Non-invasive prediction of persistent villous atrophy in celiac disease

Celiac disease (CD) is an immune-mediated enteropathy that is primarily treated with a gluten-free diet (GFD). Mucosal healing is the main target of the therapy. Currently, duodenal biopsy is the only way to evaluate mucosal healing, and non-invasive markers are challenging. Persistent elevation of anti-tissue transglutaminase antibodies (aTTG) is not an ideal predictor of persistent villous atrophy (VA). Data regarding prediction of atrophy using anti-deamidated gliadin peptide antibodies (aDGP) and abdominal ultrasonography are lacking.

To evaluate the ability of aTTG, aDGP, small bowel ultrasonography, and clinical and laboratory parameters in predicting persistent VA determined using histology.

Patients with CD at least 1 year on a GFD and available follow-up duodenal biopsy, levels of aTTG and aDGP, and underwent small bowel ultrasonography were included in this retrospective cohort study. We evaluated the sensitivity, specificity, and positive and negative predictive values of aTTG, aDGP, small bowel ultrasonography, laboratory and clinical parameters to predict persistent VA. A receiver operating characteristic (ROC) curve analysis of antibody levels was used to calculate cut off values with the highest accuracy for atrophy prediction.

Complete data were available for 82 patients who were followed up over a period of four years (2014-2018). Among patients included in the analysis, women (67, 81.7%) were predominant and the mean age at diagnosis was 33.8 years. Follow-up biopsy revealed persistent VA in 19 patients (23.2%). The sensitivity and specificity of aTTG using the manufacturer's diagnostic cutoff value to predict atrophy was 50% and 85.7%, respectively, while the sensitivity and specificity of aDGP (using the diagnostic cutoff value) was 77.8% and 75%, respectively. Calculation of an optimal cutoff value using ROC analysis (13.4 U/mL for aTTG IgA and 22.6 U/mL for aDGP IgA) increased the accuracy and reached 72.2% [95% confidence interval (CI): 46.5-90.3] sensitivity and 90% (95%CI: 79.5-96.2) specificity for aDGP IgA and 66.7% (95%CI: 41.0-86.7) sensitivity and 93.7% (95%CI: 84.5-98.2) specificity for aTTG IgA. The sensitivity and specificity of small bowel ultrasonography was 64.7% and 73.5%, respectively. A combination of serology with ultrasound imaging to predict persistent atrophy increased the positive predictive value and specificity to 88.9% and 98% for aTTG IgA and to 90.0% and 97.8% for aDGP IgA. Laboratory and clinical parameters had poor predictive values.

The sensitivity, specificity, and negative predictive value of aTTG and aDGP for predicting persistent VA improved by calculating the best cutoff values. The combination of serology and experienced bowel ultrasound examination may achieve better accuracy for the detection of atrophy.