Meta-analysis: preoperative transcatheter arterial chemoembolization does not improve prognosis of patients with resectable hepatocellular carcinoma

Background Long-term outcomes of partial liver resection of hepatocellular carcinoma (HCC) remain satisfactory due to high incidences of recurrence. This study was intended to see whether preoperative transcatheter arterial chemoembolization (TACE) reduces postoperative tumor recurrences and prolongs survival of patients with resectable HCC. Methods A computerized literature search was performed to identify relevant articles. The quality of nonrandomized comparative studies (NRCTs) was assessed using the methodological index for nonrandomized studies (MINORS). Data synthesis was performed using Review Manager 5.0 software. Results Twenty-one studies (4 randomized controlled trials and 17 NRCTs) with a total of 3,210 participants were suitable for analysis. There was no significant difference in disease-free and overall survival at 5-year (32.1% vs. 30.0% and 40.2% vs. 45.2%), and intra- and extra-hepatic recurrence (51.2% vs.53.6% and 12.9% vs.10.3%) between patients with and without preoperative TACE. Postoperative morbidity (28.9% vs. 26.8%) and in-hospital mortality (4.1% vs. 3.1%) were also similar between the two groups. Conclusions Preoperative TACE does not seem to improve prognosis and therefore it is prudent to recommend it as a preoperative routine procedure for resectable HCC.


Background
Hepatocellular carcinoma (HCC) is the fifth most common malignancy worldwide, and ranks the third leading cause of cancer-related death [1]. Hepatectomy is considered the main curative treatment for HC with a 5-year survival of 23.8-54.8% as reported in the most recent series [2][3][4][5]. Tumor recurrence even after radical surgery affects 75-100% HC patients, accounting for the major cause of death of HC patients [6]. The development of appropriate strategies to prevent tumor recurrence is therefore critical for improving long-term outcomes of HC patients after surgical resection.
Transcatheter arterial chemoembolization (TACE), which induces tumor ischemic necrosis by arterial injection of chemotherapeutic drugs and embolizing agents, is recommended as the first-line palliative treatment for inoperable HCC in the 2005 practice guidelines issued by the American Association for the Study of Liver Diseases [7]. Some researchers believed that it may reduce the viability of HCC cells before surgery and reduce postoperative tumor recurrence [8]. Although several studies have demonstrated the survival benefits of preoperative TACE for patients with HCC [9,10], others have failed to show any significant survival benefit [11][12][13][14][15][16]. Therefore, the role of preoperative TACE for HCC remains a contradictory issue. A recent review of three randomized controlled trials (RCTs) showed that preoperative TACE did not significantly improve survival [17]. However, their study only involved 257 participants, and therefore may not be convincing enough to confirm the effect conclusively [18]. To produce more reliable evidence for clinical decision-making, meta-analytic techniques could be applied to nonrandomized comparative studies (NRCTs) to ensure both the number and sample size of RCTs [19], knowing that meta-analysis of well-designed NRCTs is feasible and the results are remarkably similar to those of contemporaneous RCTs [20]. The present meta-analysis should be able to provide an updated evaluation on preoperative TACE for resectable HCC by taking into account all the currently evidence from RCTs and well-designed NRCTs.

Inclusion and exclusion criteria
The present meta-analysis was performed by following the recommendations of the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) Statement. Primary studies that evaluated the efficacy of preoperative TACE vs. no-preoperative TACE for resectable HCC were considered for inclusion. For duplicate publications reported by the same authors, either the one of higher quality or the most recent publication was selected. Abstracts, letters, editorials and expert opinions, reviews without original data, case reports, and uncontrolled studies were excluded. Study populations including recurrent HCC or unresectable diseases were also excluded.

