Screening of mitochondrial-related biomarkers connected with immune infiltration for acute respiratory distress syndrome through WGCNA and machine learning.

Summary

Using WGCNA and multiple machine-learning algorithms across public datasets, the authors identified five mitochondria-related genes upregulated in sepsis-induced ARDS, built a diagnostic nomogram with good internal performance, and linked the signature to increased phenylalanine metabolism. In silico drug predictions suggested chlorzoxazone, ajmaline, and clindamycin as potential modulators.

Key Findings

• Three immune cell types (macrophages, neutrophils, monocytes) differed significantly between sepsis alone and sepsis-induced ARDS.
• Five mitochondria-related biomarkers were upregulated in ARDS and formed a diagnostic signature with a nomogram showing good internal performance.
• Gene set enrichment linked the signature to increased phenylalanine metabolism; in silico screening suggested chlorzoxazone, ajmaline, and clindamycin as candidate drugs.

Clinical Implications

Not yet ready for clinical use, but may guide development of blood-based diagnostic panels and stratified trials in sepsis-induced ARDS; highlights phenylalanine metabolism as a potential pathway target.

Limitations

• Lack of external prospective validation and clinical utility testing
• Retrospective, in silico design susceptible to batch effects and confounding; causal mechanisms not established

Future Directions

Prospective validation of the 5-gene panel in multi-center sepsis cohorts; mechanistic studies on mitochondrial-immune interactions and phenylalanine metabolism; preclinical testing of candidate drugs.