A network meta-analysis on the efficacy and prognosis of different interventional therapies for early-stage hepatocellular carcinoma

Abstract

Purpose: It is unclear which kind of interventional therapies is the best when treating early-stage hepatocellular carcinoma (HCC). We conducted Bayesian network meta-analyses to compare local tumor progression (LTP), total tumor recurrence and survival rates and to rank the best intervention arm.

Materials and methods: A literature search of Pubmed, Embase, Cochrane library and Clinicaltrials.gov was conducted and randomized controlled trials (RCTs) comparing the outcomes of interventional therapies on early-stage HCC were enrolled. The quality assessment was conducted using Cochrane Collaboration’s tool, while the outcome synthesis of the network meta-analysis was conducted using R-3.3.4 software.

Results: A total of 35 RCTs were enrolled for further analysis. Using network meta-analysis, it was demonstrated that radiofrequency ablation (RFA) plus adjuvant therapies achieved the best performance in decreasing the LTP rate in early-stage HCC, while hepatic resection ranked as the best arm among all the interventional techniques for LTP at 3 years. Meanwhile, hepatic resection and RFA plus adjuvant therapies were the top two best arms in decreasing total recurrence. Furthermore, RFA plus adjuvant therapeutics ranked the best in achieving overall survival outcome, followed by hepatic resection. For disease-free survival, hepatic resection was the best, while for LTP-free survival, the difference among the included treatments was not significant.

Conclusions: Our network meta-analysis showed that RFA-based adjuvant therapies might be the most effective interventions in achieving the best outcomes, while hepatic resection exhibited the best performance in several situations in treating early-stage HCC. More RCTs are needed to draw more solid conclusions.

Keywords:

• Hepatocellular carcinoma
• interventional therapy
• network meta-analysis
• total recurrence survival rate

Introduction

Hepatocellular carcinoma (HCC) is the most common primary cancer of the liver and ranks fifth in incidence and is the second leading cause of death among all kinds of cancer worldwide with a 5-year survival of <12% [1,2]. With the development of diagnostics techniques, patients are more frequently be diagnosed as early-stage HCC (single HCC, ≤5 cm; or up to 3 nodules, ≤3 cm). Currently, the most commonly used treatment strategies for early-stage HCC are liver transplantation, hepatic resection and local ablative techniques [3,4]. However, the high cost and a donor shortage limit the availability of liver transplantation, and it was reported that only 15–20% of HCCs are resectable at diagnosis [3].

Over the past two decades, with the development of medical engineering and interventional technology, local ablative therapeutics, such as radiofrequency ablation (RFA), microwave ablation (MWA), percutaneous ethanol injection (PEI), laser ablation, and cryoablation, were also used as important treatments for early-stage HCC patients who were not eligible for hepatic resection treatments or as a bridge to transplantation, according to guidelines from both the European Association for the Study of the Liver and the American Association for the Study of Liver Diseases. Furthermore, other adjuvant therapies, such as concomitant agents, combination therapies with transcatheter arterial chemoembolization (TACE), or other local ablative techniques, have also been applied extensively. The mechanisms of these interventions mainly induce thermal coagulation, rapid freezing, cell dehydration, and killing cells by chemotherapies with different postablative effects, of which little is known to most doctors [5–8].

According to the latest studies and the National Comprehensive Cancer Network (NCCN) guidelines, the treatment strategies containing hepatic resection, TACE, RFA and PEI were effective and recommended for the early-stage HCC [9,10]. However, the problem surrounding the determination of which is the best choice is always conflicting and controversial. This has been especially true in recent years, during which time several molecular studies have further indicated that HCC has the characteristics of easy for recurrence and a resistance to chemotherapy [11,12]. Therefore, additional investigation into the best the treatment therapeutics for early-stage HCC is needed. Although several meta-analyses have been conducted to determine the best choice for early-stage HCC, some were not up-to-date [8] and may not be comprehensive [13,14]. In this meta-analysis, we searched the main database and used R software to compare the efficacy of different interventional techniques on the outcomes of local tumor progression (LTP), total recurrence, overall survival, disease-free survival and LTP-free survival.

Methods

Data sources and searching strategies

An electronic literature searching was conducted to identify candidate articles published through June 2018. This search was based on the published articles in Pubmed (MEDLINE), EMBASE, Cochrane Library and unpublished data from the clinical trials website (www.clinicaltrials.gov). Medical subject heading (MESH) terms and other terms were used for literature searching using different combinations. The terms used included ‘carcinoma, hepatocellular’, ‘hepatocellular carcinoma’, ‘liver cell carcinoma’, ‘liver cancer’, ‘ablation techniques’, ‘microwave ablation’, ‘radiofrequency ablation’, ‘percutaneous ethanol injection’, ‘cryosurgery’, ‘cryoablation’, ‘laser ablation’, ‘laser therapy’, ‘external beam radiotherapy’, ‘high-intensity focussed ultrasound ablation’, ‘transcatheter arterial chemoembolization’ and ‘early stage’. The search was limited to randomised clinical studies in humans, but there were no language restrictions. Two authors (Y.Q. Lin and Q. Wen) conducted the literature searching independently and a third investigator (Z.X. Sun) resolved any discrepancies. The final searching strategy was reached and conducted.

