Why This Study Matters
Colorectal cancer is the most commonly diagnosed gastrointestinal cancer and a leading cause of cancer-related death in the United States, with approximately 154,270 new cases and 52,900 deaths in 2025, according to the paper. The authors note that there have been few advances in treatment options for locally advanced colorectal cancer since the introduction of FOLFOX in the adjuvant or preoperative setting.
Immune checkpoint blockade has so far been effective mainly in mismatch repair-deficient (dMMR)/microsatellite instability-high cancers, which comprise only 5% to 15% of all colorectal cancer cases. For mismatch repair-proficient (pMMR) disease, the paper’s abstract states that effective immunotherapy is lacking. NEST tests whether botensilimab (BOT), an Fc-enhanced anti–CTLA-4 antibody, combined with the anti–PD-1 antibody balstilimab (BAL), can change that picture when given before surgery, while the primary tumor and regional lymph nodes are still intact.
Study Design
NEST (NCT05571293, registered in October 2022) was a single-center, nonrandomized, open-label, phase II exploratory study conducted at Weill Cornell Medicine/NewYork-Presbyterian Hospital. All patients received intravenous BOT 75 mg on day 1 plus BAL 240 mg, followed by surgical resection:
| Cohort | Regimen | Population | Resection timing |
|---|---|---|---|
| A (NEST-1) | BOT 75 mg day 1; BAL 240 mg days 1 and 15 | n = 10 eligible (7 pMMR, 3 dMMR) | 1 to 6 weeks after the second BAL dose |
| B (NEST-2) | BOT 75 mg day 1; BAL 240 mg days 1, 15, 29 and 43 | n = 13, pMMR | 1 to 6 weeks after the fourth planned BAL dose |
| C (NEST-2) | Same as cohort B | n = 1, dMMR | Same as cohort B |
Objectives. The primary objectives were safety, feasibility (based on surgery delay) and efficacy (pathologic response). Secondary endpoints included ctDNA changes before and after resection, and a prespecified secondary analysis of pathologic response in combined pMMR and dMMR tumors. Exploratory work examined the tumor microenvironment. Pathologic response was graded with the Mandard system; MPR was defined as ≤10% residual viable tumor cells (≥90% response).
Patient Population
Adults 18 years or older with nonmetastatic, histologically proven colon adenocarcinoma awaiting surgical resection were eligible; rectal cancer was eligible if radiotherapy was not planned and up-front resection was recommended. Prior immune checkpoint inhibitors or other systemic therapy, and contraindications to immunotherapy, were exclusions.
24 eligible patients treated at a single center, with 26 tumors evaluated (22 pMMR, 4 dMMR).
Median age 64 years (range, 35–75); 14 patients (58%) had T3 or higher tumors.
Of the 24 patients, 13 (54%) were male and 11 (46%) female; 12 were White, 4 Asian, 4 Black and 4 of other race. Two patients in cohort B had synchronous primary tumors, which is why 26 tumors were evaluated in 24 patients. Median follow-up at the March 31, 2026 data cutoff was 32.2 months for cohort A and 23.5 months for cohorts B/C. Enrollment closed for an administrative reason (a principal investigator left the institution), so cohort C enrolled only one patient.
Primary Endpoint Results: Safety, Feasibility and Pathologic Response
“Neoadjuvant BOT/BAL was safe and did not delay planned surgery in any patient.”
— Source: abstract, Shah et al., Clin Cancer Res 2026
The median time to surgery from the start of treatment was 29 days (range, 21–37) in cohort A and 55 days (range, 21–104) in cohorts B and C. All patients underwent minimally invasive resection, with one conversion to open surgery.
| Pathologic response | Result | 95% CI |
|---|---|---|
| pMMR tumors, MPR (cohorts A + B; n = 22) | 41% (9/22) | 21%–64% |
| pMMR tumors, pCR | 32% (7/22) | 14%–55% |
| pMMR tumors, pathologic response (>50% tumoral response) | 59% (13/22) | 36%–79% |
| dMMR tumors, MPR (n = 4) | 100% (4/4); two pCR, one each 99% and 98% treatment effect | 40%–100% |
| Cohort A pMMR tumors, MPR | 29% (2/7) | 4%–71% |
| Cohort B pMMR tumors, MPR | 47% (7/15) | 21%–73% |
At the patient level, the pathologic overall response (pOR) rate was 70% (7/10) in cohort A overall and 57% (4/7) for pMMR patients in that cohort, and 62% (8/13) in cohort B. The authors state the study was not designed to compare the two treatment schedules.
ctDNA and recurrence. Nine patients had baseline ctDNA collected and all were positive (range, 0.06–54.65 mean tumor molecules/mL). Of eight with a presurgical sample, seven (88%) had ctDNA clearance, and ctDNA remained undetectable after resection in all patients. At the data cutoff, no patient had colorectal cancer recurrence.
