Hepatocellular carcinoma, the predominant form of primary liver cancer, is not a single disease but a collection of biologically distinct tumors that can behave very differently in the same clinical setting. This molecular diversity is one of the central challenges in modern liver-cancer care. Although targeted drugs and immunotherapies have expanded treatment options, physicians still lack reliable methods for predicting which tumors will spread rapidly, which patients are likely to respond to a particular therapy, and which metabolic processes are driving tumor progression. A new study from researchers at Chongqing Medical University suggests that lipid droplets—specialized intracellular structures once regarded mainly as passive fat-storage depots—may provide an important route toward solving these problems.
Published in Genes & Diseases, the study developed a molecular classification system for hepatocellular carcinoma based on lipid droplet-associated genes, or LDAGs. Lipid droplets are dynamic organelles composed of a neutral-lipid core surrounded by a phospholipid monolayer and decorated with regulatory proteins. They store fatty acids and cholesterol, but they also participate in energy production, membrane synthesis, oxidative-stress control, inflammatory signaling, and interactions with other organelles. Cancer cells can remodel lipid-droplet biology to survive nutrient deprivation, resist cellular stress, support rapid proliferation, and adapt to changing conditions inside a tumor. Until now, however, the broader contribution of lipid droplet-associated genes to clinically meaningful HCC subtypes had not been comprehensively defined.
The investigators integrated transcriptomic and clinical information from 1,834 patients across several independent HCC cohorts. Their analysis began with a curated set of 122 genes linked to lipid droplet biology. These genes were evaluated across publicly available datasets, including the Cancer Genome Atlas liver hepatocellular carcinoma cohort, commonly referred to as TCGA-LIHC. Rather than assigning tumors according to a single mutation or one altered pathway, the researchers used unsupervised consensus clustering, a computational approach that repeatedly analyzes gene-expression patterns to identify groups of samples with stable molecular similarities. This strategy divided HCC tumors into three reproducible LDAG-associated subtypes, designated Cluster 1, Cluster 2, and Cluster 3.
The three groups displayed sharply different clinical and biological characteristics. Cluster 1 emerged as the most aggressive subtype. Patients in this group were more likely to have advanced-stage disease, vascular invasion, and extensive inflammatory infiltration within the tumor microenvironment. These features are clinically important because invasion into blood vessels increases the likelihood that cancer cells will disseminate to other sites. Survival analysis showed that patients with Cluster 1 tumors had significantly poorer overall survival than those assigned to the other two groups. The findings indicate that a lipid droplet-associated expression pattern may capture multiple dimensions of tumor behavior at once, linking altered metabolism with invasion, immune activity, and disease progression.
The genomic data provided further evidence that the subtypes were biologically distinct rather than merely statistical groupings. TP53 mutations were enriched in Cluster 1, consistent with the aggressive clinical profile of this class. TP53 encodes a major tumor-suppressor protein that helps coordinate DNA-damage responses, cell-cycle arrest, senescence, and apoptosis. Loss of its activity can allow genetically unstable cells to continue dividing and can promote more invasive tumor phenotypes. By contrast, CTNNB1 mutations predominated in Cluster 3. The CTNNB1 gene encodes β-catenin, a central component of the Wnt signaling pathway, which regulates cell fate, proliferation, and tissue organization. The contrasting mutation patterns suggest that lipid metabolism is intertwined with established oncogenic networks rather than operating as an isolated feature of HCC biology.
Pathway and functional-enrichment analyses revealed additional differences in the cellular programs active within each subtype. The expression of LDAGs was associated with distinct metabolic circuits and signaling pathways, indicating that tumors may use lipid droplets in different ways depending on their underlying molecular state. Some cancer cells may rely on lipid droplets as reservoirs of energy and membrane-building materials, while others may use them to buffer toxic fatty acids, regulate reactive oxygen species, or coordinate inflammatory signaling. Such adaptations could influence not only tumor growth but also how malignant cells respond to therapy. The study therefore places lipid-droplet remodeling within the larger framework of metabolic reprogramming, a hallmark of cancer that enables tumor cells to survive under conditions that would normally restrict healthy tissue growth.
