Abstract / Summary
Lenvatinib plus PD-1 inhibitor combination therapy has emerged as a frontline treatment for advanced hepatocellular carcinoma (HCC), but substantial interpatient variability in therapeutic responses underscores the urgent need for reliable predictive biomarkers. Here, we enrolled 86 HCC patients treated with lenvatinib plus tislelizumab and integrated digital spatial profiling (DSP), multiplex immunofluorescence (mIF), and immunohistochemistry (IHC) to dissect intratumoral heterogeneity and identify spatially resolved biomarkers associated with treatment outcomes. DSP analysis demonstrated distinct molecular features between epithelial cell-enriched (EP-enriched) and immune cell-enriched (IM-enriched) regions, effectively capturing spatial intratumoral heterogeneity in HCC. Furthermore, we found that FAP-α expression was significantly elevated in non-responders, predominantly localized to IM-enriched regions, and correlated with M2-type macrophage infiltration accompanied by suppressed T cell infiltration. Using an mIF validation cohort of 22 patients, we confirmed that FAP-α outperformed immune cell markers as a prognostic biomarker. IHC validation in 54 patients revealed FAP-α+ cell proportion as a robust predictor (sensitivity: 80%; specificity: 71.4%), with 85.7% of patients above the optimal threshold exhibiting treatment non-response. Upregulation of FAP-α during HCC progression and association with shorter overall survival were also observed by using public data analysis. Collectively, leveraging large clinical cohorts and cross-validation across multiple technologies, we systematically dissected the spatial molecular landscape of HCC and identified FAP-α as a clinically translatable biomarker for lenvatinib plus PD-1 inhibitor therapy, providing a rationale for clinical patient stratification and guiding personalized treatment strategies.