LETIM Robustly Predicts Immune Checkpoint Blockade Efficacy for Liver Cancer Patients Using Multiple Immune‐Related Genesets

BACKGROUND AND AIMS: Immune checkpoint blockade (ICB) is an important liver cancer treatment but shows obvious variability in clinical response due to the tumor immune microenvironment (TIME) heterogeneity. Reliable models that identify TIME characteristics to predict potential beneficiaries remain lacking. METHODS: We analyzed single-cell RNA sequencing (scRNA-seq) data from ICB responsive and nonresponsive liver cancer patients to establish genesets with both intragroup stability and intergroup variability for immune cell status evaluation. These genesets were applied to bulk RNA sequencing (RNA-seq) data from large cohorts to score and construct an ICB response prediction model. RESULTS: The study identified immune-related genesets characterizing the influence of 21 immune cell types on ICB response. Using these genesets, we developed the Liver Cancer ExtraTrees Immune-response Model (LETIM). LETIM effectively predicted response by distinguishing TIME features. LETIM surpassed existing models in predictive performance and exhibited excellent generalizability in other large liver cancer sequencing cohorts (area under the receiver operating characteristic curve: 0.88-0.92). CONCLUSIONS: Through comprehensive analysis of the effects of multiple immune cells on ICB response, LETIM achieved accurate prediction of ICB outcomes in liver cancer. Its cross-cohort applicability and robust predictive performance may support clinical translation to enable precision medicine and maximize patient benefits.