Transcriptome-based profiling of programmed cell death signatures identifies two osteoarthritis subtypes and the hub...

Osteoarthritis (OA) poses a significant global health risk, with programmed cell death (PCD) playing a key role in its development.

This study utilized integrated bioinformatics tools to discover new biomarkers and therapeutic targets for OA. A comprehensive set of 1,254 PCD-related genes encompassing 12 programmed cell death modalities was identified through literature review and pathway analysis. Gene expression data for OA patients and healthy controls were obtained from the GEO database, leading to the classification of patients into two subtypes, Cluster 1 and Cluster 2, based on ssGSEA scores. Immune infiltration analysis and LASSO regression were employed to construct classifiers for these subtypes, followed by validation with external datasets. SVM and RF machine learning techniques were used to identify hub genes. Western blotting and qRT-PCR were performed to validate hub gene expression in OA synovial tissues.

The analysis revealed two OA subtypes with distinct biological processes. A robust classifier was developed, and DNM1L emerged as the sole hub gene linked to PCD in OA. The LASSO classifier achieved an AUC of 1.000 in the training cohort but showed reduced performance in external validation cohorts, indicating the need for further validation. Both immunoblot and qRT-PCR confirmed increased DNM1L expression in OA patients.

This study successfully classified OA into two subtypes and identified DNM1L as a PCD-associated candidate gene that may contribute to OA pathogenesis.