Transarterial chemoembolization vs bland embolization in hepatocellular carcinoma: A meta-analysis of randomized trials

Abstract

Background

Although transarterial chemoembolization is considered the standard of care for intermediate hepatocellular carcinoma patients, robust data in favor of a clear superiority of chemoembolization (with chemotherapy injection) over bland embolization are lacking.

Objective

The objective of this article is to systematically analyze the results provided by randomized controlled trials comparing these two treatments in hepatocarcinoma patients.

Methods

A computerized bibliographic search on the main databases was performed. Survival rates assessed at one, two, and three years, objective response, one-year progression-free survival, and severe adverse event rate were analyzed. Comparisons were performed by using the Mantel-Haenszel test in cases of low heterogeneity or DerSimonian and Laird test in cases of high heterogeneity.

Results

Six trials with 676 patients were included. No difference in one-year (risk ratio: 0.93, 0.85–1.03, p = 0.16), two-year (risk ratio: 0.88, 0.74–1.06, p = 0.18) and three-year survival (risk ratio: 0.97, 0.74–1.27, p = 0.81) was observed. Objective response and one-year progression-free survival showed no significant difference between the two treatments (p = 0.36 and p = 0.40, respectively).

A statistically significant increase in severe toxicity after chemoembolization was found (risk ratio: 1.44, 1.08–1.92, p = 0.01), although this result could be affected by the heterogeneity of techniques adopted.

Conclusions

Our meta-analysis demonstrates a non-superiority of transarterial chemoembolization with respect to bland embolization in hepatocarcinoma patients.

Introduction

Hepatocellular carcinoma (HCC) represents the sixth most common cancer worldwide and the leading cause of mortality among patients with liver cirrhosis.^1,2 Current guidelines call for transarterial chemoembolization (TACE) as first-line treatment in patients with large or multinodular HCC and relatively preserved liver function, as long as neither cancer-related symptoms nor vascular invasion/extrahepatic spread occur.^2–4

The efficacy of TACE was established by a meta-analysis published a decade ago that included six randomized controlled trials (RCTs), only two of which, however, were positive.⁵

Furthermore, robust data in favor of a clear superiority of TACE over transarterial embolization (TAE) without any drug injection are still lacking.⁶ An RCT comparing conventional TACE (cTACE), TAE and best supportive care (BSC) was prematurely terminated because of the superiority of cTACE over BSC and this unfortunately prevented the possibility of verifying the efficacy of TAE, which could only be hypothesized based on the trend observed in overall survival (OS).⁷

On the other hand, other RCTs published more recently on this topic seem to provide discordant results.^8,9

These discrepancies could be due to a number of unsolved issues concerning TACE, since for instance the optimal chemotherapeutic agent to inject or the more effective drug-eluting and embolic agent to adopt during the procedure are still matters of debate.^10,11

The aim of this meta-analysis is to retrieve and systematically analyze the results provided by all the published RCTs comparing TACE and TAE in terms of OS, progression-free survival (PFS), objective response rate (ORR), and safety profile.

Methods

This meta-analysis is performed on the basis of Cochrane’s guidelines¹² and conducted according to the Preferred Reporting Items for Systematic reviews and Meta-Analysis (PRISMA) statements.¹³

Search strategy and selection criteria

Figure 1 reports the search strategy followed in the meta-analysis.

Figure 1.

Flowchart summarizing study selection. RCTs: randomized controlled trials.

A computerized bibliographic search was performed on PubMed/Medline, Embase, Google Scholar, and Cochrane library databases independently by two authors (AF, RV) using the following keywords: “transarterial chemoembolization,” “embolization,” “hepatocellular carcinoma,” “HCC” and “liver cancer.” A complementary manual search was performed by checking the references of all the main review articles on this topic, in order to identify possible additional studies.

