Transcatheter arterial chemoembolisation followed by three‐dimensional conformal radiotherapy versus transcatheter arterial chemoembolisation alone for primary hepatocellular carcinoma in adults

Liming Lu<sup>a</sup>; Jingchun Zeng; Zehuai Wen; Chunzhi Tang<sup>a</sup>; Nenggui Xu

doi:10.1002/14651858.CD012244.pub2

Quimioembolización arterial transcatéter seguida de radioterapia conformacional tridimensional versus quimioembolización arterial transcatéter solo para el carcinoma hepatocelular primario en adultos

Declaraciones de intereses de los autores

Versión publicada: 16 febrero 2019 Historial de versiones

https://doi.org/10.1002/14651858.CD012244.pub2

Contraer todo Desplegar todo

Resumen

disponible en

Antecedentes

El carcinoma hepatocelular, también conocido como hepatoma maligno, es una neoplasia maligna primaria del hígado. A pesar de la monitorización periódica realizada en las poblaciones de alto riesgo, la mayoría de los pacientes con carcinoma hepatocelular se diagnostican en un estadio avanzado. En consecuencia, sólo una minoría de los pacientes con la enfermedad son candidatos para la resección quirúrgica tras el diagnóstico.

Objetivos

Comparar los efectos beneficiosos y perjudiciales de la quimioembolización arterial transcatéter (QEAT) seguida de radioterapia conformacional tridimensional (RTC‐3D) versus la QEAT sola en pacientes adultos con carcinoma hepatocelular primario, que no es candidato para la resección quirúrgica.

Métodos de búsqueda

Se realizaron búsquedas en el registro de ensayos controlados del Grupo Cochrane Hepatobiliar (Cochrane Hepato‐Biliary Group Controlled Trials Register), CENTRAL, MEDLINE, Embase, LILACS, Science Citation Index Expanded, y en Conference Proceedings Citation Index – Science hasta el 31 mayo 2018. Se verificaron las listas de referencias de todos los estudios incluidos y las revisiones relacionadas para obtener otros artículos relevantes.

Criterios de selección

Se incluyeron todos los ensayos clínicos aleatorios que comparan la QEAT seguida de la RTC‐3D versus la QEAT sola en pacientes con carcinoma hepatocelular primario.

Obtención y análisis de los datos

Se utilizaron los procedimientos metodológicos estándar recomendados por Cochrane. Se presentaron los resultados del modelo de efectos fijos dada la falta de heterogeneidad estadística. De lo contrario, se presentaron los resultados del metanálisis con un modelo de efectos aleatorios. Los riesgos de sesgo de los ensayos incluidos se evaluaron con el uso de dominios de riesgo de sesgo, y los resultados de la revisión se presentaron con la calidad metodológica de los ensayos según los criterios GRADE. Las conclusiones principales se basaron en el análisis hasta tres años de seguimiento.

Resultados principales

Se identificaron ocho ensayos clínicos aleatorios (632 participantes) que cumplieron los criterios de inclusión. El riesgo de sesgo fue alto en los ocho ensayos, y se clasificó la evidencia como de certeza baja a muy baja. La media de edad varió de 16 a 78 años. La proporción de hombres varió de un 60% a un 75% y la proporción de pacientes con carcinoma hepatocelular primario en estadio III varió de un 22% a un 85%. La mediana de duración del seguimiento fue de 12 meses (dos a 38 meses).

La QEAT seguida de RTC‐3D en comparación con la QEAT sola puede haber reducido la mortalidad por todas las causas en el seguimiento de tres años (cociente de riesgos [CR] 0,80; intervalo de confianza [IC] del 95%: 0,73 a 0,88; 552 participantes; siete ensayos; evidencia de baja certeza). La QEAT seguida de RTC‐3D en comparación con la QEAT sola puede reducir la proporción de participantes sin respuesta tumoral (respuesta completa más respuesta parcial) (CR 0,49; IC del 95%: 0,39 a 0,61; 632 participantes; ocho ensayos; evidencia de baja certeza). Los datos de un ensayo sobre la calidad de vida relacionada con la salud favorecieron la QEAT seguida de RTC‐3D, aunque los datos presentados estaban pobremente definidos (evidencia de muy baja certeza). Ninguno de los ensayos informó eventos adversos graves. Los resultados de los eventos adversos no graves fueron como sigue: la QEAT seguida de RTC‐3D en comparación con la QEAT sola no mostró ninguna diferencia en los resultados de la proporción de participantes con leucopenia (CR 1,12; IC del 95%: 0,92 a 1,34; 438 participantes; cinco ensayos; evidencia de muy baja certeza) y elevación de las transaminasas séricas (CR 1,67; IC del 95%: 0,66 a 4,27; 280 participantes; cuatro ensayos; evidencia de certeza muy baja). Sin embargo, la proporción de participantes con aumento de la bilirrubina total fue mayor en el grupo de QEAT seguida de RTC‐3D que en el grupo de QEAT sola (CR 2,69; IC del 95%: 1,34 a 5,40; 172 participantes; dos ensayos; evidencia de muy baja certeza). La tasa de pacientes con alfafetoproteína sérica (AFP) sin disminución o normalización fue significativamente inferior en el grupo de QEAT seguida de RTC‐3D que en el grupo de QEAT, pero estos datos derivan de un solo ensayo (Chi² = 7,24; p = 0,007; evidencia de muy baja certeza).

Conclusiones de los autores

La QEAT seguida de RTC‐3D puede asociarse con menos mortalidad por todas las causas y mayor respuesta tumoral, a pesar de la mayor toxicidad manifestada por un mayor incremento de la bilirrubina total. Los resultados de la revisión deben considerarse con cuidado dados los defectos metodológicos de los ensayos incluidos, que resulta en evidencia de certeza baja a muy baja. Faltan datos sobre los eventos adversos graves y la calidad de vida relacionada con la salud. Tampoco hay mucha seguridad sobre los resultados de los eventos adversos no graves que se informaron. Se necesitan ensayos de alta calidad que evalúen aún más la función de la QEAT seguida de la RTC‐3D para el carcinoma hepatocelular inoperable.

