Between January, 2007 and January, 2012, 131 consecutive patients with infiltrating HCC underwent TACE as initial treatment at the Cancer Center, Sun Yat-sen University. During the same period, 3,914 patients with HCC were treated at the hospital. The patient and tumor characteristics and the presence of underlying liver diseases are summarized in Table I.

i) Patient age, 18–75 years; ii) Child-Pugh class A or B liver function (19); iii) Eastern Cooperative Oncology Group (ECOG) performance score ≤2; and iv) HCC with no previous treatment.

i) Severe coagulopathy (prothrombin activity <40% or a platelet count <40,000/mm³); ii) Child-Pugh class C liver function or evidence of hepatic decompensation, including ascites, esophageal or gastric variceal bleeding, or hepatic encephalopathy; iv) ECOG scores 3–4; and v) concomitant serious diseases of other organs.

Contrast-enhanced computed tomography (CT) and magnetic resonance imaging (MRI) scans were used to diagnose infiltrating HCC, as ultrasound was inadequate (20). The diagnosis of infiltrating HCC was established by agreement between two radiologists coming from the two centers participating in this study who performed independent reviewing of the cross-sectional imagings of all the patients.

TACE was performed as previously described (21). In brief, a selective 5 Fr catheter was introduced and visceral angiography was performed to assess the arterial blood supply to the liver and to confirm patency of the portal vein. All the patients underwent a distal super-selective catheterization of the hepatic arteries using a coaxial technique and 2.9 Fr microcatheters (Terumo Corporation, Tokyo, Japan). Subsequently, three chemotherapeutic agents at the same dosage were used throughout this study, regardless of tumor number and size. Hepatic artery infusion chemotherapy was first performed using carboplatin 300 mg (Bristol-Myers Squibb, New York, NY, USA), followed by chemolipiodolization using epirubicin 50 mg (Pharmorubicin; Pfizer, Wuxi, China) and mitomycin C 8 mg (Zhejiang Hisun Pharmaceutical Co., Ltd., Taizhou, China) mixed with 5 ml lipiodol (Lipiodol Ultra-Fluide; Andre Guerbet Laboratories, Aulnay-sous-Bois, France). If the territory of the chemolipiodolized artery did not show stagnant flow, pure lipiodol was then injected. For all cases, embolization was finally performed with absorbable 1–2-mm gelatin sponge particles (Gelfoam; Hangzhou alc Ltd., Hangzhou, China) or 350–560-μm polyvinyl alcohol particles (Alicon Pharm SCT & TEC Co., Ltd., Hangzhou, China) until stasis was achieved in the tumor-feeding arteries.

During the same study period (January, 2007–January, 2012), 156 consecutive patients with infiltrating HCC who had declined sorafenib treatment received conservative treatment (best supportive care) at another cancer center. During the same period, 3,845 patients with HCC were treated in The First Affiliated Hospital of Sun Yat-sen University. The inclusion, exclusion and diagnostic criteria were identical to those in the TACE group. The patient and tumor characteristics and the presence of underlying liver diseases are summarized in Table I.

The response of the tumors to TACE was evaluated using contrast-enhanced CT or MRI at 1 month after treatment. The presence of non-enhanced tumoral areas reflected tissue necrosis. The modified Response Evaluation Criteria in Solid Tumors on CT or MRI were used to measure tumor response (22).

Patients in the TACE and conservative treatment groups were followed up monthly for the first year and once every three months thereafter in the outpatient setting using clinical examination, biochemistry and serum α-fetoprotein (AFP) measurements. Contrast-enhanced CT or MRI scans were performed once every 1–2 months for the first year and every 2–3 months thereafter. Bone metastases were excluded by bone scintigraphy on clinical suspicion. In addition, data on the patients’ Child-Pugh class and ECOG scores were recorded.

In the TACE group, hepatocellular injury was monitored by serum bilirubin, alanine transaminase, serum albumin (ALB) and prothrombin time. TACE-related complications were evaluated at the end of the first month after treatment. Complications were reported using the National Cancer Institute Common Toxicity Criteria grading, version 4.0 (23). Another session of TACE was performed once every 2–3 months until one of the following end points was reached: i) complete devascularization of the tumor; ii) technical impossibility to embolize the residual tumor, e.g., tumor only supplied by extrahepatic collateral arteries; iii) contraindications to TACE; and iv) total resection or ablation of tumor by subsequent surgery or local ablation. Hepatic resection or local ablation were performed as previously described (24,25). In cases with ii) or iii), it was recommended that the patients received sorafenib. If they refused, conservative treatment was administered.

Statistical analyses were performed using the SPSS 10.0 statistical software (SPSS, Inc., Chicago, IL, USA). Comparisons between the two groups were performed using the Student’s t-test for continuous data and the Chi-square test for categorical data. Overall survival was calculated using a life table method and compared with the Mantel-Cox test. The survival curves were constructed with the Kaplan-Meier method and compared using the log-rank test. The relative prognostic significance of the variables in predicting overall survival rates was assessed using the multivariate Cox proportional hazards regression analysis. The results are presented as means ± standard deviation, or median and range. All the statistical tests were two-sided and P<0.05 was considered to indicate a statistically significant difference.

A total of 287 patients were recruited in this study (TACE group, n=131; and conservative treatment group, n=156). The characteristics of the patients are summarized in Table I. The two groups were comparable regarding patient characteristics, preoperative liver function and general condition (Table I). Following treatment, 47 patients in the TACE and 62 patients in the conservative treatment group received nucleoside-analog treatment for HBV (P=0.735).

