Comparing Efficacy of Guselkumab versus Ustekinumab in Patients with Psoriatic Arthritis: An Adjusted Comparison Using Individual Patient Data from the DISCOVER and PSUMMIT Trials

Included studies were of similar trial design: DISCOVER-1, DISCOVER-2, PSUMMIT 1, and PSUMMIT 2. Study overviews and treatment doses are presented in the Supplementary Material.

A flow diagram outlining the analytical procedures used is presented in Supplementary Figure S1. The primary comparative analyses between guselkumab 100 mg Q8W and 100 mg Q4W versus ustekinumab (45 and 90 mg) were performed for biologic-naïve (n = 1170) and biologic-experienced (n = 197) populations separately. Patient-level data for biologic-naïve patients were pooled for all active treatment arms from DISCOVER-1 and DISCOVER-2 and from PSUMMIT 1 and PSUMMIT 2, and IPD for biologic-experienced patients were from DISCOVER-1 and PSUMMIT-2. Pooled data included baseline characteristics and ACR 20 and PASI 90 outcomes. Notably, data for ustekinumab 45 mg and ustekinumab 90 mg from PSUMMIT 1 and PSUMMIT 2 were combined into ustekinumab 45/90 mg given that the weight-based dosing for ustekinumab is based on the specific pharmacokinetics of the molecule (e.g., drug absorption, distribution, metabolism). Two scenario analyses were also conducted to explore the separate ustekinumab doses at weeks 16, 24, and 52 versus both guselkumab doses. Scenario 1 examined only ustekinumab 45 mg as it is the recommended dose by the Food and Drug Administration (FDA) and the European Medicines Agency (EMA) for patients without concomitant moderate-to-severe psoriasis. Scenario 2 examined the recommended dose of ustekinumab 45 mg for those with a body weight ≤ 100 kg and ustekinumab 90 mg only for patients with a body weight > 100 kg, as per the FDA and EMA label. Guselkumab data for the Q8W and Q4W doses were not combined because the dose options are not related to pharmacokinetics; instead, the dose is prescribed dependent on whether or not a patient is at high risk for joint damage, according to clinical judgement.

The primary outcomes of interest, ACR 20 and PASI 90 response rates, were analyzed through 52 weeks of follow-up, using multivariable logistic regression adjusting for differences in patient characteristics between both guselkumab and ustekinumab treatment cohorts. The model was estimated with pre-specified covariates selected based on clinical input from a trained and practicing rheumatologist. Observed prognostic factors at baseline in the trial included: gender, age, body weight, PsA duration, number of swollen and tender joints, Health Assessment Questionnaire-Disability Index (HAQ-DI), C-reactive protein (CRP), dactylitis status, enthesitis status, PASI level, body surface area (BSA) affected by psoriasis, and methotrexate (MTX) use. The adjusted odds ratios (OR) for the baseline characteristics for biologic-naïve and biologic-experienced populations are presented in Supplementary Figures S2–S5.

Results are expressed as ORs with 95% confidence intervals (CI) and were considered statistically significant if the 95% CI for ORs excluded the null value of 1. Odds ratios resulting from the multivariable logistic regression model were translated into predicted response rates for ustekinumab, simulating the same patient population as the one enrolled in the guselkumab trials. Non-responder imputation was used for missing response data in the analyses. Analyses were performed using SAS software (version 9.4, SAS Institute Inc., Cary, NC, USA).

A propensity score-based modeling approach was used to conduct a sensitivity analysis for ACR 20 and PASI 90 responses at week 52; results are presented in Supplementary Figures S6 to S19. The propensity score is the conditional probability of being assigned to the guselkumab Q8W treatment arm given a set of covariates. Inverse probability weighting (IPW) aims to create a pseudo-population where the guselkumab Q8W arm and the ustekinumab arm being compared are homogeneous. The resulting weighted logistic regression model is where the single observations are up-weighted or down-weighted according to their propensity score. Analyses used different weighting schemes based on propensity scores, including the average treatment effect in the treated population (i.e., guselkumab Q8W) (ATT),¹ standardized average treatment effect (sATT), and IPW/average treatment effect (ATE).²

The primary studies (i.e., DISCOVER-1, DISCOVER-2, PSUMMIT 1, and PSUMMIT 2) were conducted in accordance with the Declaration of Helsinki and the International Committee on Harmonization of Good Clinical Practice Guidelines. Protocols were reviewed and approved by each site’s governing institutional review board or ethics committee. All participants provided written informed consent.

Tables 1 and 2 present the baseline characteristics for biologic-naïve and biologic-experienced patients on guselkumab Q8W and Q4W and ustekinumab 45/90 mg.

Significant differences across treatment groups (Table 1) were observed between both biologic-naïve treatment cohorts for number of swollen joints > 15 (19.4 vs. 29.1%), number of tender joints > 15 (51 vs. 66.9%), HAQ-DI score > 2 (4.5 vs. 13.6%), CRP ≥ 2 (24.1 vs. 93.8%), enthesitis status (60.9 vs. 70.3% without enthesitis), and BSA affected by psoriasis (19.6 vs. 25.5% with < 3% affected). Significant differences across treatment groups (Table 2) were observed between both biologic-experienced treatment cohorts for gender (38.1 vs. 58.5% male), number of swollen joints > 15 (13.2 vs. 47.5%), number of tender joints > 15 (47.4 vs. 83.1%), HAQ-DI > 2 (4.9 vs. 27.1%), CRP ≥ 2 (26.3 vs. 98.3%), enthesitis status (39.5 vs. 21.2% without enthesitis), and use of MTX at baseline (43.2 vs. 68.3% using MTX). Other characteristics examined were comparable. The imbalances between trials and prognostic values for all commonly available baseline characteristics are presented in Supplementary Table S1 and S2.