Systematic review of neoadjuvant transarterial chemoembolization for resectable hepatocellular carcinoma


David L. Morris, Department of Surgery, University of New South Wales, St George Hospital, Kogarah, Sydney, NSW 2217, Australia
Tel: +(02) 9113 2070
Fax: +(02) 9113 3997


Resection of hepatocellular carcinoma (HCC) offers the only hope for cure. However, in patients undergoing resection, recurrences, in particular, intrahepatic recurrence are common. The effectiveness of transarterial chemoembolization (TACE) as a neoadjuvant therapy for unresectable HCC was exploited by numerous liver units and employed preoperatively in the setting of resectable HCC with an aim to prevent recurrence and prolong survival. A systematic literature search of databases (Medline and PubMed) to identify published studies of TACE administered preoperatively as a neoadjuvant treatment for resectable HCC was undertaken. A systematic review by tabulation of the results was performed with disease-free survival (DFS) as the primary endpoint. Overall survival (OS), rate of pathological response, impact on surgical morbidity and mortality and pattern of recurrences were secondary endpoints of this review. Eighteen studies; three randomized trials and 15 observational studies were evaluated. This comprised of 3927 patients, of which, 1293 underwent neoadjuvant TACE. The median DFS in the TACE and non-TACE group ranged from 10 to 46 and 8 to 52 months, respectively, with 67% of studies reporting similar DFS between groups despite higher extent of tumour necrosis from the resected specimens indicating a higher rate of pathological response (partial TACE 27–72% vs. non-TACE 23–52%; complete TACE 0–28% vs. non-TACE zero), with no difference in surgical morbidity and mortality outcome. No conclusion could be drawn with respect to OS. Both randomized and non-randomized trials suggest the use of TACE preoperatively as a neoadjuvant treatment in resectable HCC is a safe and efficacious procedure with high rates of pathological responses. However, it does not appear to improve DFS.

Hepatocellular carcinoma (HCC) is a fatal malignancy of the liver that is estimated to be diagnosed in about 21 370 or 6.4 per 100 000 individuals annually in the USA. Of which, the death rate is about 5.0 per 100 000 per year and the 5-year survival rate is 12% (1). Majority of HCC develops in individuals with cirrhosis secondary to viral hepatitis (hepatitis B or C) or alcohol abuse. Treatment options of HCC are largely dependent on the tumour size and stage, uni or multifocality, and the underlying liver function (2). The only curative option is surgical resection or transplantation. Owing to the complexities, availability and patients' suitability of transplantation as a curative procedure, liver resection is the preferred option and remains the mainstay of treatment. Nonetheless, the long-term survival remains poor with high incidence of recurrences and metastasis after hepatectomy. Intrahepatic recurrences are the most common and are seen in up to 68–96% of patients. There is presently no effective neoadjuvant therapy or adjuvant therapy to complement the surgical and ablative treatments to improve outcomes for HCC (3). To date, the only adjuvant therapy shown to decrease the rate of recurrence and increase the disease-free and overall survival (OS) is intra-arterial 131I-lipiodol administered after a curative hepatectomy (4).

Transarterial chemoembolization (TACE) is a regional therapy performed under radiological guidance by infusion of chemotherapeutic agents, lipiodol and gel foam particles by an interventional radiologist. TACE exploits the tumour vasculature of hepatomas which are preferentially supplied by the hepatic artery. The portal vein is responsible for the perfusion of the normal hepatic parenchyma. By direct infusion of the lipiodol and chemotherapy through the hepatic artery, it allows a high dose of chemotherapy to be delivered directly to the tumour. This is followed by embolization using the gel foam particles. Together, this procedure is aimed at inducing tumour necrosis to prevent tumour cell dissemination during surgical resection.

Transarterial chemoembolization was recently established as a palliative therapy for unresectable HCC. The administration of TACE preoperatively in patients with unresectable HCC has been shown to results in tumour down-sizing to fit the criteria for either resection or transplantation (5). Owing to the recurring nature of HCC and the lack of effective adjuvant systemic therapies, it was hypothesized that neoadjuvant use of TACE in the setting of resectable HCC may improve survival by reducing recurrences. There is now a body of literature, comprising of observational studies and randomized trials, all of which show conflicting outcomes of TACE as a neoadjuvant therapy. This systematic review aims to provide an extensive review of all published studies of neoadjuvant TACE in the setting of resectable HCC as a prelude towards establishing an evidence-based practice of this procedure in the management of resectable HCC.


