Résumé
Multiple sclerosis’ (MS) treatment evolution in the past three decades has raised concerns regarding liver injury (LI) associated with disease-modifying therapies. This study aims to identify high-risk profiles for LI among MS patients by using an artificial intelligence-based subgroup discovery algorithm.
This retrospective cohort study utilized Optum Market Clarity electronic health record (EHR) data. Adult patients with first known MS diagnosis between January 1st, 2015, and December 31st, 2019 were included. LI event was defined either as meeting the international lab criteria or the presence of an acute LI diagnosis. Q-Finder, a supervised non-parametric subgroup discovery algorithm, was used to identify subgroups with increased LI risk.
The analysis involved 5986 MS patients, of whom 233 developed a LI. The Q-Finder identified 6 risk factors associated with elevated levels of liver enzymes and 11 clinical risk factors unrelated to liver enzymes summarized into 6 themes: substance abuse and mental health disorders, anaemia, anaemia and kidney, immune disorders, male patients and diabetes, and chronic obstructive pulmonary disease.
The Q-Finder allowed the discovery of new risk factors that may help predict MS patients at higher risk of LI and provides insights to improve patient screening criteria for clinical trials.