Prophylactic Medication for the Prevention of Endoscopic Recurrence in Crohn's Disease: A Prospective Study Based on Clinical Risk Stratification

Abstract

Background: To prevent recurrence after ileocolonic resection [ICR] in Crohn's disease [CD], postoperative prophylaxis based on risk stratification is recommended in international guidelines. This study aimed to evaluate postoperative CD recurrence after implementation of a clinical management algorithm and to determine the predictive value of clinical and histological risk factors [RFs]. Methods: In this multicentre, prospective cohort study, CD patients [≥16 years] scheduled for ICR were included. The algorithm advised no postoperative medication for low-risk patients, and treatment with prophylaxis [immunosuppressant/biological] for high-risk patients [≥1 RF: active smoking, penetrating disease, prior ICR]. Clinical and histological RFs [active inflammation, granulomas, plexitis in resection margins] for endoscopic recurrence [Rutgeerts' score ≥i2b at 6 months] were assessed using logistic regression and ROC curves based on predicted probabilities. Results: In total, 213 CD patients after ICR were included [age 34.5 years; 65% women] (93 [44%] low-risk; 120 [56%] high-risk: 45 [38%] smoking; 51 [43%] penetrating disease; 51 [43%] prior ICR). Adherence to the algorithm was 82% in low-risk [no prophylaxis] and 51% in high-risk patients [prophylaxis]. Endoscopic recurrence was higher in patients treated without prophylaxis than with prophylaxis in both low [45% vs 16%, p = 0.012] and high-risk patients [49% vs 26%, p = 0.019]. Clinical risk stratification including the prescription of prophylaxis corresponded to an area under the curve [AUC] of 0.70 (95% confidence interval [CI] 0.61-0.79). Clinical RFs combined with histological RFs increased the AUC to 0.73 [95% CI 0.64-0.81]. Conclusion: Adherence to this management algorithm is 65%. Prophylactic medication after ICR prevents endoscopic recurrence in low- and high-risk patients. Clinical risk stratification has an acceptable predictive value, but further refinement is needed.