Objective of literature search

The primary endpoint was to determine if TACE as a preoperative/neoadjuvant treatment of resectable HCC is associated with improved disease-free survival (DFS). The secondary endpoints were to determine the rates of OS, surgical morbidity and mortality, pathological response and pattern of recurrence.

Literature search strategy

Original published studies on the use of TACE as a neoadjuvant procedure in hepatectomy for HCC were identified by searching the Medline database (1966 to February 2009) and PubMed (January 1980 to February 2009) using the key words; ‘Transarterial chemoembolization’, ‘Therapeutic chemoembolization’, ‘Hepatocellular carcinoma’, ‘Hepatectomy’, ‘Liver resection’, ‘Preoperative’ and ‘Neoadjuvant’. The reference lists of all retrieved articles were manually reviewed to further identify potentially relevant studies. All relevant articles identified were assessed with application of a predetermined selection criterion.

Selection criteria

Publications where the use of TACE was preoperative as a neoadjuvant treatment before liver resection for resectable HCC were identified for inclusion. Resectability between studies was not controlled; based on extent of intrahepatic disease, but was determined as per the reported data from each study to accommodate for surgical expertise between treatment centres. TACE involved insertion of a vascular catheter into the hepatic artery where an angiography was performed with or without lipiodol. The tumour feeding vessel is identified whenever technically possible before the vascular catheter is advanced. The cytotoxic agent is then infused into the feeding artery selectively. To infarct the tumour, embolization is performed with a gelatine sponge. Inclusion criteria were studies published after year 1990, with 40 patients, human articles and papers published in the English language. Abstracts, letters, editorials and expert opinions were excluded. Studies reporting the effectiveness of ablative techniques, be it cryotherapy or radiofrequency for HCC were also excluded. It is acknowledged that all of the studies included in the present review aimed to demonstrate the use of TACE in the previously described setting.

Data extraction and critical appraisal

Two reviewers (T. C. C. and A. S.), with knowledge in biostatistics, independently appraised each article using a standard protocol. Data extracted include the methodology, quality criteria, setting of use of TACE, results of TACE on the lesion, survival and morbidity and mortality outcomes. All data were extracted and tabulated from the relevant articles texts, tables and figures. Discrepancies were resolved by discussion and consensus. Major hepatectomy was defined as resection of four or more Couinaud's segments. Response to TACE was evaluated from the studies based on the extent of tumour necrosis following pathological assessment of the resected specimen. The RECIST criteria was used to categorize the response; complete response (CR) refers to total disappearance of all lesions, partial response (PR) refers to a 30% decrease in sum of the longest diameter of target lesions. The heterogeneity in this study group reported precluded a full meta-analysis. Systematic review was performed by full tabulation of the results. Studies were selected for evaluation if they were level I evidence: randomized controlled trials (RCTs); level II evidence: non-randomized controlled clinical trials or well-designed cohort studies; level III evidence: observational studies, as described by the US Preventive Services Task Force.


Quantity and quality of evidence

Literature search using the above-described search strategy through both Medline and PubMed databases identified 126 studies. After applying the general selection criteria, 67 studies comprising of case reports, review articles, no chemoembolization and involving liver transplantation were excluded. The remaining 30 studies were individually reviewed through careful analysis of the abstract and study methods with reference to fulfilment of the systematic review's primary and secondary endpoints. As a result, a further 12 studies were excluded; six studies did not fulfil primary or secondary endpoints and six studies did not have a comparison group or inappropriate comparison (palliative treatments) (Figs 1 and 2). In total, 18 studies (6–23) which compared the addition of TACE as a neoadjuvant treatment before hepatectomy for resectable HCC were reviewed. The level of evidence of 15 studies were low (mainly level II/III). They comprised mainly of observational studies, non-randomized case–control analytic studies. There were three randomized trials (20, 21, 23).

Figure 1.

 Flowchart of publication identification and quality appraisal. HCC, hepatocellular carcinoma; TACE, transarterial chemoembolization.

Figure 2.

