This post hoc analysis included data from a 12-month, randomized, double-blind Phase 3 study (NCT01877668; OPAL Broaden) [7] and a 36-month, open-label LTE study (NCT01976364; OPAL Balance) [9]. Details of the study design and eligibility criteria have been described previously.

Briefly, the Phase 3 study enrolled patients aged ≥ 18 years with a diagnosis of PsA for at least 6 months and active arthritis and plaque psoriasis [7]. Patients had an inadequate response to ≥ 1 csDMARD and had not previously received a TNFi. Eligible patients were randomized to receive tofacitinib 5 mg twice daily (BID), tofacitinib 10 mg BID, ADA 40 mg once every 2 weeks, placebo with a switch to tofacitinib 5 mg BID at month 3, or placebo with a switch to tofacitinib 10 mg BID at month 3, all taken with a csDMARD.

Patients who completed the Phase 3 study or discontinued for reasons other than an AE related to treatment were eligible to participate in the LTE study, in which all patients received tofacitinib 5 mg BID. The dose could be increased to 10 mg BID after 1 month for inadequate symptom control, and thereafter reduced to 5 mg BID for safety reasons. Patients who entered the LTE study from the ADA group of the Phase 3 study received their first dose of tofacitinib ≥ 1 week after their last ADA injection. The LTE study also enrolled patients from OPAL Beyond (NCT01882439), but those patients were not included in this analysis.

Both studies were conducted in accordance with the Good Clinical Practice guidelines of the International Council for Harmonisation and with the principles of the Declaration of Helsinki. All patients provided written informed consent and the protocols were approved by the institutional review board or independent ethics committee at each investigational site.

The proportions of patients achieving ≥ 20%, ≥ 50%, and ≥ 70% improvements in American College of Rheumatology response criteria (ACR20, ACR50, and ACR70 responses, respectively), ≥ 75% Psoriasis Area and Severity Index improvement from baseline (PASI75) response, Health Assessment Questionnaire-Disability Index (HAQ-DI) response (decrease from baseline ≥ 0.35 for patients with baseline HAQ-DI ≥ 0.35), Psoriatic Arthritis Disease Activity Score (PASDAS) ≤ 3.2, minimal disease activity (MDA) response, and very low disease activity (VLDA) response were reported. Mean changes from the Phase 3 study baseline in Functional Assessment of Chronic Illness Therapy-Fatigue (FACIT-F) total scores, HAQ-DI scores, Physician’s Global Assessment of Psoriasis (PGA-PsO) scores, Patient Assessment of Arthritis Pain scores (on a visual analog scale [VAS]), and PASDAS were also assessed.

Efficacy outcomes were assessed at 3 months before the last visit and at the last visit in the Phase 3 study, and at month 3 (or month 6 for PASDAS and FACIT-F outcomes) in the LTE study.

Safety outcomes of interest included proportions of patients with events and incidence rates (IRs; patients with first events/100 patient-years [PY]) for treatment-emergent AEs (TEAEs), serious AEs (SAEs), serious infections, infections (assessed as a System Organ Class [SOC] from the Medical Dictionary for Regulatory Activities [MedDRA]), herpes zoster, opportunistic infections, tuberculosis, malignancies excluding non-melanoma skin cancer (NMSC), NMSC, lymphoma, major adverse cardiovascular events (MACE), venous thromboembolism, and all-cause deaths. The most frequently occurring TEAEs (occurring in ≥ 5% of patients in any group) were tabulated by MedDRA Preferred Term. AEs were coded using MedDRA v22.0. IRs and proportions for safety outcomes were assessed from months 0–3 and 0–12 in both studies.

Changes from baseline in aspartate aminotransferase (AST), alanine aminotransferase (ALT), low-density lipoprotein cholesterol, high-density lipoprotein cholesterol, total cholesterol, triglycerides, lymphocytes, neutrophils, and hemoglobin were assessed at months 3, 6, and 12 in the Phase 3 and LTE studies. Proportions of patients exceeding upper limit of normal (ULN) thresholds in total bilirubin, AST, ALT, and gamma glutamyl transferase (GT) were tabulated.

For this post hoc analysis, patients were analyzed in two groups. The ADA→tofacitinib 5 mg BID group included patients who received ADA in the Phase 3 study and then directly switched to tofacitinib 5 mg BID and maintained this dose in the LTE study. The continuing tofacitinib 5 mg BID group included patients who received tofacitinib 5 mg BID in the Phase 3 study and then continued tofacitinib 5 mg BID and maintained this dose in the LTE study.

The LTE safety analysis set included patients from the Phase 3 study who had at least one dose of treatment in the LTE study. Patients with a > 14-day gap in treatment between the last observation in the Phase 3 study and the first dose of tofacitinib in the LTE study were excluded from any ‘by-visit’ summaries of continuous data (i.e., laboratory variables), and evaluation of all efficacy endpoints also followed this 14-day gap rule.

Demographic and disease characteristics at Phase 3 study baseline were summarized descriptively. For efficacy outcomes, proportions were calculated based on the number of patients with non-missing responses at each visit. Least squares mean changes from baseline were analyzed using a repeated measures model with the fixed effects of treatment, visit, treatment by visit interaction, geographic location, and baseline value. An unstructured covariance matrix was used. No formal statistical testing was conducted for this post hoc analysis; consequently, p-values for the comparisons between treatment groups are nominal and are provided for descriptive purposes only. As an additional descriptive analysis, for dichotomous response outcomes, patients in the ADA→tofacitinib 5 mg BID group were categorized based on their response to ADA at the last Phase 3 visit and their subsequent response to tofacitinib at month 3 in the LTE study.

For IRs of safety outcomes, follow-up time was calculated up to the day of the first event, subject to a risk period of 28 days beyond the last dose or to the data cut-off date. Gaps in dosing between treatment switches or between the Phase 3 and LTE studies were included up to 28 days or to the data cut-off date. Exact Poisson (adjusted for PY) 95% confidence intervals (CIs) were calculated for the crude IRs. Laboratory parameters were reported using descriptive statistics.

Data were reported as observed, with no imputation for missing data.