Eligibility criteria

Clinical trials were included in this meta-analysis if they met the following criteria: (i) Participants: adults older than 18 and diagnosed with small HCC or early-stage HCC (single HCC, ≤5 cm; or up to 3 nodules, ≤3 cm); HCC patients with BCLC stage 0-A; or patients within the Milan Criteria, with hepatic function Child-Pugh-Turcotte classification A or B and without vascular invasion, extrahepatic metastasis and history of treatment; (ii) Intervention: the treatment strategies of the studies contained at least two of the following intervention techniques: RFA, microwave ablation, hepatic resection, PEI, laser ablation, cryoablation, external beam radiotherapy, TACE and different combination of these interventions; (iii) Comparisons: studies that compared the outcomes of different intervention techniques on early-stage HCC; (iv) Outcomes: LTP, total recurrence rates, overall survival rates and complications with a follow-up period of the included studies longer than 1 year and (v) Study design: only RCTs were included for further quality assessment and data synthesis.

The studies were excluded if: (i) they were published in the forms of case reports, reviews, editorials, conference abstracts, observational clinical studies, case-control studies, or retrospective or prospective cohort studies; (ii) they were not written in English or Chinese; (iii) they focused on large HCC or small HCC with vascular invasion, recurrence or extrahepatic metastasis of Child-Pugh classification as C grade; (iv) there was a lack of the important outcomes in the clinical trials or (v) it include duplicate data because a recent study had been included.

Data extraction and quality assessment

After ascertaining the RCTs enrolled in this study, data were extracted by two of the researchers. The data extracted mainly contained (i) Study information: first author, publication date, journal, study size and design; (ii) Patient-related information: HCC stage, tumor stage, basic characteristics, intervention, follow-up, etc.; and (iii) Outcome information: LTP, total recurrence rate, overall survival and complications. The quality assessment was conducted using Cochrane Collaboration’s tool according to the Cochrane Handbook.

Outcome measures and statistical analysis

This meta-analysis was performed according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement [15]. The main outcomes in this meta-analysis were LTP, total tumor recurrence and overall survival. LTP is defined as the appearance of tumor foci at the edge of the ablation zone after at least one contrast-enhanced follow-up study has documented adequate ablation and an absence of viable tissue in the target tumor and surrounding ablation margin by using imaging criteria [16]. The definition of total recurrence rate was the incidence of all types of tumor recurrence including local, intrahepatic and extrahepatic regions during the follow-up period [17]. Survival time was defined as the interval between the first treatment and either death or the last follow-up visit [17]. For the network meta-analysis, R-3.3.4 software (R Foundation for Statistical Computing, Vienna, Austria) was used to perform a Bayesian network meta-analysis was performed using the ‘gemtc’ and ‘rjags’ packages of R-3.3.4 software with the random effect [18]. The consistency of network meta-analysis was assessed using the node-splitting models to detect whether the results of direct and indirect comparison agreed within treatment loops [19], and the risk ratio (RR) values and 95% credible intervals (CrIs) were used to show the results of indirect comparisons. For the network analysis of survival data, hazard ratios (HR) were pooled for synthesis and to compare the effects of different interventions on the overall survival, disease-free survival and LTP-free survival rates. For studies that did not provided the HR values, Engauge Digitizer (version 4.1, M Mitchell, http://markummitchell.github.io/engauge-digitiser/) software was used to extract data from the Kaplan–Meier plot.

