Mismatch repair deficiency (dMMR) is present in approximately 15% of non-metastatic colon cancers and is characterized by a defective DNA mismatch repair system, leading to hypermutated and highly immunogenic tumors. While chemotherapy offers limited benefit to patients with dMMR colon cancer, neoadjuvant anti-PD-1 plus anti-CTLA-4 immunotherapy has demonstrated exceptional responses. This study hypothesized that the combination of anti-PD-1 (nivolumab) and anti-LAG-3 (relatlimab) would exhibit high efficacy with a favorable toxicity profile in patients with locally advanced dMMR colon cancer, building upon data from melanoma studies showing the efficacy and safety of this combination. The rationale stems from the observation that LAG-3, an immune checkpoint frequently expressed by exhausted T cells, could be a promising therapeutic target in this context, particularly given prior evidence of high LAG-3 expression in tumor-infiltrating lymphocytes within dMMR colon cancers. The NICHE-3 study aimed to evaluate this hypothesis using a short regimen of two cycles of nivolumab and relatlimab before surgery, with the primary endpoint being pathologic response (≤50% residual viable tumor).

Immune checkpoint blockade (ICB) has become a cornerstone treatment for various malignancies, including metastatic dMMR colorectal cancer. The rapid integration of immunotherapy into earlier disease stages reflects its success across multiple tumor types. Previous studies, notably NICHE-2, showed exceptional pathologic response rates (98%) in patients with locally advanced dMMR colon cancer treated with neoadjuvant nivolumab/ipilimumab. Several smaller studies corroborated these high pathologic and clinical complete response rates. However, a contrasting finding emerged from the FOXTROT study, which demonstrated a significantly lower pathologic response rate (7%) in dMMR tumors following neoadjuvant chemotherapy, highlighting the potential advantage of immunotherapy in this setting.

The NICHE-3 study (NCT03026140) was an investigator-initiated, multi-center, non-randomized, open-label, phase 2 trial employing a Simon's two-stage design. Between December 15, 2022, and April 4, 2024, 67 patients were screened, with 59 enrolled and treated. Inclusion criteria included age ≥18 years, locally advanced (≥T3 and/or N+) resectable dMMR colon adenocarcinoma, no distant metastases, WHO performance status 0 or 1, and adequate organ function. Exclusion criteria encompassed obstruction/perforation, prior ICB or chemotherapy, other malignancies (except negligible risk), immunodeficiency, autoimmune disease, and conditions requiring >10 mg prednisone daily. Patients received two 4-weekly cycles of nivolumab (480 mg) and relatlimab (480 mg) followed by surgery within 6-8 weeks. The primary endpoint was pathologic response (≤50% residual viable tumor). Secondary endpoints included safety (irAEs graded by CTCAE v4.0), major pathologic response (MPR; ≤10% RVT), pathologic complete response (PCR; 0% RVT), disease-free survival (DFS), and overall survival (OS). Pathological assessments were centrally reviewed, and staging followed the AJCC 8th edition. Statistical analyses included descriptive statistics, Clopper-Pearson method for proportions, Kaplan-Meier method for survival analyses, and R and SAS software.

Of the 59 enrolled patients, 57 (97%; 95% CI: 88–100%) achieved a pathologic response, meeting the primary endpoint. Fifty-four (92%; 95% CI: 81–97%) patients had an MPR, and 40 (68%; 95% CI: 54–79%) had a PCR. At a median follow-up of 8 months, only one patient experienced disease recurrence. All-grade irAEs occurred in 80% of patients, with grade 3-4 irAEs in 10%. The most frequent irAEs were infusion-related reactions (29%), thyroid dysfunction (22%), and fatigue (20%). Pathologic response rates were consistent across clinical stages (cT2-3 vs. cT4a/b; cN0 vs. cN+), although a numerically lower PCR rate was observed in patients with elevated baseline CEA levels (≥5 µg/L). Two patients had lymph node metastases in the resection specimen, but both showed additional nodes with complete tumor regression. Postoperative adverse events (AEs) of any grade occurred in 37% of patients, with grade 3-4 AEs in 7%.

This study demonstrates the exceptional efficacy of a short neoadjuvant regimen of nivolumab/relatlimab in patients with locally advanced dMMR colon cancer, far surpassing the response rates observed with neoadjuvant chemotherapy. The high MPR and PCR rates align with previous findings associating these with excellent outcomes in various tumor types. While PCR was observed across tumor stages, patients with cT4 tumors and higher baseline CEA levels exhibited numerically lower PCR rates, suggesting a potential link between tumor burden and response. The high MPR rates could indicate a dynamic treatment effect, where a longer interval to surgery might yield higher PCR rates in advanced tumors. Although combination therapy often increases toxicity, the nivolumab/relatlimab regimen displayed an acceptable safety profile. The observed higher rates of certain irAEs compared to previous studies using different regimens may be attributed to the relatlimab dose. Ongoing studies within the NICHE platform are investigating alternative treatment schedules to optimize the balance between efficacy and toxicity. The results support the need for larger, randomized trials comparing neoadjuvant immunotherapy regimens to adjuvant chemotherapy and exploring organ preservation strategies for dMMR colon cancer. While radiographic assessment of response after neoadjuvant immunotherapy is inaccurate, complementary modalities like circulating tumor DNA are being explored to enhance response assessment accuracy.

This phase 2 trial provides compelling evidence for the high efficacy of neoadjuvant nivolumab/relatlimab in locally advanced dMMR colon cancer, achieving remarkable pathologic response rates and an acceptable safety profile. The findings support the need for larger studies to confirm these results and to further explore optimal treatment schedules and strategies for organ preservation. Future research should also focus on refining response assessment methodologies and investigating long-term outcomes.

Limitations of this study include the inherent inaccuracies of radiographic tumor staging in neoadjuvant trials for colon cancer, particularly for lymph node metastases, potentially leading to overtreatment in some patients. The relatively short follow-up period may also limit the definitive assessment of long-term outcomes. The non-randomized nature of the study prevents definitive conclusions about causality. Furthermore, the study may not be generalizable to the entire population of patients with locally advanced dMMR colon cancer due to the inclusion/exclusion criteria.