Alternative Endoscopy Reading Paradigms Determine Score Reliability and Effect Size in Ulcerative Colitis

Area of Science:

• Gastroenterology
• Clinical Trials
• Medical Imaging Analysis

Background:

• Regulatory agencies advocate central reading of endoscopy for ulcerative colitis (UC) clinical trials.
• The role of local/site readers in conjunction with central readers remains uncertain.
• Endoscopic Mayo subscores (eMS) are key outcome measures in UC trials.

Purpose of the Study:

• To evaluate the reliability of locally- and centrally-determined eMS.
• To assess the impact of different reading paradigms on the reliability of eMS.
• To determine the effect of reading paradigms on the study's effect size.

Main Methods:

• Retrospective analysis of eMS data from the TURANDOT study (NCT01620255).
• Examination of seven reading paradigms involving local readers (LR) and central readers (CR1, CR2), including consensus reads (ConCR).
• Assessment of inter-rater reliability using intraclass correlation coefficient (ICC) and impact on endoscopic improvement.

Main Results:

• Over 40% of eMS scores between two trained central readers were discordant.
• Central readers exhibited significant variability (CR1 vs CR2 ICC: 0.475).
• Concordance between central and local readers was variable (LR vs CR1 ICC: 0.682; LR vs CR2 ICC: 0.526).
• Reading paradigms incorporating consensus between LR and CR enhanced ICC estimates to >0.8.
• At Week 12, consensus-based approaches yielded ICC estimates >0.9, proving most favorable for detecting treatment differences.

Conclusions:

• The ICC between two trained central readers for eMS is unexpectedly low.
• Current central reading processes in UC trials have notable weaknesses.
• Consensus-based reading paradigms enhance reliability and are superior for detecting treatment effects.