Results

Study selection

After the literature search, a total of 1202 studies were identified. After removal of the duplicates and those that did not meet the inclusion criteria, a total of 36 randomized clinical trials were enrolled for data analysis and further synthesis. Those contained 11 treatment arms, namely, RFA, MWA, hepatic resection, PEI, laser ablation, cryoablation, RFA plus radiotherapy, RFA plus PEI, RFA plus TACE, RFA plus sorafenib and PEI plus TACE (Figure 1). Among them, five studies compared the efficacy of RFA with MWA [20–24], seven studies compared the efficacy of RFA with hepatic resection [25–31], six RCTs compared the effects of RFA with PEI [32–37], three RCTs compared the effects of RFA versus laser ablation [7,38,39], only one study compared the effects of RFA versus cryoablation [6], two studies compared the effects of combined therapy of RFA plus ¹³¹I radiotherapy versus RFA treatment alone [17,40], three studies compared the effects of RFA with RFA plus PEI [32,41,42] and four RCTs were enrolled that compared the effects of RFA combined with TACE to RFA alone [43–46]. Meanwhile, Huang et al. compared the efficacy between hepatic resection and PEI [47], one study compared the effect of hepatic resection with RFA plus sorafenib [48] and one study compared hepatic resection with RFA plus TACE [49]. Furthermore, two studies compared the efficacy of PEI plus TACE versus PEI alone [50,51] (Figure 2). Among them, the sample sizes ranged from 20 to 403, with the follow-up duration ranging from 1 year to 10 years. All the studies concentrated on early stage or small HCC (Table 1). For quality assessment, the Cochrane Collaboration Tool was used according to the guidance of the Cochrane Handbook. As a whole, the randomization and the follow-up procedures were conducted well in most of the enrolled studies; however, due to the specificity of hepatic surgeries and interventional treatment, it is quite difficult to maintain double-blindness. In several studies with better double-blindness [23,39,40], the intervention techniques were very similar and were both percutaneous. For studies that compared the efficacy of RFA and HR, double-blindness seemed quite impossible. As a result, for these comparisons, blindness was only available for the statisticians (Figure 3).

Figure 1. PRISMA-NMA diagram of the literature search evaluating interventional RCTs for early-stage HCC through a selection process. Studies were identified from PubMed, Embase, Cochrane Library and Clinicaltrial.gov databases. There were 1202 references identified from the databases and a total of 35 RCTs were included in this study.

Figure 2. Network plot of enrolled studies. The number labeled beside each line is the number of studies comparing these two intervention therapies.

Figure 3. Quality assessment of included studies using Cochrane Collaboration’s tool. (A) Overall and (B) study-level risk of bias were assessed in these seven aspects.

Table 1. Baseline characteristics of included studies.