Eligible studies were RCTs published until February 2016 that fulfilled the following inclusion criteria: (1) compared TACE and TAE in HCC patients; (2) reported at least one of the following data: response rate, survival and adverse events; (3) reported relative risk ratio (RR) and hazard ratio (HR) or provided data for their calculation; (4) articles written in English. Case reports and abstracts or studies with insufficient data were excluded. When incomplete information was available, attempts were made to contact the corresponding authors for additional data. Disagreements were solved by discussion and following a third opinion (GS).

The quality of the included studies was assessed by two authors independently (AF, RV) according to the Cochrane Collaboration’s tool for assessing the risk of bias.¹²

Statistical analysis

The primary endpoint was the comparison of patient survival between the two treatments, expressed in terms of one-year, two-year and three-year survival rate (SR). Other outcomes evaluated were ORR (complete +partial) and severe (grade 3/4) adverse event (SAE) rate. Dichotomous data were pooled and analyzed in terms of RR and 95% confidence intervals (CIs). Comparisons between the two treatment groups across all the included studies were performed by using the Mantel-Haenszel test for fixed-effects models (in the case of low heterogeneity) or the DerSimonian and Laird test for random-effects models (in the case of high heterogeneity).

Heterogeneity between estimates was assessed by means of Cochrane’s chi-square test and a p value of < 0.10 was considered significant. Then, I² statistic was calculated and a value of >50% was suggestive of significant heterogeneity.¹²

Publication biases were assessed using funnel plots visually and performing Begg and Mazumdar’s test.

Sensitivity analysis was finally performed in order to assess the effect of lower quality articles on the final results of the meta-analysis.

All calculations were performed using Review Manager (version 5.0 for Windows; the Cochrane Collaboration, Oxford, UK) and R 3.0.2 (R Foundation for Statistical Computing, Vienna, Austria), metafor package.

Results

Literature search

As shown in Figure 1, 362 studies were initially identified. After preliminary exclusion of abstracts or papers not fulfilling the search criteria, 23 potentially relevant articles were examined. Among these studies, 17 were excluded because of incomplete data or because they were not RCTs (systematic reviews or retrospective reports). Finally, six RCTs^{7–9,14–16} with 676 patients (342 treated with TACE and 334 with TAE) were included in the meta-analysis.

Characteristics of included studies

Main characteristics of included studies are reported in Table 1.

Table 1.

Characteristics of the included studies.

Study	Arm	Drug	Sample size	Study period	Region	CP (A/B/C)	ECOG PS (0/≥1)	Mean tumor size (mm)	Number nodules (S/M)	Quality
Kawai 199214	cTACE TAE	Doxorubicin	148 141	1988–1989	Japan	107/33/7 102/25/3	71/51 77/41	33 28	103/24 96/32	Mo
Chang 199415	cTACE TAE	Cisplatin	22 24	1991–1993	China	13/9/0 17/17/0	NA NA	NA NA	9/13 11/13	Mo
Llovet 20027	cTACE TAE	Doxorubicin	40 37	1996–2000	Spain	31/9/0 27/10/0	35/5 28/9	49 52	13/27 9/28	H
Malagari 20109	DEB-TACE TAE	Doxorubicin	41 43	2005–2009	Greece	23/18/0 26/17/0	26/15 28/15	83.5 81	12/29 15/28	Mo
Meyer 20138	DEB-TACE TAE	Cisplatin	44 42	2003–2008	England	38/6/0 33/9/0	31/13 27/15	NA NA	15/29 13/29	H
Brown 201616	DEB-TACE TAE	Doxorubicin	50 51	2007–2012	USA	45/5/0 41/10/0	43/7 44/7	35 34	12/38 12/39	H

TACE: transarterial chemoembolization; TAE: transarterial embolization; CP: Child Pugh; ECOG PS: Eastern Cooperative Oncology Group Performance Status; Mo: moderate; H: high; S: single nodule; M: multinodular; USA: United States of America; NA: not available.