 Graphical depiction of 18 studies reporting on the % of studies based on (no effect, improve, worsen, not reported) for the primary endpoint (disease-free survival) and secondary endpoints of this systematic review with the summative conclusion for each endpoint.

As two of the randomized trials (20, 21) were published before the release of the CONSORT statement in 2001 (24, 25), results of the estimated effect size and precision (i.e. 95% confidence interval) were not available. Hence, meta-analysis was not appropriate. The results were extracted and clinical effectiveness was synthesized through a narrative review with full tabulation of results of all included studies (Table 1). Sixteen of 18 studies fulfilled the primary endpoint of reporting results on DFS (6–12, 14–18, 20–23). The two studies that did not fulfil the primary endpoint were included as secondary endpoints of OS, surgical morbidity and mortality and rates of pathological response were reported (13, 19).

Table 1.   Characteristics of the studies and patients from the various publications comparing neoadjuvant transarterial chemoembolization and non-transarterial chemoembolization before hepatectomy for resectable hepatocellular carcinoma
YearLevel of
patients (n)
(TACE/non-TACE) (n)
% of Child's
Grade A (TACE/non-TACE)
% of major
Mean preoperative
tumour size (cm)
  1. Results of comparison that are significantly different are marked in bold.

  2. HCC, hepatocellular carcinoma; TACE, transarterial chemoembolization; NR, not reported.

Zhou (23)2009I10852/5684/89100/1009.0/9.5
Yamasaki (21)1996I9750/47NR28/343.1/3.3
Wu (20)1995I5224/2892/8646/5014.3/14.5
Gerunda (9)2000II4020/2040/5520/03.7/4.1
Lu (12)1999II12044/7670/20NR7.2/6.1
Paye (16)1998II4824/24NR50/427.8/7.3
Kim (11)2008III33497/237NR56/625.0/4.6
Chen (7)2007III24689/15788/90100/1009.5/9.9
Choi (8)2007III273120/15398/9855/384.7/4.9
Sasaki (17)2006III235109/12670/4834/573.5/3.9
Ochiai (15)2003III148100/48NR30/294.4/4.4
Sugo (18)2003III227146/8181/91NRNR
Luo (13)2002III12662/64NR39/44NR
Zhang (22)2000III1457120/1337NRNRNR
Majno (14)1997III7649/2788/6321/05.1/4.0
Harada (10)1996III140105/35NR26/344.2/3.8
Uchida (19)1996III12860/6862/7313/193.7/NR
Adachi (6)1993III7246/26NR24/272.6/2.1

Patient characteristics

In total, this review comprised of 3927 patients, of which, 1293 underwent neoadjuvant TACE. Majority of patients reviewed had mild cirrhosis with the percentage of patients with Grade A Child–Pugh reported in the TACE group ranging from 62 to 98% and in the non-TACE group ranging between 20 and 98% in 11 of 18 studies. The extent of major hepatectomy in the TACE group ranged between 20 and 100% and in the non-TACE group between 0 and 100%, with two studies reporting significantly lower rates of major hepatectomy (17, 20) and two studies reporting a higher rate of major hepatectomy in the TACE group (8, 14) as reported in 15 of 18 studies. The mean size of tumour preoperatively before any treatment ranged from 2.6 to 14.3 cm in the TACE and from 2.1 to 14.5 cm in the non-TACE group as reported in 14 of 18 studies. Two studies reported larger tumours in the TACE group (mean tumour size TACE vs. non-TACE; 5.0 vs. 4.1 cm, 2.6 vs. 2.1 cm) (6, 14). Cytotoxic agents used for TACE comprised of monotherapy or combinations of cisplatin, fluorouracil, doxorubicin, mitomycin C and epirubicin. The number of TACE procedure performed in the studies ranged from one to three with time interval to hepatectomy ranging between 26 and 126 days. These patient and TACE characteristics are described in Tables 1 and 2.

Table 2.   Transarterial chemoembolization protocol, clinical data and response to treatment
Type of
cytotoxic agent used
number of TACE
performed per
patient (n)
Average time
delay between
TACE and
hepatectomy (days)
Tumour necrosis as an indication of response to TACE (TACE/non-TACE)Differences in morbidity and mortality (TACE/non-TACE)
  1. CDDP, cisplatin; DOX, doxorubicin; EPI, epirubicin; 5FU – f-fluorouracil; MC, Mitomycin C; NR, not reported; TACE, transarterial chemoembolization.