Study	Year	Journal	Group	Stage, tumor size	sample size	Age	Sex (male/total)	Child-Pugh (A/B)	Follow-up	Main outcomes	Complications
Abdelaziz	2014	Endo Surg	RFA versus MWA	BCLC Stage A, <5 cm	45 versus 66	56.8 ± 7.3 versus 53.6 ± 5	31 versus 48%	24/21 versus 25/41	2 years	LTP, OS	5/45 in RFA group versus 2/66 in MWA group
Qian	2012	Eur Radiol	RFA versus MWA	—; <3 cm	20 versus 22	56.0 ± 11.0 versus 52.0 ± 12.0	95 versus 91%	—	1 year	LTP	Pain, fever, etc
Shibata	2002	Radiology	RFA versus MWA	—; <4 cm	36 versus 36	63.6 versus 62.5	26/10 versus 24/12	21/15 versus 19/17	2 years	LTP	No significant difference of major complications
Yu	2017	Gut	RFA versus MWA	—; <5 cm	200 versus 203	—	—	—	5 years	LTP, OS, disease-free survival	No data
Chen	2006	Ann Surg	RFA versus surgery	—; Solitary, <5 cm	71 versus 90	51.9 ± 11.2 versus 49.4 ± 10.9	78.9% versus 83.3%	71/0 versus 90/0	5 years	OS, disease-free survival	Ascites, pain, etc. surgery > RFA
Fang	2014	J Gastroenterol Hepatol	RFA versus surgery	—; <3 cm	60 versus 60	51.4 ± 8.1 versus 53.5 ± 1 1.0	70% versus 76.7%	32/23 versus 43/17	40 months	Total recurrence rate, OS, disease-free survival	Major complication, 14/60 in surgery versus 1/60 in RFA
Feng	2012	J Hepatol	RFA versus surgery	—; <4 cm	84 versus 84	51 versus 47	94 versus 89%	39/45 versus 43/41	3 years	Total recurrence, OS, disease-free survival	8/84 in RFA versus 18/84 in surgery
Huang	2010	Radio Med	RFA versus surgery	Met the Milan criteria; <5 cm	115 versus 115	56.6 ± 14.3 versus 55.9 ± 12.7	68.7 versus 73.9%	110/5 versus 106/9	5 years	LTP, total recurrence, disease-free survival, OS	5/115 in RFA versus 32/115 in surgery
Lee	2018	Ann Surg Treat Res	RFA versus surgery	—; <4 cm	34 versus 29	56.1 ± 7.4 versus 55.6 ± 7.9	70.6 versus 79.3%	31/3 versus 29/0	5 years	LTP, OS, disease-free survival	9/34 versus 11/29, no significant difference
Lv	2006	Natl Med J China	RFA versus surgery	Milan criteria; <5 cm	51 versus 54	55 ± 13 versus 49 ± 14	82.4 versus 68.5%	46/5 versus 50/4	46 months	OS, disease-free survival	4/51 versus 6/54, no significant difference
Ng	2017	Br J Surg	RFA versus surgery	—; <5 cm	109 versus 109	57 versus 55	78.9 versus 81.7%	104/5 versus 107/2	10 years	LTP, tumor recurrence, OS, disease-free survival	No significant difference of postoperative complications between RFA and surgery
Azab	2011	Arab J Gastroenterol	RFA versus PEI	—, <5 cm	33 versus 32	—	—	—	1.5 years	OS	Pain, fever, portal vein thrombosis, etc, without significant difference
Brunello	2008	Scand J Gastroenterol	RFA versus PEI	—; <3 cm	70 versus 69	69.0 ± 7.7 versus 70.3 ± 8.1	61.4 versus 71.0%	39/31 versus 38/30	4 years	OS	10/70 in RFA versus 12/69 in PEI, without significant difference
Giorgio	2011	Anticancer Res	RFA versus PEI	—; <3 cm	142 versus 143	70 ± 2 versus 72 ± 6	73.9 versus 71.3%	70/72 versus 75/68	5 years	LTP, OS, costs	1/142 in RFA versus 3/143 in PEI, without significant difference
Lencioni	2003	Radiology	RFA versus PEI	—; <5 cm	52 versus 50	67 ± 6.0 versus 69 ± 7.4	69 versus 60%	45/7 versus 35/15	3 years	LTP, LTP-free survival, disease-free survival	Pain, fever, no severe complications were observed
Lin	2004	Gastroenterol	RFA versus PEI	—; <4 cm	52 versus 52	62 ± 11 versus 59 ± 10	67 versus 65%	41/11 versus 39/12	42 months	LTP, OS, disease-free survival	No severe complications were observed
Lin	2005	Gut	RFA versus PEI	—; <3 cm	62 versus 62	61 ± 10 versus 60 ± 8	65 versus 63%	46/16 versus 47/15	42 months	LTP, OS, disease-free survival	Major complication was 3/62 in RFA versus 0/125 in PEI and PAI group
Shiina	2005	Gastroenterol	RFA versus PEI	—; <3 cm	118 versus 114	—	67 versus 76%	85/33 versus 85/29	4 years	OS, Tumor recurrence rate	No significant difference between RFA versus PEI
