Use of a Bioinformatics‐Based Toxicity Scoring System to Assess Serotonin Burden and Predict Population‐Level Adverse Drug Events from Concomitant Serotonergic Drug Therapy

Study Objective Numerous medications interact at serotonin (5‐hydroxytryptamine [5‐ HT ]) receptors directly or through off‐target interactions, causing mild to severe serotonergic adverse drug events ( ADE s), particularly among older adults. Our objective was to develop a novel molecular‐based toxicity scoring system to assess serotonergic burden resulting from concurrently administered drugs. Quantitative methods to assess serotonergic burden may provide a useful clinical tool for improving pharmacotherapy. Design Retrospective cohort study. Data Sources PharMetrics Legacy health claims database (January 2001–December 2013) and Ch EMBL bioactivity database. Patients A 2–serotonergic drug exposure cohort (78,172 patients) and a 3–serotonergic drug exposure cohort (19,900 patients) were generated, and population‐level statistics were collected. Nonexposure cohorts were created for each drug exposure cohort and matched in a 4:1 ratio for age, sex, and length of enrollment. Measurements and Main Results Eight 5‐ HT medications were screened against multiple bioactivity databases to identify their off‐target interactions at 5‐ HT receptors and serotonin reuptake transporter protein. A computational serotonin burden score ( SBS ) was derived from the receptor‐specific interaction propensities reported from the comprehensive bioactivity screen. Linear regression was used to characterize associations between SBS s and combined total ADE incidence rate detected by International Classification of Diseases, Ninth Revision, Clinical Modification, diagnosis codes. A significantly greater incidence of 17 potential 5‐ HT– related ADE s was seen in exposed serotonergic drug cohorts (p<0.05). A positive correlation between SBS and overall ADE incidence rate in the 2–serotonergic drug exposure cohort ( R 2 = 0.69, p<0.34) and 3‐drug cohort ( R 2 = 0.85, p<0.01) was observed. When both drug cohorts were combined, total drug SBS s strongly correlated with the composite 5‐ HT adverse event rate ( R 2 = 0.92, p<0.0001). Despite an increasing burden of illness, these data suggest that drug combinations with higher SBS s are associated with a higher rate of potential serotonergic ADE s. Conclusion In this test of concept, positive associations between SBS s and serotonin‐related ADE s suggest that it may offer a pharmacologic‐based foundation for developing risk assessment tools to assist in optimizing pharmacotherapy.