Drugs for discoid lupus erythematosus

Description of the condition

Discoid lupus erythematosus (DLE) is a chronic and common form of cutaneous lupus — a form of skin inflammation of unknown cause. It occurs particularly on sun-exposed skin, such as the face, ears, and scalp, but occasionally is much more extensive, involving large areas of skin. Some people with DLE have associated disease in other parts of the body, but the majority of people with DLE are otherwise healthy. People with DLE are seen frequently in skin clinics throughout the world.

There have been no accurate studies of the prevalence of DLE, although all forms of lupus erythematosus are particularly common in women of childbearing age. The diagnosis of DLE is usually made easily on clinical grounds, characteristic clinical findings being erythema (redness), follicular plugging (blocked and enlarged openings to hair follicles), pigmentary disturbances, telangiectasia (dilated capillaries), and atrophy (skin thinning). When the diagnosis is in doubt, a skin biopsy (microscopic examination) may be required to confirm the diagnosis (Kuhn 2014).

In an attempt to improve consistency in the reporting of clinical trials, the CLASI (Cutaneous Lupus Erythematosus Area and Severity Index) scoring system has been developed (Albrecht 2005). The CLASI allows assessment of both active inflammation (erythema, scale, and hypertrophy) and damage (dyspigmentation, scarring, atrophy), in different body areas, but does not include a quality of life score. It has not been tested in different populations.

Early effective treatment may lead to total clearing of the skin lesions, but failure of treatment results in permanent scarring. The depressed scars, hair loss, and pigmentary changes are often extremely disfiguring, particularly in darker-skinned people. Permanent scarring is the rule if treatment is delayed or inadequate. Great emotional distress, social isolation, and difficulty obtaining work are problems frequently suffered by these patients. Involvement of the skin of the fingers and toes may markedly impair hand function and limit walking (Vasquez 2013). In addition, people with DLE are at risk of developing squamous cell carcinoma in the scarred areas (Fernandes 2015).

Description of the intervention

The anti-malarial drug chloroquine is the traditional therapy for DLE (Goldman 1953; Brodthagen 1959; Callen 1982). It is cheap and readily available worldwide. Flare-ups of systemic lupus were noted to occur with increased frequency after chloroquine was stopped (Rothfield 1963). However, some people cannot tolerate the drug and it is known to fail in others. Other anti-malarial agents have been used, sometimes in combination with chloroquine, but are not widely available (Kierland 1953; Maguire 1962; Feldman 1994).

In recent years, hydroxychloroquine has been used in place of chloroquine in many countries, because ocular side-effects were reported to be less frequent with hydroxychloroquine, but retinal toxicity has subsequently also been recognised with hydroxychloroquine (Geamanu 2014). The efficacy of the two forms of the drug is thought to be the same, but no trials exist to confirm this in DLE (Kreuter 2009). In discussing the treatment of DLE in 1961, Prakken commented that: "This [lack of evidence] is illustrated by a survey of our literature on the treatment of chronic lupus erythematosus. Although many papers agree on their reports of favourable results, the value of our therapy remains doubtful because the influence of the placebo effect and the publications effect has not been eliminated" (Prakken 1961). However, Rees felt that discoid lupus responded so dramatically to chloroquine that "double-blind studies are not required" (Rees 1963). In 1964, Kraak performed a literature search and concluded that available information was insufficient to evaluate the many therapies that are recommended: "The literature leaves room for the alarming assumption that none of the drugs might do any good" (Kraak 1964). His group performed a non-randomised double-blind trial comparing hydroxychloroquine with placebo (Kraak 1965). Spontaneous improvement or clearing occurred in some of their patients, particularly in the winter months (Kraak 1965). The group on chloroquine showed a greater tendency to improve (Kraak 1965). Other workers have been so concerned about possible ocular side-effects (thought to be very rare on current low dosage regimens) that chloroquine was excluded from their trial (Mackey 1974). Smoking may reduce the efficacy of chloroquine in people with lupus: a recent review reports studies that demonstrate an inverse relationship between number of cigarettes smoked and chloroquine blood levels (Chasset 2015). Unfortunately many of the other agents, reported anecdotally to be of value in discoid lupus erythematosus, have considerable potential side-effects and some, such as thalidomide, may produce major malformations in the foetus if taken during pregnancy (Cortés-Hernández 2012; Zhou 2013).