Study selection
A computerized search of the literature was performed by searching Medline, EMBASE, OVID, and Cochrane database from the time of inception to June 2012. The following medical subject heading (MeSH) terms were used: "hepatectomy," "hepatocellular carcinoma," and "transarterial chemoembolization". Only studies on humans and in the English language were considered for inclusion. Reference lists of all identified articles were manually searched for additional studies. Publication bias was assessed visually using a funnel plot.

Data extraction
Two reviewers (LW and YZ) independently extracted the following parameters from each study: first author, year of publication, study population characteristics, study design, inclusion and exclusion criteria, the number of subjects in each arm, survival, recurrence, morbidity and mortality. All relevant texts, tables and figures were reviewed for data extraction.

Qualitative analysis
The RCTs were scored using the Jadad composite scale [21], which evaluates studies based on appropriate randomization, double blinding, and an adequate description of withdrawals and drop-outs. For evaluation of NRCTs, the methodological index for nonrandomized studies (MINORS) with some modifications were applied [22]. The following 12 items were evaluated for each study: a clearly stated aim, consecutive patients, prospective data collection, reported endpoints, unbiased outcome evaluation, adequate length of follow-up, loss to follow up <5%, ≥ 20 patients in each arm, adequate control group, contemporary groups, controls equivalent to cases, and adequate statistical analyses. Studies achieving more than 16 points (maximum 24) were defined as well-designed and were included in the meta-analysis [20]. Those with less than 16 points were excluded.

Outcomes of interests
Items for assessing long-term outcomes included diseasefree and overall survival rate at 5 year, total recurrence, hepatic recurrence, and extrahepatic recurrence.
Items for assessing short-term outcomes included postoperative morbidity and in-hospital mortality.

Statistical analysis and synthesis
Dichotomous variables were tested by odds ratio (OR) with a 95% confidence interval (95% CI), and continuous variables were tested by the weighted mean difference (WMD) with a 95% CI. Between-study heterogeneity was evaluated by χ 2 and I 2 . Data that were not significantly heterogeneous (P > 0.1) were calculated using a fixed effects model, and heterogeneous data (P < 0.1) were calculated using a random-effects model. Sensitivity analysis was undertaken by using the following subgroups: (1) RCTs, (2) NRCTs, and (3) studies with matched clinicopathological parameters. Statistical analyses were performed with Review Manager version 5.0 (The Cochrane Collaboration, Software Update, Oxford). A value of P < 0.05 was considered statistically significant.
The funnel plot (Figure 1) for 5-year disease-free survival in the included studies demonstrated asymmetry, indicating an insignificant sign of publication bias.

Overall meta-analysis
The results from overall meta-analysis are summarized in Table 2.
Further pooled analysis of studies providing information found that the percentages of both intra-and extrahepatic recurrence were also similar in the two groups (51.2% vs.53.6%, P = 0.12; 12.9% vs. 10.3%, P = 0.19, respectively). No significant heterogeneity was detected between the groups in reporting these two outcomes (Figures 4, 5).

Sensitivity analysis
As Table 3 shows, the results of sensitivity analysis derived from three subgroups were all consistent with those derived from overall meta-analysis.

