Background: Mismatch repair (MMR) deficient tumors harbor hundreds to thousands of mutations that may produce neoantigens that can be recognized and targeted by T cells. We have shown that MMR deficiency can serve as a predictive biomarker for selection of tumors across tumor histologies that may respond to programmed death-1 (PD-1) blockade. MMR deficiency is present in 15% of colorectal (CRC) however is also detected in 2-20% of gastric, small bowel, and hepatobiliary cancers. Methods: We conducted a phase II study to evaluate anti-PD-1, pembrolizumab, in patients with previously-treated, progressive, advanced cancer. Twenty-one patients with MMR deficient tumors were enrolled onto the non-CRC cohort with an additional 50 patient expansion underway. The pembrolizumab dose is 10mg/kg intravenously every 2 weeks. Objective response rate (ORR), progression-free survival (PFS), and overall survival (OS) will be reported in at least the first 17 patients with non-CRC gastrointestinal (GI) cancers. Results: As of September 18, 2015, 17 patients with non-CRC GI cancers have been enrolled on the protocol with additional patients identified. The diseases represented are ampullary (N=4), pancreas (N=4), biliary (N=3), small bowel (N=3), and gastric (N=3) cancers. For the 10 evaluable patients at the time of abstract admission, ORR is 50 % (N=5/10), disease control rate is 70% (N=7/10), OS is 21 months and the median duration of response (range 5.5-17+ months) and PFS have not been reached. The median follow up duration is 7.6 months. Conclusions: PD-1 blockade shows promising activity in mismatch repair deficient GI cancers. Clinical trial information: NCT01876511