A Model Based Meta Analysis of Ulcerative Colitis for Clinical Remission

Objectives: Ulcerative Colitis (UC) is a chronic, intermittent relapsing inflammatory disease involving the mucosal and submucosal layers of the large intestine and marked by continuous inflammation from the rectum to more proximal areas of the colon [ Kobayashi, T., et al 2020]. UC clinical trials are typically designed to have an induction period where participants are randomized into either the placebo arm or an active arm. Followed by a maintenance period, where only participants with clinical response from the active arm are re-randomized to either the placebo group or active treatment. Usually, the primary measured endpoint is clinical remission (CR) which has changed definitions over the years [ Sood et al. 2023]. To account for the trial design and different definitions, Model-based meta-analysis (MBMA) methodology using aggregate level data from randomized controlled trials will be conducted. The MBMA model is aimed at quantifying the relative odd ratios of the treatments in respect to placebo rates. The odds ratio against comparators on CR will systematically categorize treatment’s odds of achieving CR against other treatments. Methods: A systematic literature review was conducted to construct a UC database that consists of publicly available summary level safety and efficacy data from 62 trials investigating biologics, Janus kinases inhibitors, & sphingosine-1-phosphate agonists. CR response in the MBMA model was described as the sum of a trial specific non-parametric (unstructured) placebo effect and a parametric drug effect depending on CR definition. A parametric placebo model was developed following the initial model to allow for the inclusion of trials with null placebo rates. Both models were conducted on induction data only and CR data from both full-Mayo and modified Mayo definitions are kept separate. Dose-response was estimated where possible, with drug specific Emax and potency (ED50). These models were developed using the generalized nonlinear least squares (gnls) and/or non-linear mixed effects (nlme) functions from the nlme package in R. Results: Sufficient data was found to create a fully connected network of drugs with CR rates from similar definitions and were well described by the Emax model. However, the non-parametric placebo model was unable to estimate the drug effect of some drugs, like Upadacitinib, due to no reported CR rates in their trial’s placebo arm. A parametric placebo model allowed us to estimate the drug effect of these drugs. Furthermore, the results from the two CR definitions were observed from 14 drugs and it illustrated the need to keep the two definitions separate. Conclusions: Ulcerative Colitis is a chronic, intermittent relapsing inflammatory disease whose onset typically occurs during adolescence and young adulthood, though it has been reported in all age groups. It was noted that the typical UC clinical trial consists of two periods and multiple definitions of CR. A study only on the induction period, showed that 14 drugs had both definitions reported. Separate MBMA models for the two definitions of CR showed that the two versions should not be combined, consistent with Naegeli et al, 2020.