PICO

Population

Intervention

Comparison

Outcome

El uso y la enseñanza del modelo PICO están muy extendidos en el ámbito de la atención sanitaria basada en la evidencia para formular preguntas y estrategias de búsqueda y para caracterizar estudios o metanálisis clínicos. PICO son las siglas en inglés de cuatro posibles componentes de una pregunta de investigación: paciente, población o problema; intervención; comparación; desenlace (outcome).

Para saber más sobre el uso del modelo PICO, puede consultar el Manual Cochrane.

Resumen en términos sencillos

disponible en

Quimioembolización arterial transcatéter seguido de radioterapia conformacional tridimensional para el carcinoma hepatocelular primario

Antecedentes

El carcinoma hepatocelular, también conocido como hepatoma maligno, es un cáncer primario de hígado. A pesar de la vigilancia periódica en las poblaciones de alto riesgo, la mayoría de los pacientes con carcinoma hepatocelular se diagnostican en un estadio avanzado. En consecuencia, una pequeña parte de los pacientes con la enfermedad son candidatos para la resección quirúrgica (extracción). Desde que se introdujo la quimioembolización arterial transcatéter (QEAT; un procedimiento que restringe el suministro de sangre a un tumor) como tratamiento paliativo (para aliviar los síntomas y mejorar la calidad de vida) en pacientes con cáncer hepático inoperable, se ha convertido en una de las formas más frecuentes de intervención. Más recientemente, se ha utilizado la tecnología de radiación moderna de la radioterapia conformacional tridimensional (RTC‐3D), que configura los haces de rayos según la forma del tumor, para mejorar los efectos adversos de la radioterapia convencional. Se predice que la combinación de QEAT seguida de RTC‐3D podría mejorar el efecto del tratamiento para el carcinoma hepatocelular. Hasta la fecha, el conocimiento acerca de los efectos beneficiosos y perjudiciales de la combinación de QEAT seguida de RTC‐3D es escaso, y los estudios actuales aún son polémicos con respecto a la eficacia de la combinación de QEAT seguida de RTC‐3D en comparación con QEAT sola. El objetivo de esta revisión sistemática Cochrane fue comparar los efectos beneficiosos y perjudiciales de la QEAT seguida de la RTC‐3D versus la QEAT sola en pacientes con carcinoma hepatocelular primario, para el cual la extracción quirúrgica no se considera una posibilidad.

Características de los estudios

Los autores de revisión realizaron búsquedas en la bibliografía médica para aclarar la función de la combinación de QEAT seguida de RTC‐3D en el tratamiento del carcinoma hepatocelular primario y para comparar los efectos beneficiosos y perjudiciales de la QEAT sola. Se recopilaron y analizaron los datos de los ensayos clínicos aleatorios (estudios clínicos en que los participantes son asignados al azar a uno de dos o más grupos de tratamiento) de los pacientes con carcinoma hepatocelular primario que podían ser sometidos a QEAT o RTC‐3D. La evidencia está actualizada hasta mayo 2018.

Resultados clave y calidad de la evidencia

La revisión incluyó ocho ensayos con 632 participantes. Todos los ensayos presentaron un alto riesgo de sesgo. La QEAT seguida de la RTC‐3D pareció ser superior a la QEAT en cuanto a la mejoría de la muerte por cualquier causa y la respuesta tumoral, a pesar de una mayor toxicidad evidenciada por un aumento más significativo de la cantidad total de bilirrubina (análisis de sangre que evalúa cómo funciona el hígado). Ningún ensayo informó de efectos secundarios graves. Un ensayo informó de la calidad de vida relacionada con la salud (una medida de la satisfacción del paciente con su vida y su salud), aunque fue definida de modo deficiente. Los resultados de la revisión eran inciertos porque los ensayos incluidos tenían defectos metodológicos. Se necesitan más ensayos clínicos aleatorios de alta calidad para confirmar o completar los resultados de la revisión.

Authors' conclusions

Implications for practice

Transcatheter arterial chemoembolisation (TACE) followed by three‐dimensional conformal radiotherapy (3‐DCRT) may be associated with lower all‐cause mortality and increased tumour response, despite the increased toxicity expressed by a higher rise of total bilirubin. Our review findings should be considered with caution because of the methodological weaknesses in the included trials, resulting in low‐ to very low‐certainty evidence. Data on serious adverse events and health‐related quality of life are lacking. We are also very much uncertain in the results of the reported non‐serious adverse events.

Implications for research

This review identifies the need for conducting high‐quality randomised clinical trials to evaluate the efficacy of TACE followed by 3‐DCRT versus TACE alone. Randomised clinical trials assessing further the role of TACE followed by 3‐DCRT in people with primary hepatocellular carcinoma are needed. The trials should be performed in people from different countries, with different aetiologies of the chronic liver disease, and the clinical outcomes should be prespecified. In addition, the trials should cover different drugs for TACE, and their dosages, frequency, and range in radiation. Such trials ought to be designed according to the SPIRIT Statements and reported according to the CONSORT Statements. Such trials ought to consider to stratify the participants according to etiology of hepatocellular carcinoma and disease severity. The different classifications of 3‐DCRT should also be studied.

