Mismatch repair-deficient/microsatellite instability-high (dMMR/MSI) esophagogastric cancers represent a distinct molecular subgroup that appears to derive limited benefit from conventional perioperative chemotherapy while showing sensitivity to immune checkpoint inhibition. Several phase II studies have therefore explored immunotherapy in the localized setting, although the optimal regimen and treatment duration remain uncertain.
The study, titled “Is pembrolizumab as safe and efficient as neoadjuvant treatment in localized dMMR/MSI esophagogastric cancers? Results from the IMHOTEP phase II study,” was published in ESMO Open in September 2026.
Authors: C. Coutzac, A. Zaanan, A. de Montfort, E. Soularue, R. Cohen, S. Le Sourd, G. Piessen, V. Hautefeuille, M. Ben Abdelghani, E. Blanc, M. Svrcek, E. Samalin, D. Pérol, and C. de la Fouchardière.
Why IMHOTEP Was Conducted
dMMR/MSI tumors account for a minority of esophagogastric adenocarcinomas and have shown limited benefit from conventional perioperative chemotherapy. At the same time, their high tumor mutational burden and immune infiltration provide a strong rationale for immune checkpoint blockade.
Previous phase II studies, including NEONIPIGA and INFINITY, reported substantial pathological responses with combinations targeting PD-1/PD-L1 and CTLA-4. IMHOTEP took a different approach, evaluating a pragmatic strategy based on single-agent pembrolizumab with relatively short neoadjuvant exposure.
Trial Design
IMHOTEP, NCT04795661, is a multicenter, open-label, phase II basket trial that included colorectal, esophagogastric, endometrial, and other digestive cancers. The esophagogastric cohort enrolled adults with histologically confirmed, locally advanced, nonmetastatic dMMR/MSI adenocarcinoma of the esophagus, gastroesophageal junction, or stomach. Eligible disease was defined as cT2-T4 N0M0 or any cT N+M0, and patients had an ECOG performance status of 0 or 1. dMMR/MSI status was confirmed using immunohistochemistry together with molecular MSI testing. All eligible patients were considered candidates for surgery.
Patients received pembrolizumab 400 mg every 6 weeks. Investigators could administer one or two neoadjuvant infusions before surgery. Postoperative pembrolizumab could subsequently be given to complete up to nine total cycles, depending on postoperative stage, performance status, and tumor response. The primary endpoint was pathological complete response, defined as ypT0N0. Pathological response was assessed locally without central review. Secondary endpoints included event-free survival and safety.
Patient Population
Between May 2021 and February 2025, 41 patients with dMMR/MSI esophagogastric adenocarcinoma were treated with pembrolizumab. The median age was 74 years, 59% of patients were male, and 61% had an ECOG performance status of 1. The stomach was the most frequent primary tumor site, accounting for 24 patients, or 58.5%. Most tumors were locally advanced, with 78% classified as at least clinical T3 and 61% having clinically positive lymph nodes.
Five patients were excluded from the efficacy population because eligibility criteria were not met or because they received more than two neoadjuvant pembrolizumab infusions before the decision regarding surgery. This left 36 patients in the efficacy population. Of these, 27 underwent surgery after one or two pembrolizumab infusions and were evaluable for the primary pCR endpoint.
Pathological Response and Surgery
All 27 patients who underwent surgery achieved an R0 resection. Thirteen had received one neoadjuvant pembrolizumab infusion and 14 had received two. Four of the 27 operated patients achieved a pathological complete response, corresponding to a pCR rate of 14.8%. One pCR occurred after a single neoadjuvant infusion and three after two infusions.
The median number of perioperative pembrolizumab infusions among operated patients was six. Seventeen subsequently received postoperative pembrolizumab. Among the remaining 10 patients, five with ypN+ disease received adjuvant chemotherapy and five received no adjuvant treatment.
Nine of the 36 patients in the efficacy population did not undergo surgery. Reasons included patient decision, pembrolizumab-related toxicity, inoperability, and one death from septic shock. During follow-up, five of these nine patients experienced disease progression, while three patients eventually achieved complete tumor regression after eight or nine pembrolizumab infusions. One patient had stable disease after a single infusion but discontinued treatment because of treatment-related pneumonitis.
Event-Free Survival
After a median follow-up of 25.8 months, 25 of the 27 patients who underwent surgery had experienced no progression, recurrence, or death. The estimated 24-month event-free survival rate in the operated population was 93% (95% CI 74%-98%). Two events were reported: one non-cancer-related death and one disease recurrence. When all 36 patients in the efficacy population were considered, including both operated and nonoperated patients, the 24-month EFS rate was 80% (95% CI 63%-90%). Median overall survival had not been reached at the time of analysis.
Safety
Among all 41 treated patients, 26 experienced treatment-related adverse events. Grade 3 treatment-related adverse events occurred in nine patients, corresponding to 22% of the safety population. Reported events included diarrhea, hepatitis, polymyalgia rheumatica, fatigue and general health deterioration, vomiting, decreased weight, metabolism disorders/cell death, ulcerative keratitis, nephritis, and pneumonitis. No pembrolizumab-related grade 5 adverse events were reported, and the investigators identified no new safety signals or unexpected perioperative complications.
Interpreting the Findings
The 14.8% pCR rate was lower than anticipated and lower than rates previously reported with dual immune checkpoint blockade in localized dMMR/MSI esophagogastric cancer. The investigators suggested that the short duration of neoadjuvant pembrolizumab exposure—only one or two infusions over approximately 6 to 12 weeks—may have contributed to the modest pathological response rate.
At the same time, the 93% 24-month EFS rate among patients who underwent surgery was encouraging. However, the investigators emphasized that this result should be interpreted cautiously because of the small sample size. In addition, 17 of the 27 operated patients received postoperative pembrolizumab, making it difficult to determine how much of the observed disease control was attributable to neoadjuvant treatment, postoperative treatment, or their combination.
Other limitations included post-enrollment exclusions, investigator discretion regarding whether patients received one or two neoadjuvant infusions, and the number of treated patients who ultimately did not undergo surgery. Pathological responses were assessed locally without central review.
Conclusion
The phase II IMHOTEP study showed that perioperative pembrolizumab monotherapy is feasible in patients with localized dMMR/MSI esophagogastric adenocarcinoma. Among the 27 patients who underwent surgery after one or two neoadjuvant pembrolizumab infusions, the pCR rate was 14.8%, while the 24-month EFS rate was 93%.
The findings support further investigation to define the optimal duration and strategy of neoadjuvant immunotherapy. The contribution of postoperative pembrolizumab, the potential value of combination approaches, and the role of nonoperative management remain open questions.
The full article is available in ESMO Open.
Find more videos on OncoDaily YouTube TV.
