AUTHOR=Luo Chao , He Songzhe , Zhang Haibo , He Shuhua , Qi Huan , Wei Anyang TITLE=Clinical and Biological Significance of DNA Methylation-Driven Differentially Expressed Genes in Biochemical Recurrence After Radical Prostatectomy JOURNAL=Frontiers in Genetics VOLUME=13 YEAR=2022 URL=https://www.frontiersin.org/journals/genetics/articles/10.3389/fgene.2022.727307 DOI=10.3389/fgene.2022.727307 ISSN=1664-8021 ABSTRACT=

Background: Biochemical recurrence (BCR) after radical prostatectomy indicates poor prognosis in patients with prostate cancer (PCA). DNA methylation (DNAm) is a critical factor in tumorigenesis and has attracted attention as a biomarker for the diagnosis, treatment, and prognosis of PCA. However, the predictive value of DNAm-derived differentially expressed genes (DMGs) in PCA with BCR remains elusive.

Methods: We filtered the methylated genes and the differentially expressed genes (DGEs) for more than 1,000 clinical samples from the TCGA cohort using the chAMP and DESeq2 packages of R language, respectively. Next, we integrated the DNAm beta value and gene expression data with the Mithymix package of R language to obtain the DMGs. Then, 1,000 times Cox LASSO regression with 10-fold cross validation was performed to screen signature DMGs and establish a predictive classifier. Univariate and multivariate cox regressive analyses were used to identify the prognostic factors to build a predictive model, and its performance was measured by receiver operating characteristic, calibration curves, and Harrell’s concordance index (C-index). Additionally, a GEO dataset was used to validate the prognostic classifier.

Results: One hundred DMGs were mined using the chAMP and Methymix packages of R language. Of these, seven DMGs (CCK, CD38, CYP27A1, EID3, HABP2, LRRC4, and LY6G6D) were identified to build the prognostic classifier (Classifier) through LASSO analysis. Moreover, univariate and multivariate Cox regression analysis determined that the Classifier and pathological T stage (pathological_T) were independent predictors of BCR (hazard ratio (HR 2.2), (95% CI 1.4–3.5), p < 0.0012, and (HR 1.8), (95% CI 1.0–3.2), p < 0.046). A nomogram based on the Classifier was constructed, with high prediction accuracy for BCR-free survival in TCGA and GEO datasets. GSEA enrichment analysis showed that the DMGs were mainly enriched in the metabolism pathways.

Conclusion: We identified and validated the nomogram of BCR-free survival for PCA patients, which has the potential to guide treatment decisions for patients at differing risks of BCR. Our study deepens the understanding of DMGs in the pathogenesis of PCA.