Abstract Introduction The optimal infliximab (IFX) dosing strategy for acute severe ulcerative colitis (ASUC) remains unclear. Though intensified IFX dosing is frequently employed, evidence supporting its efficacy has been mixed and it is uncertain if intensified IFX more effectively captures remission than standard-dose IFX. We conducted an updated meta-analysis comparing standard- vs intensified-dose IFX in ASUC, incorporating PREDICT-UC trial data and applying a Bayesian framework to better integrate prior evidence and quantify uncertainty. Methods Electronic databases were systematically searched from inception to December 2024 (PROSPERO ID: CRD42024616359). We conducted a 2-step meta-analysis. First, retrospective cohort studies were assessed with a frequentist random-effects meta-analytic approach. Second, because there is only 1 trial (PREDICT-UC) comparing intensified vs standard IFX in ASUC, we applied a Bayesian meta-analytic approach by using historical trials of standard-dose IFX in ASUC to define the prior probability distribution and then compared that to PREDICT-UC. Results Ten retrospective cohort studies (530 standard-dose and 406 intensified-dose patients) were included in the frequentist analysis. We applied a Bayesian binomial regression model with the historical randomized controlled trials of standard-dose IFX as the reference group. The estimated 3-month colectomy rate for standard-dose IFX was 27% (95% credible interval, 22%-33%). In the PREDICT-UC trial, observed colectomy rates were 15% in patients receiving rescue-dose IFX at 5 mg/kg and 6% in those receiving upfront intensified dosing at 10 mg/kg. Posterior estimates for both intensified dosing strategies in PREDICT-UC were credibly lower than the standard-dose IFX reference. Conclusion Integrating observational and trial data via Bayesian methods suggests that intensified IFX dosing reduces early colectomy in ASUC. These findings support prospective, adequately powered trials to confirm benefit, identify subgroups most likely to respond, and refine individualized accelerated-dosing algorithms.
Goyal et al. (Wed,) studied this question.