Research Paper Volume 16, Issue 20 pp 13076—13103
A lactate metabolism-related gene signature to diagnose osteoarthritis based on machine learning combined with experimental validation
- 1 Department of Pain Medicine, Yuebei People’s Hospital, Wujiang, Shaoguan 512000, Guangdong Province, China
- 2 Department of Traditional Chinese Orthopedics and Traumatology, Yuebei People’s Hospital, Wujiang, Shaoguan 512000, Guangdong Province, China
- 3 Department of Pediatric Orthopedics, Guangzhou Women and Children’s Medical Center, Tianhe, Guangzhou 510623, Guangdong Province, China
Received: October 10, 2023 Accepted: March 18, 2024 Published: October 16, 2024
https://doi.org/10.18632/aging.205873How to Cite
Copyright: © 2024 Yang et al. This is an open access article distributed under the terms of the Creative Commons Attribution License (CC BY 4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Abstract
Background: Lactate is gradually proved as the essential regulator in intercellular signal transduction, energy metabolism reprogramming, and histone modification. This study aims to clarify the diagnosis value of lactate metabolism-related genes in osteoarthritis (OA).
Methods: Lactate metabolism-related genes were retrieved from the MSigDB. GSE51588 was downloaded from the Gene Expression Omnibus (GEO) as the training dataset. GSE114007, GSE117999, and GSE82107 datasets were adopted for external validation. Genomic difference detection, protein-protein interaction network analysis, LASSO, SVM-RFE, Boruta, and univariate logistic regression (LR) analyses were used for feature selection. Multivariate LR, Random Forest (RF), Support Vector Machine (SVM), and XGBoost (XGB) were used to develop the multiple-gene diagnosis models. 12 control and 12 OA samples were collected from the local hospital for re-verification. The transfection assays were conducted to explore the regulatory ability of the gene to the apoptosis and vitality of chondrocytes.
Results: Through the bioinformatical analyses and machine learning algorithms, SLC2A1 and NDUFB9 of the 273 lactate metabolism-related genes were identified as the significant diagnosis biomarkers. The LR, RF, SVM, and XGB models performed impressively in the cohorts (AUC > 0.7). The local clinical samples indicated that SLC2A1 and NDUFB9 were both down-regulated in the OA samples (both P < 0.05). The knockdown of NDUFB9 inhibited the viability and promoted the apoptosis of the CHON-001 cells treated with IL-1beta (both P < 0.05).
Conclusions: A lactate metabolism-related gene signature was constructed to diagnose OA, which was validated in multiple independent cohorts, local clinical samples, and cellular functional experiments.