UQCR10 and TNNT1: novel biomarkers for sarcopenia identified through integrated transcriptomic analysis and machine learning.

BACKGROUND: Sarcopenia, the age-related loss of muscle mass and function, significantly impacts health outcomes in older adults. Despite its prevalence, molecular diagnostics and targeted therapies remain limited due to incomplete understanding of its pathophysiology. METHODS: We analyzed the GSE226151 dataset comparing skeletal muscle transcriptomes from sarcopenic (n = 20) and healthy adults (n = 20) using differential expression analysis and weighted gene co-expression network analysis (WGCNA). Key genes were identified through the intersection of differentially expressed genes and hub genes. Machine learning feature selection (LASSO, XGBoost, and Random Forest) was employed to identify optimal biomarkers, followed by validation in an independent cohort (GSE111016, n = 40, 20 sarcopenia vs. 20 healthy controls). Comprehensive functional enrichment analysis was performed using Gene Ontology, KEGG, and Reactome databases. We employed fivefold cross-validation to account for modest sample size, with performance metrics reported with 95% confidence intervals obtained via bootstrapping (1000 iterations). RESULTS: Integration of differential expression analysis (97 DEGs) and WGCNA (269 hub genes from the darkolivegreen module) identified five key genes in sarcopenia: COX6C, ETFB, TNFAIP3, TNNT1, and UQCR10. Multi-method machine learning feature selection consistently ranked UQCR10 and TNNT1 as the most important biomarkers. A two-biomarker panel (UQCR10 + TNNT1) achieved superior performance with Random Forest modeling (AUC = 0.738 [95% CI 0.559-0.875], PRAUC = 0.643 [95% CI 0.405-0.832], sensitivity = 0.850, specificity = 0.600) in the validation cohort, substantially outperforming single-biomarker approaches. Three-group analysis (Healthy/Pre-sarcopenia/Sarcopenia) revealed continuous dose-response patterns: all five genes showed significant linear trends across disease progression (p < 0.005), with UQCR10 (mitochondrial) and TNNT1 (contractile) exhibiting strongest relationships (R(2) = 0.281 and 0.192 respectively), suggesting continuous pathophysiological progression rather than discrete disease stages. Functional enrichment analysis revealed three major dysregulated biological processes: energy metabolism and mitochondrial function (UQCR10, COX6C, ETFB), muscle structure and contractile function (TNNT1), and inflammatory response regulation (TNFAIP3). CONCLUSIONS: Our integrated transcriptomic approach identified a promising two-biomarker panel (UQCR10 + TNNT1) for sarcopenia detection and elucidated key molecular mechanisms underlying the condition. While these findings are hypothesis-generating and require validation in larger, more diverse cohorts, they provide mechanistic insights and candidate biomarkers warranting further development. The identified genes and pathways offer potential targets for novel therapeutic interventions aimed at preserving muscle health during aging.