Discussion
The design of TACE is based on the principle that primary HCC is supplied almost exclusively (90%) by the hepatic arteries. The obstruction of the feeding  arteries can induce tumor ischemic necrosis. A combination of chemotherapy can drastically increase the local concentration of the chemotherapeutic agent and may improve the benefit of therapy. In 2003, a review of 7 RCTs showed that TACE significantly improved 2-year overall survival compared with nonactive treatment in patients with unresectable HCC [41]. In 2005, this therapy was recommended as standard intervention for unresectable patients with large/multifocal HCC who do not have vascular invasion or extrahepatic spread [7]. In contrast, the results of present pooled analysis of 21 trials do not support the use of preoperative TACE in the management of patients with resectable HCC.
Although TACE is effective for main tumors, intrahepatic metastases, tumor thrombi in the portal veins, and capsular invasion, which are considered risk factors contributing to HCC recurrence, are more unresponsive to TACE because of collateral and portal vein blood supply [8,13,16]. In addition, TACE mainly affects welldifferentiated HCC without completely killing poorly differentiated cells [42], which harbour a high grade of malignancy and ready spread within the portal venous system [43]. Furthermore, hematogeneous and lymphatic spread dissemination of cancer cells can precede TACE treatment. It is therefore reasonable to conclude that TACE is unable to reduce the risk of postoperative recurrence, or confer a survival advantage.
Adachi et al. [13] and Harada et al. [28] reported that the TACE subgroup with complete tumor necrosis had a better survival rate than the group without TACE. This is partly due to residual confounding, because tumors with complete necrosis are strongly associated with favorable tumor-related factors, such as smaller tumor size and tumor encapsulation or less portal involvement. On the other hand, several other reports failed to make the same conclusion [15,28,30,31]. It was found in the present study that approximately 20.1% of HCC tumors responded completely to TACE, suggesting that most of the HCC tumor cells were viable even when treated with TACE. The labeling index of proliferating cell nuclear antigen (a most widely used proliferation-associated marker) was significantly higher in the TACE group, indicating that residual HCC cells following preoperative TACE exhibit more aggressive behavior [44]. In support of this observation, Liou et al. [45] indicated that incomplete HCC necrosis after TACE (especially combined with necrotic area >50% main tumor size) was associated with the development of lung metastasis that has a strong adverse impact on patient survival. Adachi et al. [13] and Kim et al. [37] found that subjects with partial tumor necrosis had the lowest disease-free survival rate among the TACE subgroups and tended to have a lower survival rate than the group without TACE. Zhou et al. [16] noted that five patients lost the chance of potentially curative liver resection because of progression of disease with metastases  (n = 4) and liver failure (n = 1) during intervals between last TACE hepatic resection. The mechanism underlying accelerated tumor progression by TACE is unclear. Intratumoral necrosis was found to weaken the adhesive potential of the tumor and subsequently facilitate the release of cancer cells from the primary tumor and dislodgment into the bloodstream [46]. Xiao et al. [47] reported that mutated p53 could enhance the proliferation of HCC cells and suppress the apoptosis of HCC cells after TACE. In addition, vascular endothelial growth factor (VEGF), the most specific known angiogenic factor that plays a critical role in tumor growth, invasion, and metastasis, was up-regulated by tumor tissue ischemia and hypoxia after TACE [48].
In the present meta-analysis, in-hospital mortality did not differ significantly between TACE group patients and non-TACE group patients. However, Gerunda et al. [25] reported that three patients died of liver failure during 2-5 months after surgery in the TACE group. Sasaki et al. [34] and Uchida et al. [29] reported that late death due to liver failure was significantly higher in the TACE group than that in the non-TACE group. Uchida et al. [29] considered that hepatic function impairment induced by TACE could be repaired easily in the noncirrhotic liver, but hepatic function may gradually and progressively deteriorate due to TACE in some cirrhotic patients.
This study has several limitations. First, much of the evidence comes from NRCTs that could either exaggerate or underestimate the measured magnitude of effect size [20]. To minimize this effect, we limited the analysis to well-designed studies. As a matter of fact, the estimates from overall meta-analysis were consistent with those derived from RCTs, suggesting that the magnitude of the effect was not affected by the inclusion of NRCTs. Second, significant heterogeneity was present in some outcomes. Variability in the surgeon experience and the chemoembolization schedule may have introduced potential bias. In addition, clinicopathological factors associated with recurrence, such as hepatitis status, cirrhotic liver, and tumor staging, might be another source of potential heterogeneity. The use of random-effects models partially mitigates this concern. Third, although 21 studies involving more than 3,000 patients were enrolled for analysis, funnel plot analysis revealed the sign of publication bias. This may relate to the use of published English data only. Fourth, it is important to note that 18 of the 21 studies were from