fig1

Figure 1. Non-proportional hazards in HCC trials: patterns, examples, and implications for interpretation. (A) Illustrates the three main patterns of non-proportional hazards observed in HCC trials: diminishing effect, delayed effect, and crossing hazards; (B) maps these patterns onto pivotal immunotherapy-based trials, including IMbrave050, LEAP-012, HIMALAYA, and CheckMate 9DW, highlighting the distinct clinical implications of each pattern; (C) summarizes a practical interpretive framework: test the proportional hazards assumption, avoid relying on the hazard ratio alone when NPH are present, confirm data maturity, and apply methods better suited to time-varying treatment effects, including MaxCombo testing, restricted mean survival time, and piecewise hazard ratios. Created in BioRender. Villa, E. (2026) https://BioRender.com/fclev2z. HCC: Hepatocellular carcinoma; HR: hazard ratio; NPH: non-proportional hazards; RMST: restricted mean survival time.