  2. Result of comparison that are significantly different are marked in bold.

Zhou (23)5FU, CDDP1.54672/5215/0More perihepatic adhesions and longer operation time in TACE group0/0
Yamasaki (21)DOX1NR27/NR16/NRNo6/9
    Mean % necrosis 75/7   
Wu (20)DOX312615% reduction in tumor size in 67% of patients, no change 17%, increased 17% No4/7
Gerunda (9)EPI148NRNRNo0/0
Lu (12)CDDP, DOX22971/230/0NR 
Paye (16)DOXNR4567/424/0More perihepatic adhesions and longer operation time in TACE group13/8
Chen (7)5FU, MC, CDDP22666/521/0No3/1
Choi (8)DOX24970/NR28/0No3/1
Sasaki (17)5FU, MC, DOX133NRNRNR 
Ochiai (15)MC, DOX136NR14/NRNR 
Sugo (18)EPINRNRNR16/0No6/1
Luo (13)CDDP, EPI1.772NRNRMore perihepatic adhesions, porta hepatis occlusion, haemorrhage and abdominal drainage in TACE group0/0
Zhang (22)5FU, MC, DOX2NRNR38/0.4NR 
Majno (14)DOX, CDDP2NRNR50/4No4/7
Harada (10)MC, DOX17344/NR29/NRNo5/1
    25% reduction in tumor size in 50% of patients   
Uchida (19)MC, DOX, EPI1NR17/0 No8/6
Adachi (6)MC, DOX13341/NR22/NR NR

Impact of neoadjuvant transarterial chemoembolization on disease-free survival

Following the use of TACE as a neoadjuvant treatment before hepatectomy, four of 18 studies (22%) reported an improvement in DFS (7, 9, 16, 22), 12 of 18 studies (67%) reported no difference in DFS (6, 8, 10–12, 14, 15, 17, 18, 20, 21, 23) and results were not available in two studies (11%) (13, 19). Six of 12 studies (50%) in addition to reporting no difference in DFS further demonstrated a poorer median DFS (in months) in the TACE group, however, this difference when statistically compared were not significant (6, 8, 11, 12, 20, 23). The median DFS in the TACE and non-TACE group ranged from 10 to 46 months and 8 to 52 months respectively. In the four studies that reported an improvement in DFS, three studies reported a median DFS of 14, 20 and 36 months in the TACE group compared 8, 16 and 12 months in the non-TACE group respectively (7, 16, 22). One study reported a 5-year DFS of 57% with 21% in the non-TACE group (Table 3) (7, 9, 16, 22). Of the four studies that report an improved DFS, only one study further confirmed a improvement in OS (7). Overall, the summation of evidence suggests that the neoadjuvant use of TACE in the setting of resectable HCC does not improve DFS (Table 3; Fig. 2).

Table 3.   Survival outcomes and patterns of recurrence between transarterial chemoembolization and non-transarterial chemoembolization groups after hepatectomy
First authorMedian disease-free
survival of (TACE/non-TACE) (months)
Median overall survival (TACE/non-TACE)
5-year survival
Patterns of intrahepatic
recurrence (TACE/non-TACE)
Patterns of extrahepatic
recurrence (TACE/non-TACE)
Solitary (%)Multiple (%)Lung (%)Others (%)
  • Results of comparison that are significantly different are marked in bold.

  • *

    Reported as 5-year disease-free survival.

  • NR, not reported; TACE, transarterial chemoembolization.

