Related Experiment Videos
A Metabolic Cluster-Derived Risk Model for Prognostic Stratification, Immunometabolic Profiling, and Therapeutic
Wensi Liu1, Cong Wang2, Tao Su1
1Key Laboratory of Endocrine Glucose & Lipids Metabolism and Brain Aging, Ministry of Education, Institute of Laboratory Animal Models, Shandong Provincial Hospital Affiliated to Shandong First Medical University, Science and Technology Innovation Center, Shandong First Medical University & Shandong Academy of Medical Sciences, Jinan, Shandong, People's Republic of China.
Purpose:
Metabolic reprogramming is a hallmark of prostate cancer; however, the clinical utility of metabolism-related genes for prognostic stratification remains limited.
Patients And Methods:
Prostate cancer patients from The Cancer Genome Atlas (TCGA) and Gene Expression Omnibus (GEO) databases were stratified into metabolism-associated clusters based on metabolism-related gene expression. Genes differentially expressed between clusters were used to construct a prognostic risk model via least absolute shrinkage and selection operator (LASSO) regression. Associations with clinicopathological features, genomic alterations, metabolic activity, immune status, and therapeutic response were systematically evaluated. Computational analyses included tumor immune dysfunction and exclusion (TIDE) and the immunophenoscore for predicting immunotherapy response, pRRophetic and CellMiner for assessing drug sensitivity, and the Connectivity Map for screening candidate compounds. Exploratory assessments were performed using immunohistochemistry (IHC) in a 20-patient cohort with follow-up data and quantitative polymerase chain reaction (qPCR) in an independent 14-patient cohort.
Results:
Two metabolism-associated clusters with distinct survival outcomes and molecular features were identified. A three-gene signature stratified patients into high- and low-risk groups, with the high-risk group showing poorer survival, greater tumor aggressiveness, elevated tumor mutational burden and microsatellite instability, and enhanced metabolic pathway activity. This group also exhibited an immunosuppressive microenvironment and was predicted to have reduced responsiveness to some immunotherapies and endocrine therapies, along with increased resistance to specific chemotherapeutic agents. In silico analyses suggested several candidate small molecules for further evaluation in the high-risk group. The IHC- and qPCR-based exploratory experiments showed associations between the derived scores and clinicopathological features, providing preliminary support for the transcriptomic signature.
Conclusion:
This metabolic cluster-derived three-gene model may support prognostic stratification and characterization of immunometabolic heterogeneity in prostate cancer. The observed associations between metabolic heterogeneity and computationally predicted therapeutic sensitivity provide a rationale for further investigation of potential treatment strategies.