Summary of findings

Open in table viewer

Summary of findings for the main comparison. TACE followed by 3‐DCRT compared to TACE for primary hepatocellular carcinoma

TACE followed by 3‐DCRT compared to TACE for primary hepatocellular carcinoma
Patient or population: primary hepatocellular carcinoma Setting: hospitalised in China Intervention: TACE followed by 3‐DCRT Comparison: TACE
Outcomes	*Anticipated absolute effects (95% CI)**		Relative effect (95% CI)	№ of participants (studies)	Quality of the evidence (GRADE)	Comments
	Risk with TACE	Risk with TACE+3‐DCRT
All‐cause mortality: at 3 years Follow‐up: mean 17 months	Study population		RR 0.80 (0.73 to 0.88)	552 (7 RCTs)	⊕⊕⊝⊝ Low^a	—
	853 per 1000	683 per 1000 (623 to 751)
Proportion of participants without tumour response (CR+PR) Follow‐up: mean 18 months	Study population		RR 0.49 (0.39 to 0.61)	632 (8 RCTs)	⊕⊕⊝⊝ Low^a	—
	495 per 1000	243 per 1000 (193 to 302)
Serious adverse events	None of the trials reported data on serious adverse events.				⊕⊝⊝⊝ Very low^a,b	—
Health‐related quality of life	Health‐related quality of life was significantly better in the TACE followed by 3‐DCRT group than in the TACE alone group (Chi² = 4.479, P = 0.034)			66 (1 RCT)	⊕⊝⊝⊝ Very low^a,b	—
Non‐serious adverse events: leukopenia Follow‐up: mean 13.2 months	Study population		RR 1.12 (0.92 to 1.34)	438 (5 RCTs)	⊕⊝⊝⊝ Very low^a,c	—
	475 per 1000	532 per 1000 (437 to 636)
Non‐serious adverse events: serum transaminases elevation Follow‐up: mean 7.5 month	Study population		RR 1.67 (0.66 to 4.27)	280 (4 RCTs)	⊕⊝⊝⊝ Very low^a,d,e,f	—
	328 per 1000	549 per 1000 (217 to 1000)
Non‐serious adverse events: total bilirubin elevation Follow‐up: mean 6 months	Study population		RR 2.69 (1.34 to 5.40)	172 (2 RCTs)	⊕⊝⊝⊝ Very low^a,g	—
	108 per 1000	292 per 1000 (145 to 586)
Proportion of participants without serum AFP normalisation	The rate of participants with serum AFP without decline or normalisation was a significantly lower in the TACE followed by 3‐DCRT group than in the TACE alone group (Chi² = 7.24, P = 0.007)			96 (1 RCT)	⊕⊝⊝⊝ Very low^a,b	—
*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). 3‐DCRT: three‐dimensional conformal radiotherapy; AFP: alpha fetoprotein; CI: confidence interval; CR: complete response; PR: partial response; RCT: randomised clinical trial; RR: risk ratio; TACE: transcatheter arterial chemoembolisation.
GRADE Working Group grades of evidence High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: 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 certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.
^aDowngraded two levels for risk of bias: most of the included RCTs had unclear risk of concealment of allocation, non‐blinded assessment of outcomes, attrition bias, and other bias. ^bDowngraded one level for imprecision: we were unable to combine the data in an overall analysis due to lack of data. ^cDowngraded one level for heterogeneity: the heterogeneity test showed that variation existed in point estimates due to among‐study differences. ^dDowngraded one level for heterogeneity: the heterogeneity test showed that large variation (I² = 88%) existed in point estimates due to among‐study differences. ^eDowngraded two levels for imprecision: the sample size was less than 300 (280 participants), the number of events not high (108 events), and the confidence intervals of the pooled RR clearly crossed the line of no effect and appreciable harm. ^fDowngraded one level for publication bias; based on the funnel plot. ^gDowngraded one level for imprecision: the sample size was less than 300 (172 participants).

Background

Description of the condition

Hepatocellular carcinoma, also called malignant hepatoma, is a primary malignancy of the liver. It is the fifth most common neoplasm worldwide and its incidence is increasing (Venook 2010; Ferlay 2015). It is the second leading cause of cancer‐related death (El‐Serag 2014). Most people with hepatocellular carcinoma develop malignancy secondary to either viral hepatitis infections (hepatitis B or hepatitis C) or cirrhosis (alcoholism being the most common cause of hepatic cirrhosis) (Kumar 2004).

Hepatocellular carcinoma exhibits two main global patterns: one in North America and Western Europe where the prevalence of chronic hepatitis C virus is increasing, and another in non‐Western countries, such as those in sub‐Saharan Africa, Central and Southeast Asia, and the Amazon basin where the prevalence of chronic hepatitis B is high (El‐Serag 2001; Lodato 2006). Epidemiological data show that the incidence of hepatocellular carcinoma is changing around the world following aetiology; it is increasing in many high‐income countries, whereas it is declining in low‐income countries (McGlynn 2001). Usually, there are more men than women who develop hepatocellular carcinoma, and the women are usually between 30 years and 50 years of age (Kumar 2004). Yearly, 610,000 people die of hepatocellular carcinoma worldwide (WHO 2009), and about half of these deaths occur in China. Hepatocellular carcinoma is one of the deadliest cancers in China, where chronic hepatitis B is found to be the cause in 90% of the deaths. With the introduction of hepatitis B virus vaccination, the incidence of hepatocellular carcinoma has been decreasing in low‐income countries (Lodato 2006). In Japan, 90% of hepatocellular carcinomas are associated with hepatitis C. Food infected with Aspergillus flavus (especially peanuts and corns stored during prolonged wet seasons), which produces aflatoxin, poses another risk factor for hepatocellular carcinoma. However, most malignant tumours of the liver discovered in people from Western countries are metastases from tumours elsewhere, and hepatocellular carcinoma is generally seen as a rare cancer (Lodato 2006). Hepatocellular carcinoma is one of the few types of cancer that has increased in frequency and mortality in the USA (Mittal 2013) and Europe (Deuffic 1998).

