work on this review.

The Cochrane Skin editorial base wish to thank Sue Jessop, who was the Cochrane Dermatology Editor for this review; Ben Carter, who was the statistical editor; Esther van Zuuren, who was the methods editor; the clinical referees, Brigitte Dréno and Gloria Sanclemente; and the consumer referee, Jack Tweed, as well as Denise Mitchell, who copy-edited the review.

Data and analyses

Download statistical data

Comparison 1. Blue-red light versus placebo
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Participant's and investigator's global assessment of improvement at final treatment1 Risk Ratio (M-H, Fixed, 95% CI)Totals not selected
1.1 Participant's global assessment of improvement1 Risk Ratio (M-H, Fixed, 95% CI)0.0 [0.0, 0.0]
1.2 Investigator's global assessment of improvement1 Risk Ratio (M-H, Fixed, 95% CI)0.0 [0.0, 0.0]
Analysis 1.1.

Comparison 1 Blue-red light versus placebo, Outcome 1 Participant's and investigator's global assessment of improvement at final treatment.

Comparison 2. Blue-red light versus topical benzoyl peroxide
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Participant's and investigator's global assessment of improvement at final treatment1 Risk Ratio (M-H, Fixed, 95% CI)Totals not selected
1.1 Participant's global assessment of improvement1 Risk Ratio (M-H, Fixed, 95% CI)0.0 [0.0, 0.0]
1.2 Investigator's global assessment of improvement1 Risk Ratio (M-H, Fixed, 95% CI)0.0 [0.0, 0.0]
Analysis 2.1.

Comparison 2 Blue-red light versus topical benzoyl peroxide, Outcome 1 Participant's and investigator's global assessment of improvement at final treatment.

Comparison 3. Blue-red light versus blue light alone
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Participant's and investigator's global assessment of improvement at final treatment1 Risk Ratio (M-H, Fixed, 95% CI)Totals not selected
1.1 Participant's global assessment of improvement1 Risk Ratio (M-H, Fixed, 95% CI)0.0 [0.0, 0.0]
1.2 Investigator's global assessment of improvement1 Risk Ratio (M-H, Fixed, 95% CI)0.0 [0.0, 0.0]
Analysis 3.1.

Comparison 3 Blue-red light versus blue light alone, Outcome 1 Participant's and investigator's global assessment of improvement at final treatment.

Comparison 4. Vehicle + 1000 s blue light versus vehicle + 500 s blue light
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Participant's global assessment of improvement at 6 weeks1 Risk Ratio (M-H, Fixed, 95% CI)Totals not selected
2 Investigator's global assessment of improvement1 Risk Ratio (M-H, Fixed, 95% CI)Totals not selected
2.1 1000 s blue light versus 500 s blue light at 3 weeks1 Risk Ratio (M-H, Fixed, 95% CI)0.0 [0.0, 0.0]
2.2 1000 s blue light versus 500 s blue light at 6 weeks1 Risk Ratio (M-H, Fixed, 95% CI)0.0 [0.0, 0.0]
Analysis 4.1.

Comparison 4 Vehicle + 1000 s blue light versus vehicle + 500 s blue light, Outcome 1 Participant's global assessment of improvement at 6 weeks.

Analysis 4.2.

Comparison 4 Vehicle + 1000 s blue light versus vehicle + 500 s blue light, Outcome 2 Investigator's global assessment of improvement.

