Gastroesophageal junction (GEJ) cancer has limited treatment advancements, with most patients experiencing disease progression within 5 years. Adjuvant nivolumab is now FDA-approved following complete resection and residual pathologic disease after neoadjuvant chemoradiotherapy (CRT). Early-stage tumors often show high programmed cell death ligand 1 (PD-L1) expression, suggesting immune resistance potentially reversed by anti-PD-1 antibodies. Neoadjuvant CRT may prime PD-L1 expression, justifying its combination with immune checkpoint inhibitors (ICIs). Lymphocyte-activation gene 3 (LAG-3) is another co-inhibitory receptor highly expressed in GEJ cancers. Combined anti-LAG-3 and anti-PD-1 therapy could potentially enhance antitumor responses by reinvigorating exhausted T cells. However, data on neoadjuvant PD-1 inhibitors with CRT in GEJ cancer are limited and conflicting. This study aimed to evaluate the safety, feasibility, and efficacy of neoadjuvant nivolumab or nivolumab plus relatlimab (an anti-LAG-3 inhibitor) combined with CRT in resectable stage II/III GEJ cancer, alongside pathological response, circulating tumor DNA (ctDNA) burden contraction, and systemic neoantigen-specific T cell responses. The integration of immunotherapy into GEJ cancer treatment is crucial, but its broad efficacy remains elusive. Neoadjuvant immunotherapy's application relies on the primary tumor as a source of tumor antigens to prime tumor-specific T cells and address micrometastatic disease driving recurrence. Analyzing ctDNA can help assess systemic tumor burden and response to neoadjuvant ICIs, potentially identifying patients who will achieve a pathological complete response (pCR) and those with minimal residual disease (MRD) at higher risk of relapse.

Several studies have explored the use of immunotherapy in combination with neoadjuvant chemoradiotherapy for various cancers, including lung cancer and melanoma, demonstrating potential benefits in terms of improved pathological response and overall survival. However, the results for esophageal or gastroesophageal junction (E/GEJ) cancers have been more inconsistent, with some trials showing promising results, while others have shown limited benefit or even increased toxicity. The CheckMate 577 study highlighted the role of adjuvant nivolumab after neoadjuvant CRT for patients with resected E/GEJ cancer with residual pathologic disease, leading to FDA approval. Preclinical and clinical evidence supports the idea that neoadjuvant CRT can enhance PD-L1 expression, making combined immunotherapy and CRT a promising approach. The expression of LAG-3, another immune checkpoint, further strengthens the rationale for combining anti-PD-1 and anti-LAG-3 therapies. Previous trials like RELATIVITY-047 demonstrated the potential synergistic effect of combining anti-PD-1 and anti-LAG-3 therapies in advanced melanoma. Several other phase II studies involving different ICIs in combination with neoadjuvant chemoradiotherapy for E/GEJ cancer have yielded varying results, highlighting the need for further research to determine the optimal immunotherapy strategy and its impact on long-term survival. Liquid biopsy techniques, specifically ctDNA analysis, have emerged as a promising tool for non-invasive monitoring of tumor response to therapy, offering valuable insights into minimal residual disease and predicting recurrence risk. However, the utility of ctDNA in the context of neoadjuvant immunotherapy combined with CRT for E/GEJ cancers needs further investigation.