Many other agents have been used to treat this condition (Callen 1997). Sunscreens and sun avoidance reduce flares of cutaneous lupus, but are not usually sufficient alone (Kuhn 2011a). Other possible treatments include dapsone (Coburn 1982; Lindskov 1986), topical and intralesional steroids (please note that the term 'steroid' has been used in this review in place of the term 'corticosteroid', although both terms have the same meaning) (Jansen 1965; Marsden 1968; Reyman 1974; Smith 1986), gold (Dalziel 1986), clofazimine (Mackey 1974), retinoids (Ruzicka 1988; Shornick 1991), methotrexate (Goldstein 1994), azathioprine (Tsokos 1985), lenalidomide (Okon 2014), thalidomide (Hasper 1983; Knop 1983; Naafs 1985), topical salbutamol (Jemec 2009), sulphasalazine (Artuz 1996), phenytoin (Rodriguez-Castellanos 1995), interferon alpha-2a (Martinez 1992), topical calcineurin blockers (pimecrolimus and tacrolimus) (Wollina 2007; Tsellos 2008), and biological agents such as abatacept (Merrill 2010a), belimumab (Manzi 2012), efalizumab (Usmani 2007), etanercept, infliximab, rituximab, sifalimumab (Merrill 2011) and sirukumab (Szepietowski 2013)

How the intervention might work

Steroids have anti-inflammatory and immunosuppressive effects, acting on genetic material in the cell nucleus, via steroid receptors in the cell (Norris 2005). Tacrolimus and pimecrolimus, which are calcineurin inhibitors, inhibit the activity of T lymphocytes (via calcium channels), a key cell in the development of cutaneous lupus (Khandpur 2004). Salbutamol may exert an anti-inflammatory effect by activating receptors on the surface of cells, particularly lymphocytes and eosinophils (Jemec 2009). Evidence suggests that chloroquine and hydroxychloroquine which are antimalarials, have an effect on inflammation by reducing antigen presentation and the production of inflammatory cytokines (Rainsford 2015). The anti-inflammatory mechanism of action of retinoids, such as acitretin, are not fully understood. Retinoid receptors allow access to the cell nucleus, and retinoids promote normal cell growth (Schroeder 2007).

Why it is important to do this review

Permanent scarring is likely if treatment is delayed or inadequate. People with DLE often experience great emotional distress. Social isolation and difficulty obtaining work are frequent problems. The relative efficacies of possible forms of treatment compared to placebo or each other need to be clearly established. Some of these agents (such as the cytotoxics) are potentially toxic (Weinblatt 2013; Goldberg 2015). Both thalidomide and the retinoids are highly teratogenic (they may cause birth defects when given to pregnant women), which renders their use in fertile women problematic (Ortiz 2013; Zhou 2013). It is this combination of uncertainty of relative effectiveness and some potential to cause considerable harm that has fuelled the need for this systematic review. This is an update of a Cochrane Review first published in 2000 (Jessop 2000); and previously updated in 2009 (Jessop 2009).

Criteria for considering studies for this review

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).

Data collection and analysis

SJ and DW, with PJ's assistance, extracted the data independently and resolved differences by discussion.

MG examined and described the statistical information in individual trials and edited or wrote the sections on measures of treatment effect, data synthesis and unit of analysis issues and created the 'Summary of findings' tables.

Description of studies

Results of the search

We retrieved 71 records from the searches for the 2009 review (Jessop 2009), of which two met the criteria for inclusion (Roenigk 1980 and Ruzicka 1992). Literature searches in 2014, 2015, and 2016 yielded a further 155 studies. We identified 12 studies from our searches of trials registers. In total we screened 238 records. We excluded 202 records on the basis of titles and abstracts. Of the remaining 36 records, one article was not available in full text and has been added to studies awaiting classification (Pothinamthong 2012) along with eight trials registry records (see Characteristics of studies awaiting classification). One study is ongoing (see Characteristics of ongoing studies). The remaining 26 articles were assessed in full text for eligibility.

Of the 26 full texts, we excluded 23 studies (14 did not describe outcomes in the DLE group specifically and others were excluded because they referred to photoprotection trials and not drug trials or were not RCTs) (see Characteristics of excluded studies). We included three new studies in this update (Barikbin 2009, Jemec 2009, and Kuhn 2011), giving a total of five included studies altogether (see Characteristics of included studies). We noted that diagnosis of DLE was generally based on clinical judgement in the early studies, but investigators required histological (skin biopsy) confirmation in the three new studies.