Subgroup and Exploratory Analyses
| Analysis | Finding (as reported) |
|---|---|
| Tumor mutational burden, PD-L1 combined positive score | No associations with treatment response (no statistics given in the text) |
| KRAS G12D (n = 2) or activating BRAF (n = 1, V600R) | No pathologic response (0%–10% response) |
| Other KRAS/BRAF mutations or wild-type (n = 19) | 3 of 19 (14%) had no pathologic response (P = 0.048) |
| Prominent Crohn disease–like reaction | 73% of pMMR cases with >50% pathologic response vs 13% without response (P = 0.028) |
| Immune cell density at resection, responders vs nonresponders | 3.1 vs 0.52 immune cells/μm (P = 0.0048) |
| CD8+/Treg ratio, baseline to resection (all tumors) | 0.286 to 6.20; CD8+ T-cell density 13-fold higher, CD8+ share of infiltrate 5% to 30% (P = 0.077) |
The authors caution that the RAS/RAF observation is limited by the small sample size and needs validation. The multiplex imaging analysis covered a subset of samples (6 baseline biopsies and 11 resection specimens), and tumors that responded showed organized, co-localized CD8+ T-cell and B-cell infiltrates, while nonresponding tumors showed spatially isolated infiltrates with a greater proportion of FOXP3+ regulatory T cells.
Safety Profile
| Cohort | Reported events |
|---|---|
| All cohorts | No grade 4 toxicities; no deaths or study discontinuations; 2 of 24 patients (8%) had a Clavien–Dindo grade 1 complication |
| Cohort A (n = 10) | One patient with grade 2 diarrhea/colitis resolved with TNFα inhibition; five patients with fever, chills and fatigue within 2 weeks of dosing (one with grade 3 fatigue); 2 of 10 required high-dose steroids and one required TNFα inhibition |
| Cohorts B and C (n = 14) | Colitis/diarrhea n = 5 (grade 2, n = 2; grade 3, n = 3); grade 2 myositis/myocarditis n = 1 requiring high-dose steroids and infliximab. Nine received all planned treatment; 3 received 3 BAL doses (colitis); 2 received 2 BAL doses (colitis n = 1, myositis n = 1) |
The authors write that more patients had immune-mediated adverse events in NEST-2 (5/14 vs 2/10 in NEST-1) and that this toxicity suggests more than two doses of immunotherapy in the neoadjuvant setting may not be feasible.
Interpretation and Broader Context
The authors place NEST alongside the NICHE trials of nivolumab/ipilimumab (external source: cross-study context from the paper’s Discussion; not a within-trial comparison, and the trials differ in design and size). They report that NICHE-1 showed a median time to surgery of 33 days and an MPR of 19% (n = 6/31) in a similarly staged pMMR colon cancer population, versus 29 days and a 41% MPR (n = 9/22) in NEST. They also note a 3-year disease-free survival rate of 93% in NICHE-2 and a 2-year recurrence rate of 16.9% with neoadjuvant chemotherapy in FOxTROT, versus a 100% disease-free survival rate in NEST at 24 to 33 months of follow-up.
The authors conclude that the data support continued development of BOT/BAL in the neoadjuvant setting for pMMR colorectal cancer. They list open questions: whether therapy duration affects local and systemic outcomes, which biomarkers predict resistance, and why immune activation is effective in localized disease but diminished once metastases develop.
Limitations
Limitations noted by the authors: small sample sizes in both NEST-1 and NEST-2, relatively short follow-up (so durable efficacy is unknown), and a study underpowered to determine the optimal duration of neoadjuvant therapy versus number of immunotherapy treatments before surgery. Separately, the study was single-center and nonrandomized. The MPR for NEST-1 appears as 27% in the Discussion and as 2 of 7 (29%) in the Results; the Results figure is used here.
Funding and disclosures: This trial was developed as an investigator-initiated study and was supported by Agenus Inc. BOT and BAL were provided by Agenus Inc., which the paper states had no role in the study design, data analysis, or interpretation of results. The paper’s author disclosures list multiple relationships with Agenus Inc. and other companies.