Hepatocellular carcinoma at its early stages is often non‐symptomatic. As the cancer grows, symptoms may include pain in the upper abdomen on the right side, which may extend to the back and shoulder, or cause swollen abdomen (bloating), weight loss, loss of appetite, loss of the sensation of being full, fatigue, nausea and vomiting, jaundice, or fever. Mostly, these symptoms happen in stages III or IV of the disease. Diagnosis of hepatocellular carcinoma may involve physical examination such as examination of the liver, spleen, or any lumps; ascites; and jaundice. It may also entail an alpha‐fetoprotein (AFP) test, computer tomography scan, ultrasound, magnetic resonance imaging, angiogram, or biopsy (El‐Serag 2008). Despite regular surveillance conducted in high‐risk populations, most people with hepatocellular carcinoma are diagnosed at an advanced stage. Consequently, a minority of people with the disease are suitable for surgical resection. The recurrence rates are as high as 65% to 80% within five years, even for those people who undergo surgical resection (Li 2013), which results in a five‐year survival of about 40% (Cha 2005).

Description of the intervention

People with early‐stage cancer (20% to 30% of people with hepatocellular carcinoma) are considered suitable for treatments such as liver resection, liver transplantation, percutaneous ablation, percutaneous ethanol injection, and radiofrequency ablation (Llovet 2004; Cabrera 2010). Since transcatheter arterial chemoembolisation (TACE) was introduced as a palliative treatment in people with unresectable hepatocellular carcinoma, it has become one of the most common forms of intervention (Takayasu 2006); and it is considered a standard treatment option for people with unresectable hepatocellular carcinoma (Kothary 2007). In addition, TACE is usually performed as a temporary treatment while waiting for a liver transplant, or for people for whom surgical or percutaneous ablative treatment is contraindicated. TACE involves the injection of anticancer drugs (doxorubicin, epirubicin, or cisplatin) and iodised oil (Lipiodol Ultra‐Fluide, Laboratoires Guerber, Aulnay‐sous‐Bois, France) (¹³¹I‐lipiodol radiotherapy) into the hepatic artery, followed by the administration of embolic agents (Nakamura 1990; Bronowicki 1994). Currently, TACE is considered the standard of care for people with intermediate‐stage hepatocellular carcinoma presenting with Child‐Pugh class A and B liver function, and large or multinodular hepatocellular carcinoma without cancer‐related symptoms, macrovascular invasion, or extrahepatic metastasis (Murata 2014).

These treatment recommendations occur irrespective of the fact that a Cochrane systematic review has been unable to identify high‐quality evidence in support of TACE (Oliveri 2011).

Since the early 2000s, the modern radiation technology of three‐dimensional conformal radiotherapy (3‐DCRT) has been applied in clinics to improve the shortcomings of conventional radiotherapy. Radiotherapy for hepatocellular carcinoma has resulted in unsatisfactory outcomes since the late 1980s because of the liver's poor tolerance to irradiation (Liang 2005). 3‐DCRT is aided by a computerised treatment‐planning system which has enabled the tight conformation of a high‐dose volume to hepatocellular carcinoma lesions in three dimensions. Thus, 3‐DCRT has made it possible to escalate the irradiation dose to focal hepatocellular carcinoma without causing undue dose‐limiting toxicity in neighbouring non‐cancerous liver tissues. Therefore, it spares non‐cancerous liver tissue from excess damage, and has increasingly been recognised as a potentially curative option for people with hepatocellular carcinoma (Lawrence 1990; Feng 2011).

How the intervention might work

TACE is appropriate for hepatocellular carcinoma, as the hepatic artery delivers 99% of the blood supply to hepatic tumours (Murata 2014). TACE seems to improve survival compared with the best supportive care in meta‐analyses of randomised trials (Cammà 2002; Llovet 2003); and in two individual clinical trials (Llovet 2002; Lo 2002). The antitumour effect of TACE is greater than that of other anticancer drugs (Yoshikawa 1994); or iodised oil alone (Takayasu 1987; Yamagami 2014). As stated above, we lack high‐quality evidence in support of TACE (Oliveri 2011).

With advances in 3‐DCRT, local radiation of the liver has become safer (Robertson 1993); and its efficacy is better than in conventional radiotherapy (Matsuura 1998). 3‐DCRT has shown favourable outcomes in local control and survival, with a median survival time of 10 months to 25 months (Yu 2014); and a three‐year survival of around 30% for people with hepatocellular carcinoma (Lee 2013). Several series that employed 3‐DCRT have reported a dose‐response relation in radiotherapy for liver cancers with better response rates and prolonged hepatic control in groups that received higher radiotherapy doses (Robertson 1993; Seong 2000).

The inadequacy of single TACE in inducing complete tumour necrosis has also been well documented (Sasaki 1987), and TACE is usually repeated at regular intervals. Nevertheless, repeated TACE frequently becomes ineffective due to tumour progression. As most primary liver tumours have dual blood supplies, it is easy to re‐form the collateral circulation in lesions with the residual tumour cells after TACE (Ikeda 1991; Cheng 2000). It is predicted that the combination of TACE and 3‐DCRT could enhance treatment effects for hepatocellular carcinoma. It was reported that the one‐year survival rate for the combination of TACE followed by 3‐DCRT was 73%, two‐year survival rate was 53%, and three‐year survival rate was 35%, which was higher than the TACE alone group (one‐year survival rate 60%, two‐year survival rate 31%, and three‐year survival rate 14%) (Zou 2014).

