Real-World Effectiveness and Safety of Upadacitinib Plus Vedolizumab vs Upadacitinib Monotherapy During Induction in Moderate-to-Severe Ulcerative Colitis

This multicenter retrospective comparative cohort study was conducted at six tertiary inflammatory bowel disease referral centers in China. Consecutive adults with moderate-to-severe ulcerative colitis who initiated upadacitinib induction therapy, either as monotherapy or with concomitant vedolizumab, between January 2023 and September 2025 were screened for eligibility. The index date was defined as the date of upadacitinib initiation. Treatment exposure was classified according to the regimen initiated at the index date: upadacitinib monotherapy or upadacitinib plus vedolizumab. The same eligibility criteria, treatment exposure definitions, assessment windows, outcome definitions, and statistical procedures were applied to both treatment groups. Patients in the combination-therapy group received oral upadacitinib 45 mg once daily for 8 weeks and intravenous vedolizumab 300 mg at weeks 0, 2, and 6. Patients in the monotherapy group received oral upadacitinib 45 mg once daily for 8 weeks without concomitant vedolizumab or another advanced therapy. Baseline clinical and biochemical assessments were defined as the most recent eligible assessments obtained before or on the index date. Week-8 clinical, biochemical, and endoscopic assessments were defined as the eligible assessments closest to day 56 after upadacitinib initiation. Endoscopic recordings were centrally reviewed by blinded gastroenterologists, and disagreements were resolved by a third blinded adjudicator. The primary outcome was modified Mayo clinical remission at week 8. Secondary outcomes were modified Mayo clinical response, C-reactive protein normalization among patients with an elevated baseline C-reactive protein concentration, and endoscopic remission. Safety was assessed from the index date through the week-8 assessment. The primary analysis used propensity score-based stabilized inverse probability of treatment weighting, targeting the average treatment effect in the study population. Adjusted risk differences were estimated using survey-weighted linear probability models, and adjusted risk ratios were estimated using survey-weighted quasi-Poisson regression models with robust design-based standard errors. Sensitivity analyses included stabilized weights truncated at the 1st and 99th percentiles, overlap weighting, and conventional multivariable logistic regression without propensity-score weighting. The protocol was reviewed and approved by the Research Ethics Committee of The Sixth Affiliated Hospital, Sun Yat-sen University (approval number 2026ZSLYEC-079). The requirement for informed consent was waived because the study involved retrospective analysis of de-identified clinical data.