Of the nine drug trials of DLE listed in Characteristics of studies awaiting classification, the interventions included etanercept, other biologics, R-salbutamol, pimecrolimus, and an oral retinoid, alitretinoin. Half of the studies were described as complete but results have not yet been published. We attempted to contact two of the investigators, without success. One investigator reported that their study did not include analysis of the subset with DLE.

For a summary of our screening process see Figure 1.

Risk of bias in included studies

Overall, risk of bias was considered low or unclear for most parameters for the included studies.

See Figure 2 for the risk of bias summary of the judgements for each included study.

Effects of interventions

See: Summary of findings for the main comparison Fluocinonide 0.05% compared with hydrocortisone 1%; Summary of findings 2 Acitretin 50 mg daily compared with hydroxychloroquine 400 mg daily; Summary of findings 3 Pimecrolimus 1% compared with betamethasone 17-valerate 0.1%; Summary of findings 4 R-salbutamol 0.5% topical cream compared with placebo; Summary of findings 5 Tacrolimus 0.1% cream compared with placebo

Summary of main results

Five trials are reported in this version of the review, with a total of 197 participants. Three of the studies are new to this update (Barikbin 2009; Jemec 2009; Kuhn 2016). Two of the trials included small numbers of participants (10 in Barikbin 2009 and 14 in Kuhn 2011).

Interventions in the five trials were topical steroids, hydroxychloroquine, acitretin, topical salbutamol, topical tacrolimus, and topical pimecrolimus.

The study of topical steroids provides some evidence that a potent steroid cream, fluocinonide 0.05%, may be superior to low-potency hydrocortisone 1% in the complete resolution of discoid lupus erythematosus skin lesions (Summary of findings for the main comparison) (Roenigk 1980). The ideal duration of treatment or likelihood of recurrences following treatment has not been established. Our second primary outcome (percentage of people with clearing of erythema in at least 50% of lesions) was not measured.

In the comparative trial of hydroxychloroquine 400 mg and acitretin 50 mg (Ruzicka 1992), no difference in efficacy was demonstrated between the two drugs in terms of the primary outcome, with complete resolution of skin lesions reported in 46% of people taking acitretin and in 50% of those on hydroxychloroquine. Regarding the second primary outcome, 42% of participants taking acitretin showed improvement in erythema in at least 50% of lesions compared with 68% of the hydroxychloroquine group (Summary of findings 2).

The two studies of calcineurin inhibitors (pimecrolimus 1% and tacrolimus 0.1%) (Barikbin 2009; Kuhn 2011), and the study of R-salbutamol 0.5% topical cream (Jemec 2009), unfortunately included only 61 participants in all and did not report the primary outcomes of the review (Summary of findings 3; Summary of findings 4; Summary of findings 5).

There is some evidence that the trial drugs were free of serious side-effects, but the short duration of most of the trials does not allow any conclusion about long-term adverse effects of topical steroids. Adverse events were reported in the 12-week tacrolimus 0.1% study, which was compared to placebo (Kuhn 2011), but none were serious. Roenigk 1980 (fluocinonide cream 0.05% vs hydrocortisone cream 1%) was short (six weeks in each arm), and thus side-effects of long-term topical steroid use could not be assessed. Ruzicka 1992 (acitretin 50 mg vs hydroxychloroquine 400 mg) was the only study that reported discontinuation of participants due to adverse events (four people receiving acitretin discontinued treatment due to dry lips and gastrointestinal symptoms). Jemec 2009 compared topical salbutamol 0.5% cream with placebo (vehicle), and both groups reported minor adverse events but no serious adverse events. Barikbin 2009 did not present data on adverse events.

We are uncertain about the effect of the interventions on patient satisfaction/quality of life, as this primary outcome was not measured by any of our included studies. The secondary outcomes of preventing new lesions and implications for healthcare costs were not assessed in any study either. Relapse rate when medication was stopped or reduced was reported in Barikbin 2009 and Kuhn 2011. There was no relapse reported in Barikbin 2009 at eight weeks, after stopping the trial medication at four weeks; whilst in Kuhn 2011, participants relapsed by 12 weeks.

Overall completeness and applicability of evidence

The studies in this review have not addressed all the objectives of the review and do not identify the most effective treatment for DLE in all types of participants.

The studies identified in this review did not address (or addressed inadequately) the relative benefits of many currently used interventions, against each other or against placebo, including hydroxychloroquine (or chloroquine), retinoids, topical steroids, calcineurin inhibitors, methotrexate, azathioprine, lenalidomide or biological agents. Outcome measures failed to address quality of life measures or, in most cases, relapse rate after medication was stopped. Most studies did not report on complete resolution of skin lesions or clearing of erythema either. Although different populations participated in the trials, there were no trials from Asia, Africa, Australia or South America.