The researchers also investigated whether the LDAG-based classification could provide clues about treatment response. Drug-sensitivity analyses suggested that tumors in the aggressive Cluster 1 group might be more responsive to sorafenib, a multikinase inhibitor historically used in advanced HCC. Sorafenib targets several signaling proteins involved in tumor proliferation and angiogenesis, including RAF kinases and receptors associated with vascular endothelial growth factor and platelet-derived growth factor signaling. The prediction does not establish that every Cluster 1 patient will benefit from sorafenib, nor does it replace prospective clinical testing. It does, however, raise the possibility that molecular features associated with an unfavorable prognosis could simultaneously reveal a therapeutic vulnerability. In precision oncology, that distinction is critical: a high-risk tumor may not simply require more intensive treatment but may possess specific metabolic dependencies that can be exploited.
To identify genes that might be driving these aggressive behaviors, the investigators constructed interaction and expression-based networks and highlighted five hub genes: PLIN3, SET, CKAP4, RAP1B, and PISD. Among them, PLIN3 showed the strongest prognostic value. PLIN3 encodes perilipin 3, a lipid-droplet coat protein that helps regulate the formation, maintenance, and mobilization of lipid droplets. By controlling access to stored neutral lipids, perilipin proteins can influence fatty-acid availability, energy balance, organelle communication, and protection against lipotoxic stress. In HCC cells, the researchers found that reducing PLIN3 expression lowered intracellular lipid accumulation, inhibited proliferation and migration, and suppressed tumor growth in experimental models. Conversely, increasing PLIN3 expression promoted more aggressive cellular behavior. These experiments move the study beyond correlation and support a functional role for PLIN3 in the metabolic remodeling that sustains HCC progression.
The findings do not yet make PLIN3 a clinically approved biomarker or treatment target, and several questions remain open. The cohorts used for classification were assembled from existing datasets, while the predicted treatment associations will need validation in prospective patient studies. It will also be necessary to determine whether PLIN3 acts directly through lipid-droplet storage, fatty-acid oxidation, oxidative-stress regulation, or communication with other organelles such as mitochondria and the endoplasmic reticulum. Nevertheless, the study offers a framework for connecting intracellular lipid handling with tumor evolution and patient outcome. By classifying HCC according to 122 lipid droplet-associated genes and identifying PLIN3 as a potential driver of malignant behavior, the researchers provide a new view of liver-cancer heterogeneity—one in which the machinery that stores and manages fat may help determine how a tumor grows, spreads, and responds to therapy.
Subject of Research: Lipid droplet-associated genes, metabolic subtypes, prognosis, and therapeutic vulnerabilities in hepatocellular carcinoma
Article Title: Lipid droplet-associated gene signatures classify metabolic subtypes and identify PLIN3 as a key driver in hepatocellular carcinoma
Web References: https://doi.org/10.1016/j.gendis.2026.102067
References: Genes & Diseases. DOI: 10.1016/j.gendis.2026.102067
Image Credits: Huiying Gu, Qiumin Wu, Haibei Zhao, JingLong Du, Zhenzhen Zhang, Juan Chen
Keywords: Hepatocellular carcinoma; lipid droplets; lipid droplet-associated genes; LDAGs; PLIN3; metabolic reprogramming; molecular subtypes; sorafenib; TP53; CTNNB1; precision oncology; liver cancer
Tags: intracellular lipid storage and cancer cell survivallipid droplet dynamics in tumor progressionlipid droplet remodeling in hepatocellular carcinomalipid droplet signatures forlipid droplet-associated genes in liver cancerLipid droplet-based molecular classification of hepatocellular carcinomalipid droplets as biomarkers for liver cancer prognosislipid metabolism and therapeutic targets in liver cancermetabolic heterogeneity in hepatocellular carcinomarole of PLIN3 in liver cancer therapytargeting lipid droplet pathways in liver cancer treatment