The rationale for combined TACE and 3‐DCRT was based on the following three considerations. First, with the iodised oil injection by TACE, the deposit of iodine would have made the margin of the median gross tumour volume clearer, making gross tumour volume delineation more accurate, and 3‐DCRT plan verification easier. Second, after TACE, the tumour burden becomes less and the number of tumour cells decreases. This would make it easier for 3‐DCRT to control the malignancy. Third, the 3‐DCRT radiosensitivity of hepatocellular carcinoma still exists when 3‐DCRT begins several weeks after TACE (Zhou 2007).

Why it is important to do this review

To date, little is known about the benefits and harms of the combination of TACE and 3‐DCRT, and only few clinical studies have been conducted. Some were in favour of the combination of TACE and 3‐DCRT (Shim 2005; Koo 2010), while another showed that the survival rates of people with combined TACE and 3‐DCRT were similar to those with TACE alone (Chia‐Hsien 2001). Despite the publication of further studies on the use of TACE followed by 3‐DCRT in people with primary hepatocellular carcinoma, we found no systematic reviews or meta‐analyses with randomised clinical trials comparing the combination of TACE followed by 3‐DCRT versus TACE alone.

Objectives

To compare the beneficial and harmful effects of transcatheter arterial chemoembolisation (TACE) followed by three‐dimensional conformal radiotherapy (3‐DCRT) versus TACE alone in adults with primary hepatocellular carcinoma, considered unsuitable for surgical removal.

Methods

Criteria for considering studies for this review

Types of studies

We considered all randomised clinical trials investigating a combination of TACE and 3‐DCRT versus TACE alone for inclusion, whether they were double‐blind, single‐blind, or open‐label, and regardless of publication status, language, and length of the trial. In addition, we scanned quasi‐randomised and other observational studies which were retrieved with the searches for randomised clinical trials to identify reports on harm. By not searching specifically for harms in observational studies, we are aware that we, in the present systematic review, may have been biased towards assessing benefits and ignoring harms (see Storebø 2018).

Types of participants

Participants older than 18 years diagnosed with hepatocellular carcinoma based on their pathological findings in at least one lesion with or without other laboratory evidence such as B‐ultrasound, computed tomography, or AFP. Furthermore, diagnosis had to conform to the following criteria.

Participants had not received any anticancer therapy.
Karnofsky score was 69 or less (Park 2014).
Child‐Pugh grade of liver function was A or B (Liang 2015).
Number of white blood cell 4.0 × 10⁹/L or greater.
Model for end‐stage liver disease (MELD) score less than 10.
There were no contraindications (vascular or adjacent organ involvement, involvement with lymph nodes, distant metastasis, jaundice and ascites, cardiopulmonary dysfunction, coagulation disorders) of TACE and 3‐DCRT.

Types of interventions

We included trials comparing TACE followed by 3‐DCRT versus TACE alone in people with primary hepatocellular carcinoma.

Types of outcome measures

We sought to measure the following outcomes at the end of treatment, as well as at maximal follow‐up.

Primary outcomes

All‐cause mortality (death from any cause). We calculated one‐year, two‐year, and three‐year all‐cause mortality. We drew primary conclusions based on three‐year all‐cause mortality, as the longer the follow‐up period, the stronger the evidence.
Proportion of participants without tumour response: according to the World Health Organization (WHO) Handbook for reporting the results of cancer treatment (Spieth 2003), the responses were assessed as follows:
- complete response (CR), complete disappearance or 100% necrosis of all tumours with no evidence of new lesions;
- partial response (PR), more than 50% reduction or more than 50% necrosis (or both) of all measurable lesions with no evidence of new lesions;
- progressive disease (PD), more than 25% enlargement of all measurable lesions or appearance of new lesions;
- stable disease (SD), no change.
Serious adverse events: we used the International Conference on Harmonisation (ICH) Guidelines for Good Clinical Practice's definition of a serious adverse event (ICH‐GCP 1997); that was, any untoward medical occurrence that resulted in death, was life threatening, required hospitalisation or prolongation of existing hospitalisation, resulted in persistent or significant disability or incapacity, or was a congenital anomaly or birth defect. We considered all other adverse events as non‐serious.

Secondary outcomes

Health‐related quality of life as reported in the trials.
Non‐serious adverse events, such as abdominal pain, fatigue, poor appetite, nausea, vomiting, fever, leukopenia, thrombocytopenia, MELD score, etc. We analysed the following non‐serious adverse events separately: proportion of participants with leukopenia, with serum transaminases elevation, and with total bilirubin elevation.
Proportion of participants without serum AFP normalisation.

Search methods for identification of studies

Electronic searches

We searched the The Cochrane Hepato‐Biliary Group Controlled Trials Register (May 2018; Cochrane Hepato‐Biliary Group Module), the Cochrane Central Register of Controlled Trials (CENTRAL) in the Cochrane Library (2018, Issue 4), MEDLINE Ovid (1946 to May 2018), Embase Ovid (1974 to May 2018), LILACS (1982 to May 2018; Bireme), Science Citation Index Expanded (1900 to May 2018; Web of Science), and Conference Proceedings Citation Index – Science (1990 to May 2018; Web of Science) (Royle 2003). We checked reference lists of all included studies and related reviews manually for further related articles. Appendix 1 provided the search strategies with the time spans of the searches.

Searching other resources

We searched the reference lists of the identified trials to identify further relevant trials.

We also searched online trial registries such as ClinicalTrials.gov (clinicaltrials.gov/), European Medicines Agency (EMA; www.ema.europa.eu/ema/), WHO International Clinical Trials Registry Platform (www.who.int/ictrp), and the Food and Drug Administration (FDA; www.fda.gov), as well as pharmaceutical company sources, for ongoing or unpublished trials.

Data collection and analysis

We performed the review according to the recommendations of the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011) and the Cochrane Hepato‐Biliary Group Module. We used the Cochrane statistical software Review Manager 5, for data entry and analysis (Review Manager 2014).

