Introduction Coeliac disease (CD) remains underdiagnosed. Previous studies have shown that 5–13% of patients with CD have undergone a previous gastroscopy where the opportunity to take duodenal biopsies and make a diagnosis had been missed. Clinicians may rely on the presence of endoscopic markers of CD to guide biopsy however these have been shown to lack the required sensitivity. A routine duodenal biopsy approach may solve this problem but this is time consuming and expensive. Methods to improve the macroscopic detection of CD at endoscopy to guide biopsy would seem advantageous. Magniview endoscopes are capable of 136x optical zoom without loss of definition. At this level of magnification it should be possible to identify blunting and absence of villi. We aimed to assess the feasibility of magniview in patients with suspected CD.
Method Patients with attending for a duodenal biopsy to diagnose CD were invited to take part. Appearances were graded on a 3 point scale: normal, partial villous atrophy or marked villous atrophy. All patients received 4 biopsies from the second part of the duodenum and at least 1 biopsy from the bulb. Concurrently serum for endomysial (EMA) and tissue transglutaminase (tTG) antibodies were taken at the time of endoscopy. Macroscopic markers of CD are compared VA on histology as the gold standard. Sensitivity of endoscopic markers was compared to a control group CD who had standard endoscopy using the Fisher exact test.
Results 27 patients (52% female, mean age 49.6) have been recruited to date. 14 (51.9%) patients tested had positive serology. In total 13 (48%) patients were diagnosed with CD. All 13 patients diagnosed with CD were correctly identified on Magniview. 4 patients had a false positive Magniview assessment. Of these 3 (75%) were felt to have partial villous atrophy. A single further patient with a positive EMA and tTG was felt to have marked villous atrophy in the bulb but normal D2 however biopsies were normal from both D1 and D2. In summary Magniview to detect VA demonstrated sensitivity, specificity, PPV and NPV of 100% (72–100), 71% (42–90), 76% (50–92) and 100% (66–100) respectively. The sensitivity of Magniview for detecting VA was superior to a standard endoscopy cohort of 89 patients with CD (100% vs. 41.6%, p < 0.0001).
Conclusion The addition of Magniview to standard endoscopy to aid the diagnosis of CD is feasible and currently has 100% sensitivity and negative predictive value. However a larger study in a lower prevalence population is required to fully elucidate the utility of Magniview in clinical practice.
Disclosure of interest None Declared.