Hepatic resection versus transarterial chemoembolization for the initial treatment of hepatocellular carcinoma: A systematic review and meta-analysis.

BACKGROUND & AIMS
According to the Barcelona Clinic Liver Cancer (BCLC) staging system, hepatic resection and transarterial chemoembolization (TACE) should be recommended in patients with hepatocellular carcinoma (HCC) within and beyond the BCLC stage A, respectively. We conducted a systematic review and meta-analysis to compare the overall survival between HCC patients undergoing hepatic resection and TACE.


METHODS
PubMed, EMBASE, and Cochrane library databases were searched. All relevant studies were considered, if they reported the survival data in HCC patients undergoing hepatic resection and TACE. Hazard ratios (HRs) with 95% confidence intervals (CIs) were calculated for the comparison of cumulative overall survival. Odds ratios (ORs) with 95%CIs were calculated for the comparison of 1-, 3-, and 5-year survival rates. Subgroup analyses were performed according to the BCLC stages and portal vein tumor thrombus (PVTT). Sensitivity analyses were performed in moderate- and high-quality studies and in studies published after 2005.


RESULTS
Fifty of 2029 retrieved papers were included. One, 15, and 34 studies were of high-, moderate-, and low-quality, respectively. The overall meta-analysis demonstrated a statistically significantly higher overall survival in hepatic resection group than in TACE group (HR=0.60, 95%CI=0.55-0.66). Additionally, 1-, 3-, and 5-year survival rates were statistically significantly higher in hepatic resection group than in TACE group (OR=1.82, 95%CI=1.56-2.14; OR=3.09, 95%CI=2.60-3.67; OR=3.48, 95%CI=2.83-4.27). The subgroup meta-analyses confirmed the statistical significance in HCC within the BCLC stage A (HR=0.72, 95%CI=0.64-0.80), in HCC beyond the BCLC stage A (HR=0.60, 95%CI=0.51-0.69), in HCC within the BCLC stage B alone (HR=0.48, 95%CI=0.25-0.90), and in HCC with PVTT (HR=0.78, 95%CI=0.68-0.91). The statistical significance was also confirmed by sensitivity analyses in moderate- and high-quality studies (HR=0.62, 95%CI=0.53-0.71) and in studies published after 2005 (HR=0.59, 95%CI=0.53-0.66).


CONCLUSIONS
Based on a systematic review and meta-analysis, hepatic resection may be considered in HCC beyond the BCLC stage A. However, given the limitations of study quality, more well-designed randomized controlled trials should be warranted to confirm these findings.


IntroductIon
Nowadays, Barcelona Clinic Liver Cancer (BCLC) stage is the sole system approved by the European Association for Study of the Liver (EASL) and American Association for the Study of Liver Disease (AASLD) guidelines for the prognostic classification and treatment selection of hepatocellular carcinoma (HCC) [1][2].According to this staging system, hepatic resection should be recommended in the BCLC stage 0 or A HCC with a single nodule (i.e., "the patients do not have liver cirrhosis or have liver cirrhosis but still have well preserved liver function, normal bilirubin and hepatic vein pressure gradient < 10 mmHg"), and transarterial chemoembolization (TACE) should be recommended in the BCLC stage B HCC (i.e., "the patients have large/multifocal HCC but without vascular invasion or extrahepatic spread").
Recently, based on the real-world data, several large scale studies suggested that hepatic resection might be also appropriate in HCC cases beyond the BCLC stage A. First, Farinati et al. analyzed the treatment selection and prognosis of 405 HCC cases in the BCLC stage B who were enrolled between 1986 and 2008 by the Italian Liver Cancer group [3].Only 40% of HCC cases in the BCLC stage B underwent TACE.However, TACE achieved a significantly shorter survival time than hepatic resection in such patients (median: 27 months versus 37 months).Second, Vitale et al. analyzed the outcomes of 2090 HCC cases in the different BCLC stages who were enrolled between 2000 and 2012 by the Italian Liver Cancer group [4].In the BCLC stages 0, A, and B, the net survival benefit of hepatic resection over non-surgical treatments was statistically significant.Third, Roayaie et al. analyzed the survival of 8,656 cases diagnosed with HCC between 2005 and 2011 in BRIDGE study [5].Hepatic resection not only achieved a significantly better survival than other treatments in ideal candidates for resection, but also achieved a significantly better survival than TACE in nonideal candidates for resection.
Herein, we conducted a systematic review and metaanalysis of available literatures to clarify the survival benefits of hepatic resection over TACE in HCC patients.

results systematic review
Overall, 2028 papers were initially retrieved via the three major databases, including 1219 papers in PubMed, 758 in EMBASE, and 51 in Cochrane library databases.One eligible paper was also manually identified.

