Assessment of Use and Safety of Edaravone for Amyotrophic Lateral Sclerosis in the Veterans Affairs Health Care System

Question What is the real-world experience with edaravone in patients with amyotrophic lateral sclerosis (ALS) within a national integrated health care system? Findings In this cohort study of data from 369 US veterans with documented or probable ALS, a significantly greater proportion of acute all-cause hospitalizations was associated with edaravone treatment and, among users receiving at least 6 months of treatment, an increased likelihood of ALS-related hospitalization was associated with edaravone treatment compared with riluzole-only treatment. Meaning While these findings should be interpreted with caution, in this evaluation, edaravone (mostly used with riluzole) was associated with more hospitalizations compared with riluzole-only therapy; more evidence is needed to evaluate edaravone treatment outcomes in real-world settings. This cohort study uses real-world data to assess patient characteristics and use as well as the safety and effectiveness of edaravone among patients with amyotrophic lateral sclerosis in the US Department of Veterans Affairs Health Care System. Importance Using real-world data, the US Department of Veterans Affairs (VA) initiated a surveillance evaluation of edaravone after its approval for amyotrophic lateral sclerosis (ALS) in 2017. The use and safety of edaravone for patients with ALS in the VA health care system remain to be assessed. Objective To describe a pharmacovigilance surveillance initiative with edaravone to monitor patient characteristics, utilization (edaravone cycles and riluzole use), and safety and to evaluate safety/effectiveness. Design, Setting, and Participants This propensity score-matched cohort study used data on 369 patients with documented definite or probable ALS in the Veterans Health Administration (VHA) with at least 1 prescription for edaravone between August 1, 2017, and September 30, 2019. The analysis compared edaravone (alone or with riluzole) with riluzole only. For chronic users (>= 6 months of drug), a time-to-event model evaluated ALS-related outcomes, with censoring at outcome, death, or end of evaluation. Patients with Parkinson disease, dementia, schizophrenia, or significant respiratory insufficiency per diagnosis codes within 2 years before prescription initiation were excluded. In overall matched cohorts, 223 patients treated with edaravone were 1:3 propensity score matched based on predefined confounders. For the chronic user subgroup analysis, 96 patients receiving edaravone and 424 patients receiving riluzole only were included. Exposures Edaravone (alone or with riluzole) vs riluzole only. Main Outcomes and Measures Patient characteristics, ALS drug use, and mortality. Acute outcomes (within 6 months of index) included proportion and mean time to event for death, discontinuation, or all-cause hospitalization, and outcomes for chronic users (receiving >6 months of treatment) included hazard ratios of outcomes related to disease-state progression. Results Of 369 patients who received edaravone, most were older (mean [SD] age, 64.6 [11.3] years), male (346 [93.8%]), and White (261 [70.7%]). As of September 2019, 59.9% of edaravone patients had discontinued treatment; of those, 49.5% (108 of 218) received only 1 to 3 treatment cycles. Approximately 30% (110 patients) died. In a matched evaluation, significantly more acute all-cause hospitalization events occurred with edaravone (35.4% vs 22.0% for riluzole only); 72.6% of the edaravone cohort received edaravone with riluzole. Among chronic users, edaravone patients (70.8% edaravone with riluzole) had an increased hazard ratio of ALS-associated hospitalization (2.51; 95% CI, 1.18-8.16). The death rate was lower with edaravone but the difference was not statistically significant. Conclusions and Relevance Early edaravone discontinuation was common in the VA. Although outcomes favored use of riluzole only in the matched analysis, results should be interpreted with caution, as unmeasured bias in observational data is likely.