Selection of studies

Two review authors (LL and JZ) independently identified the trials for inclusion. We listed the excluded studies with their reasons for exclusion. We resolved disagreements by discussion with the other author (WZ). We collated multiple reports of the same study, so that each study rather than each report was the unit of interest in the review.

Data extraction and management

Two review authors (LL and JZ) independently extracted the following data from each trial.

Year and language of publication.
Country.
Year trial was conducted.
Inclusion and exclusion criteria.
Sample size.
Population characteristics such as age, sex ratio, Karnofsky score and Child‐Pugh grade of liver function.
For TACE, use of drugs for chemotherapy and embolisation.
For 3‐DCRT, use of irradiation such as dosage, frequency, and range.
Treatment measures for adverse effects.
Outcomes (see Primary outcomes; Secondary outcomes).
Methodological quality and bias risk.
Sample size calculation.
Intention‐to‐treat (ITT) analysis.

We sought missing information or clarification of unclear information by contacting the authors of the individual trials. We resolved any differences in opinion through discussion.

Assessment of risk of bias in included studies

Two review authors (LL; JZ) independently assessed the risk of bias for each included trial according to the recommendations in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011), the Cochrane Hepato‐Biliary Group Module, and methodological studies (Schulz 1995; Moher 1998; Kjaergard 2001; Wood 2008; Savović 2012a; Savović 2012b; Lundh 2017; Savović 2018).

We used the following definitions in the assessment of risk of bias.

Allocation sequence generation

Low risk of bias: sequence generation was achieved using computer random number generation or a random number table. Drawing lots, tossing a coin, shuffling cards, or throwing dice were adequate if performed by an independent person not otherwise involved in the trial.
Unclear risk of bias: method of sequence generation was not specified.
High risk of bias: sequence generation method was not random.

Allocation concealment

Low risk of bias: participant allocations could not have been foreseen in advance of, or during, enrolment. Allocation was controlled by a central and independent randomisation unit. Allocation sequence was unknown to the investigators (e.g. whether the allocation sequence was hidden in sequentially numbered, opaque, and sealed envelopes).
Unclear risk of bias: method used to conceal the allocation was not described, so that intervention allocations may have been foreseen in advance of, or during, enrolment.
High risk of bias: allocation sequence was likely known to the investigators who assigned the participants.

Blinding of participants and personnel

Low risk of bias: any of the following: no blinding or incomplete blinding, but the review authors judged this outcome unlikely to have been influenced by lack of blinding (mortality) (Wood 2008; Savović 2012a; Savović 2012b); or blinding of participants and key study personnel was ensured, and it was unlikely that the blinding could have been broken.
Unclear risk of bias: any of the following: insufficient information to permit judgement of ‘low risk’ or ‘high risk’; or the trial did not address this outcome.
High risk of bias: any of the following: no blinding or incomplete blinding, and the outcome was likely to have been influenced; or blinding of key study participants and personnel was attempted, but likely could have been broken, thus influencing the outcome.

Blinding of outcome assessment

Low risk of bias: any of the following: no blinding of outcome assessment, but the review authors judged that the outcome measurement was unlikely to have been influenced by a lack of blinding; or blinding of outcome assessment was ensured, and it was unlikely that the blinding could have been broken.
Unclear risk of bias: any of the following: insufficient information to permit judgement of ‘low risk’ or ‘high risk’; or the trial did not address this outcome.
High risk of bias: any of the following: no blinding of outcome assessment, and the outcome measurement was likely to have been influenced by a lack of blinding; or the outcome assessment was blinded, but it was likely that the blinding could have been broken, and the outcome measurement was therefore likely to have been influenced.

Incomplete outcome data

Low risk of bias: missing data were unlikely to have made treatment effects depart from plausible values. Sufficient methods, such as multiple imputation, were employed to handle missing data.
Unclear risk of bias: there was insufficient information to assess whether missing data, in combination with the method used to handle missing data, were likely to have induced bias in the results.
High risk of bias: the results were likely to have been biased due to missing data.

Selective outcome reporting

Low risk of bias: the trial reported the following predefined outcomes: all‐cause mortality, tumour response assessments, and serious adverse events. If the original trial protocol was available, the outcomes should be those called for in that protocol. If the trial protocol was obtained from a trial registry (e.g. clinicaltrials.gov/), the outcomes sought should have been those enumerated in the original protocol if the trial protocol was registered before or at the time it began. If the trial protocol was registered after the trial began, those outcomes were not considered reliable.
Unclear risk of bias: not all predefined outcomes were reported in full, or it was unclear whether data on these outcomes had been recorded or not.
High risk of bias: one or more predefined outcomes was not reported.

Other bias

Low risk of bias: the trial appeared to be free of other bias that could put its integrity at risk.
Unclear risk of bias: the trial may or may not have been free of other domains that could have put it at risk of bias.
High risk of bias: there were other factors in the trial that could have put it at risk of bias.

We judged trials to be at an overall low risk of bias if assessed with a low risk of bias in all above domains. We judged trials to be at an overall high risk of bias if assessed with unclear risk of bias or high risk of bias in one or more of the above domains.

Measures of treatment effect

We performed the meta‐analyses according to Cochrane recommendations (Higgins 2011) and the Cochrane Hepato‐Biliary Group Module. We used the Review Manger 5 software package provided by Cochrane (Review Manager 2014). For dichotomous variables, we calculated the risk ratio (RR) with a 95% confidence interval (CI). For continuous variables, we calculated the mean difference with a 95% CI.

Unit of analysis issues

We took into account the group of participants per intervention group in the randomised clinical trials with parallel‐group design. In the case of cross‐over trials, we planned to use the data from the first trial period only. We did not expect to find cluster‐randomised trials. For trials with multiple intervention groups, we planned to include the groups in which our experimental and control interventions were compared. We planned to divide the control group into two or more to avoid double‐counting in case it was a common comparator.

