Abstract / Summary
Abstract Background Early treatment-related physiological changes are increasingly recognized as prognostic indicators in advanced hepatocellular carcinoma (HCC), yet they are rarely integrated into routine response assessments. This study aimed to characterize early dynamic physiological phenotypes during systemic therapy and evaluate their clinical associations and prognostic value using a landmark analysis framework. Methods We retrospectively analyzed 387 patients with advanced HCC receiving tyrosine kinase inhibitor monotherapy or combination systemic regimens. A prespecified 6-week landmark approach evaluated changes in skeletal muscle index (ΔSMI) and ECOG performance status (ΔECOG). Patients were stratified into three hierarchical phenotypes: Stable Response (P1), Muscle-predominant Attrition (P2), and Functional Collapse (P3). Statistical evaluations included structural equation modeling (SEM), multivariable Cox regression, Fine–Gray competing-risk models, and sensitivity analyses. Results Patients receiving combination regimens exhibited more pronounced early physiological decline compared to those on monotherapy ( P < 0.001). SEM confirmed significant bidirectional temporal associations between skeletal muscle changes and functional status over the initial 6 weeks (standardized β = 0.14–0.18). P2 demonstrated the highest objective response rate (46.4%) despite marked muscle loss, whereas P3 was characterized by poor tumor response alongside rapid functional deterioration. In multivariable landmark analyses, P3 remained independently associated with shorter overall survival (adjusted hazard ratio, 2.35; 95% confidence interval, 1.75–3.18; P < 0.001), as did higher 30-day rates of change in the neutrophil-to-lymphocyte ratio. Competing-risk and lag-time sensitivity analyses yielded consistent findings. Conclusions Early dynamic physiological phenotypes offer valuable prognostic insights beyond baseline characteristics for patients with advanced HCC undergoing systemic therapy. Supplementing conventional radiological evaluations with early physiological assessments could enhance risk stratification and guide individualized patient monitoring.