Discovering and Validating Cuproptosis-Associated Marker Genes for Accurate Keloid Diagnosis Through Multiple Machine Learning Models

Abstract

Background

Keloid is a common condition characterized by abnormal scarring of the skin, affecting a significant number of individuals worldwide.

Objective

The occurrence of keloids may be related to the reduction of cell death. Recently, a new cell death mode that relies on copper ions has been discovered. This study aimed to identify novel cuproptosis-related genes that are associated with keloid diagnosis.

Methods

We utilized several gene expression datasets, including GSE44270 and GSE145725 as the training group, and GSE7890, GSE92566, and GSE121618 as the testing group. We integrated machine learning models (SVM, RF, GLM, and XGB) to identify 10 cuproptosis-related genes (CRGs) for keloid diagnosis in the training group. The diagnostic capability of the identified CRGs was validated using independent datasets, RT-qPCR, Western blotting, and IHC analysis.

Results

Our study successfully categorized keloid samples into two clusters based on the expression of cuproptosis-related genes. Utilizing WGCNA analysis, we identified 110 candidate genes associated with cuproptosis. Subsequent functional enrichment analysis results revealed that these genes may play a regulatory role in cell growth within keloid tissue through the MAPK pathway. By integrating machine learning models, we identified CRGs that can be used for diagnosing keloid. The diagnostic efficacy of CRGs was confirmed using independent datasets, RT-qPCR, Western blotting, and IHC analysis. GSVA analysis indicated that high expression of CRGs influenced the gene set related to ECM receptor interaction.

Conclusion

This study identified 10 cuproptosis-related genes that provide insights into the molecular mechanisms underlying keloid development and may have implications for the development of targeted therapies.

Graphical Abstract

Keywords:

• machine learning
• cuproptosis
• keloid
• novel biomarker

Data and Code Availability

The data and code used for analyses in this study can be found at https://github.com/wxmm20230126/Computational-Framework/blob/main/Public%20code_keloid.R.

Ethical Approval and Consent to Participate

The work was approved by the Clinical Research Ethics Committee of Huizhou First Hospital. Informed consent forms are not required for patient data extracted from public databases.

Acknowledgments

We are grateful to everyone who provides and builds public data.

Informed Consent Statement

Informed consent was received from each patient participating in the study.

Consent for Publication

All authors gave consent to publish.

Disclosure

The authors declare that there are no conflicts of interest regarding the publication of this study.