Introduction: Immune checkpoint inhibitors (ICIs) have emerged as effective treatment options for many advanced malignancies. These agents have been associated with several immune related adverse events (irAEs) that may lead to discontinuation of therapy. Gastrointestinal (GI) side effects are amongst the most commonly reported irAEs. Currently, limited data exists on upper GI bleed (UGIB) in patients exposed to ICIs. The aim of this study was to evaluate the characteristics and risk factors associated with UGIB among cancer patients treated with ICIs. Methods: We performed a retrospective study on all cancer patients from Cleveland Clinic health system who received treatment with ICI’s till June 2021. Only the patients with clinical and endoscopic evidence of upper GI bleed were included in the study. The study group (GI Bleed) was matched with control group (non-Bleeders) in 1:3. We compared characteristics of both groups (Bleeders and Non-Bleeders) using univariate and multivariate analysis. Results: Of the 6820 patients that received ICIs over the study period, 401 patients developed GI bleed. 30 patients had clinical and endoscopic evidence of upper GI Bleed. The most common endoscopic findings were gastric and duodenal ulcers (63%). Compared with the non-bleeding cohort, patients who developed UGIB bleeding had lower BMI (p= 0.0203). Non-small cell lung cancer was the most frequently treated malignancy in the bleeding group. Patients in the bleeding group had higher exposure to pembrolizumab compared to the non-bleeding group (80% vs. 45.6%, p= 0.001). The incidence of ICI induced colitis and hepatotoxicity were more frequent in the bleeding group compared to the non-bleeding cohort (53.5% vs. 27.8%, p= 0.0106). The presence of cirrhosis, diabetes and chronic kidney disease were similar in both groups. On the multivariant analysis, exposure to pembrolizumab (OR=3.62, p=0.0211), aspirin (OR= 10.55, p=0.0002) and with history of ICI induced colitis and hepatotoxicity (OR=3.66 p=0.0108) were predictive factors in development of bleeding. (Table) Conclusion: This is a novel study that evaluates the risk of UGIB in cancer patients treated with ICIs. In this univariate and multivariate analysis, exposure to aspirin, pembrolizumab and overall incidence of IrAEs were associated with increased risk of bleeding. Although UGIB is a relatively rare irAE, more studies are needed for risk stratification of patients exposed to ICIs. Table 1. - Baseline Clinical Characteristics Of Cancer Patients Exposed To ICIs Variable Non-Bleeding (n=90) Bleeding (n=30) p-value Age, mean ± SD 70.6 ± 11.0 69.6 ± 11.3 0.6854 Gender (Male), n (%) 56 (62.2) 19 (63.3) 0.9133 Race (White), n (%) 76 (84.4) 23 (76.7) 0.3316 Hispanic, n (%) 1 (1.1) 2 (6.7) 0.1539 Medication Type, n (%) PembrolizumabOther 41 (45.6)49 (54.4) 24 (80.0)6 (20.0) 0.0010 BMI, n (%)< 26≥ 26 38 (42.2)52 (57.8) 20 (66.7)10 (33.3) 0.0203 History of Upper GI Bleed, n (%) 7 (7.8) 0 (0.0) 0.1902 On Aspirin prior to procedure, n (%) 5 (5.6) 12 (40.0) < 0.0001 On Plavix prior to procedure, n (%) 1 (1.1) 1 (3.3) 0.4391 On Warfarin prior to procedure, n (%) 0 (0.0) 1 (3.3) 0.2500 Chronic kidney disease, n (%) 34 (37.8) 13 (43.3) 0.5893 Diabetes, n (%) 33 (36.7) 9 (30.0) 0.5073 Cirrhosis, n (%) 4 (4.4) 1 (3.3) 1.0000 Type of cancer treated, n (%)Non-small cell lung cancerOther 25 (27.8)65 (72.2) 15 (50.0)15 (50.0) 0.0253 Evidence of metastasis to the GI Tract, n (%) 4 (4.4) 0 (0.0) 0.5710 ICPI induced side effect, n (%) 25 (27.8) 16 (53.3) 0.0106 Treatment with steroids, n (%) 21 (23.3) 11 (36.7) 0.1527 Treatment with Infliximab or Entyvio, n (%) 1 (1.1) 1 (3.3) 0.4391 Current alive, n (%) 29 (32.2) 8 (26.7) 0.5682 PPI, n (%)NoYes 53 (58.9)37 (41.1) 17 (56.7)13 (43.3) 0.8307 Footnote: PPI (protein pump inhibitor).