, according to new research presented at the annual meeting of the American College of Gastroenterology.
An endoscopy score cutoff of 4 or higher had a specificity of 82.8% across all colitis grades, and a cutoff of 5 or higher had a specificity of 87.6%, said Yinghong Wang, MD, PhD, a gastroenterologist at the University of Texas MD Anderson Cancer Center, Houston.
Immune-mediated colitis (IMC) is a common immune-related adverse event associated with immune checkpoint inhibitors. Dr. Wang and colleagues previously reported on endoscopic presentations of IMC, including severe inflammation with deep ulcerated mucosa; moderate to severe inflammation with diffuse erythema, superficial ulcers, exudate, and loss of vasculature; and mild inflammation with patchy erythema, aphtha, edema, or normal mucosa associated with histological inflammation.
Dr. Yinghong Wang
Endoscopic scoring systems haven’t been established for IMC, but previous studies have shown benefits from early endoscopic evaluation. The current Common Terminology Criteria for Adverse Events (CTCAE) grading system for clinical symptoms alone has been poorly correlated with endoscopic findings and unable to provide accurate assessments, Dr. Wang said.
“There is a critical and urgent need to develop a new scoring system that could provide accurate and comprehensive assessment for IMC severity to better predict the requirement of more aggressive selective immunosuppressive therapy (SIT), which includes infliximab and vedolizumab,” she said.
Dr. Wang and colleagues conducted a retrospective international study across 14 centers to develop a new comprehensive endoscopic scoring system to assess the severity of IMC and explore its utility in predicting the need for aggressive treatment with SIT. They included 674 adult cancer patients in the United States, United Kingdom, Germany, and Australia with IMC who underwent endoscopic evaluation between 2010 and 2020.
All patients had received immune checkpoint inhibitors, an IMC diagnosis, and endoscopy and histology evaluations for IMC. In addition, all patients had diarrhea, including 92% who had grade 2 diarrhea and higher and 80% who had grade 2 colitis and higher. About 85% were treated with corticosteroids, 31% were treated with infliximab, 10% were treated with vedolizumab, and 5% were treated with both treatment types, corticosteroids and SIT.
Based on endoscopic reports, the research team looked at 10 endoscopic features and assigned one point each for erythema, edema, loss of vasculature, friability, erosions, exudate, any ulcers, large ulcers, deep ulcers, and more than two ulcers. The median IMC endoscopic score was 2.
The scoring system was devised by measuring the specificity of a selected score cutoff in predicting the need for SIT based on clinical consensus from the study group.
The researchers divided the cohort into a training set and a validation set. In the training set, an IMC endoscopy score cutoff of 4 or more had a specificity of 82.8% across all colitis grades and 96.4% among grade 1 colitis to predict SIT use. A cutoff of 5 or more had a specificity of 87.6% across all colitis grades and 98.2% among grade 1. These specificities were comparable to those of the validation sets.
At the same time, the CTCAE score was poorly associated with prediction of future SIT use, with a specificity of 27.4% for clinical colitis grading and 12.3% for diarrhea grading.
In addition, an IMC endoscopic score of 4 or 5 plus ulcer factors had a numerically higher specificity than a Mayo Endoscopic Score of 3. The IMC endoscopic score had a specificity of 85% at a cutoff of 4 and 88.2% at a cutoff of 5, as compared with 74.6% for the Mayo score.
Early endoscopic evaluation in disease course was associated with early SIT use, with a P value of less than .001.
“Implementation of this novel endoscopic scoring system could guide future IMC treatment more precisely,” Dr. Wang said.
The study funding was not disclosed. The authors reported consultant roles, advisory roles, and research support from several pharmaceutical companies.