Quality of the evidence

Five trials, involving 197 participants were included. Some primary outcome measures in this review (per cent with complete resolution, percentage with at least 50% improvement in erythema) were inadequately addressed in the majority of included trials. Furthermore, quality of life was not assessed in any of the five included studies.

Where quantitative results were provided for primary outcomes, the overall quality of the evidence was judged to be low owing to the small size of most of the studies and high risk of bias on important domains such as incomplete outcome data and non-comparability of treatment arms at baseline.

All trials were judged to be at low risk of bias for blinding as they were all double-blind and interventions were identical in appearance (although for two studies, Roenigk 1980 and Ruzicka 1992, the methods used to ensure blinding were not fully described). In four studies the method used to generate the allocation sequence was adequate; however, one study was a quasi-randomised trial where patients were numbered based on time of referral and odd numbered participants were allocated to one treatment and even numbered patients to the other. However, as this study did not include any of our pre-specified primary or secondary outcome measures (instead a total mean severity score was calculated before and after treatment for each participant), no data for this study were presented in our 'Summary of findings' tables. The method used to conceal allocation was also unclear in three of the studies (Jemec 2009, Roenigk 1980; Ruzicka 1992); however, this was judged to be adequate for the other two studies (Barikbin 2009; Kuhn 2011), For four studies, either the numbers analysed were the same as the number recruited or the number of dropouts was very small, so these were judged to be at low risk of bias for incomplete outcome data. However in the Roenigk 1980 study, 15 of the 78 participants did not contribute data towards the primary outcome of clearing or excellent improvement for reasons which may relate to outcome (Summary of findings for the main comparison). Therefore, the quality of evidence on whether there is a difference between fluocinonide and hydrocortisone (low-potency steroid) with respect to clearing or excellent improvement was downgraded by one level. In the trial by Ruzicka 1992, there were more participants with DLE in the hydroxychloroquine group. This created unbalanced treatment groups, which could have influenced the outcomes of clearing or excellent improvement and more than 50% reduction in erythema (Summary of findings 2). We identified no other potential sources of bias that might have had a direct effect on any risk measures presented, However, in many instances our ability to assess these was unclear.

Sample size was judged to be inadequate in all of the included studies, except in the case of Roenigk 1980 with 78 participants (others ranged from 10 to 58 participants per study). This resulted in imprecise measures of effect in instances where these were presented, and evidence was therefore downgraded accordingly in instances where assessment of the quality of evidence is provided (Summary of findings for the main comparison; Summary of findings 2). The small numbers in Ruzicka 1992 may weaken the validity of this study, as does the lack of a control arm, as neither intervention is of established efficacy (Summary of findings 2). The inclusion of people with subacute cutaneous lupus erythematosus is an even greater source of possible error and the results require confirmation.

We also assessed the quality of evidence for adverse events leading to discontinuation or significant morbidity in four out of five comparisons in the 'Summary of findings' tables. The quality of evidence was moderate, with downgrading for imprecision in all comparisons because the evidence was based on only one study with a small number of participants. Furthermore, only Ruzicka 1992 reported participants discontinuing, while the remaining studies only reported minor adverse events. The overall sample size may not have been large enough to identify serious adverse events.

All of the comparisons included in this review were assessed in single studies, which did not allow us to assess consistency of results across studies. As such it was not possible in our review to assess what factors could potentially contribute to differences in findings between studies (heterogeneity).

Indirectness cannot be assessed with confidence in the included studies. As the studies were performed in specialist clinics and most people with DLE are likely to be treated initially at that level, the setting is not out of line with clinical practice.

Potential biases in the review process

We attempted to conduct a comprehensive search for studies, but the fact that one study has not yet been incorporated may be a source of potential bias (Pothinamthong 2012). Studies in which participants had SLE or undefined cutaneous LE were excluded from the review, but two studies that included participants with DLE and SCLE, (Ruzicka 1992 and Kuhn 2011), were included and this may have introduced a bias. There were 12 studies identified in trials registries, but unpublished and most without contact details. The reviewers were unable to establish the outcome of these trials. After the publication of the original protocol and review (Jessop 2000), a scoring measure was published, and has been widely used for reporting outcomes in DLE (Albrecht 2005). One included study (Jemec 2009) reported outcomes using this scoring measure, but the data were not included in our systematic review because this type of outcome measure was not specified in the review protocol.

