(044) Utilizing Real-World Data to Create an External Control Arm for a Phase 3 Trial of Deucravacitinib for Treatment of Plaque Psoriasis

Background: Continued effort to use real-world evidence to accelerate new drug development has been made. The US Food and Drug Administration (FDA) has released guidance on how real-world data (RWD) will be considered in regulatory decision making. One such application is the use of RWD in clinical trials to create external control arms (ECAs).

Objectives: To use RWD to create an ECA for the Phase 3 POETYK PSO-1 study of deucravacitinib versus apremilast vs placebo for treatment of moderate-to-severe plaque psoriasis (PsO) and to compare outcomes in the ECA to results observed in the placebo arm.

Methods: Electronic health record data (2017-2024) from 6 specialty dermatology networks and 6 integrated delivery networks in the OMNY Health RWD platform was accessed. The POETYK PSO-1 study was a randomized active comparator-controlled trial with two investigational treatment arms and one placebo arm (n = 166). Coprimary efficacy endpoints were achievement of Psoriasis Area and Severity Index (PASI) of 75 and physician global assessment (PGA) of 0 or 1 at week 16. Patients meeting the available eligibility criteria from the trial were included in the ECA. Only the PGA endpoint was evaluated because of low availability of PASI in the real-world setting. The first visit where the patient met the severity definition was used. Topical medications were used to exclude patients in the 2 weeks prior to the index severity visit; patients could receive topicals in the 16-week study period. Percentages of patients achieving PGA of 0 or 1 at week 16 were computed overall and by various patient characteristics.

Results: A total of 969 patients were included. The ECA and trial placebo arm had similar distributions in weight (mean: 91.3 vs 89.1 kg), BMI (mean: 31.3 vs 30.2 kg/m2), PGA (moderate/severe: 78.2%/21.8% vs 77.1%/22.3%), and body surface area (mean: 23.7 vs 25.3). On average, ECA patients were older (mean age: 54.9 vs. 47.9 years) and more likely to be white (88% vs 77%). At the end of the study period, 18.5% of patients in the ECA had achieved a PGA of 0 or 1, compared to 7.2% of patients in the placebo arm. Stratifying by previous biologic experience, weight (≥ 90 kg, < 90 kg), gender, and previous systemic experience did not result in notable changes in the study endpoint. Among patients with race data in the ECA, 20.1% of white patients achieved a PGA of 0/1 compared to 9.3% of nonwhite patients.

Conclusions: While many baseline characteristics were similar between the ECA and placebo arm, the proportion of patients achieving the primary endpoint in the ECA was approximately twice that of the placebo arm. Differences in racial distribution may account for part of the discrepancy. Future research may include understanding other differences in these patient populations as they relate to the study endpoint.