Lysosome-related biomarkers LUM and SERPINA3 for dilated cardiomyopathy: a diagnostic signature linked to immune-cell infiltration
Zhuqing Li, Chunbo Ai, Feng Zhang, C LuAbstract
Background
Dilated cardiomyopathy (DCM) is a major cause of heart failure and sudden death. Lysosomes regulate multiple forms of cell death, but their direct association with DCM remains unclear. We aimed to identify and validate lysosome-related biomarkers for DCM.
Methods
Public left-ventricular transcriptomes were analysed, with GSE57338 as the training dataset and GSE5406 as the external validation dataset. Differentially expressed genes were intersected with lysosome-related genes to identify candidates. Biomarkers were prioritised using protein–protein interaction analysis and machine-learning feature selection, evaluated by receiver operating characteristic analysis, and integrated into a nomogram. Immune-cell infiltration, regulatory network construction, gene set enrichment analysis, in silico drug screening and molecular docking were performed to explore potential mechanisms. Biomarker expression was validated by RT-qPCR in human cardiac tissues.
Results
LUM and SERPINA3 were identified as lysosome-related biomarkers. Across datasets and RT-qPCR validation, LUM was significantly upregulated and SERPINA3 was significantly downregulated in DCM versus controls (P<0.05). Diagnostic performance was high (LUM AUC=0.916 in GSE57338 and 0.965 in GSE5406; SERPINA3 AUC=0.964 in GSE57338 and 0.9536 in GSE5406), and the two-gene nomogram achieved an AUC of 0.978 in the training dataset (GSE57338). SERPINA3 expression correlated positively with central memory CD8 T-cell enrichment scores. A transcription factor (TF)–mRNA–microRNA (miRNA) network implicated 18 TFs and 14 miRNAs linked to LUM. Gene set enrichment analysis identified 29 enriched pathways for LUM and 64 for SERPINA3. In silico screening suggested lurasidone as a candidate SERPINA3-interacting compound, supported by molecular docking.
Conclusion
LUM and SERPINA3 are lysosome-related biomarkers for DCM and support a parsimonious two-gene diagnostic model, providing insight into immune involvement and potential therapeutic exploration.