The recruitment period ranged from 1988 to 2012. Three studies were conducted in Western Europe,^7–9 one in the United States (US)¹⁶ and two in Asia.^14,15 The three older studies adopted cTACE^7,14,15 whereas in the other three RCTs drug-eluting beads (DEB)-TACE was performed.^8,9,16 Doxorubicin was the most frequently injected drug in TACE cohorts (see Table 1). In all the studies but that by Kawai et al.,¹⁴ patients enrolled were within Child-Pugh B score and presented prevalently with multinodular tumoral disease. None of the studies reported statistically significant differences in baseline demographic, clinical and tumoral parameters between the two treatment groups. Three RCTs^7,8,16 were considered high quality and three^9,14,15 moderate quality. More details on the methodological characteristics and quality of included articles are shown in Supplementary Table 1.

Other technical characteristics of the trials included in this meta-analysis (such as selectivity of drug delivery, tumor response criteria adopted, timing of response assessment, and treatment repetition schedule) are reported in Supplementary Table 2.

Survival

Concerning patient survival, the results of our meta-analysis confirm previous findings⁸ on the non-superiority of TACE over TAE. In fact one-year SR, reported in all the included studies, was similar between the two treatment regimens (RR 0.93, 95% CI 0.85–1.03, p = 0.16). As described in Table 2, this finding was confirmed at two and three years (RR 0.88, 0.74–1.06, p = 0.18 and RR 0.97, 0.74–1.27, p = 0.81, respectively). Forest plots of one-year, two-year and three-year SR are reported in Supplementary Figures 1–3.

Table 2.

Risk ratios and heterogeneity of one-year, two-year and three-year survival rate.

Survival estimate	No. of studies	No. of patients	RR (95% CI)	p value	Heterogeneity I2 p
One-year SR	6	676	0.93 (0.85–1.03)	0.16	0% .60
Two-year SR	5	592	0.88 (0.74–1.06)	0.18	0% .55
Three-year SR	3	460	0.97 (0.74–1.27)	0.81	0% .54

SR: survival rate; RR: risk ratio; CI: confidence interval.

Noteworthy is the fact that no evidence of heterogeneity was found with regard to SR at any time points evaluated (Table 2) and no publication bias was detected by visual examination of funnel plots (Supplementary Figures 4–6) and with Begg and Mazumdar’s test (p = 0.34, 0.25, and 0.52, respectively).

Sensitivity analysis conducted both by restricting analysis to high-quality studies and by excluding each article once per time did not change the main summary estimate (data not shown).

Therefore, the absence of heterogeneity and the rigorous sensitivity analysis strengthen the conclusion that TACE is not superior to TAE in the treatment of HCC patients.

ORR

ORR was reported in four studies.^7–9,16 Since a moderate level of heterogeneity was found (χ²= 5.75, degree of freedom (df) = 3, I²= 48%; p = 0.12), the DerSimonian and Laird test for random-effects models was used. Pooled data from these articles showed no significant difference between the two treatments with an RR of 1.14 (0.86–1.49, p = 0.36; Figure 2). No evidence of publication bias was found (p = 0.54).

Figure 2.

Forest plot comparing objective response rate for TACE to that for TAE. Random-effect DerSimonian Laird model showed a summary risk ratio slightly higher after TACE without reaching statistical significance. Moderate heterogeneity was found. TACE: transarterial chemoembolization; TAE: transarterial embolization.

Unfortunately, meta-regression analysis aimed at exploring sources of heterogeneity was not feasible because of the low number of available RCTs and sensitivity analysis conducted either by study quality (moderate vs high) and drug injected (doxorubicin vs cisplatin) was unable to detect these parameters as responsible for the moderate heterogeneity found.

Sensitivity analysis performed by means of elimination of single studies did not show significant deviations in overall summary estimate (data not shown). Additional sensitivity analyses conducted according to those technical parameters known to influence tumor response, namely selectivity of drug delivery (non-selective/lobar vs selective), tumor response criteria (Response Evaluation Criteria In Solid Tumors (RECIST)/World Health Organization (WHO) vs European Association for the Study of the Liver (EASL)), timing of response assessment (1 month vs >1 month), and treatment repetition schedule (pre-defined vs on-demand), confirmed the non-significant difference between the two treatments (Supplementary Table 3).

