This was a retrospective cohort study conducted in Seoul St. Mary’s Hospital, Seoul, Korea. We reviewed the medical records of patients who were treated with MET chemotherapy or sorafenib for advanced HCC with PVTT between June 2005 and August 2013. The diagnosis and stage of HCC were assessed according to the American Association for the Study of Liver Disease guidelines and the BCLC staging classification. Inclusion criteria were (1) age 18-80 years, (2) an Eastern Cooperative Oncology Group performance status of zero to two, (3) Child-Pugh Class A or B, (4) advanced HCC according to the BCLC staging classification (confirmed PVTT in the main [Vp4], first [Vp3], or second branch [Vp2] of the portal vein or extrahepatic metastasis), (5) acceptable blood cell counts (absolute neutrophil count≥1.0×10⁹/L and platelet count≥50×10⁹/L), and (6) vascular access to the lesion for implantation of a chemoport.

Exclusion criteria were (1) evidence of hepatic decompensation including uncontrolled ascites, gastrointestinal bleeding, or hepatic encephalopathy, (2) concurrent serious medical condition(s) such as underlying cardiac or renal disease, (3) other concurrent primary malignancy, (4) another chemotherapy such as hepatic arterial infusion chemotherapy (HAIC) or systemic chemotherapy, (5) adjuvant MET chemotherapy or sorafenib after curative treatment, or (6) absence of an intrahepatic lesion. The trial was conducted according to the Declaration of Helsinki, and the protocol was approved by the Institutional Ethics Committee of The Catholic University of Korea.

The total number of HCC patients in our institute between June 2005 and August 2013 was 2,381. A total of 303 patients of advanced HCC treated with MET chemotherapy or sorafenib were enrolled in this period. In these patients, 61 patients were MET group and 242 patients were sorafenib group. In MET group, seven patients were excluded because they did not have PVTT. In sorafenib group, 151 patients were excluded because they did not have PVTT. In remaining 91 patients, three patients were excluded by post-liver transplantation usage of sorafenib, 18 patients were excluded by combined with other treatment modality, one patient was excluded by adjuvant usage of sorafenib after curative surgery, and 16 patients were excluded because they did not follow up after their first prescriptions. Therefore, 54 patients with advanced HCC and PVTT who had undergone MET were assigned to the MET group, and 53 patients who had undergone sorafenib treatment were assigned to the sorafenib group (Fig. 1).

Chemotherapeutic agents were delivered via hepatic arterial infusion chemoport with an arterial infusion pump. Catheter insertion and placement were performed as previously described [15]. The metronomic chemotherapeutic agents were cisplatin and 5-fluorouracil (5-FU), which were infused weekly for three weeks with one week break between each cycle.

Single doses of cisplatin and 5-FU were 15 mg/m² and 50 mg/m², respectively. Cisplatin and 5-FU were infused via hepatic arterial infusion chemoport with an arterial pump for two hours. Along with the metronomic chemotherapy, epirubicin (30 mg/m²) was infused monthly via hepatic arterial infusion chemoport without the use of lipiodol or gelfoam embolization. This protocol was repeated every four weeks and continued until disease progression or unacceptable toxicity was evident or the patient refused to continue (Fig. 2) [12]. In instances including cisplatin infusion, intravenous hydration was performed to prevent nephrotoxicity, and an appropriate anti-emetic agent was administered to all patients. Dose adjustments were made depending on the toxicity observed in the preceding treatment cycle. The following cycle of treatment was reduced by 30% in cases of repeated grade two toxicity during the preceding cycle. Treatment was stopped in all grade three or grade four toxicity cases and was not reinitiated until symptom resolution. If a patient required a delay longer than four weeks for recovery, they were excluded from the study.

Principally, patients who were treated with sorafenib received standard doses of sorafenib (400 mg twice a day, orally). Dose adjustments of sorafenib were made depending on clinically significant toxicity (more severe than grade two based on the National Cancer Institute Common Terminology Criteria for Adverse Events Version 4.0) or the clinician’s determination of patient tolerance. During sorafenib treatment, the patients visited the outpatient clinic every three or four weeks for safety and tolerability assessments.

During the follow-up period, laboratory tests including alphafetoprotein (AFP), albumin, bilirubin, aspartate aminotransferase, alanine aminotransferase, and prothrombin time and a liver dynamic computed tomography (CT) scan (non-enhanced, arterial, portal, and delayed venous phases) or a liver dynamic magnetic resonance imaging (MRI) scan with hepatocyte specific contrast were performed to evaluate treatment response and reserved liver function every four to eight weeks after treatment. The treatment response was assessed using the modified Response Evaluation Criteria in Solid Tumors (mRECIST) [16]. An objective response was defined as complete response, and partial response and disease control included complete response, partial response, and stable disease.

The primary endpoints were OS and time to progression in each treatment group. The OS was calculated from the initiation of treatment to death or final follow-up visit. The time to progression was calculated from the initiation of treatment to radiologic progression. The secondary endpoints were objective response rate and disease control rate at eight weeks of treatment. The treatment-related toxicity assessment was based on the National Cancer Institute Common Terminology Criteria for Adverse Events Version 4.0.

