Background: E/GEJ adenocarcinoma has a high mortality rate despite curative intent therapy. The use of immune checkpoint inhibition is beneficial for treatment of this cancer in the metastatic and adjuvant settings but the role of these agents in the perioperative setting remains unclear. Here we report the results of an initial safety run-in of nivolumab when given in combination with neoadjuvant chemoradiation. Methods: Pts with a localized T1N1-3M0 or T2-3N0-2M0 E/GEJ adenocarcinoma with an ECOG PS of 0-1 and whom were deemed surgical candidates for an esophagectomy by a qualified surgeon were eligible. In step 1, pts were randomized to neoadjuvant therapy with carboplatin AUC 2 and paclitaxel 50 mg/m2 intravenously (IV) weekly x 5 along with 41.4-50.4 Gy radiation without (Arm A) or with (Arm B) nivolumab 240 mg IV during weeks 1 and 3 of treatment, followed by esophagectomy. Pts underwent a second randomization (step 2) to adjuvant nivolumab 240 mg IV every 2 weeks x 12 cycles with or without ipilimumab 1 mg/kg IV every 6 weeks during cycles 1, 4, 7 and 10. For the safety run-in, 30 pts were planned for accrual to allow for 12 evaluable pts per arm. Pts were followed for safety during neoadjuvant therapy through surgery and toxicities monitored per CTCAEv5. Pre-specified early stopping rules were defined to allow halting of the trial if deemed unsafe. Planned study accrual is 278 pts. Neoadjuvant primary endpoint is pathologic complete response rate, adjuvant primary endpoint is disease-free survival. Results: A total of 31 pts were enrolled to the safety run-in element of the study (Arm A, n = 16; Arm B n = 15). Male, 94%; White, 100%; median age, 62; esophageal adenocarcinoma, 52%; GEJ, 48%. Grade (G) 3 events occurring in more than one pt on Arm A—decreased lymphocytes (n = 5). G4 events occurring on Arm A—decreased lymphocytes (n = 1). G3 events occurring in more than one pt on Arm B—decreased lymphocytes (n = 2); anemia (n = 2); leukopenia (n = 4); hypotension (n = 2). G4 events occurring on Arm B—decreased lymphocytes (n = 3); cardiac tamponade and pericardial effusion (n = 1). Cardiac events were thought to be secondary to tumor location, not neoadjuvant treatment. On Arm B, notable G3 events seen in one pt each included colonic obstruction, wound infection and esophageal anastomotic leak. Of pts who have reached the time for surgery, 12/14 pts on Arm A and 13/13 pts on Arm B have proceeded to surgery. Of pts who have completed step 1, 7/14 pts on Arm A and 8/11 pts on Arm B have registered to step 2. Conclusions: The addition of nivolumab to carboplatin, paclitaxel and radiation in the neoadjuvant setting appears to be safe with no disproportionate level of toxicity observed between the two treatment arms. Accrual to the remainder of the trial continues with 43/278 patients accrued. Clinical trial information: NCT03604991