However, these findings should be interpreted with caution because of the heterogeneous characteristics of the populations enrolled in the trials and the different technical devices adopted throughout the long time span in which the included studies were retrieved (1988–1989 to 2007–2012) that could not be taken into account only through sensitivity analysis.

PFS

Three studies reported one-year PFS data.^8,9,16 Meta-analysis of such an outcome found a non-significant trend in favor of TACE (RR = 1.35, 0.67–2.71, p = 0.40) (Figure 3).

Figure 3.

Forest plot comparing one-year progression-free survival between TACE and TAE. Meta-analysis of one-year PFS found a non-significant trend in favor of TACE (RR = 1.35, 0.67–2.71, p = 0.40). High heterogeneity was present. PFS: progression-free survival; TACE: transarterial chemoembolization; TAE: transarterial embolization; RR: risk ratio.

Overall RR was computed by means of the DerSimonian-Laird test because of the high grade of heterogeneity found among the included studies (χ²= 8.0, df = 2, p = 0.02, I²= 75%) (Figure 3).

In order to assess the influence of the single lower-quality study on final results and to explore the sources of heterogeneity, RR was re-calculated considering only high-quality papers. After performing this kind of sensitivity analysis, RR decreased near to 1 (RR = 0.97, 0.59–1.59, p = 0.90) with a considerably lower grade of heterogeneity (χ²= 1.47, df = 1, p = 0.23, I²= 22%).

Funnel plot and Begg and Mazumdar’s test (p = 0.59) did not show any evidence of publication bias (data not shown).

Despite the sensitivity analysis and the absence of publication bias, we think the PFS findings are supported only by a very low grade of evidence because of the small number of trials reporting this outcome and the considerable heterogeneity. Further trials are needed in the future to properly assess this aspect.

Safety

Meta-analysis of five studies^7–9,15,16 with 394 patients recording data on SAE rate reported a statistically significant increase in severe toxicity after TACE as compared to TAE (RR = 1.44, 1.08–1.92, p = 0.01; Figure 4). There was no evidence of heterogeneity (χ²= 0.59, df = 4, I²= 0%; p = 0.96) nor of publication bias (p = 0.23). Sensitivity analysis conducted according to the drug injected confirmed the aforementioned result, whereas analysis restricted only to DEB-TACE studies showed a slightly decreased RR barely above the significance threshold (RR = 1.30, 0.99–1.70, p = 0.06). Additional sensitivity analyses conducted according to the aforementioned procedural characteristics of the trials (drug delivery, tumor response criteria, timing of response assessment, and repetition schedule) confirmed the superiority of TAE in terms of safety profile (Supplementary Table 3).

Figure 4.

Forest plot comparing severe adverse event (SAE) rate between TACE and TAE. Meta-analysis of SAE rate reported a statistically significant increase in severe toxicity after TACE as compared to TAE (RR = 1.33, 1.03–1.73, p = 0.03), with no evidence of heterogeneity. TACE: transarterial chemoembolization; TAE: transarterial embolization.

Unfortunately meta-analysis stratified according to each side effect was not possible because of heterogeneous and incomplete reporting; however, as shown in Supplementary Table 4, the incidence of post-embolization syndrome (when reported) was similar between the two treatment regimens, and among SAEs acute cholecystitis, liver failure, embolism and sepsis were the most common.

However, we think that the heterogeneity in (a) reporting the adverse events, (b) the doses of the chemotherapeutical agent, (c) the type of chemotherapeutical agent and delivery modalities, and (d) the different relevance of the various adverse events impede drawing any firm conclusions about the real superiority of TAE vs TACE also in terms of adverse events.

Discussion

Although TACE is considered the standard of care for intermediate HCC patients, robust data in favor of a clear superiority of TACE (with chemotherapy injection) over TAE (bland embolization) are still lacking.

