Single-Cycle Pembrolizumab in Localized DMMR Colon Cancer

Key Takeaways

• Pathologic complete response was observed in 44% and major pathologic response in 57% after one pembrolizumab cycle before surgery.
• Pathologic complete response was higher in stage I-II disease than stage III disease, at 61% versus 33%.
• Surgery was usually completed within about a month, grade 3 adverse events occurred in 11%, two surgery-related deaths occurred, and endoscopy outperformed biopsy while ctDNA remained nonspecific for pCR.

In the RESET-C phase II trial, a single preoperative pembrolizumab cycle was followed by pathologic complete response in 44% of 84 patients who underwent surgery for dMMR colon cancer. Major pathologic response was observed in 57%, and curative-intent surgery proceeded in nearly the entire cohort. Investigators also tracked perioperative timing and examined how endoscopy, biopsies, and circulating tumor DNA aligned with postoperative pathology. Together, those surgical and assessment data placed perioperative feasibility at the center of the trial readout.

RESET-C was a Danish national, investigator-initiated, multicenter phase II study that enrolled 85 adults between February 2023 and March 2024. Eligible participants had histologically confirmed localized dMMR colon carcinoma, stage I to III from cT1N0M0 to cT4N2M0, and were scheduled for elective curative-intent surgery. One patient with stage I disease declined surgery, leaving 84 patients evaluable for pathologic response and surgical outcomes. The regimen used one pembrolizumab cycle at 4 mg/kg up to 400 mg every 6 weeks, followed by endoscopy with biopsies and surgery 3 to 5 weeks later. Median age was 74 years, 72% were women, 60% had clinical stage III disease, and pCR was defined as Mandard TRG1 without lymph node status.

Among surgical patients, pCR was 44% (37/84; 95% CI, 33 to 55) and MPR was 57% (48/84; 95% CI, 46 to 68). Pathologic complete response rates were 67% in stage I, 56% in stage II, and 33% in stage III disease. The stage I-II versus stage III comparison was 61% versus 33%, with P = .02 and adjusted OR 3.1 (95% CI, 1.2 to 8.1). The stage I-II versus stage III MPR comparison was 70% versus 49% with P = .07, and 82 patients achieved R0 resection, or 98% (95% CI, 92 to 100). Twelve resection specimens were node-positive, and 35 of 84 patients had both pCR and negative nodes, showing stage-dependent regression without conflating pCR with nodal clearance.

The median interval from pembrolizumab to surgery was 32 days, with seven patients beyond the predefined 5-week window and two beyond 7 weeks. Preoperative endoscopy was completed in 84 of 85 patients at a median of 23 days after treatment, and surgery followed 8 days later. Surgical complications occurred in 31 of 84 patients, including eight Clavien-Dindo grade 3b or higher events, four readmissions, and a median hospital stay of 4 days. Grade 3 adverse events affected 9 of 85 patients, or 11% (95% CI, 5 to 19), with no grade 4 or 5 adverse events. Five treatment-related serious adverse events included hepatitis, colitis, alanine aminotransferase increase, adrenal insufficiency, and asymptomatic myositis or myocarditis, while two deaths within 30 days were surgery-related.

In the centrally reviewed preoperative endoscopy analysis, 81 patients had images available and 76 had adequate quality. Optical assessment predicted pCR with 77% sensitivity, 93% specificity, 90% PPV, 83% NPV, and 86% accuracy; stage I-II versus stage III values were 100%, 92%, 97% versus 50%, 93%, 78%. Biopsies in 81 patients showed 68% sensitivity, 75% specificity, 66% PPV, 76% NPV, and 72% accuracy. In 42 baseline samples and 40 postimmunotherapy samples, 22 were ctDNA-positive before treatment and 9 after; in paired samples, ddPCR showed a -0.15 change in percent mutated alleles (95% CI, -0.27 to -0.04; P = .011), and post-treatment ctDNA negativity had 28% specificity for pCR. At 18.4 months median follow-up, OS was 98%, DFS 96%, and one recurrence occurred; no comparator arm, no watch-and-wait arm, short endoscopy-to-surgery timing, and need for longer follow-up limited interpretation.