Agreements and disagreements with other studies or reviews

We are not aware of any other relevant reviews or studies.

Appendix 1. CENTRAL (Cochrane Library) search strategy

#1 discoid lupus erythematosus
#2 (cutaneous or skin) and lupus erythematosus
#3 MeSH descriptor: [Lupus Erythematosus, Cutaneous] explode all trees
#4 MeSH descriptor: [Lupus Erythematosus, Discoid] explode all trees
#5 #1 or #2 or #3 or #4

Appendix 2. MEDLINE (Ovid) search strategy

1. discoid lupus erythematosus.mp.
2. Lupus Erythematosus, Cutaneous/
3. cutaneous lupus.mp.
4. Lupus Erythematosus, Discoid/
5. or/1-4
6. randomised controlled trial.pt.
7. controlled clinical trial.pt.
8. randomized.ab.
9. placebo.ab.
10. clinical trials as topic.sh.
11. randomly.ab.
12. trial.ti.
13. 6 or 7 or 8 or 9 or 10 or 11 or 12
14. exp animals/ not humans.sh.
15. 13 not 14
16. 5 and 15

[Lines 6-15: Cochrane Highly Sensitive Search Strategy for identifying randomised trials in MEDLINE: sensitivity- and precision-maximizing version (2008 revision)]

Appendix 3. Embase (Ovid) search strategy

1. discoid lupus erythematosus/
2. discoid lupus erythematosus.mp.
3. skin lupus erythematosus/
4. cutaneous lupus erythematosus.mp.
5. or/1-4
6. crossover procedure.sh.
7. double-blind procedure.sh.
8. single-blind procedure.sh.
9. (crossover$ or cross over$).tw.
10. placebo$.tw.
11. (double$ adj blind$).tw.
12. allocat$.tw.
13. trial.ti.
14. randomised controlled trial.sh.
15. random$.tw.
16. or/6-15
17. exp animal/ or exp invertebrate/ or animal experiment/ or animal model/ or animal tissue/ or animal cell/ or nonhuman/
18. human/ or normal human/
19. 17 and 18
20. 17 not 19
21. 16 not 20
22. 5 and 21

Appendix 4. LILACS search strategy

(discoid$ and lupus and (erythematosus or eritematoso))

These terms combined with the Controlled clinical trials topic-specific query filter.

Disclaimer

This project was supported by the National Institute for Health Research (NIHR), via Cochrane Infrastructure funding to the Cochrane Skin Group. The views and opinions expressed herein are those of the authors and do not necessarily reflect those of the Systematic Reviews Programme, NIHR, National Health Service or the Department of Health.

Internal sources

• Nottingham University, UK.

External sources

• The National Institute for Health Research (NIHR), UK.

  The NIHR, UK, is the largest single funder of the Cochrane Skin Group.

Differences between the protocol and the current update

For differences between other published versions, please see the 'Differences between protocol and review section' within the original publications.

Note: many of the methods section headings are missing from the protocol and previously published versions. In this update, we have tried to rectify this, in line with the current Cochrane Handbook and RevMan software.

Background: this section has been updated.

Objectives: the sentence 'To identify the need for further study to make rational clinical decisions possible when treating cutaneous lupus' was omitted from this update in line with current Cochrane Handbook for Systematic Reviews of Interventions and RevMan 5 software and the previously published versions of the review as it was considered redundant.

Types of interventions: we planned to include other anti-malarial quinines other than chloroquine and hydroxychloroquine, but did not find any RCTs for these interventions. Although not planned in the protocol, in this update we did additionally include lenalidomide, biological agents, (including abatacept, adalimumab, belimumab, etanercept, efalizumab, infliximab, rituximab, sifalimumab and sirukumab), topical calcineurin antagonists, (tacrolimus and pimecrolimus) and topical salbutamol as these agents had become available and trials had been performed in connective tissue disorders. We also clarified the types of comparators we would accept, as this detail was not specified in the protocol. We expanded the list of excluded interventions to include phototherapy and photoprotection, which were not originally listed in the protocol, for clarification only. This does not represent a change in methodology.

Types of outcome measures: in a previous version of the review, under Primary outcomes, we added 'percentage of people with' to our 'complete resolution of skin lesions…' and 'clearing of erythema…' outcomes, and these have been retained (Jessop 2009). Note that we accepted the following terms to describe complete resolution: complete clearing, clearing, marked improvement and excellent improvement.