Dealing with missing data

We considered participants with completely missing data as treatment failures and performed ITT analyses. If data for any participant were obtained at any point before the measured time point, this observation was carried forward.

Assessment of heterogeneity

We explored heterogeneity using the Chi² test with significance set at a P value of 0.10 or less, and measured the extent of heterogeneity using the I² statistic (Higgins 2002). We interpreted I² values as follows:

probably not important: 0% to 40%;
possible moderate heterogeneity: 30% to 60%;
possible substantial heterogeneity: 50% to 90%;
considerable heterogeneity: 75% to 100%.

Assessment of reporting biases

We used visual asymmetry on a funnel plot to explore reporting bias when at least 10 randomised clinical trials were identified in a particular field (Egger 1997; Macaskill 2001). In addition, we performed the linear regression approach described by Egger to determine the funnel plot asymmetry if the result of a funnel plot was unclear (Egger 1997).

Data synthesis

Meta‐analysis

We performed our meta‐analyses in accordance with Cochrane's recommendations (Higgins 2011), and used Review Manager 5 software for our analyses (Review Manager 2014). We evaluated all missing data using ITT analyses. We treated missing data as treatment failures.

We expressed binary outcomes using RR with 95% CI. If the results were statistically significant according to our Trial Sequential Analysis (see 'Trial Sequential Analysis' below), we calculated the number needed to treat for an additional beneficial outcome (NNTB) and the number needed to treat for an additional harmful outcome (NNTH) using fixed‐effect and random‐effects models (DerSimonian 1986; DeMets 1987). We interpreted the results according to Jakobsen 2014. If there is absence of statistical heterogeneity or only one trial was included, the fixed‐effect and the random‐effects models would show identical results. We presented the results of the fixed‐effect model in this situation. If there was substantial statistical heterogeneity, we reported the results from the random‐effects meta‐analysis. We presented unavailable data and inappropriate data using descriptive means.

Trial Sequential Analysis

We applied Trial Sequential Analysis as cumulative meta‐analyses are at risk of producing random errors due to sparse data and repetitive testing of the accumulating data (Brok 2008; Wetterslev 2008; Brok 2009; Thorlund 2009; Wetterslev 2009; Thorlund 2010). To minimise random errors, we calculated the required information size (i.e. the number of participants needed in a meta‐analysis to detect or reject a certain intervention effect) (Wetterslev 2008). We calculated the required information size adjusted for diversity, since the heterogeneity adjustment with the I² statistic underestimated the required information size (Wetterslev 2008; Wetterslev 2009). In our meta‐analysis, we performed Trial Sequential Analysis to maintain an overall 2.5% risk of a type I error and 20% of type II error (a power of 80%) (Wetterslev 2009). On the basis of the required information size, we constructed trial sequential monitoring boundaries (Lan 1983; Wetterslev 2008; Thorlund 2011). These boundaries determined the statistical inference one may draw regarding the cumulative meta‐analysis that had not reached the required information size; if the trial sequential monitoring boundary for benefit or harm was crossed before the required information size was reached, firm evidence may have been established, and further trials may have turned out to be superfluous. In contrast, if the boundary had not been surpassed, it was probably necessary to continue doing trials in order to detect or reject a certain intervention effect. This could be determined by assessing whether the cumulative Z‐curve crossed the trial sequential boundaries for futility. If futility boundaries had been crossed, then further trials may have been unnecessary (TSA 2011). We conducted Trial Sequential Analysis using software from The Copenhagen Trial Unit (Thorlund 2011; TSA 2011; Wetterslev 2017).

Subgroup analysis and investigation of heterogeneity

We performed the following subgroup analyses.

Trials at low bias risk compared to trials at high bias risk.
With TACE, different drugs used in chemotherapy and embolisation.
With 3‐DCRT, different dosage, frequency, and range in irradiation.
Presence or absence of chronic liver disease.
Aetiology of the chronic liver disease.

Sensitivity analysis

We assessed the robustness of our analyses by performing a sensitivity analysis, excluding studies from the overall analysis of high risk of bias due to lack of allocation concealment, blinding, or incomplete reporting of primary outcome. We compared the GRADE assessment of imprecision with that obtained with Trial Sequential Analysis (Jakobsen 2014; Castellini 2018).

'Summary of findings' tables

We presented the evidence in summary of findings Table for the main comparison using GRADEpro software in accordance with the principles of the GRADE system (Guyatt 2011a). This was done to assess the certainty of the body of evidence associated with specific outcomes such as all‐cause mortality, recent objective response of hepatocellular carcinoma, and serious adverse events in our review.

The GRADE approach defined the certainty in a body of evidence as the extent to which one could be confident that an estimate of effect or association was close to the quantity of specific interest. According to GRADE, the certainty in a body of evidence included five factors regarding limitations in the design and implementation of available studies suggesting high likelihood of bias: indirectness of evidence (population, intervention, control, outcomes); unexplained heterogeneity or inconsistency of results (including problems with subgroup analyses); imprecision of results (wide CIs); and high probability of publication bias (Balshem 2011; Guyatt 2011a; Guyatt 2011b; Guyatt 2011c; Guyatt 2011d; Guyatt 2011e; Guyatt 2011f; Guyatt 2011g; Guyatt 2011h; Guyatt 2013a; Guyatt 2013b; Guyatt 2013c; Guyatt 2013d; Mustafa 2013; Guyatt 2017).

Results

Description of studies

The search identified 550 reports (Figure 1).

Figure 1

Study flow diagram.

