Importance of endoscopic and histological evaluation in the management of immune checkpoint inhibitor-induced colitis

Immune checkpoint inhibitors (ICPI) are efficacious treatments for advanced malignancies but can result in immune mediated diarrhea and colitis (IDC). Currently, the guidelines for the treatment of IDC depend only on clinical symptoms. Endoscopic and histologic features of such adverse events are not well studied in a manner that can help to gauge treatment plans. We aimed to characterize endoscopic and histologic features of IDC and to assess their association with clinical outcomes. Our study included patients who had undergone endoscopy for IDC (1/2010 to 3/2018). Patients with GI infection at time of onset were excluded. High-risk endoscopic features were ulcers deeper than 2 mm, larger than 1 cm, and extensive colonic involvement. Univariate and multivariate logistic regression were performed to assess the association of endoscopic and histological features with clinical outcomes. A total of 182 patients was included; most were white (92%), males (65%) with a mean age of 60 years. Median time from ICPI initiation to IDC was 7 weeks. Fifty-three percent had grade 3-4 diarrhea, and 32% grade 3-4 colitis. Forty-nine patients had mucosal ulcerations, 66 non-ulcerative inflammation and 67 normal endoscopy. Calprotectin was higher in patients with ulceration (P = 0.04). The sensitivity of lactoferrin to detect histologic and endoscopic inflammation was 90% and 70% respectively. Patients who underwent endoscopy earlier than 7 days after IDC onset had shorter duration of IDC symptoms and duration of steroid treatment than those who underwent endoscopy after 7 days of IDC onset (P = 0.026 and P = 0.053, respectively). Patients who underwent endoscopy > 30 days of symptom onset required longer duration of steroids (P = 0.02), had more recurrent symptoms (P < 0.01) and received later infliximab/vedolizumab add-on therapy than did those who underwent endoscopy ≤30 days (P = 0.03). High-risk features were associated with more frequent (P = 0.03) and longer duration (P = 0.02) hospitalization and infliximab/vedolizumab requirement (P < 0.01). Patients with active histological inflammation had more recurrence (P < 0.01) and repeat endoscopy (P < 0.01). Repeat endoscopy was required in 47 patients. A multivariate logistic regression revealed that longer ICPI treatment was associated with more frequent hospitalizations (OR 1.00; 95%CI 1.00-1.01; P < 0.01) and high-risk endoscopic features were associated with the requirement of infliximab/vedolizumab (OR 3.89; 95%CI 1