Zhou (23)10/12NR30/2135/50  33/27
Yamasaki (21)39/31*NR63/62NR  NR
Wu (20)16/2719/110NR69/42 43/1913/81
Gerunda (9)57/21*NR43/38NR  NR
Lu (12)19/2236/38NRNR  NR
Paye (16)14/817/14NR71/86 7/07/14
Kim (11)20/34NR44/62NR NR 
Chen (7)20/1658/4046/3245/2255/7867/4633/54
Choi (8)46/52NRNR77/7523/2520/20 
Sasaki (17)15/1534/5029/49NR NR 
Ochiai (15)25/2558/45NR39/4846/4515/6 
Sugo (18)30/26NRNR47/1753/8310/29 
Zhang (22)36/12NRNRNR NR 
Majno (14)18/11*NRNR51/74 7/04/9
Harada (10)24/15NR53/46NR NR 
Uchida (19)NRNR24/63NR NR 
Adachi (6)34/38NRNRNR NR 

Impact of neoadjuvant transarterial chemoembolization on overall survival

In total, 12 of 18 studies reported long-term outcomes in terms (either median OS or 5-year survival). Four studies of 18 studies (22%) reported a poorer long-term outcome when the TACE and non-TACE group were compared in terms of median OS (19 vs. 110 months, 34 vs. 50 months) and 5-year survival (44 vs. 62%, 29 vs. 49%, 24 vs. 63%) (11, 17, 19, 20). One study reported a longer median OS (58 vs. 40 months) and 5-year survival (48 vs. 32%) in the TACE compared with the non-TACE group (7). Seven studies of 18 studies (39%) reported no difference in long-term outcomes in the TACE and non-TACE groups (9, 10, 12, 15, 16, 21, 23). Six studies (33%) did not report on long-term outcomes (6, 8, 13, 14, 18, 22). Based on the heterogeneous results of the studies reviewed, the efficacy of neoadjuvant TACE in resectable HCC is inconclusive (Table 3; Fig. 2).

Impact of neoadjuvant transarterial chemoembolization on surgical morbidity and mortality

Morbidity outcomes were available in 12 studies (67%) (7–10, 13, 14, 16, 18–21, 23). Nine studies of 18 studies (50%) report no significant difference in postoperative complications between the TACE and non-TACE groups (7–10, 14, 18–21). Three of 18 studies (17%) report increase rates of postoperative complications (13, 16, 23). Common complications among the three studies were an increased in operative time owing to perihepatic adhesions in the TACE group (13, 16, 23). Luo et al. (13) specifically assessed the influence of TACE on perioperative outcomes and additionally reported higher volume of intraoperative haemorrhage (TACE 829±949 ml vs. and non-TACE 560±616 ml; P<0.05) and that increased the difficulty of operating in the TACE group owing to chemical hepatitis of the hepatic parenchyma, oedema, liable haemorrhage and adhesions. This further translated into increased abdominal drain output during the first three postoperative days (TACE 159±149 ml vs. and non-TACE 110±86 ml; P<0.05). Comparison of mortality outcomes between TACE and non-TACE groups were reported in 12 of 18 studies (67%), all 12 studies demonstrate similar rates of mortality between groups with the mortality rate ranging between 0 and 13% (7–10, 13, 14, 16, 18–21, 23). Overall, majority of studies did not demonstrate any increase in morbidity (nine studies) and mortality (12 studies) in the TACE group, therefore, suggesting that preoperative use of TACE as a neoadjuvant treatment in resectable HCC is safe (Table 2; Fig. 2).

Impact of neoadjuvant transarterial chemoembolization on rates of tumour necrosis

The response to TACE could be evaluated from the extent of tumour necrosis at pathological assessment with rates of necrosis approximated according to the RECIST criteria. Fourteen of 18 studies (78%) reporting the extent of tumour necrosis in the resected specimens that allowed evaluation of the rates of necrosis (6–8, 10, 12, 14–16, 18–23). In the TACE group, the rates of partial necrosis ranged between 27 and72% and the rates of complete necrosis ranged between 0 and 28%. In the non-TACE groups, the rates of partial necrosis ranged between 23 and 52% and the rates of complete necrosis were zero. Six of 14 studies (43%) reported significantly increased rates of partial necrosis (7, 12, 16, 19, 21, 23) and nine of 14 studies (64%) reported significantly increased rates of complete necrosis (7, 8, 14, 16, 18, 19, 21–23). Overall, TACE results in higher rates of partial and complete necrosis as evaluated by the extent of necrosis evaluated pathologically compared with no preoperative therapy from the resected specimens (Table 2; Fig. 2).

