Tofacitinib Efficacy/Safety in Patients with Ankylosing Spondylitis by Baseline Body Mass Index: A Post Hoc Analysis of Phase 2/3 Trials

Abstract

Introduction: We assessed tofacitinib efficacy and safety in ankylosing spondylitis (AS) by body mass index (BMI) category.

Methods: Data were pooled from phase 2/3 trials; analyses included patients with active AS randomized (1:1) to tofacitinib 5 mg twice daily or placebo, who were stratified by baseline BMI into < 25, ≥ 25 to < 30, and ≥ 30 kg/m² categories. Efficacy was assessed at week 12 and safety to week 16.

Results: Of 370 patients, 153, 131, and 86 had a baseline BMI of < 25, ≥ 25 to < 30, and ≥ 30 kg/m², respectively. At baseline, patients with BMI < 25 kg/m² were younger and more likely to be current smokers/Asian, and patients with BMI ≥ 30 kg/m² had higher mean waist circumference/swollen joint count (SJC) and were more likely to have enthesitis, high-sensitivity C-reactive protein (hsCRP) > 5 mg/L, an inadequate response to tumor necrosis factor inhibitors (TNFi), and prior biologic disease-modifying anti-rheumatic drug (bDMARD) use versus other categories. Across categories, tofacitinib responses/improvements were greater than with placebo, except for ≥ 40% Assessment of SpondyloArthritis international Society improvement (ASAS40), ASAS partial remission, 50% improvement from baseline in the Bath Ankylosing Spondylitis Disease Activity Index score (BASDAI50), and Ankylosing Spondylitis Disease Activity Score using C-reactive protein (ASDAS-CRP) inactive disease rates, which were similar for tofacitinib and placebo in the BMI ≥ 30 kg/m² category. Treatment effects were similar across categories, except for BASDAI50, which was smaller in the BMI ≥ 30 category versus the < 25 kg/m² category. More adverse events (AEs) and serious adverse events (SAEs) with tofacitinib were reported in the BMI < 25 kg/m² category, which had a higher proportion of current smokers versus other categories.

Conclusions: Regardless of baseline BMI category, efficacy was greater with tofacitinib versus placebo in patients with AS, and no treatment effect differences between categories were observed, with exceptions for BMI ≥ 30 kg/m² (more active/treatment-refractory disease and a smaller sample size). Overall, tofacitinib safety was generally comparable across categories; however, AE/SAE rates with tofacitinib were higher in the BMI < 25 kg/m² category (which had more current smokers). This post hoc analysis demonstrates that tofacitinib can be considered as a treatment option for AS, regardless of baseline BMI category; however, interpretation was limited by small sample sizes and differences in sample sizes and baseline characteristics across categories.

Trial registration: ClinicalTrials.gov identifiers, NCT01786668 and NCT03502616.

Keywords: Body mass index; Randomized controlled trial; Rheumatology; Therapeutics.

Fig. 1

Rates at week 12 for a ASAS20 response, b ASAS40 response, c ASAS partial remission, d BASDAI50 response, and e ASDAS-CRP clinically important improvement response by baseline BMI category. Analyses were conducted using logistic regression models which included treatment, BMI category, treatment by BMI interaction, and prior treatment history unless stated otherwise. Non-responder imputation was used to account for missing values, where patients with missing values were considered as non-responders. ^†95% CI for OR versus placebo excluded 1. ^‡95% CI for OR versus BMI < 25 kg/m² excluded 1. ^aUnivariable logistic regression models were performed to analyze the association of selected baseline covariates and ASAS20 response at week 12; however, no additional variables were included in the final multivariable logistic regression model, which was used to evaluate treatment effects. ^bUnivariable logistic regression models were performed to analyze the association of selected baseline covariates and ASAS40 response at week 12; geographical region, disease duration since diagnosis, and baseline MASES and total back pain scores were selected in the final multivariable logistic regression model. ASAS Assessment of SpondyloArthritis international Society, ASAS20/40 ≥ 20%/40% Assessment of SpondyloArthritis international Society improvement, ASDAS-CRP Ankylosing Spondylitis Disease Activity Score using C-reactive protein, BASDAI50 ≥ 50% improvement from baseline in the Bath Ankylosing Spondylitis Disease Activity Index, BID twice daily, BMI body mass index, CI confidence interval, MASES Maastricht Ankylosing Spondylitis Enthesitis Score, n number of patients with a response, N number of patients in the full analysis set, N1 number of patients with observation at each visit, OR odds ratio

Fig. 2

LS mean (SE) changes from baseline at week 12 in a ASDAS-CRP, b BASDAI, c BASMI, d BASFI, e total back pain, and f nocturnal spinal pain scores and in g CRP levels by baseline BMI category. Analyses were conducted using ANCOVA models, which included baseline outcome values, treatment, BMI category, treatment by BMI interaction, and prior treatment history unless stated otherwise. Data were as observed, and missing values were not imputed. ^†95% CI for differences between treatment LS means excluded 0. ^aUnivariable regression models were performed to analyze the association of selected baseline covariates and ΔASDAS-CRP at week 12; however, no additional variables were included in the final multivariable ANCOVA model. Δ change from baseline, ANCOVA analysis of covariance, ASDAS-CRP Ankylosing Spondylitis Disease Activity Score using C-reactive protein, BASDAI Bath Ankylosing Spondylitis Disease Activity Index, BASFI Bath Ankylosing Spondylitis Functional Index, BASMI Bath Ankylosing Spondylitis Metrology Index, BMI body mass index, CI confidence interval, CRP C-reactive protein, LS least squares, N number of patients in full analysis set, N1 number of patients with observation, SE standard error