Comparison 5. 20% ALA-PDT versus vehicle plus blue light
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Participant's global assessment of improvement at 6 weeks1266Risk Ratio (M-H, Random, 95% CI)0.87 [0.72, 1.04]
1.1 20% ALA-PDT (1000 s) versus vehicle plus blue light (1000 s)1135Risk Ratio (M-H, Random, 95% CI)0.94 [0.72, 1.22]
1.2 20% ALA-PDT (500 s) versus vehicle plus blue light (500 s)1131Risk Ratio (M-H, Random, 95% CI)0.81 [0.63, 1.03]
2 Investigator's global assessment of improvement at 3 weeks1266Risk Ratio (M-H, Random, 95% CI)0.92 [0.56, 1.52]
2.1 20% ALA-PDT (1000 s) versus vehicle plus blue light (1000 s) at 3 weeks1135Risk Ratio (M-H, Random, 95% CI)0.85 [0.44, 1.65]
2.2 20% ALA-PDT (500 s) versus vehicle plus blue light (500 s) at 3 weeks1131Risk Ratio (M-H, Random, 95% CI)1.02 [0.47, 2.18]
3 Investigator's global assessment of improvement at 6 weeks1266Risk Ratio (M-H, Random, 95% CI)0.81 [0.51, 1.29]
3.1 20% ALA-PDT (1000 s) versus vehicle plus blue light (1000 s) at 6 weeks1135Risk Ratio (M-H, Random, 95% CI)0.92 [0.50, 1.71]
3.2 20% ALA-PDT (500 s) versus vehicle plus blue light (500 s) at 6 weeks1131Risk Ratio (M-H, Random, 95% CI)0.70 [0.35, 1.39]
Analysis 5.1.

Comparison 5 20% ALA-PDT versus vehicle plus blue light, Outcome 1 Participant's global assessment of improvement at 6 weeks.

Analysis 5.2.

Comparison 5 20% ALA-PDT versus vehicle plus blue light, Outcome 2 Investigator's global assessment of improvement at 3 weeks.

Analysis 5.3.

Comparison 5 20% ALA-PDT versus vehicle plus blue light, Outcome 3 Investigator's global assessment of improvement at 6 weeks.

Comparison 6. 20% ALA-PDT 30 min incubation plus IPL versus 20% ALA-PDT 3 h incubation plus IPL
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Participant's and investigator's global assessment of improvement at 12 weeks1 Risk Ratio (M-H, Fixed, 95% CI)Totals not selected
1.1 Participant's global assessment of improvement1 Risk Ratio (M-H, Fixed, 95% CI)0.0 [0.0, 0.0]
1.2 Investigator's global assessment of improvement1 Risk Ratio (M-H, Fixed, 95% CI)0.0 [0.0, 0.0]
Analysis 6.1.

Comparison 6 20% ALA-PDT 30 min incubation plus IPL versus 20% ALA-PDT 3 h incubation plus IPL, Outcome 1 Participant's and investigator's global assessment of improvement at 12 weeks.

Comparison 7. 20% ALA-PDT plus 560 nm IPL versus 560 nm IPL alone
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Participant's global assessment of improvement at 8 weeks1 Risk Ratio (M-H, Fixed, 95% CI)Totals not selected
Analysis 7.1.

Comparison 7 20% ALA-PDT plus 560 nm IPL versus 560 nm IPL alone, Outcome 1 Participant's global assessment of improvement at 8 weeks.

Comparison 8. 20% ALA-PDT 1000 s versus 500 s
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Participant's global assessment of improvement at 6 weeks1 Risk Ratio (M-H, Fixed, 95% CI)Totals not selected
2 Investigator's global assessment of improvement1 Risk Ratio (M-H, Fixed, 95% CI)Totals not selected
2.1 Investigator's global assessment of improvement at 3 weeks1 Risk Ratio (M-H, Fixed, 95% CI)0.0 [0.0, 0.0]
2.2 Investigator's global assessment of improvement at 6 weeks1 Risk Ratio (M-H, Fixed, 95% CI)0.0 [0.0, 0.0]
Analysis 8.1.

Comparison 8 20% ALA-PDT 1000 s versus 500 s, Outcome 1 Participant's global assessment of improvement at 6 weeks.

Analysis 8.2.

Comparison 8 20% ALA-PDT 1000 s versus 500 s, Outcome 2 Investigator's global assessment of improvement.

Comparison 9. 20% ALA-PDT versus 15% ALA-PDT