Di Costanzo	2015	J Gastroenterol Hepatol	RFA versus laser ablation	—; <5 cm	70 versus 70	70 versus 70	75.7 versus 67.1%	63/7 versus 67/3	5 years	LTP, LTP-free survival, OS	No significant difference between RFA versus laser ablation
Ferrari	2007	Radio Med	RFA versus laser ablation	—; <4 cm	40 versus 41	70.5 ± 5.3 versus 68.3 ± 6.4	70.0 versus 68.3%	28/12 versus 31/10	5 years	LTP, total recurrence, OS	No data
Orlacchio	2014	Radio Med	RFA versus laser ablation	—; <4 cm	15 versus 15	71.5 ± 4.6 versus 73.5 ± 6.7	73.3 versus 66.7%	9/6 versus 10/5	1 year	Total recurrence, OS	No major complications, laser ablation has less minor complications compared with RFA
Wang	2015	Hepatology	RFA versus cryoablation	—; <4 cm	180 versus 180	53.3 ± 8.9 versus 53.9 ± 9.6	83.3 versus 77.8%	109/71 versus 121/59	5 years	LTP, OS, disease-free survival	Major complication was 7/180 in RFA group versus 6/180 in cryoablation group
Bian	2014	J Natl Cancer Inst	RFA versus RFA + radiotherapy	BCLC 0-B; —	65 versus 62	55 versus 55	86.2 versus 88.7%	51/14 versus 56/6	3 years	Tumor recurrence rate, OS	No serious adverse events or treatment-related deaths were observed
Chen	2014	J Hepatol	RFA versus RFA + radiotherapy	BCLC 0-A; -	68 versus 68	48.9 ± 7.3 versus 50.8 ± 6.8	77.9 versus 82.4%	47/21 versus 42/26	5 years	Tumor recurrence rate, OS	No significant difference was observed
Chen	2005	Chin J Oncol	RFA versus RFA + PEI	—; <5 cm	41 versus 45	50 versus 48	90.2 versus 80%	—	29 months	LTP, OS	No data
Azab	2011	Arab J Gastroenterol	RFA versus RFA + PEI	—; <5 cm	33 versus 33	—	—	—	1.5 years	OS	Pain, fever, portal vein thrombosis, etc, without significant difference
Zhang	2007	Radiology	RFA versus RFA + PEI	—; <7 cm	67 versus 66	52.2 ± 10.3 versus 53.3 ± 11.3	86.6 versus 86.4%	37/24 versus 42/20	5 years	LTP, total recurrence, OS	No procedure-related mortalities or major complications
Kobayashi	2008	Liver International	RFA versus RFA + TACE	—; <3 cm	10 versus 10	63 versus 67	80 versus 70%	3/5 versus 4/5	4 years	Tumor recurrence rate	No severe complication was observed
Morimoto	2010	Cancer	RFA versus RFA + TACE	—; <5 cm	18 versus 19	73 versus 70	67 versus 79%	16/2 versus 18/1	45 months	Tumor recurrence rate, OS	No severe complication was observed, no significant difference as to minor complications
Peng	2013	J Clin Oncol	RFA versus RFA + TACE	—; <7 cm	95 versus 94	55.3 ± 13.3 versus 53.3 ± 11.0	74.7 versus 79.8%	90/5 versus 90/4	5 years	Tumor recurrence rate, OS, recurrence-free survival	No significant difference
Shibata	2010	Radiology	RFA versus RFA + TACE	—; <3 cm	43 versus 46	69.8 ± 8.0 versus 67.2 ± 8.9	76.7 versus 67.4%	33/10 versus 32/14	4 years	LTP; OS; LTP-free survival; Event-free survival	Major complications: 1/43 in RFA group versus 1/46 in combined group
Yan	2016	Oncol Lett	Surgery versus RFA + Sorafineb	—; <3 cm	60 versus 60	62.6 ± 2.5	56.7%	—	5 years	LTP, tumor recurrence rate, OS, disease-free survival	Surgery has more adverse reactions compared with RFA
Huang	2005	Ann Surg	Surgery versus PEI	—; <3 cm	38 versus 38	59 ± 11.4 versus 63 ± 10.9	71.1 versus 50%	28/0 versus 29/3	5 years	Recurrence rate, OS	No significant complications in surgery and PEI groups
Liu	2016	Br J Surg	Surgery versus RFA + TACE	Met the Milan criteria; <5 cm	100 versus 100	49 versus 52	94 versus 86%	98/2 versus 96/4	5 years	LTP, OS	23/100 in surgery versus 11/100 in RFA + TACE group
Koda	2011	Cancer	PEI versus TACE + PEI	—; <3 cm	26 versus 26	66.4 ± 6.6 versus 66.2 ± 8.0	69.2 versus 53.8%	14/8 versus 19/5	7 years	LTP, OS	No significant difference was observed
Mizuki	2010	Oncol Lett	PEI versus TACE + PEI	—; <4 cm	14 versus 13	63.6 ± 6.2 versus 65.8 ± 7.3	50 versus 69.2%	–	5 years	LTP, OS, disease-free survival	No significant difference was observed

