Predicting fibrostenotic complications prior to endoscopy in eosinophilic esophagitis (EoE) is challenging and esophageal strictures and narrowing are commonly missed on endoscopy. To develop and validate a score to predict fibrostenosis in EoE patients prior to endoscopy. We leveraged a large database of newly diagnosed EoE patients. Fibrostenosis was defined as esophageal stricture, luminal narrowing, or dilation performed during the diagnostic endoscopy. Patients were randomly divided into a development and validation set. We compared features between patients with and without fibrostenosis to inform the initial model and assess predictive ability, as measured by area under curve (AUC). We tested the model in the independent validation set and generated a score to predict low, medium and high fibrostenosis risk. In 655 newly diagnosed EoE patients in the development set, fibrostenosis was associated with age ≥ 18years (OR 10.64; 95% CI 5.61-20.17), symptoms for ≥ 5years prior to diagnosis (OR 2.07; 1.32-3.24), dysphagia (OR 3.72; 1.68-8.22), food impaction (OR 1.68; 1.07-2.62), and lack of abdominal pain (OR 0.28; 0.14-0.60). The model predicted fibrostenosis (AUC = 0.841). In the validation set (n = 654), AUC was preserved (0.831). A scoring system was generated, with scores of ≤ 2 being low risk (< 10% chance of stricture), 2.5-4.5 medium risk (10-50% stricture), and 5-6 high risk (> 50%). We developed and validated the PICK-UP-STRICS score to increase suspicion and detection of fibrostenotic disease in EoE using readily available clinical features prior to endoscopy. This score may guide clinical decisions on the need of endoscopic dilation.
Read full abstract