Comprehensive Summary
Amyotrophic lateral sclerosis (ALS) is a fatal neurodegenerative disease that destroys motor neurons, resulting in a drastic loss of function and eventual death. In this paper, Wang et al. investigated biomarkers of endoplasmic reticulum (ER) stress, a process linked to neuroinflammation and protein aggregation that correlates with ALS. To identify differentially expressed ER stress-related genes (DE-ERSGs), the authors trained three machine learning (ML) algorithms: the least absolute shrinkage and selection operator (LASSO), support vector machine-recursive feature elimination (SVM-RFE), and random forest (RF) analysis. Using the NCBI Gene Expression Omnibus, two datasets of gene profiles of ALS and non-ALS patients were collected. By intersecting the data from all three algorithms, 6 hub DE-ERSGs that collaborated in promoting ER stress that induced ALS-related symptoms were identified (ABCA1, CKAP4, TOR1AIP1, MMP9, EDC4, and ALPP) and proposed as potential therapeutic targets. Utilizing the Enricher web platform and DSigDB database, Wang et al identified 4 potential drugs (nitroglycerin, diazepam, fenretinide, and adaravone) that may target the DE-ERSGs. Further research with these drugs can pave promising avenues to develop therapies for patients with ALS.
Outcomes and Implications
ALS is a neurodegenerative disease that progresses very swiftly, inflicting respiratory failure, paralysis, and death within five years of its onset. The ML models incorporated into this study provide the potential for physicians to detect biomarkers that warn of ALS before the appearance of symptoms. By examining the pathways by which ER stress can augment protein malformation and lead to degenerative disease, this study opens avenues for therapeutic intervention. Although clinical translation will require further study, the identification of hub genes and candidate drugs provides a roadmap for earlier detection and personalized therapies, establishing a critical step toward more effective ALS management.