Neoadjuvant immunotherapy produced deep responses in localized dMMR/MSI gastroesophageal cancer
Localized dMMR/MSI gastroesophageal adenocarcinomas sit in a biologically distinct category. Their mismatch repair failure creates a high burden of somatic mutations, which in turn produces a more visible target for the immune system and helps explain why these tumors have often responded poorly to fluoropyrimidine-based chemotherapy but strikingly well to immune checkpoint blockade.
A French retrospective case series from Paoli-Calmettes Institute adds another layer to that story. Investigators reviewed consecutive patients with localized dMMR/MSI gastric or gastroesophageal junction adenocarcinoma treated with neoadjuvant immunotherapy between January 2024 and July 2025, then tracked what happened on endoscopy, imaging, pathology, and follow-up. The central question was no longer whether the tumors could shrink. It was whether some patients might avoid surgery altogether without immediate loss of disease control.
Surgery was not the only path to disease clearance
Ten patients entered the series. The median age was 70.5 years, and the cohort included six men and four women. Nine patients received dual immunotherapy before local treatment, while one received chemoimmunotherapy. The small sample size does not allow broad generalization, but the pattern of response was hard to ignore.
Six patients underwent surgery. Four had a pathological complete response, one had partial regression classified as TRG1, and one had TRG2. In a disease where residual microscopic tumor can define the next clinical step, those postoperative findings matter. A pCR after neoadjuvant treatment means the resected specimen contains no viable tumor, a result that speaks to both biologic sensitivity and the possibility of overtreatment if every responder is taken to the operating room by default.
Four other patients were managed non-operatively. Two achieved complete clinical response, and all four ultimately reached clinical complete response after individualized treatment continuation. In practical terms, the absence of detectable disease on endoscopy or imaging became the working endpoint for these patients. For surgeons and oncologists, this is a difficult threshold because radiographic silence does not always equal histologic sterilization. Still, in a subgroup already known to be unusually immunoresponsive, the combination of serial endoscopy, imaging, and clinical assessment is beginning to support a watchful strategy for carefully selected cases.
Follow-up suggested durable control after immune-based treatment
After a median follow-up of 19.5 months, ranging from 10 to 22 months, every patient was alive and disease-free. For a case series of localized gastroesophageal cancer, that is a meaningful outcome, especially because it includes patients who never went to the operating room. The follow-up is still short for a disease in which recurrence can emerge later, but the early signal is favorable enough to justify serious attention from multidisciplinary teams.
The biological logic behind these responses is straightforward even if the management implications are not. dMMR/MSI tumors generate numerous neoantigens, which can make them more visible to activated T cells once inhibitory immune signaling is blocked. Dual checkpoint blockade can therefore produce rapid tumor regression and, in some patients, complete eradication of detectable disease. In gastroesophageal adenocarcinoma, where surgical morbidity can be substantial, the possibility of avoiding gastrectomy or esophagogastric resection after a complete response has obvious appeal. The burden is not abstract: surgery may still be curative, but it also carries nutritional, functional, and quality-of-life consequences that can persist long after the cancer is removed.
The study also illustrates how carefully non-operative management must be defined. cCR is not a single scan or a single scope. It is a composite judgment built from endoscopic appearance, imaging, and ongoing reassessment. In the absence of randomized data, that composite decision will remain highly dependent on local expertise, patient preference, tumor location, and the willingness to accept close surveillance.
Real-world practice is now pressing against the surgery-first model
These findings align with the broader direction set by recent phase II studies such as NEONIPIGA and INFINITY, which reported unusually high pCR rates with neoadjuvant immune checkpoint blockade in dMMR/MSI gastric and gastroesophageal junction cancers. The present series extends that logic into routine practice and suggests that the therapeutic conversation is shifting from “Can immunotherapy work?” to “Who still needs an operation after it works?”
There is no reason to pretend the answer is simple. Pathologic response remains the most reliable proof of treatment effect, but it is unavailable in patients who never undergo resection. Clinical response assessment can miss residual disease, while surgery may overtreat patients who have already cleared their cancer. The tension between those two risks defines the next phase of care for this subgroup.
One sentence captures the clinical pressure point: complete responders may no longer fit neatly into a mandatory surgery pathway.
For now, the series supports feasibility rather than a definitive change in standard management. The cohort is small, treatment approaches were not uniform, and follow-up remains limited. Yet the signal is strong enough to justify prospective trials focused on response-adapted strategies, longer surveillance, and carefully measured non-operative management in localized dMMR/MSI gastroesophageal adenocarcinoma. The key unanswered issue is not whether immunotherapy can produce profound regressions; it already can. The unresolved question is how safely clinicians can trust those regressions when deciding whether surgery is still necessary.