Meta-analysis
Four of 50 papers were not included in the metaanalysis, because they reported only the survival times, but not the survival rates or Kaplan-Meier curves [17,25,32,37].The remaining 46 papers were included in the metaanalysis.
Additionally, the meta-analyses demonstrated that 1-, 3-, and 5-year survival rates were statistically significantly higher in hepatic resection group than in TACE group (Table 2).There were statistically significant heterogeneities in all of the 3 meta-analyses.

subgroup analysis in patients with different bclc stages
In HCC patients within the BCLC stages 0 and A, the subgroup meta-analysis demonstrated a statistically significantly higher overall survival in hepatic resection group than in TACE group (HR=0.72,95%CI=0.64-0.80,P<0.00001) (Figure 3).The heterogeneity was not statistically significant (P=0.92;I 2 =0%).Funnel plots demonstrated that all studies laid within 95%CI (Supplementary Figure 2).
Additionally, the meta-analyses demonstrated that 1-, 3-, and 5-year survival rates were statistically significantly higher in hepatic resection group than in TACE group (Table 2).There was a statistically significant heterogeneity in the meta-analysis of 5-year survival rate, but not in those of 1-and 3-year survival rates.
Additionally, the meta-analyses demonstrated that 1-, 3-, and 5-year survival rates were statistically significantly higher in hepatic resection group than in TACE group (Table 2).There were statistically significant heterogeneities in all of the three meta-analyses.

subgroup analysis in patients with bclc stage b alone
In HCC patients with BCLC stage B alone, the subgroup meta-analysis demonstrated a statistically significantly higher overall survival in hepatic resection group than in TACE group (HR=0.48,95%CI=0.25-0.90,P=0.02) (Figure 4).The heterogeneity was statistically significant (P<0.00001;I 2 =92%).Funnel plots demonstrated that not all studies laid within 95%CI (Supplementary Figure 4).
Additionally, the meta-analyses demonstrated that  1-year survival rate was statistically similar between the two groups, but 3-and 5-year survival rates were statistically significantly higher in hepatic resection group than in TACE group (Table 2).There were statistically significant heterogeneities in the meta-analyses of 1-and 3-year survival rates.The heterogeneity could not be evaluated in the meta-analysis of 5-year survival rate.
Additionally, the meta-analyses demonstrated that 1-and 3-year survival rates were statistically significantly higher in hepatic resection group than in TACE group, but 5-year survival rate was statistically similar between the two groups (Table 2).There was a statistically significant heterogeneity in the meta-analysis of 5-year survival rate, but not in those of 1-and 3-year survival rates.

sensitivity analyses in moderate-and high-quality studies
In 16 moderate-and high-quality studies, the sensitivity analysis demonstrated a statistically significantly higher overall survival in hepatic resection group than in TACE group (HR=0.62,95%CI=0.53-0.71,P<0.00001) (Figure 6).The heterogeneity was statistically significant (P<0.00001;I 2 =83%).Funnel plots demonstrated that not all studies laid within 95%CI (Supplementary Figure 6).
Additionally, the meta-analyses demonstrated that 1-, 3-, and 5-year survival rates were statistically significantly higher in hepatic resection group than in TACE group (Table 2).There were statistically significant  heterogeneities in all of the 3 meta-analyses.
Additionally, the meta-analyses demonstrated that 1-, 3-, and 5-year survival rates were statistically significantly higher in hepatic resection group than in TACE group (Table 2).There were statistically significant heterogeneities in all of the 3 meta-analyses.

