Amyotrophic Lateral Sclerosis (ALS) is a fatal progressive neurodegenerative disease. Each year approximately 5,000 new cases of ALS are diagnosed in the U.S., and the life expectancy from the time of diagnosis is less than five years. Despite decades of research and dozens of clinical trials, only two Food and Drug Administration (FDA)-approved drugs exist for ALS, i.e., Riluzole and Edaravone, and they only have small effects on length of survival and only for a small portion of patients with ALS. Thus, identifying novel and more effective therapeutic options for ALS is critical.

Rationale: An incomplete understanding of the cause of ALS, the absence of reliable animal models, the relatively small number of affected individuals, and the variability in the disease have hampered traditional approaches to identifying drugs to treat this disease. For unknown reasons, Veterans are one and a half times more likely to be affected by ALS than the general population. Our access to the electronic health records (EHRs) of the large cohort of over 13,000 Veterans who have been diagnosed with ALS and received care through the VA Health Care System between 2010 and 2019 offers a unique opportunity to identify prescribed medications that influence (1) the likelihood of developing ALS and/or (2) the rate at which the disease progresses in an individual. While medications will vary among individuals, we will be able to analyze the effects of hundreds of different medications in this large set of patients with ALS. Importantly, recent breakthroughs in machine learning (ML) will allow us to uncover confounding variables and determine whether any observed effects are due to a true causal effect of a given drug.

Objective: The goal of our study is to use a ML-based analysis to determine whether prescribed medications either have an effect on the risk of developing ALS or influence how rapidly the disease progresses. If any medications appear to have a causal effect in our analysis, we will further determine potential mechanisms by which the drug may be acting on the disease process using a newly developed algorithm, PathFX. PathFX analyzes protein networks surrounding a drug's target(s) in a data-driven fashion and uses these networks to discover drug mechanisms. Our aim is to identify currently available medications that prevent ALS or slow progression of the disease and to also identify potential new targets for development of drugs to treat ALS.

Potential Clinical Applications: Our study's focus on prescribed drugs allows repurposing of currently available drugs to the treatment of ALS, based on our results. Because our study includes all patients with a diagnosis of ALS in the large VA Healthcare system, we expect that our findings will apply broadly to all ALS patients. We anticipate that our results will lead to standard controlled clinical trials or even possible adoption of new therapeutic approaches, either limiting drugs that may be harmful or adding drugs that may slow progression of ALS. Our study may also identify drugs that increase long-term risk of developing ALS and should be avoided or drugs that appear to protect individuals from developing ALS. Our efforts are directed toward drugs that are already in clinical use and therefore adopting strategies based on our findings should be low-risk. Importantly, we may identify new targets for treating ALS. For example, multiple drugs appearing to slow progression of ALS with off-target effects that converge on a unique molecular pathway would imply that this pathway is potentially important for ALS and can be subjected to drug discovery studies.