In the phase 3 SHARP and AP trials, sorafenib (SOR) significantly improved overall survival (OS) vs placebo (PLC) in patients with hepatocellular carcinoma (HCC) (SHARP: HR 0.69; p
In both trials, tumors were assessed at baseline and every 6 weeks. Early TGR, defined as the relative change per month in the sum of target lesion diameters (SLD) from start of treatment, determined at start of treatment, was derived with a parabola-like 3-parametric model describing the SLDs as a function of time during the treatment period. TGR was examined both for prognostic and predictive characteristic for OS with uni- (UV) and multivariate (MV) models. Baseline SLD was included to compare the roles of a tumor-related baseline and post baseline variable.
544/602 patients (90%)/SHARP and 190/226 (84%)/AP patients were evaluable for TGR and OS. Mean early TGR was significantly different between treatment arms (SHARP: PLC 0.042, SOR 0.007, p = 0.035; AP: PLC 0.115, SOR 0.028, p = 0.005). Early TGR as categorical covariate in a Cox model analysis of OS was a strong prognostic factor for time to death of similar strength in both treatment arms. The table shows hazard ratios (HR) and p-values in UV and MV analyses of OS with TGR for PLC and SOR separately and for TGR together with treatment. Interaction terms with treatment were always non-significant; thus, TGR is prognostic in both treatment arms, but does not predict whether patients respond differentially to SOR and PLC.