BCLC: Barcelona Clinic Liver Cancer; MWA: microwave ablation; PEI: percutaneous ethanol injection; LTP: local tumor progression; OS: overall survival; RFA: radiofrequency ablation; TACE: transcatheter arterial chemoembolization.

Network meta-analysis

The network meta-analyses of interventional techniques for early-stage HCC using R-software were presented in Figures 4–6. The main results are listed as follows.

Figure 4. Network meta-analysis for analysing the rate of different interventional treatments on LTP at 1, 2, 3 and longer than 3 years. (A) Network plots of the included studies and the connections among them. The four panels indicate the subgroup analysis of LTP at 1, 2, 3 and longer than 3 years. Abbreviations: RFA: radiofrequency ablation; MWA: microwave ablation; PEI: percutaneous ethanol injection; TACE: transcatheter arterial chemoembolization. (B) The relative forest plots of different interventional arms on LTP compared with RFA, using risk ratio (RR) values and 95% credible intervals (CrIs). (C) The results of treatment ranks using R-software. The abbreviations were the same as in (A). The dark bars indicate the high probability of being the best interventional arms, whereas the light bars suggest being the worst arms in this outcome.

LTP rate

LTP represents the appearance of tumor foci at the edge of the ablation zone. Most of the included studies reported the results of LTP rates. In this meta-analysis, the LTP rates of 1, 2, 3, and longer than 3 years were calculated, respectively, and displayed in Figure 4. As to LTP rate at 1 year, nine studies involving seven intervention arms were included for the network meta-analysis (Figure 4(A)). The results showed that local ablative techniques (MWA, PEI) and RFA plus TACE had similar LTP rates when treating early-stage HCC, whereas other RFA plus adjuvant therapies (RFA plus PEI and PEI plus TACE) were superior in decreasing the LTP incidence, exhibiting the best performance compared with most of the other treatments in this subgroup (Figure 4(B)). Furthermore, using rank probabilities tests, it showed that PEI plus TACE and RFA plus PEI ranked the first and second best treatment arms on the parameter LTP at 1 year (Figure 4(C)). For LTP at 2 years, a total of six studies involving six intervention arms were enrolled (Figure 4(A)). It showed that the intervention techniques had similar performance in the outcome of LTP rate at 2 years (Figure 4(B)). The rank probability test indicated that RFA plus adjuvant therapies (PEI, TACE) and PEI plus TACE showed better efficacy (Figure 4(C)). Due to the limited number of studies enrolled, the LTP at 2 years conferred limited information. Meanwhile, for the LTP rate at 3 years, 11 studies containing seven treatment arms were enrolled (Figure 4(A)). The pooled RR estimates showed that the intervention of hepatic resection was superior (RR 0.078, 95% CrI 0.0023, 0.61) and PEI was inferior (RR 2.1, 95% CrI 1.2, 3.9), compared with RFA intervention (Figure 4(B)). However, local ablative techniques (MWA, laser ablation) and RFA or PEI plus adjuvant therapies (TACE) had almost the same efficacy (Figure 4(B)). Using the rank probability test, it showed that hepatic resection had the highest probability of being the best treatment arm (Figure 4(C)).

Studies reporting the LTP rate longer than 3 years mainly contained the data of LTP at 5 years; a total of 12 studies including 10 treatment arms were enrolled for data synthesis (Figure 4(A)). The results showed that RFA plus adjuvant therapies (radiotherapy, PEI and sorafenib) had a lower LTP rate compared with RFA alone, whereas other interventions showed similar efficacy in decreasing the LTP (Figure 4(B)). The rank probability results showed that RFA plus adjuvant therapies are the best treatment arms, with RFA plus PEI being the best, followed by RFA plus sorafenib and RFA plus radiotherapy (Figure 4(C)). Therefore, the results of LTP showed that RFA plus adjuvant therapy and hepatic resection had the highest probability to be the best treatment arms. In detail, for studies with less than 3 years of follow-up data, the intervention of hepatic resection may decrease the LTP incidence most efficiently, while RFA plus adjuvant therapies may become the most effective measures for longer than 3 years.

Total recurrence

Concerning the outcome synthesis of total recurrence rate, the comparison was also divided into the 1-, 2-, 3-, and longer-than-3 year subgroups for further analysis. For the outcome of the total recurrence rate at 1 year, seven studies containing five intervention arms were enrolled for data synthesis (Figure 5(A)). It showed that the interventions of hepatic resection and RFA plus adjuvant therapies (RFA plus radiotherapy and RFA plus TACE) had decreased risk of total recurrence rate at 1 year (Figure 5(B)). The rank probability test results were consistent with the forest plot, showing that RFA plus adjuvant therapies (radiotherapy and TACE) also had the highest probability to be the best treatment arm, followed by hepatic resection (Figure 5(C)).

Figure 5. Network meta-analysis for analysing the rate of different interventional treatments on total recurrence at 1, 2, 3 and longer than 3 years. (A) Network plots of the included studies and the connections among them. The four panels indicate the subgroup analysis of total recurrence at 1, 2, 3 and longer than 3 years. Abbreviations: MWA: microwave ablation; PEI: percutaneous ethanol injection; RFA: radiofrequency ablation; TACE: transcatheter arterial chemoembolization. (B) The relative forest plots of different interventional arms on total recurrence rates compared with RFA, using risk ratio (RR) values and 95% credible intervals (CrIs). (C) The results of treatment ranks using R-software. The abbreviations were the same in (A).

For the subgroup analysis of the total recurrence rate at 2 years, six studies involving four treatment arms were enrolled for network meta-analysis (Figure 5(A)). It showed that the treatments of hepatic resection and RFA plus radiotherapy had a lower risk of total recurrence at 2 years (Figure 5(B)). Using a rank probability test, the results were consistent with the forest plot, suggesting that hepatic resection and RFA plus radiotherapy had the high probability of being the best treatment arms (Figure 5(C)).

For the subgroup analysis of total recurrence rate at 3 years, eight studies containing 5 treatment arms were included (Figure 5(A)). The relative forest plot showed that the treatments of hepatic resection and RFA plus radiotherapy decreased the total recurrence rate compared with other treatment arms (Figure 5(B)). The rank probability test showed that RFA plus radiotherapy had the highest probability of becoming the best treatment arm, followed by hepatic resection (Figure 5(C)).

