Multi-omics biomarkers aid prostate cancer prognostication

Abstract

Effective biomarkers and diagnostic tools are urgently needed in clinical settings for improved management of prostate cancer patients, especially to reduce over-treatment of indolent tumors and for early identification of aggressive disease. Gene expression signatures are currently the "gold standard" to provide guide clinical decision, however their clinical utility and interpretability is questionable. Multi-modal molecular profiling provides an holistic approach to systematically unravel the biological complexity underlying cancer pathogenesis, hence biomarkers developed using such an integrated approach hold the potential to more accurately capture cancer-driving alterations than signatures based on a single omics modality. Currently, however, robust and reproducible multi-omics biomarkers are still lacking for prostate cancer. In this study, we analyzed transcriptomics and metabolomics profiles jointly in a prostate cancer cohort and identified two prognostic signatures with high statistical powers (signature1: EGLN3, succinate, trans-4-hydroxyprolin; and signature2: IL6, SLC22A2, histamine). Our approach leveraged a priori biological knowledge of the cellular metabolism and gene circuitry, enabling the identification of dysregulated network modules. Functional bioinformatics analyses suggest that these signatures can capture relevant molecular alterations in prostate cancer tissues, including dysregulations of cellular signaling, cell cycle progression, and immune system modulation, stratifying patients in distinct risk groups. Next, we trained two gene expression signatures as a proxy for the multi-omics ones, extending our investigation to publicly available data, further confirming their prognostic values in independent patient cohorts. In summary, the analysis of multi-modal molecular grounded in cellular network biology represents a promising approach for the development of robust prognostic biomarkers of detecting and discriminating high grade disease.

MultiNEP: a multi-omics network enhancement framework for prioritizing disease genes and metabolites simultaneously

“…We applied MultiNEP and competing methods to prioritize candidate disease-associated gene expressions and metabolites for two human cancers, the Dana-Farber Cancer Institute/Harvard Cancer Center (DF/HCC) Prostate Cancer Cohort ( Oh et al 2006 , Xu et al 2022 ), and the GEO Breast Cancer cohort (GSE37751) ( Terunuma et al 2014 ).…”

“…Available information in the DF/HCC Prostate Cancer Cohort are described in Oh et al (2006) . We used 94 tumor and 48 normal-adjacent tissue samples (with 48 with tumor and normal-adjacent pairs) with gene expression and metabolite abundance data as described in ( Xu et al 2022 ). We constructed the disease similarity matrix E with correlations among 18 009 gene expressions and 203 metabolites using omics profiles of 94 tumors and 48 normal-adjacent samples, and calculated disease association scores v using paired t-test with 48 matched tumor and normal-adjacent pairs.…”

“…DF/HCC prostate cancer cohort data used in this article is available at ( https://github.com/Karenxzr/MultiModalPC ) from Xu et al (2022) . The breast cancer cohort data used in this article can be downloaded from ( https://www.ncbi.nlm.nih.gov/geo/ ) (accesion number: GSE37751) and supplementary data from Terunuma et al (2014) .…”

“…MultiNEP outperforms competing methods and prioritizes more disease-associated genes and metabolites simultaneously using both within- and between-omics interactions in all simulation scenarios with appropriate values of the weighting parameters. In real data applications to prioritize candidate cancer-associated genes and metabolites using data from the Dana-Farber/Harvard Cancer Center (DF/HCC) Prostate Cancer Cohort ( Xu et al 2022 ) and the GSE37751 Breast Cancer cohort ( Terunuma et al 2014 ), MultiNEP consistently identifies more disease-related genes according to the DisGeNET ( Piñero et al 2017 ) database.…”

Advancements in MRI-Based Radiomics and Artificial Intelligence for Prostate Cancer: A Comprehensive Review and Future Prospects

“…As reported, we observe that the results of multi-omics studies are superior to the single omics, multi-omics are very extensive, and the specific methods are also quite different. For example, multimodal molecular analysis based on cell network biology provides robust prognostic biomarkers to detect and identify high-level diseases [139]. While in other studies, multi-omics analysis, integrating genomics, methylomics, and transcriptomics are used to assess the risk correlation between DNA methylation and PCa [140].…”