# Integrated machine learning identifies disulfidptosis-related and ferroptosis-related genes to evaluate survival prognosis and treatment efficacy in kidney renal clear cell carcinoma

## TL;DR

This study combines two cell death pathways to build a new model for predicting survival and treatment response in kidney cancer.

## Contribution

First integrated model using disulfidptosis and ferroptosis genes for kidney cancer prognosis and treatment guidance.

## Key findings

- A 5-gene signature and nomogram improved survival prediction in kidney cancer patients.
- High-risk patients showed distinct immune profiles and drug sensitivity patterns.
- RT-qPCR confirmed the expression of key genes in kidney cancer cell lines.

## Abstract

Ferroptosis and disulfidptosis, two programmed cell death pathways, critically drive tumor growth by affecting metastasis. Although the prognostic value of disulfidptosis and ferroptosis had been separately validated in kidney renal clear cell carcinoma (KIRC), prognostic effect of integrating two programmed death genes remains unclear in KIRC. Our objective is to establish an innovative prognostic model for KIRC.

We sourced KIRC patients’ information that contains clinical and genomic from The Cancer Genome Atlas (TCGA) database. We selected disulfidptosis-related and ferroptosis-related genes (DRFs) to construct a prognostic model. By combining clinical features and prognostic models, we developed the nomogram. Additionally, the mechanism of DRF was explored in KIRC, including tumor immune dysfunction and exclusion (TIDE), Kaplan-Meier (K-M) analysis, tumor microenvironment (TME) analysis, and more. Drug sensitivity analysis shows which drugs are sensitive to tumors. Experiment with RT-PCR to confirm DRFs gene expression in the cell line.

Constructing risk score with five DRFs, all tumor samples were categorized into high-risk group (HG) and low-risk group (LG). The HG samples demonstrated lower survival rates according to K-M survival curves. The nomogram with risk score demonstrated significant predictive value than nomogram without the risk score. TME analysis indicated that the proportion of T cells follicular helper and Tregs was higher in HG, while Macrophages M1 and Mast cells resting were higher in LG. GSEA analysis demonstrated Retinol metabolism pathway, drug metabolism other enzymes pathway, etc. were enriched in HG, while endocytosis-related pathway, neurotrophin signaling pathway, etc. were enriched in LG. TIDE analysis showed tumors in HG are more prone to immune evasion. The drug sensitivity analysis indicated that the HG is sensitive to antitumor drugs such as Cedrane and Osimertinib, while the LG is sensitive to antitumor drugs such as 5-Fluorouracil and Entinostat. RT-qPCR have confirmed expression of DRFs in KIRC cell lines.

Our DRFs-based prognostic model and nomogram effectively predict survival and guide treatment decisions.

•First disulfidptosis/ferroptosis integrated prognostic model for KIRC.•5-gene signature and nomogram optimize survival prediction.•Risk stratification reveals TMB-immune interplay and divergent drug sensitivity.

First disulfidptosis/ferroptosis integrated prognostic model for KIRC.

5-gene signature and nomogram optimize survival prediction.

Risk stratification reveals TMB-immune interplay and divergent drug sensitivity.