For the subgroup analysis of total recurrence longer than 3 years, seven studies containing six treatment arms were included (Figure 5(A)). The network meta-analysis showed that the treatments of hepatic resection and RFA plus adjuvant therapies had decreased total recurrence compared with RFA (Figure 5(B)). The rank probability test also showed that RFA plus sorafenib, RFA plus PEI and RFA plus radiotherapy had the highest probability of being the best arms (Figure 5(C)). In this outcome analysis, it still suggested that the treatment of hepatic resection was the most effective for decreasing total recurrence at 1 or 2 years, whereas RFA plus adjuvant therapies help to decrease total recurrence for 3 or more than 3 years.

Overall survival rates

For the analysis of survival rates after different interventions, the HR and its corresponding 95% CtrI values were calculated. The outcome of survival analysis was divided into the subgroup analysis of overall survival (OS), disease-free survival and LTP-free survival. For the data synthesis of OS, a total of 24 studies containing 11 specific interventional arms were included (Figure 6(A)). It showed that RFA plus adjuvant therapies (radiotherapy, TACE and sorafenib) and hepatic resection were the best interventions to achieve the better OS rates when compared with local ablative techniques. Among the local ablative techniques, MWA and cryoablation had better efficacy, whereas PEI and laser ablation were inferior compared with RFA in OS comparisons (Figure 6(B)). In addition, the rank probability test showed that RFA plus adjuvant therapies (radiotherapy, sorafenib and TACE) ranked the best treatment arm in achieving the highest OS rate (Figure 6(C)).

Figure 6. Network meta-analysis for analysing the rate of different interventional treatments on overall, disease-free and LTP-free survival rates. (A) Network plots of the included studies and the connections among them in the comparison of survival rates. The three panels indicate the analyses of overall survival rates at 1, 2, 3 and longer than 3 years. Abbreviations: MWA: microwave ablation; PEI: percutaneous ethanol injection; RFA: radiofrequency ablation; TACE: transcatheter arterial chemoembolization. (B) The relative forest plots of different interventional arms on overall survival, disease-free survival and LTP-free survival rates compared with RFA, using HR values and 95% credible intervals (CrIs). (C) The results of treatment ranks using R-software. The abbreviations were the same in (A).

For the analysis of disease-free survival, a total of 13 studies containing 8 specific interventional arms were enrolled (Figure 6(A)). The forest plot in Figure 6(B) exhibited similar efficacies in achieving better disease-free survival, with hepatic resection being the best among all the interventional arms (RR 0.68, 95% CrI 0.57, 0.90, compared with RFA). Using the rank probability test, it showed that hepatic resection ranked the best in the comparison of disease-free survival (Figure 6(C)).

For the analysis of LTP-free survival, five studies containing five specific treatment arms were enrolled, shown in Figure 6(A). The forest plot indicated that the efficacy of local ablative techniques (RFA and laser ablation) and RFA plus TACE were better than PEI (Figure 6(B)). In addition, the rank probability test suggested that RFA plus adjuvant therapy (TACE) ranked the first and it is the best intervention arm for in achieving better LTP-free survival (Figure 6(C)).

Discussion

To date, several comparisons of different intervention techniques on HCC treatment have been published; however, the network meta-analyses on this topic are still very rare [8,13]. This study collected the latest data and compared the most commonly used interventional techniques for early-stage HCC, which might be the most comprehensive study currently. In this network meta-analysis, we summarized the main interventional techniques for early-stage HCC and compared the rates of LTP and total recurrence, the overall, disease-free and LTP-free survival of different treatment arms on these patients. It was demonstrated that RFA plus adjuvant therapies achieved the best performance in decreasing LTP in early-stage HCC, while for LTP at 3 years, hepatic resection ranked the best of being the best arm among all the interventional techniques, with RR 0.079 and 95% CrI (0.0023, 0.61) when compared to RFA alone. Meanwhile, for the outcome of total recurrence, the results showed that hepatic resection and RFA plus adjuvant therapies were also the two best arms in decreasing total recurrence, with hepatic resection being the best when less than 3 years, while RFA plus adjuvant therapies ranked the best if the follow-up period was longer than 3 years. Furthermore, the survival data was also synthesised, demonstrating that RFA plus adjuvant therapeutics rank the best in achieving the best OS rate, followed by hepatic resection and local ablative techniques. For disease-free survival, hepatic resection was the best while for LTP-free survival, the difference between these treatments was not significant.