Impact of neoadjuvant transarterial chemoembolization on patterns of recurrence

The rates of intrahepatic recurrence in the TACE group ranged from 35 to 77% and in the non-TACE group from 17 to 86% as reported in eight of 18 studies (7, 8, 14–16, 18, 20, 23). Four studies that investigated the rates of solitary and multiple recurrences reported similar rates (7, 8, 15, 18). One study reported an overall significantly lower rate of intrahepatic recurrence in the TACE group compared with the non-TACE group (51 vs. 70%; P<0.05) (14).

The rates of extrahepatic recurrence in the TACE group ranged from 7 to 67% and in the non-TACE group from 9 to 81% as reported in eight of 18 studies (7, 8, 14–16, 18, 20, 23). Four studies that investigated the rates of pulmonary and extrapulmonary recurrence reported similar rates (7, 14, 16, 20) (Table 3).


The use of TACE as a neoadjuvant treatment for HCC was first described in the early 1990s, where its use was proposed in a variety of settings; palliatively for unresectable recurrent HCC, to improve the resectability of unresectable HCC, to downstage the primary tumour for transplantation or to delay the time to surgery (26, 27). Seven randomized clinical trials in the late 1990s investigated the role of TACE for unresectable HCC. The results of these trials were summarized in a meta-analysis which demonstrated an improved 2-year survival (odds ratio 0.53, P=0.017) compared with controls who were treated conservatively or with suboptimal management (5). This established the role of TACE as the standard of care for unresectable HCC, where it may palliate patients or facilitate resectability (5). Subsequently, investigations became directed towards the rationale of using TACE preoperatively as a neoadjuvant treatment to prevent recurrence. Numerous studies have been published and its clinical efficacy is not proven owing to the numerous conflicting reports. This systematic review effectively summarizes all reported studies of TACE preoperatively as a neoadjuvant treatment for resectable HCC to evaluate its clinical efficacy and safety.

The results of our systematic review show that there is a long history of the use of administering TACE as a neoadjuvant therapy before hepatectomy for HCC. Most of the reported studies are from Asian institutions. Apart from an increased risk of the presence of perihepatic adhesions that increase operative time, TACE is a safe procedure with no adverse effect on the perioperative outcome. In terms of clinical efficacy, two randomized trials were first conducted in year 1995 and 1996 (20, 21). Results from both these trials showed no difference in DFS following neoadjuvant TACE, and in fact, one trial reported a poorer OS in the TACE group (20). The impact of these trials on clinical practice was poor. More recently in 2009, a randomized trial comparing neoadjuvant TACE for large resectable HCC was reported (23). The results were again similar, with no difference in DFS and OS between groups that received or did not receive TACE (23). All the reviewed studies have endeavoured to employ TACE preoperatively as a loco-regional therapy to impact the course of the disease by reducing tumour size, induce tumour necrosis and prevent tumour cell dissemination during surgery. Although these theoretical rational appear to be valid, it appears convincing that there is no effect of this practice on survival. In fact, if the tumour is already resectable, administering TACE with an intention of shrinking the tumour to facilitate an easier operation may actually complicate the operation further during the process of liver mobilization owing to the presence of perihepatic adhesions following the TACE procedure (13, 16, 23). Although not investigated in this review, the administration of TACE may induce liver function impairment and increase the risk of liver failure. Further, the period taken to administer TACE, 29–126 days as shown in our review depending on the number of TACE procedure administered, results in a delay in undergoing surgery. There is also constantly no evidence of TACE making surgery easier in this review. During this time, there may be a propensity for tumour growth and dissemination. In the trial of Zhou et al. (23), the average period taken for the administration of TACE was 46 days. As a result of this delay, five patients lost the chance of undergoing a curative liver resection owing to disease-progression, extrahepatic metastasis and hepatic failure.

