Abstract 1110: Longitudinal and multi-regional analysis of cancer phenotypes in hepatocellular carcinoma and intrahepatic cholangiocarcinoma

Cancer cells exhibit abnormal metabolic behavior, a feature that allows cancer cells to utilize different metabolic pathways and substrates (such as glucose, fatty acids, and glutamine) to fulfill their energy and biosynthetic requirements. This study aims to explore the spatial and temporal variations in cancer phenotypes in hepatocellular carcinoma (HCC) and intrahepatic cholangiocarcinoma (iCCA) using single-cell RNA sequencing data. We utilized single-cell RNA sequencing data from the Sequential NCI-CLARITY cohort (GSE229772). Tumor biopsies were collected longitudinally. We also used the Multi-Regional Cohort (GSE189903), where single cells were isolated from multiple regions (tumor cores, borders, and adjacent normal tissues). Gene signatures in the anabolic and catabolic axis were applied to classify cells into four distinct metabolic clusters: O, W, W/O, and Q. To predict metabolic fluxome and variations, we used FLUXestimator, an online platform that implements single cell flux estimation analysis (scFEA). This unsupervised approach uses a neural network architecture to estimate reaction rates from transcriptomics data. Additionally, pathway enrichment analysis was conducted to corroborate our findings. Our analysis identified the four distinct metabolic clusters across both cohorts. Briefly, the O cluster is characterized by high oxidative phosphorylation and is associated with better clinical outcomes. The W cluster is defined by high glycolytic activity and is linked to aggressive tumor behavior. The W/O cluster represents a hybrid state with both high glycolytic and oxidative activities, contributing to metabolic flexibility and tumor aggressiveness. The Q cluster relies heavily on glutamine metabolism and is associated with poor prognosis and drug resistance. Normal liver cells were predominantly grouped into the O cluster, with hepatocytes displaying low HIF1, AMPK, and MYC activity, indicative of normal metabolic function. Similarly, border cancer cells were conformed in a distinct cluster, characterized by a low HIF1 score, but a higher AMPK score. Temporal analysis from the Sequential NCI-CLARITY cohort revealed dynamic changes in the distribution of these clusters over time, reflecting the evolution of cancer phenotypes. FLUXestimator predictions and pathway enrichment analysis indicated significant changes in metabolic pathways, supporting the observed phenotypic shifts. This approach validates our comprehensive network model and suggests a method for developing new metabolic networks for cancer modeling based on key features observed during cancer progression. This study reveals metabolic diversity in HCC and iCCA tumors, identifying clusters linked to tumor progression stages. These phenotypic traits may serve as a potential biomarker for better clinical outcomes. Citation Format: Javier Villela Castrejon, Dongya Jia, Jason T. George. Longitudinal and multi-regional analysis of cancer phenotypes in hepatocellular carcinoma and intrahepatic cholangiocarcinoma [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2025; Part 1 (Regular Abstracts); 2025 Apr 25-30; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2025;85(8_Suppl_1):Abstract nr 1110.