Metabolomics and lipidomics predictor of survival in hepatocellular carcinoma patients receiving tyrosine kinase inhibitor and immune checkpoint inhibitor combination therapy.

Although tyrosine kinase inhibitor and immune checkpoint inhibitor (TKI-ICI) combination therapy has emerged as a promising treatment for hepatocellular carcinoma (HCC), reliable biomarkers for predicting long-term survival remain underexplored. Here, we conducted metabolomics and lipidomics profiling in baseline plasma samples from 58 patients with HCC who received TKI-ICI therapy in a prospective phase II trial. Prognostic features were identified using an integrated machine learning framework combining random forest survival analysis, LASSO regression, and Cox modeling. Untargeted metabolomics identified 2 lipids, phosphatidylinositol lyso 18:1 and O-phosphorylethanolamine, that were associated with progression-free survival. Lipidomics further revealed 9 prognostic lipids, including cholesteryl ester (18:1), triacylglycerol (TG) (15:0/15:0/21:6), TG (18:1/20:5/20:5), TG (41:3), phosphatidylserine (42:3), phosphatidylethanolamine (38:5), sphingosine (d18:1), phosphatidylcholine (43:5), and ceramide (d18:2/22:0), as independent predictors of progression-free survival. Multivariate Cox modeling integrating metabolomic and lipidomic markers reinforced their prognostic relevance. Meanwhile, 6 lipids, including phosphatidylcholine (39:8p), phosphatidylserine (18:0/20:4), TG (18:1/20:4/22:5), TG (15:0/15:0/21:6), ph sphingomyelin (d38:5), and sphingomyelin (d41:5), were found to be associated with overall survival. Functional enrichment analysis revealed that these prognostic lipids were involved in sphingolipid-related metabolism and immune-related signaling, highlighting the importance of lipid-immune crosstalk in reshaping responses to TKI-ICI therapy in patients with HCC. As sphingolipids are also known to modulate drug metabolism enzymes and transporters, they may thereby affect interindividual variability in TKI pharmacokinetics. In conclusion, our findings demonstrated that circulating lipidomic features, particularly sphingolipid-related species, were predictive of long-term survival in patients with HCC receiving TKI-ICI combination therapy. These lipids may serve as noninvasive biomarkers for survival prediction, patient stratification, and informed therapeutic decision making, while offering insights into lipid-immune interplay in immunotherapy-based cancer treatment. SIGNIFICANCE STATEMENT: This study integrates metabolomic and lipidomic profiling of baseline plasma from patients with hepatocellular carcinoma receiving tyrosine kinase inhibitor and immune checkpoint inhibitor combination therapy, identifying sphingolipid-related lipid species as strong predictors of long-term survival. Unlike prior work focused on short-term response or monotherapy, these findings highlight lipidomic markers as noninvasive tools for survival prediction and treatment stratification, providing new insights into lipid-immune interactions and supporting the clinical utility of lipidomic signatures in guiding therapeutic decisions.