dIscussIon
In theory, the BCLC staging system needs to be persistently updated with the dramatic improvement in the understanding of HCC and the invention of novel therapeutic modalities for HCC.Accumulated evidence suggests that the optimal treatment modality of HCC in the BCLC stage B may be further refined.In the present systematic review, we collected the comparative data regarding the overall survival in HCC patients undergoing hepatic resection and TACE.The overall meta-analysis demonstrated a statistically significant survival benefit of hepatic resection over TACE.In addition, considering the potential bias of patient selection, we performed several subgroup meta-analyses.All of them confirmed statistically significant survival benefits of hepatic resection over TACE.
At present, the curative treatment options of HCC mainly include liver transplantation (LT), hepatic resection, and radiofrequency ablation (RFA).Although LT is obviously superior to hepatic resection for the complete removal of tumor tissues, it is largely restricted by the scarcity of liver donors.On the other hand, hepatic resection appears to be superior to RFA for the improvement of overall survival and recurrencefree survival in HCC within the Milan criteria [56].In clinical practices, hepatic resection is often regarded as the primary choice of therapy for early stage HCC, if the lesion is single, hepatic function is well-preserved, and portal hypertension is not severe.Under this circumstance, the results of our subgroup meta-analysis that hepatic resection could achieve a significantly better survival than TACE in HCC within the BCLC stage A were in line with our expectations.
The non-curative treatment options of HCC mainly include TACE and sorafenib.TACE is the firstline treatment option of HCC in the BCLC stage B. This recommendation is primarily attributed to the survival benefits of TACE over conservative or suboptimal treatments [57].But we are not sure about whether TACE surpasses other active treatments for the improvement of overall survival.Our subgroup meta-analyses suggested that the survival was statistically significantly better in HCC patients beyond the BCLC stage A undergoing hepatic resection than in those undergoing TACE.Thus, hepatic resection might be also considered in selected HCC cases in the BCLC stage B.
After our study was registered, Kapitanov and colleagues published a similar meta-analysis to compare the short-and long-term results of hepatic resection versus TACE in HCC patients with cirrhosis [58].They also concluded that the survival at 1 and 3 years was significantly better in patients treated with surgery than in those treated with TACE.Compared with their study, our systematic review and meta-analysis had several strengths.First, the selection of target population was broader and the number of included studies was larger.Thus, we were permitted to conduct more subgroup meta-analyses according to the study and patient characteristics.Second, the study quality was strictly assessed.Thus, we could readily understand the grade of current evidence.Third, the duplicate data were excluded, thereby avoiding the inflation of relevant information [59].In the meta-analysis by Kapitanov et al., two papers by Zhong et al. were included.However, they reported the overlapping data.In the first paper, 392 HCC patients in the BCLC stage B were enrolled between January 2000 and November 2007 [60]; and in the second paper, 860 HCC patients in the BCLC stages B and C were enrolled between January 2000 and November 2007 [55].By comparison, the first paper with a smaller sample size was excluded from our meta-analysis.Indeed, four other papers conducted by the same study team were also excluded from our metaanalysis [61][62][63][64].
The limitations of our meta-analysis should be clearly emphasized.First, only one included study was a randomized controlled trial.Additionally, a majority of included studies were of low quality.Certainly, we conducted a subgroup meta-analysis of moderate-and high-quality studies to confirm the reliability of our findings.Second, the heterogeneity among studies was statistically significant in all but one meta-analysis of HCC cases within the BCLC stage A. We employed a random-effect model to produce a relatively conservative estimate.Third, the publication bias was statistically significant in a majority of meta-analyses, despite three major English-language databases were searched.Fourth, the overall survival was the sole outcome observed in our study.Thus, we could not capture the other advantages or disadvantages of hepatic resection versus TACE.However, it should be noted that the overall survival was the most important endpoint to measure the therapeutic effectiveness in HCC [65].By contrast, the time to recurrence, progression-free survival, and disease-free survival were the secondary endpoints.We could hardly translate the improvements in these secondary endpoints into the clinical practice recommendations.
In conclusions, hepatic resection might provide a better overall survival than TACE in HCC beyond the BCLC stage A. However, we should acknowledge that the current evidence is of low-quality.Considering that the drawbacks of study designs potentially lead to the selection biases, more well-designed randomized controlled trials should be warranted to compare the survival benefit of hepatic resection versus TACE in such patients.

MeterIAls And MetHods
This work was registered with PROSPERO on December 19, 2014 (registration number: CRD42014015618).

study selection
Only clinical studies including more than 10 patients were considered in the systematic review.Accordingly, duplicate papers among databases, redundant publications, narrative or systematic reviews, study protocols, comments, experimental studies, and case reports were excluded.If two or more papers by the same study team had the overlapping data, only one paper with more adequate data and/or a longer enrollment period would be included.
The inclusion criteria should be as follows.
Participants: patients with HCC.
Interventions: hepatic resection and TACE as initial treatment modalities.
The exclusion criteria should be as follows.
3) Mixed malignancies.11) No detailed data regarding the number of observed patients in the hepatic resection or TACE group.
Notably, the major reason for exclusion of studies including patients with recurrent HCC and spontaneous rupture of HCC was the discrepancy in the treatment selection and outcomes among them.