Searching other resource: although not planned in the protocol, we searched Index Medicus by hand for studies relating to treatment of discoid lupus erythematosus for the years 1956 to 1966 because we thought it important, and we recorded all adverse events reported in the included and excluded studies as we recognise the need to report adverse events.

Search methods for identification of studies: the databases and date ranges that we planned to search in the protocol have been extended for this update review, in line with current standard search methods.

Data collection and analysis: In the original protocol, we had not made plans at all regarding how to deal with within-patient studies, missing data, sensitivity analysis, and 'Summary of findings' tables, but we now have in line with current requirements.

Where results are estimated for individual studies with low numbers of outcomes (< 10 in total) or where the total sample size is less than 30 participants, we decided to report the proportion of outcomes in each treatment group together with a P value from a Fisher’s Exact test. This follows guidance provided by the Skin Group Statistics editors on how to deal with small numbers, made available since the last update of this review.

"We tested statistical heterogeneity using the I² statistic (0% to 40%: may not be important; 30% to 60%: may represent moderate heterogeneity; 50% to 90%: may represent substantial heterogeneity; 75% to 100%: represents considerable heterogeneity) (Higgins 2011).” An assessment of how we would deal with heterogeneity was missing from the original protocol therefore we have amended this at this stage in the event that future updates of this review will contain a pooled analysis of two or more studies.

Assessment of risk of bias in included studies: the text within this section was written in the 'Quality rating of included studies' section of the protocol. The review has been amended to follow the RevMan 5 recommended headings and the new Cochrane Handbook for Systematic Reviews of Interventions guidelines.

Two additional headings have been added to this section: Comparability of the two arms and Intention to treat analysis.

According to the Cochrane Handbook for Systematic Reviews of Interventions, controlled trials that allocate participants by quasi-randomisation, or that fail to conceal allocation during recruitment, are at risk of selection bias. We have included such trials, but have indicated that there is high risk of selection bias.

Imbalance at baseline has been assessed in this version of the review and added to the risk of bias tables.

Measures of treatment effect: although not planned in the protocol, for any significant outcomes (P < 0.05) we wanted to present the number needed to treat for an additional harmful outcome (NNTH) where the first treatment in the comparison is harmful (risk ratio (RR) > 1 for safety outcomes). We did not do this as no significant harms were reported.

In the protocol, we did not pre-state the effect measures that we would report. For dichotomous outcomes (primary outcomes 1 and 2 and adverse events), results are presented as RR with 95% confidence intervals (CI). For this update, for continuous outcomes (primary outcome 3), we planned to present results in the form of mean differences (with 95% CIs). Where there were treatment comparisons where studies used different scales for the same outcome, we planned to convert results to standard mean differences (SMD) to allow pooling of results. We could not carry out these plans because the studies were few and heterogeneous.

Unit of analysis issues > Cross-over trials: in the protocol we did not plan how to analyse cross-over trials; but having included such a study, we followed the guidance provided in Section 16.4.4 of the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011). Where no evidence of a carry-over effect was present, analyses of paired differences were presented if these data could be obtained from the trial report. Otherwise, only data from the first treatment period were used.

Unit of analysis issues >Within-patient studies: although not planned in the protocol, internally controlled trials were analysed using techniques for paired designs (e.g. paired t-test, McNemar's test). Where appropriate, these were included in additional data tables and not pooled with parallel group trials.

Assessment of heterogeneity: in the protocol, we had planned to use statistical tests for homogeneity between studies. However, assessment of heterogeneity was not performed as no findings were based on pooled results from 2 or more trials.

Data synthesis: in the protocol, we planned to draw up a synthesis of included trials; however, this was not possible with the data in the studies we found, due to heterogeneity of study methods and interventions. No two trials investigated the same interventions.

Subgroup analysis and investigation of heterogeneity: in the protocol, we had planned to perform subgroup analysis on the following measures: disseminated DLE versus localised DLE; DLE with systemic lupus versus DLE without systemic lupus; histologically proven versus clinically diagnosed DLE; and effect of ethnic group on outcome. However, we were unable to address these issues, as none of the studies provided data on individual subgroups, (and two of the trials enrolled small numbers of participants).

Trials that included more than one subset of cutaneous lupus were included if most participants had DLE and the outcome data described outcomes for DLE.

Other: we planned where possible to calculate cost effectiveness ratios using quality-of life-measures; however, the studies identified did not provide these data.

'Summary of findings' table: although not planned in the protocol, we used GRADEpro 2008 to create a 'Summary of findings' table as suggested in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011). In this, we have summarised the primary outcomes for the most important comparisons. Subgroup analysis was not possible.