# Integrated machine learning based on cuproptosis and RNA methylation regulators to explore the molecular model of prostate cancer and provide novel insights to immunotherapy

## TL;DR

This study identifies a new set of biomarkers for prostate cancer using machine learning, which could help predict patient outcomes and guide treatment.

## Contribution

The study introduces a novel 9-gene biomarker model based on cuproptosis and RNA methylation regulators for prostate cancer prognosis and treatment guidance.

## Key findings

- A 9-gene model was developed using machine learning to predict prostate cancer prognosis.
- High-risk patients showed higher tumor mutation burden and lower immune infiltration.
- Two genes, C4orf48 and SLC26A1, were confirmed to be highly expressed in prostate cancer.

## Abstract

Background: As a highly prevalent tumor in males, prostate cancer (PCa) needs newly developed biomarkers to guide prognosis and treatment. However, few researches have elaborated on the function of cuproptosis-associated RNA methylation regulators (CARMRs).

Methods: We identified CARMRs based on single-sample gene set enrichment analysis and weighted gene co-expression network analyses. Subsequently, we performed 10 machine learning algorithms and 101 combinations of them to select the best model in TCGA, GSE70768, GSE70769, and DKFZ cohorts. Furthermore, we explored the potential function of CARMRs in the tumor microenvironment, immunotherapy, and tumor mutation burden (TMB). We validated the expression of the two genes with the largest regression coefficients using qRT-PCR.

Results: In our analysis, we successfully established a consensus prognostic model with 9 CARMRs based on the 101-machine learning framework. Furthermore, functional enrichment analysis revealed different metabolic and signaling pathways in the high- and low-risk groups. Notably, the high-risk group had a higher TMB, a lower level of immune infiltration, and a lower expression of immune checkpoints. Through drug sensitive analysis, we screened chemotherapy drugs suitable for different groups. Vitro experiments illustrated the high expression of C4orf48 and SLC26A1 in PCa compared with normal controls. The discovery was in concordance with bioinformatic analysis results.

Conclusion: A gene signature with 9 CARMRs was developed in our study, which served as biomarkers for PCa. This brings benefits in determining the prognosis of patients with PCa and guiding personalized treatment.