Phenotypic Response Surfaces-Guided Optimization (PRS-OPT) of Propolis-Metformin-Regorafenib Combination Therapy for MASLD-Associated Hepatocellular Carcinoma.

Objectives: Hepatocellular carcinoma (HCC) arising in metabolic dysfunction-associated steatotic liver disease (MASLD) develops under lipid-rich stress and inflammatory remodeling, which can alter therapeutic windows. We aimed to determine whether phenotypic response surface-guided optimization (PRS-OPT) can nominate hepatocyte-sparing propolis-metformin-regorafenib (PMR) dose windows that retain antitumor activity under MASLD-like fatty-acid (FA) stress and translate to an in vivo immune endpoint. Methods: PMR combinations were profiled in hepatoma cell lines (PLC/PRF/5 and HepG2) and non-malignant hepatocytes (THLE-2) under FA-free and FA-enriched conditions. Quadratic response surfaces were fitted and used for constrained dose nomination, followed by in vitro validation. Cell-death contributions were assessed by inhibitor-rescue, Annexin V/PI flow cytometry, and immunoblotting. PRS-OPT was extended to zebrafish MASLD-HCC to optimize dosing against hepatic macrophage accumulation, with longitudinal imaging and qPCR endpoints. Results: FA exposure shifted the feasible efficacy-tolerability landscape, leading PRS-OPT to nominate a distinct FA-optimized PMR window with improved hepatocyte preservation while maintaining tumor suppression. Validation confirmed tumor-preferential activity across hepatoma lines with hepatocyte sparing. Mechanistic assays supported apoptosis with context dependence and evidence consistent with a ferroptosis-related component in HepG2. In zebrafish MASLD-HCC, PRS-OPT-nominated dosing reduced hepatic macrophage accumulation and improved disease-relevant transcriptional and morphological readouts. Conclusions: PRS-OPT enables interpretable, multi-objective dose nomination for MASLD-relevant HCC contexts, and establishes PRS-OPT as an interpretable, multi-objective framework for regimen nomination that is directly extensible to additional phenotypic endpoints for preclinical evaluation.