(CSPOT19) Tree-based mining of real-world data for potential safety signals associated with the use of dupilumab in the United States (2017-2023)

Background: Dupilumab is a monoclonal antibody authorized for use in the management of atopic dermatitis, asthma, chronic rhinosinusitis with nasal polyps, eosinophilic esophagitis, and prurigo nodularis in the US. Real-world data on the safety of dupilumab are limited.

Objectives: To conduct a broad screening assessment of the safety of dupilumab using tree-based scan statistics (TBSS).

Methods: We identified new users of dupilumab aged 12+ between March 1, 2017, and June 30, 2023, in 13 sites active in the Sentinel System including national insurers, regional integrated delivery systems, and national Medicare and Medicaid. New users were required to have no evidence of dupilumab in the 183 days prior, no evidence of pregnancy, and continuous enrollment with medical and drug coverage during evaluation windows. Using a self-controlled design, we compared outcome occurrence in the 4 weeks following drug initiation (risk) to a 4-week period preceding initiation (control). We also evaluated temporal clusters of outcomes in the 6 months after initiation (risk). We categorized outcomes in a hierarchical tree based on ICD-10-CM diagnosis codes. The primary analyses evaluated outcomes in inpatient or emergency settings during risk and/or control windows, with no evidence of any diagnosis code beginning with the same 3 characters in any care setting 400 days prior. In sensitivity analyses, we expanded to outcomes in ambulatory settings. We computed TBSS to observe elevated frequencies of outcomes post-treatment using TreeScan. We used a threshold of p <=0.05 to identify statistical alerts.

Results: We observed 89,516 new episodes of dupilumab in the pre/post analysis. The cohort was majority female (56%) with average age 48.7 ± 16.5 years. A majority of the cohort (89.2%) had evidence of any dupilumab indication at baseline with the most common being atopic dermatitis (61.3%). We observed 70,688 new episodes of dupilumab in the tree-temporal cluster analysis with similar characteristics. Primary tree scan analyses identified “poisoning by, adverse effect of and underdosing of other drugs, medicaments and biological substances” (T50.99*) and “acute embolism and thrombosis of deep veins of lower extremity” (I82.4*). Primary tree-temporal scan analyses also identified T50.99* clustered 4-25 days post-initiation. Sensitivity analyses identified elevated occurrences of indications, known risks (eg, conjunctivitis, arthralgias), an open safety signal (psoriasis), and other potential safety signals (eg, cellulitis).

Conclusions: Broad screening in real-world data provides a complementary view of dupilumab’s safety profile. A few statistical alerts require further consideration to determine if they are new safety signals.