In all infiltrating HCCs, the margins of the tumors were poorly demarcated. The median infiltrating HCC diameter was 9.0 and 9.8 cm for the TACE and conservative treatment groups, respectively. The majority of the patients in the two groups had radiographic evidence of macrovascular invasion at the time of the diagnosis of infiltrating HCC (TACE vs. conservative treatment group, 89/131 vs. 126/156, respectively; P=0.364). All patients with macrovascular invasion had some degree of portal vein tumor thrombosis (PVTT). In the TACE group, 22.9% of the patients had main portal vein involvement, whereas 45% had involvement of the right and/or the left hemihepatic portal and/or sectional/segmental portal vein. In the conservative treatment group, 23.1% of the patients had main portal vein involvement, whereas 57.7% had involvement of the right and/or left hemihepatic portal and/or sectional/segmental portal vein. In addition to portal vein tumor thrombi, 12 and 16 of the patients in the TACE and conservative treatment groups exhibited tumor invasion of the hepatic vein(s), respectively (P=0.844). On further analysis, 6 and 9 of the patients had the tumor thrombi extending into the main hepatic vein(s), 6 and 7 into the inferior vena cava and 3 and 4 extended into the right atrium in the TACE and the conservative treatment groups, respectively. A total of 89 and 97 infiltrating HCC lesions displayed early arterial hyper-enhancement in the TACE and the conservative treatment groups (67.9 vs. 62.2%, respectively; P=0.706). All these lesions demonstrated washout during the portal venous phase. At the time of diagnosis of infiltrating HCC, 51.1 and 50.64% of the patients exhibited intrahepatic satellite lesions and 29.8 and 30.1% had extrahepatic metastases, respectively (P=0.999 and P=0.999, respectively). The most common metastatic sites in the TACE and conservative treatment groups were the lungs (18.3 vs. 19.2%, respectively) and lymph nodes (10.8 vs. 10.3%, respectively). Intrahepatic biliary ductal dilatation was found in 9.9 and 9.6% in the TACE and conservative treatment groups, respectively (P=0.999). There were 30 and 45 patients in the TACE and the conservative treatment groups who received a liver MRI (P=0.528). Among these patients, 22 (73.0%) in the TACE and 31 (69.0%) tumors in the conservative treatment group exhibited relative homogeneity and mild hyperintensity on T2-weighted images. The remaining tumors exhibited isointensity to the surrounding liver parenchyma (Table II).

In the TACE group, 131 patients received a mean of 1.5 sessions (range, 1–4 sessions) of TACE. Of those patients, 52 (39.7%) received one session and 79 (60.3%) received more than one sessions of TACE. The initial TACE consisted of the injection of anticancer drugs, lipiodol and gelatin sponge particles in 10 of 19 (52.6%) patients with main/hemihepatic portal vein invasion and portal vein obstruction, 43 of 70 (61.4%) patients with main/hemihepatic portal vein invasion, but without portal vein obstruction, and 27 of 59 (45.8%) patients with sectional/segmental PVTT. The remaining 42 patients received anticancer drugs and lipiodol injection only.

The tumor response and complications in the two groups are shown in Tables III and IV, respectively. All the TACE-related complications were successfully managed with conservative treatment. The 1-month mortality rate was 0.8 and 3.8% in the TACE and conservative groups, respectively (P=0.134).

Following TACE, the tumors in 6 patients were downstaged and suitable for partial hepatectomy (n=6) or local ablative therapy (radiofrequency ablation, n=1; or radiofrequency + percutaneous ethanol injection, n=1). Thirteen patients with tumor progression following TACE received sorafenib treatment.

At a median follow-up of 6.0 months (range, 1–59 months), 285 patients (94.9%) had succumbed to the disease. The overall median survival was 5.0±0.35 months [95% confidence interval (CI): 4.32–5.68 months]. The 6-, 12- and 24-month overall survival rates for all the patients were 41.9, 12.9 and 1.1%, respectively. The median survival for the TACE and the conservative treatment groups was 7.0±0.3 and 3.0±0.1 months, respectively (P<0.001). The 6-, 12- and 24-month overall survival rates for the TACE and the conservative treatment groups were 61.7, 18.5 and 2.3% vs. 22.7, 12.1 and 0%, respectively. The TACE group exhibited significantly better overall survival compared to the conservative group (P<0.001, Fig. 1).

In the TACE group, the median survival for the 8 patients who were downstaged to receive potentially curative treatments and the remaining 123 patients was 13.0±3.07 and 7.0±0.27 months, respectively. The 6-, 12- and 24-month overall survival rates for TACE + curative treatment and TACE alone were 69.2, 53.8 and 15.4 vs. 60.4, 14.4 and 1%, respectively. The difference was significant (P=0.01, Fig. 2). The median survival for the 13 patients who received sorafenib and the remaining 118 patients was 7.1±0.86 and 6.9±0.42 months, respectively (P=0.563).

On univariate analysis, 6 factors were correlated with survival, namely age, serum γ-glutamyl transpeptidase, serum ALB, PVTT type, maximum tumor size and treatment allocation (Table V). On multivariate analysis, only treatment allocation [odds ratio (OR)=1.777; 95% CI: 1.499–2.107; P<0.001] and PVTT type (OR=1.721; 95% CI: 1.504–1.907; P<0.001) were independent predictors of overall survival.