data extraction
The following data were extracted: the first author, publication year, publication form, region, enrollment period, study design, study population, follow-up time, inclusion and exclusion criteria, number of HCC cases, treatment selection, survival rate, survival times, and Kaplan-Meier curve analysis with log-rank test.If the propensity score matching analysis was performed, we just collected the survival data after the propensity score matching analyses.If both survival rates and Kaplan-Meier curves were presented, only the survival rates would be collected.If only Kaplan-Meier curves were presented, we extracted the cumulative 1-, 3-, and 5-year survival rates by using the Distance Tool in the Measurements menu of Foxit PDF Reader software version 5.4.4.1023 (Foxit Cooperation, California, USA).This software was freely downloaded.

study quality
Because both retrospective/prospective observational studies and randomized controlled trials were included in the present systematic review, we could not employ a single scale to evaluate the quality of all included studies.More importantly, because our study was designed to compare the overall survival between patients undergoing hepatic resection and those undergoing TACE, the study quality assessment should primarily focus on the comparability of patient characteristics between the two groups.According to the Newcastle-Ottawa scale and major prognostic factors of HCC [66], we developed the following 9 questions that were more specific to the study quality assessment in the present systematic review.
1) Were the patients consecutively enrolled and prospectively followed?
2) Was the age statistically similar between the two groups?
3) Was the gender statistically similar between the two groups?4) Was the Child-Pugh score/class or MELD score statistically similar between the two groups?5) Were the diameter and number of tumor statistically similar between the two groups?6) Was the BCLC stage or other HCC stage statistically similar between the two groups?7) Were the criteria for treatment selection homogeneous between the two groups?8) Was the follow-up time clearly reported?9) Was the proportion of patients lost to follow-up less than 20%?
If the answers to 7-9 questions were "Yes", the study would be considered to be of high quality.If the answers to 4-6 questions were "Yes", the study would be considered to be of moderate quality.Otherwise, it would be considered to be of poor quality.

Meta analysis
Only a minority of included studies clearly reported the hazard ratios (HRs) for the overall survival in HCC patients with hepatic resection versus TACE.Therefore, we calculated ln [HR] with standard error by using the calculation sheets which were developed by Matthew Sydes and Jayne Tierney [67].The survival rates at different time points were entered into the calculation sheet of "(2a) curve data".Accordingly, a curve was produced in the calculation sheet of "(2b) curve copy", and ln[HR] and se(ln[HR]) could be available in the calculation sheet of "(4) output information".Then, HRs with 95% confidence intervals (CIs) were pooled by using a random-effects model.Additionally, to provide the survival data in detail, we compared the 1-, 3-, and 5-year survival rates between HCC patients with hepatic resection versus TACE.The odd ratios (ORs) with 95% CIs were pooled by using a random random-effects model.In these meta-analyses, a P value of <0.05 was considered statistically significant.Heterogeneity between studies was assessed by using the I 2 statistic (I 2 > 50% was considered as having substantial heterogeneity) and the Chi-square test (P<0.10 was considered to represent significant statistical heterogeneity).Funnel plots were performed to evaluate the publication bias.Subgroup meta-analyses were performed according to the BCLC stages (within versus beyond BCLC stage A).Subgroup difference between the two groups was evaluated by using the I 2 statistic (I 2 > 50% was considered as having statistically significant difference) and the Chi-square test (P<0.10 was considered to represent statistically significant difference).Subgroup meta-analyses were also performed in patients with BCLC stage B alone and in those with portal vein tumor thrombus (PVTT).Sensitivity analyses were performed in moderate-and high-quality studies and studies published after 2005.All metaanalyses were conducted by using the statistical package Review Manager version 5.1.6(Copenhagen, The Nordic Cochrane Center, The Cochrane Collaboration, 2011).

Figure 2 :
Figure 2: the overall meta-analysis comparing the overall survival between Hcc patients undergoing hepatic resection and tAce.

Figure 3 :
Figure 3: the subgroup meta-analysis comparing the overall survival between Hcc patients within and beyond the bclc stage A undergoing hepatic resection and tAce.

Figure 4 :
Figure 4: the subgroup meta-analysis comparing the overall survival between Hcc patients with bclc stage b alone undergoing hepatic resection and tAce.

Figure 5 :
Figure 5: the subgroup meta-analysis comparing the overall survival between Hcc patients with PVtt undergoing hepatic resection and tAce.

Figure 6 :
Figure 6: sensitivity analysis in moderate-and high-quality studies.

Figure 7 :
Figure 7: sensitivity analysis in studies published after 2005.