Integrative miRNA networks predict prostate cancer recurrence

ABSTRACT: Developing an effective predictor of tumour recurrence is an important challenge in the management of localized prostate cancer. Accurate prediction of disease-free survival is essential for decision-making before or after definitive local therapy, and would allow identification of patients who may derive benefit from adjuvant therapy. Today these decisions are largely based on the clinic-pathologic prognostic factors of tumour size, extent and grade along with serum PSA abundance. MicroRNAs (miRNAs) are promising class of biomarkers to improve patient risk-stratification: they show differential abundance across prostate cancer sub-types and have both oncogenic and tumour-suppressive roles. Here, we report the miRNA abundance profiling of 319 prostate tumours with matching detailed clinical annotation, long-term follow-up and accompanying genetic and epigenetic data including mRNA abundance, copy number alterations, point mutations and methylation. We built a cancer driver-miRNA-target regulatory network to characterize the extent to which genomic, transcriptional and post-transcriptional events contribute to the miRNA abundance architecture. By linking this network with machine-learning we created a multi-modal biomarker that accurately predicts biochemical recurrence after definitive local therapy. We find that combining miRNA information with genetic and epigenetic drivers and clinicopathological parameters improves biomarker accuracy, and quantifies the value of multi-modal biomarkers that exploit the regulatory architecture of gene expression in their development.

ORGANISM(S): Homo sapiens

PROVIDER: GSE135535 | GEO | 2024/04/05

REPOSITORIES: GEO