Amongst the non-randomized trials, the largest retrospective study was by Zhang et al. (22), who reported a series of 1457 patients undergoing hepatectomy for HCC, of which 120 received two procedures of TACE as neoadjuvant therapy. In the survival analysis, with DFS as their endpoint, the authors report a 5-year DFS of 51% in the group having two TACE procedure, 35.5% in the group having one TACE procedure, and 21.4% in the group which did not receive TACE (P<0.05). The multivariate analysis identified factors such as the number of TACE procedure, the response of TACE, number of lesion, intraoperative tumour thrombus, tumour size, tumour type, daughter nodules, vascular invasion and postoperative áFP levels as prognostic factors on survival. This study by Chen and colleagues of 246 patients, of which 89 underwent two TACE procedure as a neoadjuvant therapy before mesohepatectomy for large central HCC showed a survival benefit in both DFS and OS. Paye et al. (16) who reported a series of 48 patients, of which half underwent TACE and half did not. The mean tumour size in the TACE group was 7.8 and 7.1 cm in the non-TACE group. After TACE, there was a 71% overall response rate, with the mean tumour size in the TACE group being reduced to 7.1 cm. The survival analysis showed a significant benefit in DFS in the TACE group. In another large study of 227 patients by Sugo et al. (18), of which 146 underwent preoperative TACE as neoadjuvant therapy, no significant difference on DFS was demonstrated. However, when stratified according to tumour stages (I–IV) based on the Liver Cancer Study Group of Japan, when patients with tumour stages III and IV were studied, the TACE group had a significantly longer DFS and OS compared with the non-TACE group. From these four studies, it appeared that the efficacy of TACE may be observed in patients with larger tumours only for which a DFS benefit may be derived following its use. However, the randomized trial by Zhou et al. (23), where patients with large tumours of 9 cm were investigated in their trial, for which no disease-free and OS difference were seen between groups who received and did not receive TACE. However, there was a PR rate of 72% and CR rate of 15% in the TACE group. Clearly, it appears that the tumour size may impact the response rates that would then translate to affect the survival results when studying the therapeutic efficacy of TACE.

The derivation of the primary and secondary endpoints was limited by the differences in the patients' baseline characteristics. There were different rates of severity of chronic liver disease between studies. Further, the size of tumours ranged from 2.1 to 14.5 cm, hence indicating that the extent of the disease was vastly different. This explained the rates of major hepatectomy which ranged from 0 to 100% in all studies. More importantly, the difference in tumour sizes may limit a homogenous therapeutic efficacy of TACE among the studies as there are uncertainties over the benefit of TACE on very small tumours. The selection and number of cytotoxic agents used for the TACE ranged from one to three agents with up to three TACE procedures performed in some studies. Together, the patients treated in each study were different and there were varying protocols employed at each institution.

A recent review of randomized trials from the Cochrane Hepato-Biliary Group Controlled Trials Register on neoadjuvant and adjuvant therapies with curative surgery or just surgery alone (28). Twelve trials comprising of systemic, locoregional, chemotherapy and immunotherapy were reviewed; four of 12 trials reported OS benefits and two trials reported DFS benefits at 5 years when adjuvant or neoadjuvant therapy was administered. Despite this, the authors' conclusion from undergoing this extensive review was that there were no clear evidence for efficacy of any of the adjuvant and neoadjuvant protocols reviewed (28). It must be noted that in this Cochrane review, only the two older randomized trials by Wu et al. (20) and Yamasaki et al. (21). Our review includes both randomized trials, including the most recent trial by Zhou et al. (23) and non-randomized trials; including retrospective analysis, to provide a comprehensive analysis into the specific role of TACE as a neoadjuvant therapy for which the major conclusions of this paper are based on.

The most recent consensus paper from the Japan Society of Hepatology proposed an algorithm for treatment of HCC where in the setting of hepatic only disease, with appropriate liver function (Child–Pugh A/B); tumours with vascular invasion should undergo resection with adjuvant transarterial infusional 5FU and cisplatin or TACE. In the absence of vascular invasion, a single hypovasular lesion (early HCC) may undergo strict observation or ablation. Where there are one to three tumours, resection with or without ablation may be performed when tumour sizes were ≤3 cm. When tumour sizes were >3 cm, hepatectomy should be combined with adjuvant TACE. In the presence of four or more tumours, TACE or transarterial infusional chemotherapy or a combined approach of resection and ablation may be employed. If the disease fulfilled the Milan criteria and age ≤65, transplantation may be considered (29).

In conclusion, this systematic review presents the role of TACE in the setting of a resectable HCC. Current evidence indicates that there appears to be no DFS advantage despite its safety and feasibility. A well-designed prospective multi-institutional RCT, with a clearly defined protocol for concealed allocation, eligibility criteria, TACE intervention regimen and endpoints will be potentially meaningful.