This phase Ib, open-label, multi-institution trial enrolled 32 patients with resectable clinical stage II/III distal E/GEJ adenocarcinoma or squamous cell carcinoma. Patients were consecutively enrolled into two arms: Arm A (nivolumab + CRT) and Arm B (nivolumab + relatlimab + CRT). Nivolumab was administered every 2 weeks for two induction cycles, followed by three additional doses concurrently with CRT. In Arm B, initial patients received both nivolumab and relatlimab according to the same schedule; however, due to unacceptable toxicity, the protocol was amended to deliver nivolumab and relatlimab only during the induction phase for subsequent patients. Standard-of-care CRT consisted of weekly carboplatin and paclitaxel with radiotherapy (50.4 Gy in 28 fractions). Ivor Lewis esophagectomy was performed 6-8 weeks after CRT. The primary endpoint was safety (monitored using Bayesian stopping rules for specific toxicities), and the secondary endpoint was feasibility (assessed by the proportion of patients undergoing surgery without significant treatment-related delays). Exploratory endpoints included OS, RFS, PCR, MPR, and biomarker analyses. Pathological response was assessed using a modified Ryan scheme and the CAP tumor regression score. Immunohistochemistry was performed on baseline and resected tumor specimens to evaluate PD-L1, LAG-3, HER2, and MMR protein expression. Ultra-sensitive targeted next-generation sequencing (NGS) was used to analyze serial plasma and matched WBC DNA samples for ctDNA detection and longitudinal tracking of tumor-specific alterations. Neoantigen-specific T cell responses were assessed using the MANAFEST assay in selected patients. Statistical analyses included Kaplan-Meier method for survival analysis, log-rank test for comparing survival curves, Fisher's exact test and Wilcoxon rank-sum test for other comparisons.

The study met its primary safety endpoint for Arm A, but Arm B required protocol amendment to mitigate toxicity. The most common adverse events were fatigue, nausea, thrombocytopenia, and dermatitis. Grade 3 or higher treatment-related adverse events (TRAEs) occurred in 31.3% of patients (18.8% in Arm A, 43.8% in Arm B). Immunotherapy-related adverse events (irAEs) occurred in 53.1% of patients (31.3% in Arm A, 75% in Arm B), with dermatitis being the most frequent. The PCR rate was 40% in Arm A and 21.4% in Arm B; MPR was 53.5% in Arm A and 57.1% in Arm B. Two-year RFS rates were 62.5% (Arm A) and 87.1% (Arm B); two-year OS rates were 75% (Arm A) and 93.8% (Arm B). Higher baseline PD-L1 (CPS ≥ 5) and LAG-3 expression were associated with deeper pathological responses, particularly in Arm B. ctDNA analysis revealed that patients with undetectable ctDNA post-ICI induction had significantly longer RFS and OS (log-rank, P = 0.038 and P = 0.032, respectively). ctDNA clearance mirrored neoantigen-specific T cell responses. In patients who did not achieve PCR, ctDNA negativity post-ICI induction predicted a longer RFS (log-rank, P = 0.058).

This study provides valuable insights into the safety and feasibility of combining PD-1 and LAG-3 blockade with CRT in the neoadjuvant setting for resectable E/GEJ cancers. While Arm B showed increased toxicity requiring a protocol amendment, the long-term efficacy suggests further investigation is warranted. The higher pathological response rates in Arm A, despite a lower proportion of SCC patients, highlights the potential benefit of nivolumab in E/GEJ cancer. The improved survival observed in Arm B despite a lower PCR rate and higher toxicity could be linked to the association between irAEs and improved ICI efficacy. The results underscore that PCR may not be the most reliable predictor of long-term survival in this setting. This study's most significant finding is the strong correlation between ctDNA clearance and favorable outcomes, suggesting ctDNA could be a valuable biomarker for monitoring treatment response and predicting long-term survival. The association of ctDNA kinetics with neoantigen-specific T cell responses further strengthens this observation. This study demonstrates the potential of using ctDNA to guide treatment decisions, potentially allowing for early identification of patients who would benefit most from neoadjuvant immunotherapy and those at higher risk of relapse requiring further intervention.

Neoadjuvant nivolumab with CRT is an active and safe regimen for operable E/GEJ cancer, showing superior results compared to historical controls. While relatlimab addition increased irAEs, long-term efficacy warrants further investigation on optimal sequencing. ctDNA assessment appears superior to traditional pathological assessment in predicting long-term clinical outcomes, making it a crucial tool for personalized treatment decisions and future intervention strategies. Larger studies are needed to validate these findings.

The relatively small sample size is a key limitation of this study. The high rate of undetectable ctDNA in this early-stage population may limit the generalizability of the ctDNA findings. The study's exploratory nature and the limited number of patients in each subgroup also restrict the ability to draw definitive conclusions on certain aspects of the biomarker analysis. Longer follow-up is needed to fully assess long-term outcomes and the durability of response.