Inflammation promotes resistance to immune checkpoint inhibitors in high microsatellite instability colorectal cancer

Abstract

Inflammation is a common medical complication in colorectal cancer (CRC) patients, which plays significant roles in tumor progression and immunosuppression. However, the influence of inflammatory conditions on the tumor response to immune checkpoint inhibitors (ICI) is incompletely understood. Here we show that in a patient with high microsatellite instability (MSI-H) CRC and a local inflammatory condition, the primary tumor progresses but its liver metastasis regresses upon Pembrolizumab treatment. In silico investigation prompted by this observation confirms correlation between inflammatory conditions and poor tumor response to PD-1 blockade in MSI-H CRCs, which is further validated in a cohort of 62 patients retrospectively enrolled to our study. Inhibition of local but not systemic immune response is verified in cultures of paired T cells and organoid cells from patients. Single-cell RNA sequencing suggests involvement of neutrophil leukocytes via CD80/CD86-CTLA4 signaling in the suppressive immune microenvironment. In concordance with this finding, elevated neutrophil-to-lymphocyte ratio indicates inhibited immune status and poor tumor response to ICIs. Receiver operating characteristic curve further demonstrates that both inflammatory conditions and a high NLR could predict a poor response to ICIs in MSI- CRCs, and the predictive value could be further increased when these two predictors are combined. Our study thus suggests that inflammatory conditions in MSI-H CRCs correlate with resistance to ICIs through neutrophil leukocyte associated immunosuppression and proposes both inflammatory conditions and high neutrophil-to-lymphocyte ratio as clinical features for poor ICI response.

Fig. 1. Inflammatory conditions are associated with poor response to PD-1 blockade in dMMR/MSI-H CRCs.

A CT scans were conducted on Patient 1 after 0 courses, 3 courses and 9 courses of treatment. Images of the primary tumor and metastatic lesions are shown. Tumor lesions are marked by arrows. B, C Tumor response (B) and progression-free survival (C) after PD-1 blockade were compared between MSI-H CRC patients with (N = 14) and without (N = 48) inflammatory conditions. The Wald chi-square test (B) and Kaplan-Meier method with Log-rank test (C) were used. D Bright field (Scale bar: 100 μm) images of tumor organoids from 4 patients are shown. Each experiment was repeated 3 times. E Tumor organoids were co-cultured with paired TILs or PBMC-derived T cells for 6 h and were then separated for apoptosis assays. The two-tailed unpaired Student’s t test was used. The mean values are shown, and the standard deviations are displayed by the error bars. Each experiment was repeated 3 times with 3 replicates. Source data are provided as a Source Data file. PR partial response, CR complete response, SD stable disease, PD progressed disease, TIL tumor-infiltrating lymphocyte, PBMC peripheral blood mononuclear cell.

Fig. 2. Elevated neutrophil infiltration is associated with inhibited tumor immune status.

A, B The 4489 qualified cells were divided into eight epithelial cells and five groups of immune cells using a UMAP plot (A). Cell number in each cluster is shown (B). C The expression level of neutrophil marker genes in each cluster is shown using a violin plot. D Data of MSI CRC cases in the TCGA were analyzed using ImmuCellAI. Infiltration levels of neutrophils were compared between patients who predicted poor response (N = 72) and good response (N = 117) to ICIs. The median values are shown as the center of the box, and the ranges are displayed by the whiskers. The quartiles are shown as the bounds. The two-tailed unpaired Student’s t test was used. E Correlation between total infiltration scores and neutrophil infiltration levels was analyzed. The Spearman rank correlation test was used. F Correlations between the infiltration levels of cytotoxic T cells, Th1 cells, NK cells, B cells and neutrophils were analyzed. The Spearman rank correlation test was used. Source data are provided as a Source Data file.

Fig. 3. Single-cell sequencing reveals an inhibitory role of neutrophils in tumor immune status through CD80/CD86-CTLA4 axis.

A, B Interactions between subtypes of immune cells were analyzed. Functional phenotypes and predicted interactions of myeloid cells and T cells (A), and ligand-receptor interactions in immune checkpoints between immune cells are shown (B). C Myeloid cells were divided into 2 clusters (N2 and non-N2) using a UMAP plot. D The expression level of CD80, CD86, CCL5 and CSF1R in N2 and non-N2 cluster is shown. E Murine T cells, T cells pre-stimulated with neutrophils, T cells pre-stimulated with neutrophils in the presence of CD80 + CD86 neutralizing antibodies (anti-CD80/CD86) were pre-treated with PD-1 neutralizing antibodies and then co-cultured with MC38 cells for 6 h. Apoptosis assays were used to evaluate the proportion of apoptosis in each group. The two-tailed unpaired Student’s t test was used. The mean values are shown, and the standard deviations are displayed by the error bars. Each experiment was repeated 3 times with 3 replicates. Source data are provided as a Source Data file. Neu neutrophil.

Fig. 4. Elevated NLR is associated with poor immune status and resistance to ICIs.

A In a cohort of 142 surgically resected dMMR CRCs, the CD4⁺, CD8⁺ and FOXP3⁺ TILs in the IM, TS and CN were compared between patients with NLR ≤ 3 (N = 79) and NLR > 3 (N = 63). The two-tailed unpaired Mann-Whitney U rank-sum test was used. The median values are shown as the center of the box, and the ranges are displayed by the whiskers. The quartiles are shown as the bounds. B In 58 MSI-H CRCs receiving ICIs, the ratio of local inflammation was compared between patients with NLR ≤ 3 (N = 38) and NLR > 3 (N = 20). The Wald chi-square test was used. C Tumor response to ICIs were compared between MSI-H CRC patients with NLR ≤ 3 and NLR > 3. The Wald chi-square test was used. D ROC curve analysis was conducted to evaluate the predictive performance of the prognostic features for ICI response. Source data are provided as a Source Data file. IM invasive margin, TS tumor stromal, CN cancer nest, NLR neutrophil-to-lymphocyte ratio, PR partial response, CR complete response, SD stable disease, PD progressed disease, ROC receiver operating characteristic.