Identification of diagnostic biomarkers in prostate cancer-related fatigue by construction of predictive models and experimental validation.

Background: Cancer-related fatigue (CRF) is a prominent cancer-related complication occurring in Prostate cancer (PCa) patients, profoundly affecting prognosis. The lack of diagnostic criteria and biomarkers hampers the management of CRF.

Methods: The CRF-related data and PCa single-cell data were retrieved from the GEO database and clinical data was downloaded from the TCGA database. The univariate logistic/Cox regression analysis were used to construct the prediction models. The predictive value of models was analyzed using the ROC curve and Kaplan-Meier survival. The hub genes were screened by an intersection analysis of DEGs. The mice model of PCa and PCa-related fatigue were established, and fatigue-like behaviors of mice were detected. The expression of selected hub genes was validated by RT-PCR and IHC analysis.

Results: The diagnosis and risk models showed great predictive value both in the training and validation dataset. Five genes (Baiap2l2, Cacng4, Sytl2, Sec31b and Ms4a1) that enriched the CXCL signaling were identified as hub genes. Among all hub genes, the MS4A1 expression is the most significant in PCa-related fatigue mice.

Conclusions: We identified MS4A1 as a promising biomarker for the diagnosis of PCa-related fatigue. Our findings would lay a foundation for revealing the pathogenesis and developing therapies for PCa-related fatigue.