Results of the search

Electronic literature searches revealed two hits in The Cochrane Hepato‐Biliary Group Controlled Trials Register, 12 hits in the Cochrane Central Register of Controlled Trials (CENTRAL), 104 hits in MEDLINE, 212 hits in Embase, zero hits in LILACS, and 220 hits in Science Citation Index Expanded and Conference Proceedings Citation Index – Science. Appendix 1 shows the search strategies. Thirteen reports consisted of additional reports on published reviews (Zou 2014; Bai 2016). After removing duplicates, 389 reports remained. We excluded 371 irrelevant reports based on the title, abstract, or both. We retrieved and read the full‐text of 18 reports, and finally included eight trials (eight reports).

Included studies

Eight trials satisfied our inclusion criteria (Zhao 2006; Shang 2007; Xiao 2008; Ning 2009; Liao 2010; Gong 2011; Xiao 2011; Chen 2014).

Characteristics of included studies

We summarised the characteristics of the eight included trials in the Characteristics of included studies table.

Study design

All trials were parallel group randomised clinical trials.

Funding

Only one trial was supported by a grant from the local science and technology bureau (Liao 2010). Other trials did not provide any data on funding.

Participants

There were 632 participants with primary hepatocellular carcinoma. The mean age ranged from 16 years to 78 years. The proportion of men ranged from 60% to 75%, and the proportion of people with stage III primary hepatocellular carcinoma ranged from 22% to 85%. The proportion of people with tumour size greater than 10 cm was 45% in one trial (Chen 2014), and with greater than 3 cm size ranging from 40% to 100% in five trials (Zhao 2006; Shang 2007; Xiao 2008; Liao 2010; Xiao 2011). The proportion of people with a single tumour was 68% to 75% in two trials (Xiao 2011; Chen 2014), the proportion with Child‐Pugh class A ranged from 65% to 71% in two trials (Xiao 2008; Liao 2010), and the proportion with AFP greater than 400 μg/L ranged from 41% to 100% in three trials (Zhao 2006; Shang 2007; Xiao 2011). Accordingly, these participants would mostly likely be considered unsuitable for surgical resection.

Interventions

People underwent two courses (Zhao 2006; Shang 2007; Xiao 2008; Ning 2009; Gong 2011; Xiao 2011; Chen 2014) or three to five courses (Liao 2010) of TACE with one‐month interval. 3‐DCRT was delivered one to four weeks after the last course of TACE, if liver function tests were normal. The sum of the radiation doses in 3‐DCRT received by each individual ranged from 30 Gray (Gy) to 66 Gy with 2 Gy/day to 5 Gy/day and 3 days/week to 5 days/week.

Comparisons

All eight trials compared TACE followed by 3‐DCRT versus TACE alone. The chemotherapy included 5‐fluorouracil (750 mg to 1250 mg) (Zhao 2006; Shang 2007; Xiao 2008; Ning 2009; Liao 2010; Gong 2011; Chen 2014), cisplatin (40 mg to 120 mg) (Zhao 2006; Shang 2007; Xiao 2008; Ning 2009; Liao 2010; Gong 2011; Xiao 2011; Chen 2014), adriamycin (30 mg to 100 mg) (Shang 2007; Xiao 2008; Liao 2010; Xiao 2011), hydroxyl radical (15 mg to 20 mg) (Zhao 2006; Ning 2009), and mitomycin C (6 mg to 14 mg) (Ning 2009; Chen 2014). Embolisation therapy: peripheral embolisation was performed by iodine oil emulsion, and central embolisation was performed by gelfoam.

Outcomes

All trials reported all‐cause mortality and tumour response. Duration of therapy and embolisation were similar in the two intervention groups. No trial reported serious adverse events. Only one trial reported the rate of participants without decline or normalisation of AFP (Zhao 2006), and another trial provided some information on health‐related quality of life (Ning 2009). Five trials reported non‐serious adverse events (Zhao 2006; Shang 2007; Xiao 2008; Liao 2010; Chen 2014).

Excluded studies

See Characteristics of excluded studies table.

We excluded 10 observational studies (Chia‐Hsien 2001; Zeng 2004; Guo 2005; Liang 2005; Shim 2005; Chung 2006; You 2007; Zhang 2009; Koo 2010; Lu 2015).

Risk of bias in included studies

See Figure 2 and Figure 3.

Figure 2

Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.

Figure 3

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

All trials were at high risk of bias (Zhao 2006; Shang 2007; Xiao 2008; Ning 2009; Liao 2010; Gong 2011; Xiao 2011; Chen 2014).

Allocation

Three trials performed random sequence generation using a random number table, and hence were at low risk of bias (Xiao 2008; Liao 2010; Xiao 2011). The remaining five studies had unclear risk of bias, falling into the group of high risk of bias trials.

Allocation concealment had unclear risk of bias in all trials, as the trials provided no information.

Blinding

All trials were at unclear risk of bias for blinding of participants and investigators. Also, detection bias was unclear in all studies, hence all trials were at high overall risk of bias.

Incomplete outcome data

Three trials were at low risk of attrition bias, as they did not have any missing data after randomisation (Xiao 2008; Ning 2009; Chen 2014). One trial had high risk of bias because it did not account for participants with missing outcomes (Liao 2010). Other trials were at unclear risk of bias (Zhao 2006; Shang 2007; Gong 2011; Xiao 2011).

Selective reporting

All trials were at high risk of bias for selective reporting as none reported one clinically relevant outcome (serious adverse events); not all protocols were available.

Other potential sources of bias

We assessed six trials as having a low risk of bias regarding other potential sources of bias such as demographic and baseline characteristics of the randomised participants (Zhao 2006; Shang 2007; Xiao 2008; Liao 2010; Xiao 2011; Chen 2014), and we considered the remaining two trials as having unclear risk of bias because the trial authors did not provide the demographic and baseline characteristics of the randomised participants (Ning 2009; Gong 2011).