Patient baseline characteristics
A total of 508 patients were analyzed for safety. The patients demographic and baseline characteristics by Child-Pugh score and BCLC stage at the start of therapy are shown in Table 1.
Median age was 70 years, and approximately 80 % of the patients were males; 85 % of the patients were Child-Pugh A and 11.4 % were Child-Pugh B; 54.7 % of the patients were classified as BCLC stage C. A worse Eastern Cooperative Oncology Group (ECOG) score correlated with a worse Child-Pugh score. The ECOG score was similar among patients with BCLC stages A and B but tended to be higher in patients with BCLC stage C.
Sorafenib administration by Child-Pugh score and BCLC stage at the start of therapy is shown in Table 2. Among the patients with Child-Pugh A, a similar proportion received an initial daily dose of 400 mg (47.0 %) versus 800 mg (46.3 %). A slightly higher proportion of patients with Child-Pugh B (53.4 %) than Child-Pugh A (47.0 %) received an initial daily dose of 400 mg.
Of the patients with BCLC stage B, 50.0 and 40.7 % received an initial daily dose of 400 and 800 mg, respectively. The proportion of patients with BCLC stage C (53.6 %) who received an initial daily dose of 800 mg was slightly higher than for those with BCLC stage B (40.7 %).
The average daily dose of sorafenib, 419.0 mg, was similar to that received by patients with Child-Pugh A and B scores and with BCLC stages A and B (400.0 mg); that of BCLC stage C was slightly higer (471.0 mg).
The median treatment duration with sorafenib was 15.90 weeks. Treatment duration tended to become shorter as the Child-Pugh score worsened; patients with Child-Pugh A and Child-Pugh B had median treatment durations of 17.40 and 7.60 weeks, respectively.
A total of 500 patients were analyzed for efficacy in the ITT analysis. The difference between the safety and ITT population was due to reasons such as exclusion of patients who had a history of sorafenib treatment. OS and TTP by Child-Pugh score and BCLC stage per Response Evaluation Criteria in Solid Tumors (RECIST) v1.0 are shown in Figs. 1 and 2.
The median OS in patients with Child-Pugh A (17.4 months; Fig. 1a) was longer than in those with Child-Pugh B (4.9 months), suggesting that the Child-Pugh score is a prognostic factor. Similarly, the median TTP in patients with Child-Pugh A (3.7 months; Fig. 1b) was longer than in patients with Child-Pugh B (2.3 months), but the difference was not as remarkable as that seen for OS. The TTP by modified RECIST (mRECIST) also showed a similar trend (data not shown).
Although the median OS in patients with BCLC stage A was not reached, OS tended to be shorter with more advanced BCLC stage (Fig. 2a). Median OS was longer in patients with better liver function; median OS in patients with BCLC stage B of Child-Pugh A and Child-Pugh B were 20.7 (95 % CI 15.4–unknown) and 8.9 (95 % CI 4.6–14.4) months, respectively. TTP, as measured by RECIST, tended to be shorter with more advanced BCLC stage (Fig. 2b); TTP in patients with BCLC stage A was 6.5 (95 % CI 4.1–8.8) months, 4.1 (3.4–5.0) months in patients with stage B and 3.0 (2.6–3.5) months in patients with stage C. TTP by mRECIST showed a similar tendency (data not shown).
A summary of AEs by Child-Pugh score and BCLC stage at the start of sorafenib therapy is shown in Table 3.
The incidence of AEs and drug-related AEs in patients with Child-Pugh A and Child-Pugh B were similar (94.9 % and 94.8, 88.2 and 86.2 %, respectively)
The incidence of serious AEs (SAEs) and drug-related SAEs in patients with Child-Pugh B was higher than in patients with Child-Pugh A (69.0 % and 37.0, 32.8 and 16.2 %, respectively). The incidence of AEs leading to permanent discontinuation of sorafenib was 38.7 % in patients with Child-Pugh A and 51.7 % in patients with Child-Pugh B. The incidence of treatment-emergent death occurring up to 30 days after discontinuation of sorafenib was 11.8 % in patients with Child-Pugh A and 34.5 % in patients with Child-Pugh B.
Drug-related AEs reported more frequently in patients with Child-Pugh A than with B included hand-foot skin reaction (HFSR), hypertension, alopecia, hoarseness, decreased platelet count, pruritus and rash/desquamation. However, vomiting and abnormal laboratory tests were reported more often in patients with Child-Pugh B than with A.
The drug-related AEs of the hepatic system of liver dysfunction, hypoalbuminemia, and hepatic encephalopathy were observed more frequently in patients with Child-Pugh B than Child-Pugh A. Among drug-related SAEs, the incidence rates of liver dysfunction, hepatic encephalopathy, gastric ulcer and abnormal laboratory tests were also higher in patients with Child-Pugh B.
Comparison with other geographic regions in GIDEON
Patient baseline characteristics, incidence of drug-related AEs, BCLC stage, median OS and TTP, and treatment history in the five geographic regions of the GIDEON study (Asia-Pacific, European Union, Latin America, USA and Japan) are summarized in Table 4 . In Japanese patients, the median age was higher (70 years) and a history of locoregional therapy was also higher (84.4 %) than for other regions. Particularly, TACE was conducted more frequently in Japanese patients (71.3 %). Infection with HCV was etiologically associated with 53.1 % of HCC cases in Japan, which was comparable to the USA. Japan experienced the highest incidence of drug-related AEs, including CTCAE grades 3 and 4, drug-related SAEs and AEs resulting in permanent discontinuation of sorafenib, but the lowest rate of deaths. In Japan, 43.7 % of patients had BCLC stage A at the time of initial diagnosis, but the majority of patients had progressed to stage B (31.9 %) or C (54.7 %) by the initiation of sorafenib therapy. Regardless of BCLC stage, Japanese patients showed a longer time from initial diagnosis to death than those in other regions. In addition, the median OS from the start of sorafenib therapy was longest, but the median TTP was shorter than in other regions.
Effects of the number of transcatheter arterial chemoembolization sessions on the tumor response rate and Child-Pugh status
The relationships between the number of TACE sessions and its tumor response rate before the start of sorafenib therapy and between the number of TACE sessions and Child-Pugh score at initiation of sorafenib therapy are shown in Table 5. It has been shown that there is no significant correlation between the tumor reduction rate (World Health Organization and RECIST criteria) and the pathologic necrosis rate after TACE with lipiodol . The response evaluation criteria that take account of the tumor necrosis are thus required in liver cancer treatment. Therefore, it is common in Japan to determine the treatment effect using the modified RECIST and the response evaluation criteria in cancer of the liver [23, 24].
The number of TACE sessions was higher in Japan than in other regions; however, patients with ≥6 TACE sessions tended to have lower complete and partial response rates. In addition, when the number of TACE sessions before sorafenib therapy was ≥6, the percentage of patients with Child-Pugh B was higher at initiation of sorafenib therapy.