K Gottlieb, J Requa et al. Gastroenterol 2021; 160: 710-719. Central Reading of Ulcerative Colitis Clinical Trial Videos Using Neural NetworksKey finding: A deep learning algorithm can be trained to predict levels of UC severity from full-length endoscopy videos with excellent agreement with human central readers; endoscopic healing accuracy was 97% for UC endoscopic index of severity (UCEIS) and 95.5% for endoscopic Mayo score.
The investigators enrolled 53 adults with celiac disease (CD) for at least two years and followed symptoms as well as stool/urine testing for gluten immunogenic peptide (GIP). “GIP in stool can detect gluten consumption of more than 40 mg/d and the urine tests are positive from 40 and 500 mg/d of gluten.”
Key findings:
Over the 4-week study period, weekend samples (urine) identified 70% of patients excreted GIP at least once, compared with 62% during weekdays (stool).
Patients had a median of 3 exposures during the 4 weeks.
Also, the authors noted increases in GIP excretion towards the end of the study. “This suggests a potential Hawthorne effect that could be explained by a decrease in hypervigilance that often is seen in a context of research studies.”
The authors note that GIP “excretions of greater than 2 mcg/g in stool or greater than 12 ng/mL in urine can induce mucosal damage in almost 100% of patients.”
My take: This study adds to the body of literature emphasizing the high rate of inadvertent gluten exposure.
gutsandgrowth 