Recently, as a result of technological development, the choice for the best treatment strategy became quite difficult, and it mainly depends on the doctor’s experience and the patients’ consent. Hepatic resection is the always the first-line choice in early-stage HCC treatment, but it always brings about a high morbidity of complications, such as infection, bleeding, hepatic failure, etc. Meanwhile, local ablative techniques are also widely used in the treatment of early-stage HCC. Among them, RFA is a kind of commonly used medical procedure that achieves tumor ablation through high-frequency alternation current and has now become the most commonly used local ablation therapeutic. Several studies indicated that RFA was better than PEI [52,53] but was inferior to hepatic resection for early-stage or small HCC [54]. Recent advancement also brings about new techniques such as microwave ablation, laser ablation and cryoablation, which destruct the tumor through multiple mechanisms [14], with the new classification of chemical ablation and energy-based ablation. The effects of different treatments on tumor recurrence and survival are still not clear. For microwave ablation, recent studies have concentrated on the efficacy of microwave ablation compared with RFA and hepatic resection. A recent meta-analysis showed that microwave ablation may be superior to hepatic resection as it is as effective as hepatic resection in terms of overall survival, disease-free survival, tumor recurrence and is associated with fewer complications [55], while microwave ablation is better than RFA in preventing LTP in treating large tumors [5,56]. In our network meta-analysis, we demonstrated that microwave ablation also performed well in regard to LTP. The treatment of microwave ablation had similar efficacy in achieving better survival rates, but limited data was available as to the role of microwave ablation in the outcome of total recurrence. Therefore, more importance should be attached to microwave ablation and more RCTs are needs to compare the efficacy of microwave ablation with other interventions. Retrospective studies have demonstrated that cryoablation had similar treatment successes and complication rates but had increased local recurrence rate compared with RFA [57,58]. However, Li et al. have demonstrated that cryoablation was as effective as hepatic resection in the treatment of solitary, small HCC with a short duration of hospitalisation and fewer complications [59]. In our study, the treatment of cryoablation played a better role in the outcome of LTP at 1 year compared with microwave ablation and PEI and with similar OS and disease-free survival compared with RFA. Furthermore, the treatment of laser ablation has also been studied in recent years with almost similar efficacy compared to RFA, but with fewer complications [7,38,39]. However, the number of studies concerning laser ablation is not very large. It has been shown in our study that laser ablation had inferior total recurrence and overall survival rates compared to RFA, but even so, more RCTs concerning the new techniques such as cryoablation, laser ablation and irreversible electroporation should be conducted. In our network meta-analysis, we showed that hepatic resection is better in all the aspects compared to local ablative techniques such as RFA, PEI, and microwave ablation, indicating that hepatic resection will achieve the most complete efficacy when treating early-stage HCC but with more complications and a longer recovery time.

Furthermore, TACE, which concentrates chemotherapeutic agents at the tumor site while blocking the primary artery feeding the tumor, was considered to be the most frequent loco-regional treatment for unresectable HCC [3], whereas it is reported that TACE is starting to be used in early-stage HCC and shows a prolonged survival rate. Therefore, combination therapeutics become more and more popular in HCC treatment. In addition to TACE as the combination therapy, adjuvant therapies also develop very quickly with the classification of concomitant agents, combination therapies, and concurrent therapies. A previous meta-analysis also showed that RFA-based combination treatment has the highest probability in achieving the best OS rate [8]. In our network meta-analysis, we showed that the treatments of RFA plus adjuvant therapeutics achieved the best efficacy in almost all the aspects of the outcomes. Therefore, combination therapy is strongly recommended for early-stage HCC.

The methodological quality in this meta-analysis shows that several studies included did not perform well in the procedure of randomization and blindness. It is because in some comparisons, the two treatment arms were totally different, such as RFA versus HR. Therefore, these studies had a high risk of bringing bias. Meanwhile, there were some heterogeneity for some of the outcomes in conventional meta-analysis. For example, the participants in the studies enrolled may have different viral infections, while some were cirrhotic and some were noncirrhotic. Therefore, when interpreting the results of this network meta-analysis, these factors should be taken into consideration.

There were some limitations. First, network meta-analysis has the advantage of theoretically being able to compare technologies/interventions that were not actually compared in any trial. There are many unknowns about this methodology and the results need careful interpretations. Second, we only enrolled the RCTs in this meta-analysis and some high-quality cohort studies were still excluded because this may bring great heterogeneity when synthesising data. Third, we did not detail the composition of chemotherapy because different compositions of chemotherapy were found in the limited number of studies included. Fourth, we did not synthesise the data of complications because different complications were reported, and some were rather subjective.

In conclusion, by using network meta-analysis, it was shown that RFA plus adjuvant therapeutics and hepatic resection achieved the best results in treating early-stage HCC, by decreasing the LTP and the total recurrence rate and by improving the OS, disease-free, and LTP-free survival rates. Nevertheless, more high-quality RCTs are needed to provide more solid evidence.

Disclosure statement

No potential conflict of interest was reported by the authors.