Lipid droplet-associated gene signatures classify metabolic subtypes and identify PLIN3 as a key driver in hepatocellular carcinoma

Hepatocellular carcinoma is characterized by considerable molecular heterogeneity, which complicates prognostic predictions and contributes to therapeutic resistance. This study aimed to develop a molecular classification framework grounded in lipid droplet-associated genes (LDAGs) and to comprehensively elucidate their biological significance and clinical applicability in guiding personalized treatment approaches. By leveraging multi-cohort datasets, we defined LDAG-based molecular subtypes and systematically characterized their genomic alterations, metabolic features, pathway activation patterns, and therapeutic vulnerabilities. Three distinct subtypes (C1–C3) were identified according to LDAG expression patterns, each demonstrating unique clinical outcomes, mutational profiles, and metabolic reprogramming. The C1 subtype correlated with the poorest overall survival, more advanced tumor stages, and activation of pro-proliferative signaling pathways. Therapeutic vulnerabilities were subtype-dependent, with C1 showing heightened sensitivity to sorafenib. Five pivotal LDAGs (PLIN3, SET, CKAP4, RAP1B, and PISD) were implicated in the aggressive phenotype of C1, among which PLIN3 exhibited the strongest prognostic value. Functional assays confirmed that PLIN3 knockdown reduced lipid accumulation, suppressed cell proliferation and migration, and impaired tumorigenesis, whereas its overexpression promoted aggressive tumor behavior. In conclusion, our LDAG-based classification system stratifies hepatocellular carcinoma into three clinically relevant subtypes. PLIN3 emerges as a promising prognostic biomarker and therapeutic target, thereby mechanistically linking lipid metabolism to hepatocellular carcinoma progression.