This post hoc analysis included pooled data from three phase 3 placebo-controlled RCTs (OPAL Broaden, OPAL Beyond, and a trial in China) and one LTE trial (OPAL Balance; safety only) in patients with PsA. For patients with AS, data were pooled from one phase 2 and one phase 3 placebo-controlled RCT. Trial details are published [7,8,9,10,11,12] (see the electronic supplementary material).

Patients were categorized by their reported cigarette smoking status (current/past or never) at trial screening. Efficacy and safety data from RCTs were analyzed by treatment and by current/past or never smoker status. Efficacy data from the RCTs were assessed for patients with PsA receiving tofacitinib 5 or 10 mg BID to month 6, or placebo to month 3. Efficacy data were assessed for patients with AS receiving tofacitinib 5 mg BID or placebo in the phase 3 trial, to week 16. Due to the study design, efficacy for patients with AS enrolled in the phase 2 trial was assessed to week 12 only. Safety in the RCTs was assessed in patients with PsA receiving tofacitinib 5 or 10 mg BID or placebo to month 3, and in patients with AS receiving tofacitinib 5 mg BID or placebo to week 12.

Additionally, the long-term safety of tofacitinib was assessed by current/past versus never smoker status in patients receiving ≥ 1 tofacitinib dose using data from RCTs and the LTE trial (PsA only) up to month 48 for PsA, and week 48 for AS. Data from patients with PsA were assessed as ‘average tofacitinib 5 mg BID’ (average total daily dose < 15 mg), ‘average tofacitinib 10 mg BID’ (average total daily dose ≥ 15 mg), and ‘all tofacitinib’ (received ≥ 1 tofacitinib 5 or 10 mg BID dose). In patients with AS, data were assessed as ‘tofacitinib 5 mg BID’ (received ≥ 1 tofacitinib 5 mg BID dose) and ‘all tofacitinib’ (received ≥ 1 tofacitinib 2, 5, or 10 mg BID dose). Efficacy outcomes by current, past, and never smoker status at month 3 (PsA) and week 12 (AS) were reported.

Efficacy outcomes assessed in patients with PsA included American College of Rheumatology ≥ 20% and ≥ 50% responses (ACR20 and ACR50, respectively), minimal disease activity (MDA), PsA Disease Activity Score (PASDAS) ≤ 3.2 [15], ≥ 75% improvement in Psoriasis Area and Severity Index (PASI) response (assessed in patients with baseline body surface area ≥ 3% and baseline PASI > 0), ≥ 50% improvement in pain visual analog scale, and Functional Assessment of Chronic Illness Therapy-Fatigue (FACIT-F) ≥ 40.1 rates [16], and change from baseline in Health Assessment Questionnaire-Disability Index (HAQ-DI).

Efficacy outcomes assessed in patients with AS included ≥ 20% and ≥ 40% improvement in Assessment of SpondyloArthritis international Society responses (ASAS20 and ASAS40, respectively), AS Disease Activity Score (ASDAS) < 2.1 [17], Bath Ankylosing Spondylitis Disease Activity Index 50 response (BASDAI50), ≥ 50% improvement in back pain, and FACIT-F ≥ 40.1 rates [16].

Safety outcomes assessed in patients with PsA and AS included treatment-emergent adverse events (TEAEs), serious AEs (SAEs), and discontinuations due to AEs. In addition, AEs of special interest (AESI), which included serious infections, herpes zoster infections (serious and non-serious), major adverse cardiovascular events (MACE), malignancies excluding non-melanoma skin cancer (NMSC), NMSC, venous thromboembolism (including deep vein thrombosis and pulmonary embolism), and arterial thromboembolism, were evaluated up to month 48 for patients with PsA, and up to week 48 for patients with AS.

All PsA analyses were based on the final datasets. Phase 2 AS data are from the final dataset. Phase 3 AS data are based on two databases: efficacy data are from the week 16 analysis (database cut-off: December 19, 2019; data snapshot: January 29, 2020), and all other data are from the final week 48 analysis.

Binary and continuous outcomes were assessed using longitudinal logistic regression models and longitudinal mixed-effects linear models, respectively, fitted to include treatment, visit, smoking status, median body mass index (BMI) status (i.e., BMI compared with the baseline median value; < 27.8 and ≥ 27.8 kg/m2 for patients with PsA and < 26.2 and ≥ 26.2 kg/m2 for patients with AS), age, sex, trial, prior TNFi use, and geographic regions at baseline, and all-way interactions of smoking status among median BMI status at baseline, visit, and treatment; and for continuous models, the baseline value of the dependent variable was included. BMI was included in the model due to the relationship between smoking and BMI [18], and for simplicity the two-category median BMI status was used.

Adjusted rates (standard error [SE]) and odds ratios (ORs; 95% confidence intervals [CIs]) versus placebo for binary outcomes and adjusted means (SE) and differences in adjusted means (95% CIs) for continuous outcomes were reported.

Adjusted incidence rates (IRs; number of patients with events/100 patient-years) and associated 95% CIs for safety outcomes were estimated using Poisson regression models, fitted to include treatment, smoking status, median BMI status, sex, trial, and treatment, and smoking status interaction. For several outcomes, models failed to converge; therefore, unadjusted IRs and associated 95% CIs (based upon inverting the Chi-square approximation to the Poisson mean) were also reported.

Data were as observed; missing data were not imputed. p values and 95% CIs were not adjusted for multiple comparisons and were considered descriptive. Numerical differences in adjusted response rates and means with tofacitinib and placebo were noted. The magnitude of treatment response between tofacitinib versus placebo (binary outcomes: OR [95% CI]; continuous outcomes: differences in adjusted means [95% CI]) were assessed between current/past versus never smokers; comparisons with overlapping 95% CIs were considered as similar. Adjusted response rates and means differences between current/past versus never smokers were based on nominal p < 0.05. Safety outcome comparisons between current/past and never smokers were described as higher or lower if the 95% CIs of the unadjusted or adjusted IRs did not overlap.

Additional details on the statistical analysis are in the electronic supplementary material.

All studies were conducted in accordance with the principles of the Declaration of Helsinki and Good Clinical Practice Guidelines of the International Council for Harmonisation, and were approved by the relevant Institutional Review Board and/or Independent Ethics Committee at each investigational site [7,8,9,10,11,12]. Patients provided written, informed consent.