7 In the group of patients treated with dose-adjusted sorafenib for ≥70% of the Estrogen antagonist treatment period the average received dose was 474 mg daily (469 mg daily for BCLC B and 476 mg daily for BCLC C patients). Instead, in the group of patients who maintained full dose of sorafenib for the entire treatment period or received a dose-adjusted for <70% of the whole treatment period the mean received dose was 748 mg daily (723 mg daily for BCLC B and

754 mg daily for BCLC C patients). In the SOFIA study the average actual received doses of sorafenib were 474 mg daily in the dose-adjusted group and 748 mg daily in the full-dose group. These average doses were strictly similar to the theoretical doses of 400 mg and 800 mg daily. Therefore,

we also performed analyses according to these theoretical doses. Sorafenib-based treatment strategies were evaluated according to BCLC (B or C) stage and sorafenib dose (full dose: 800 mg daily; dose-adjusted: 400 mg daily). The strategies analyzed were: (1) full or dose-adjusted sorafenib for BCLC B and C patients together (Fig. 1A); (2) full or dose-adjusted sorafenib for BCLC B patients (Fig. 1B); (3) full or dose-adjusted sorafenib for BCLC C patients (Fig. 1C). Given that there are no other agents besides sorafenib that have demonstrated significant survival benefit or have been approved for this patient population by the Food and Drug Administration (FDA), all sorafenib strategies were compared to best supportive care (BSC). BSC incorporated medical staff visits, hospitalizations, and laboratory MI-503 molecular weight and radiology tests. Survival of patients who underwent BSC was modeled by application of risk ratios from a recent meta-analysis of 30 randomized controlled trials (RCTs) of untreated HCC patients enrolled in trials of palliative treatments. 3, 4 Treatment effectiveness was modeled by application of Kaplan-Meier survival curves from the recent field practice prospective SOFIA study (6). We used a Markov model to simulate the costs and effects

associated with sorafenib treatment 上海皓元医药股份有限公司 and BSC over a 5-year time horizon. The model was designed to simulate cohorts of Caucasian male patients, 67 years old, with BCLC C HCC (75%), or BCLC B HCC who failed locoregional therapies (25%), well-compensated cirrhosis, and with performance status of 0-1, as included in the SOFIA study. The model comprised three health states: BCLC B HCC, BCLC C HC,C and death (Fig. 2). In such a model patients suffering an acute event could die during that month or survive (at least for that month). The health states were mutually exclusive, i.e., a patient could experience a single health state at any given time. For each transition, we obtained the time-dependent transition rates by assuming a Weibull distribution, parameters of which were estimated using available data (6).