In fact, while the well-known hypervascularization of HCC nodules provides the rationale for the occlusion by embolic particles that results in tumor hypoxia and necrosis, on the other hand whether the addition of local chemotherapy could have an additive anti-tumor effect is still a matter of debate.^6,17

A previous systematic review failed to find a significant difference in OS between the two treatment regimens; however, this meta-analysis was published before the introduction of DEBs and its findings apply only to conventional TACE (cTACE).¹⁸ Therefore, whether the aforementioned non-superiority of TACE over TAE still persists in light of the recently developed devices is unclear.

The aim of our meta-analysis was to perform a comprehensive and up-to-date overview of available data on the comparison between TACE and TAE in HCC patients. In order to draw a robust and reliable conclusion on this important topic, only RCTs were retrieved.

A total of six studies with 676 patients were included. The three older studies adopted cTACE^7,14,15 whereas in the other three RCTs DEB-TACE was performed.^8,9,16 Doxorubicin was the most frequently injected drug in TACE cohorts (Table 1).

Pooled RRs of SR at one, two and three years did not differ significantly between the two treatments (0.93, 0.88 and 0.97, respectively). This finding, which is in keeping with previous reports,^8,18,19 was confirmed in sensitivity analysis and strengthened by the absence of heterogeneity.

Our analysis showed no significant difference between the two treatments in terms of ORR (RR: 1.14, p = 0.36) nor of one-year PFS (RR: 1.35, p = 0.40). Therefore, response and progression data seemed to confirm the non-superiority of TACE but the high heterogeneity and the low number of trials impeded drawing definitive conclusions in this regard.

However, it is very important to highlight some critical issues in debating this topic and to remark on the difficulty of overcoming limits of the current analysis.

The literature on this topic comprises trials including different size cohorts, patient populations and tumor stage compositions. For instance, some studies also included patients with locally advanced disease, for which TACE was never proven effective, so that differences in survival vs other endovascular embolizing comparators are hardly achievable (unless the comparators were toxic).²⁰ Secondly, the embolizing agent varied over time in the last decades, hence the different size or the different embolizing techniques may lead to lack of demonstration of efficacy due to dilution bias and high heterogeneity in the meta-analysis of data.^20,21 This is particularly relevant considering that the time range in which the included studies were retrieved spanned from 1988–1989 to 2007–2012. Hence the strength of our conclusion is limited by such an intrinsic shortcoming. At the same time a more restricted selection of the literature would lead to a too-limited amount of data to be analyzed. Hence, at present there is not a better way to provide data on the comparison in survival outcomes between TACE and TAE.

The current meta-analysis revealed that TACE presents more disadvantages than TAE with regard to the incidence of SAEs (RR: 1.33, p = 0.03). Although precise stratification of data according to singular adverse events was not possible because of incomplete reports in most included studies, we can conclude that TACE resulted in a significantly higher rate of post-embolization syndrome and systemic adverse events (liver abscess, liver failure, gastrointestinal hemorrhage, leukopenia, ischemic biliary stricture, spontaneous bacterial peritonitis, septic shock, allergic dermatitis, and severe alopecia).

Unfortunately, the aforementioned heterogeneity concerning technical aspects of the procedures and safety outcomes reporting impedes drawing any firm conclusions about the real superiority of TAE vs TACE also in terms of adverse events.

To summarize, there are some limitations to our study. First of all, response rate and PFS data were affected by moderate/high heterogeneity. Second, the low number of included studies requires particular caution in interpreting our findings. However, we deliberately decided to restrict inclusion criteria only to RCTs in order to provide more robust and reliable outcome estimates.

In conclusion, despite these weaknesses, our meta-analysis supports the non-superiority of TACE with respect to TAE, which in turn appears even safer particularly when compared to conventional chemoembolization. These conclusions need to be confirmed in broad non-inferiority trials with a large number of cases, strict selection of patients in terms of tumor burden, severity of liver dysfunction (as this strongly affects OS) as well as comorbidities (as these as well affect survival) and standardized modality of endovascular tumor treatment and reporting of adverse events.²² In the lack of a better source of evidence, the present meta-analysis appears to provide the most possible solid information on the comparison of TACE with TAE.

Declaration of conflicting interests

None declared.

Funding

This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.