856 Evaluation of Paired and Separated Endoscopic Reads at Baseline and Week 14 in the Exploratory BERGAMOT Induction Cohort

Abstract

INTRODUCTION: Treatment guidelines recommend that Crohn’s disease evaluation and treatment decisions should include endoscopic evaluation in conjunction with clinical and laboratory assessment. Previously, an adjudicated central read model (CRM) for evaluating Simple Endoscopic Score for Crohn's Disease (SES-CD) that included a local reader and at least 1 central reader (CR) was found to reduce placebo rates and variability and increase treatment response rates versus single reader models (1). This study explored the impact on scoring by blinded reading of paired endoscopic reads (i.e. simultaneous assessment of pre- and postinduction treatment endoscopic videos) versus unpaired reads scored by the CRM. METHODS: Patients in this analysis (n = 63) were a subset from the exploratory BERGAMOT induction cohort, who were randomized 2:2:1 to receive subcutaneous etrolizumab 105 mg every 4 weeks (n = 24); etrolizumab 210 mg at weeks 0, 2, 4, 8, and 12 (n = 24); or placebo (n = 15). Three CRs performed the paired endoscopy readings. The unpaired reads were evaluated using the CRM and included a predefined sliding threshold to trigger adjudication of discordant scores by a second CR. All readers were blinded to treatment arm, timing of the endoscopy, and the other reader’s assessments. RESULTS: In the overall population, 48% had ileocolonic disease, 30% had colonic disease, and 22% had disease restricted to the ileum. Prior treatment for Crohn’s disease included anti-TNFs (73%), oral corticosteroids (49%), or nonbiologic immunosuppressants (27%). In general, the mean baseline and week 14 SES-CD evaluated by the paired read model were higher than those assigned by the CRM. Differences between models in assigning categorical thresholds were more varied (Table 1). The placebo rates of ≥ 50% SES-CD decrease from baseline, SES-CD ≤ 4, and SES-CD ≤ 2 were similarly low (0%-6.7%) with either the unpaired CRM or the paired endoscopy model. More patients achieved ≥ 50% SES-CD decrease from baseline with etrolizumab treatment using the CRM compared with the paired read model (21% vs 13%, respectively). CONCLUSION: Based on data generated by experienced CRs in a small subset of patients, both paired and unpaired central reading of CD endoscopic videos resulted in very low placebo response rates. However, unpaired central reading resulted in increased treatment response rates compared with a paired read paradigm.