Twitter Facebook LinkedIn
Center for Global Research Data

Extended Validation of an Ulcerative Colitis Vedolizumab Decision Support Tool

Lead Investigator: Neeraj Narula, Hamilton Health Sciences
Title of Proposal Research: Extended Validation of an Ulcerative Colitis Vedolizumab Decision Support Tool
Vivli Data Request: 6686
Funding Source: None
Potential Conflicts of Interest: Neeraj Narula holds a McMaster University Department of Medicine Internal Career Award. Neeraj Narula has received honoraria from Janssen, Abbvie, Takeda, Pfizer, Merck, Lupin and Ferring.

Summary of the Proposed Research:

Ulcerative colitis (UC) is an inflammatory bowel disease characterized by bloody diarrhea and periods of relapse and remission, with an estimated prevalance in North America of 37.5 to 238 per 100,000. Achievement of endoscopic remission and histologic remission are associated with improved long-term outcomes including preventing of hospitalization, surgery, and colon cancer. The achievement of endoscopic and histologic remission is best done with biologics such as Vedolizumab and Adalimumab. These biologics have relative strengths and weaknesses, and indirect comparative effectiveness data would suggest similaritly in treatment response between these 2 agents. A recent head-to-head clinical trial for Vedolizumab versus Adalimumab in UC has demonstrated that Vedolizumab results in higher rates of clinical and endoscopic remission. This might suggest that Vedolizumab is superior to Adalimumab and should therefore always be used first line prior to this therapy, however, sub-group analyses within this trial have observed that this superiority is not universal across all patient populations. There is currently no guidance for societies on how to consider these data and best personalize this decision within the context of this head-to-head to trial.

We have previously built a clinical decision support tool using the phase 3 clinical trial program for Vedolizumab in UC. We observed this tool to be accurate for identifying Vedolizumab treated UC patients who would achieve clinical and endoscopic remission within the clinical trial programs. During the external, routine practice validation, we were able to confirm the diagnostic performance of this tool and its ability to identify UC patients most likely to respond to Vedolizumab therapy. We were also able to further extend the value of this tool by demonstrating that this tool did not predict response to anti-tumor necrosis factor (TNF) therapy and that this tool was able to identify UC patients who were more likely to respond to Vedolizumab as compared to anti-TNF therapy. This tool therefore helps to personalize the decision in routine practice for patients who are often not included in the phase 3 and 4 clinical trial programs, and brings forward personalization of this decision within the context of the recently completed head-to-head trial for Vedolizumab. Some limitations remain, however, which include: the lack of blinded centralized endoscopy scoring in the routine practice validation cohort, the lack of balancing between Vedolizumab and anti-TNF treated patients for unmeasured confounders, and the lack of assessment for histologic outcomes which are now known to be of clinical significance are increasingly being incorporated as a key end-point in clinical trials and clinical practice

In the current proposal we aim to specifically address these gaps through a post-hoc analysis of the VARSITY clinical trial program. The clinical trial design of this study allows for balancing of measured and unmeasured confounders between Vedolizumab and Adalimumab treated UC patients, and the centralized blinded assessment of endoscopy and histology allows for further extension of our decision support tool to outcomes of growing importance.

Statistical Analysis Plan

Descriptive statistics will be used to summarize baseline characteristics (e.g. disease activity and patient demographics) as well as outcomes among patients after stratifying them based on the clinical decision support tool scoring system and categorization approach.5 Dichotomous variables will be presented as proportions or percentages. Continuous variables will be reported as means with standard deviations or medians with interquartile ranges.

First, we will calculate the clinical decision support tool score for all patients included in the trial. This scoring system is a simple numeric score as follows (6):

Absence of prior anti-TNF exposure = + 3 points
Disease duration of 2 years or more = + 3 points
Moderate (Mayo 2) baseline endoscopic activity = + 2 points
Baseline albumin = +0.65 points per 1 g/L

This will result in the creation of a linear score for the cohort which can be used for area under the curve (AUC) of the receiver operative characteristic (ROC) assessments, and this will be done for both Vedolizumab and Adalimumab treated UC patients. The AUC for the baseline score will then be assessed for several outcomes separately in Vedolizumab and Adalimumab treated patients and then compared using the concept of generalized U-statistics, as described by DeLong et al.(9) The specific outcomes to be used are defined below.

From here patients will be categorized into low (26 points or less), intermediate (27-32 points), or high (33 points or more) probability of response.(5) This will be done for Vedolizumab and Adalimumab treated patients. Binary comparisons will then be made between Vedolizumab and Adalimumab treated UC patients for the outcomes listed, across the 3 sub-groups (low, intermediate, high probability of response). Confidence intervals will be calculated at the 95% interval, and p value comparisons will be held to a 0.025 significance threshold to account for multiple hypothesis testing. Data will be analyzed using Stata.

Our apriori hypothesis is that because the clinical decision support tool is specific to predicting response to Vedolizumab in UC, then among the high probability of response group we will see significantly higher rates remission for Vedolizumab treated UC patients versus Adalimumab treated UC patients. Among the intermediate or low probability of response groups, where Vedolizumab is predicted to be less effective, we will see at minimum equal outcomes between Vedolizumab and Adalimumab treated patients and possibly superiority of Adalimumab to Vedolizumab in the low probability of group patients. This will fully define whether this decision support tool is capable of predicting who would most benefit from Vedolizumab relative to Adalimumab. Additional analyses will be done to calculate sensitivity, specificity, positive and negative predictive values, and positive and negative likelihood ratios for the tool in each treatment group.

Requested Studies:

A Randomized, Double-Blind, Double-Dummy, Multicenter, Active-Controlled Study to Evaluate the Efficacy and Safety of Vedolizumab IV Compared to Adalimumab SC in Subjects With Ulcerative Colitis
Sponsor: Takeda
Study ID: NCT02497469
Sponsor ID: MLN0002-3026

Public Disclosure:

Parambir S Dulai, MD, Emily C L Wong, BS, Walter Reinisch, MD, Jean-Frederic Colombel, MD, John K Marshall, MD, MSc, Neeraj Narula, MD, Decision Support Tool Identifies Ulcerative Colitis Patients Most Likely to Achieve Remission With Vedolizumab vs Adalimumab. Inflammatory Bowel Diseases, 2021; izab310, DOI: 10.1093/ibd/izab310