Continuous data were expressed as median and range, and categorical data as percentage. Treatment response between groups was compared using the Chi-square or Fisher’s exact test. Other variables between different groups were compared using these tests or an independent t test, as appropriate. Cumulative overall survival was calculated using the Kaplan-Meier method, and the difference between groups was compared using the log-rank test. A Cox proportional-hazards model was used to identify independent clinical factors or groups that had effects on the overall survival rate. A P-value<0.05 was considered statistically significant (SPSS ver. 20.0, SPSS Inc., Chicago, IL, USA).

A total of 107 patients were analyzed into this study between June 2005 and August 2013. Among these patients, 54 with advanced HCC and PVTT who had undergone MET were assigned to the MET group, and 53 who had undergone sorafenib treatment were assigned to the sorafenib group. Table 1 shows the baseline characteristics of all patients enrolled in this study. The mean age was 56.2±10.3 years; 89 patients were male, and 18 patients were female. The most common cause of underlying liver disease was hepatitis B virus infection (81.3%). Fifty-nine patients were Child-Pugh class A, and 48 patients were Child-Pugh class B. The HCC stage of all patients was BCLC C stage. The maximal tumor diameters of the MET group and sorafenib group were 12.5±4.6 cm and 9.2±5.1 cm, respectively (P=0.001). The median number of treatment cycles in the MET group was two (range 1-15 cycles), and the median treatment duration of the sorafenib group was 51 days (range, 11-725 days). The baseline characteristics between the two treatment groups were similar except for tumor number, maximal tumor diameter, and treatment history.

Table 2 shows the treatment response of each group at eight weeks of treatment. According to the mRECIST, complete response, partial response, stable disease, and progressive disease were noted in zero (0.0%), four (7.4%), 21 (38.9%), and 29 (53.7%) patients in the MET group and zero (0.0%), zero (0.0%), nine (22.0%), and 32 (78.0%) patients in the sorafenib group (P=0.026). The disease control rate was higher in the MET group than the sorafenib group (46.3% vs. 22.0%, P=0.014). However, the objective response rate showed no statistically significant difference between the two groups (7.4% vs. 0.0%, P=0.131).

The median OS of 107 patients during follow-up was 127 days (range, 12-2,666 days). In the MET group, 51 patients (94.4%) died, three patients (5.6%) were lost to follow-up, and only one patient (1.7%) survived. In the sorafenib group, 46 patients (86.8%) died, six patients (11.3%) were lost to follow-up, and no patient survived. The most common cause of death in both groups was tumor progression (31.4% and 58.7% in the MET group and sorafenib group, respectively). The second most common cause was hepatic failure, including hepatorenal syndrome (23.5 % and 23.9% in two groups, respectively), and other causes included variceal bleeding and infection (9.8% and 0.0%, and 5.9% and 2.2%, respectively). There was a statistically significant difference between the two groups (P=0.007) based on cause, and there were no deaths related to treatment.

The median survival time in the MET and sorafenib groups was 158 days (95% confidence interval [CI], 132-184 days) and 117 days (95% CI, 92-142 days), respectively (Fig. 3). The overall survival rate showed significantly better results in the MET treatment group than in the sorafenib group (P=0.029). The time to progression in the two groups was 62 days (95% CI, 42-82 days) and 78 days (95% CI, 63-93 days), respectively; however, the difference was not statistically significant (P=0.472).

Univariate analysis was performed using the Cox proportional hazards model, which identified three prognostic factors for OS: Child-Pugh class, treatment modality, and serum AFP level. MET was an independent and more accurate prognostic factor associated with OS (hazard ratio [HR]=0.471, 95% CI, 0.296-0.749, P=0.001) than sorafenib, and Child-Pugh class B and higher serum AFP level (≥400 mg/dL) were independent, less accurate, OS-associated prognostic factors (HR=1.856, 95% CI, 1.173-2.973, P=0.008, and HR=1.680, 95% CI, 1.112-2.538, P=0.014, respectively) (Table 3). In subgroup analysis in patients with Child-Pugh class B disease, there was no statistically significant difference in the baseline characteristics between MET and sorafenib group (Supplementary Table 1) and the median OS of the MET and sorafenib groups was 190 days and 58 days, respectively (29 and 19 patients, P<0.001) (Fig. 4). Otherwise, in subgroup analysis of patients with Child-Pugh class A disease, there was no statistically significant difference in the median OS of the MET and sorafenib groups. (157 days and 143 days, respectively, P=0.813). In subgroup analysis in naïve patients, there was no statistically significant difference in the baseline characteristics between MET and sorafenib group except for sex (Supplementary Table 2) and the median OS of the MET and sorafenib groups was 229 days and 61 days, respectively (27 and 16 patients, P<0.001) (Fig. 5).

Table 4 shows the treatment-related toxicity of both treatments. The majority of MET-related toxicity was hematological adverse events. Leukopenia developed in 48% of patients treated with MET. Thrombocytopenia, hyperbilirubinemia, and increased alanine aminotransferase level developed in 22%, 30%, and 17% of patients treated with MET, respectively. The majority of sorafenib-related toxicity was mucosal toxicity, including skin and gastrointestinal tract toxicity. Hand-foot syndrome and gastrointestinal tract toxicity developed in 28% and 40% of patients treated with sorafenib, respectively, and hyperbilirubinemia and alanine aminotransferase elevation developed in 36% and 17% of such patients. Patients who were experienced grade 3/4 toxicities in MET and sorafenib group, were 25 and 16, respectively and there was no statistically significant difference between two groups (P= 0.488).