Minimal invasive diagnostics in patients suspected of Crohn’s disease

Crohn's disease (CD) has been increasingly prevalent in countries with a Western lifestyle over the past decades, with North America and northern Europe experiencing some of the highest rates worldwide. However, the typical symptoms and signs of CD are common among the general population. The non-specific symptoms overlapping with more common conditions such as functional bowel disorders pose a diagnostic challenge, resulting in delayed diagnostics work-up. Current guidelines for suspected CD recommend a two-step diagnostic process involving an initial ileocolonoscopy followed by a small bowel examination, regardless of initial findings. This strategy can, however, lead to additional delays and unnecessary examinations. Neglecting this strategy, on the other hand, may result in missed disease cases or an incomplete understanding of the disease burden, ultimately affecting treatment decisions. There is a demand to optimize the initial diagnostic approach for patients suspected of CD, aiming for a rapid, dependable, and patient-friendly modality. In an ideal scenario, implementing a single-exam strategy could streamline the diagnostic process, enhancing efficiency and reducing unnecessary examinations. This PhD study aimed to investigate the feasibility of patient-friendly, minimally invasive diagnostic modalities and to optimize their use in patients suspected of CD.

Study 1 is a prospective blinded, multicenter study that assesses the diagnostic accuracy of pan-enteric capsule endoscopy (PCE) and MRI compared to Ileocolonoscopy as the initial diagnostic modality for newly referred patients with suspected CD. We included 153 patients and found PCE to be a highly accurate diagnostic tool far superior to MRI. PCE and MRI were also associated with less discomfort than ileocolonoscopy. Thus, PCE could be a feasible alternative to ileocolonoscopy in selected patients with suspected CD.

Study 2 examined the adherence to the bowel cleansing regimen for PCE and factors affecting the image quality. Fifty-nine patients from Study 1, whose amount of consumed bowel preparation was known, were included. The polyethylene glycol volume significantly affects the image quality, but only a few patients could consume the prescribed volume. The diagnostic yield remained unaffected by the image quality.

Study 3 examined the agreement of endoscopic severity and classification of lesions with PCE and ileocolonoscopy. The analysis included 99 patients from Study 1. There was substantial agreement for detecting ulcerations and the total SES-CD scores between the two modalities. The assessment of disease activity does not differ significantly across these endoscopic modalities, suggesting that their results are interchangeable – making implementation in clinical practice less troublesome.

Study 4 is an external validation study of Artificial intelligence (AI)-assisted analysis of PCE from Study 1. We aimed to determine the diagnostic accuracy, interobserver variance, and estimated time reduction of AI-assisted review. We found the assistance of the AXARO® framework reduced the initial review time substantially while maintaining high diagnostic accuracy for CD and inflammatory bowel disease.

The present PhD study provides a robust foundation for the future implementation of minimally invasive and patient-friendly PCE as the initial diagnostic approach for patients suspected of CD. AI assistance in PCE analysis has proven beneficial by significantly reducing analysis time without compromising diagnostic accuracy. The results of the studies suggest that PCE with AI assistance is a fast, valuable tool for ruling out inflammatory bowel disease in newly referred patients suspected of CD.