Immune Landscape of Colorectal Cancer Tumor Microenvironment from Different Primary Tumor Location

Abstract

To define differences in tumor microenvironment (TME) immune phenotypes between right and left colorectal cancers (CRCs) and explore their therapeutic implications. Gene expression profiling and clinical characteristics of patients with CRC were retrieved from The Cancer Genome Atlas data portal. Immune cell infiltration was estimated based on single-sample gene set enrichment analysis. CRCs tissue microarrays (TMAs) containing 90 consecutive cases of surgical samples were used for validation. Expression of CD8A and VEGFA was confirmed by immunohistochemistry (IHC) analysis with TMAs, and overall survival (OS) was analyzed. Expression profiling data demonstrated that CRC immune microenvironment from right side tumor was characterized as increased infiltration of immune cells with enhanced cytotoxic function, based on higher cytotoxic activity scores (CYT) and interferon-γ signatures. Expression of VEGFA, which could be neutralized by bevacizumab, was associated with decreased levels of activated CD8⁺ T-cells, Th1 cells, and PRF1 expression on the right side, but not on the left side. IHC analysis of TMAs further confirmed an inverse correlation between CD8A and VEGFA expression, and revealed a favorable OS for patients with CD8A^HiVEGFA^Lo disease among right-side CRCs. For the left side, higher CD56^bright natural killer cell infiltration and active 4-1BB/IFN-ɑ signaling, which could providing a favorable condition for cetuximab-mediated antibody-dependent cell-mediated cytotoxicity effect, was present in a cohort with extended OS. In the TME, features of immune phenotype sidedness were identified, providing an implication for differential responses to bevacizumab/cetuximab treatment. In addition, a new avenue for innovative experimental design and combinational immunotherapy to treat CRC patients was suggested.

Keywords: immunophenotype; left-side colorectal cancer; right-side colon cancer; therapeutic implications; tumor microenvironment.

Figure 1

Immune landscape of colorectal cancer. (A) Unsupervised clustering of 638 patients from the The Cancer Genome Atlas cohort using single-sample gene set enrichment analysis scores from 27 immune cell types. Primary tumor location, mutation status of EGFR, BRAF, KRAS, and TP53, metastasis, gender, MSI, polyps, survival, anatomic location, as well as stage were annotated in the lower panel. Hierarchical clustering was performed with Euclidean distance and Ward linkage. Three distinct immune infiltration clusters, here termed high infiltration, median infiltration, and low infiltration, were defined. (B) Relative cytolytic score (CYT) and expression of IFN-γ between immune infiltration high, median, and low tumors clustered by overall immune cell infiltration. Two-tailed Student’s t-tests were used for all analyses. Error bars represent the mean ± SEM. *p < 0.05; **p < 0.01; ***p < 0.001; p ≥ 0.05, not significant.

Figure 2

Heterogeneous immune cell infiltration in right-side and left-side colorectal cancer (CRC). (A) Immune infiltration in right-side and left-side CRC based on immune infiltration clustering scores [right side: high (H) = 81; median (M) = 82; low (L) = 36 vs left side: H = 102; M = 160; L = 97, p = 0.0052]; the absolute value and composition ratio of different immune infiltration statuses are show on the right. (B) Distribution of immune cell infiltration between different primary tumor locations. Higher than 1, more infiltration in left side; lower than 1, less infiltration in right side. (C) Comparison of relative cytotoxic activity scores (CYT) between right- and left-side CRC (p = 7.73e−6). (D) Relative antigen presentation machinery (APM) between right-side and left-side CRC (p = 4.16e−5). (E) Relative interferon-γ signature between right- and left-side CRC (p = 0.0004). (F) Relative T-cell infiltration score (TIS) between right- and left-side CRC (p = 6.942e−10). (G) Relative CD8 T-cell/Treg ratios between right- and left-side CRC (p = 7.51e−11). Two-tailed Student’s t-tests were used for all analyses. Error bars represent the mean ± SEM. *p < 0.05; **p < 0.01; ***p < 0.001; p ≥ 0.05, not significant.

Figure 3

Enhanced CD56^bright natural killer (NK) cell response in left-side colorectal cancer (CRC). (A) CD56^bright NK cells in right- and left-side CRC (left vs right, p = 0.012). (B) CD56^bright NK cell infiltration was shown to correlate with patient survival [left: n(CD56^bright NK^Hi) = 179, n(CD56^bright NK^Lo) = 180, p = 0.0225; right: n(CD56^bright NK^Hi) = 99, n(CD56^bright NK^Lo) = 100, p = 0.226]. (C) The 4-1BB signaling pathway and interferon-α response have different roles in right- and left-side CRC. In the left side [4-1BB alone: n(4-1BB^Hi) = 179, n(4-1BB^Lo) = 180, p = 0.00427; IFNA alone: n(IFNA^Hi) = 179, n(IFNA^Lo) = 180, p = 0.0187; 4-1BB combined with CD56^bright NK cells: n(NK^Hi4-1BB^Hi) = 89, n(NK^Hi4-1BB^Lo) = 90, n(NK^Lo4-1BB^Hi) = 90, n(NK^Lo4-1BB^Lo) = 90, p = 0.000424; IFNA combined with CD56^bright NK cells: n(NK^Hi IFNA^Hi) = 89, n(NK^Hi IFNA^Lo) = 90, n(NK^Lo IFNA^Hi) = 90, n(NK^Lo IFNA^Lo) = 90, p = 0.00973], in the right side [4-1BB alone: n(4-1BB^Hi) = 99, n(4-1BB^Lo) = 100, p = 0.961; IFNA alone: n(IFNA^Hi) = 99, n(IFNA^Lo) = 100, p = 0.549; 4-1BB combined with CD56^bright NK cells: n(NK^Hi4-1BB^Hi) = 49, n(NK^Hi4-1BB^Lo) = 50, n(NK^Lo4-1BB^Hi) = 50, n(NK^Lo4-1BB^Lo) = 50, p = 0.428; IFNA combined with CD56^bright NK cells: n(NK^Hi IFNA^Hi) = 49, n(NK^Hi IFNA^Lo) = 50, n(NK^Lo IFNA^Hi) = 50, n(NK^Lo IFNA^Lo) = 50, p = 0.594]. Two-tailed Student’s t-tests and Kaplan–Meier survival log-rank test were used for those analyses. Error bars represent the mean ± SEM. *p < 0.05; **p < 0.01; ***p < 0.001; p ≥ 0.05, not significant.

Figure 4

VEGFA hinders T-cell infiltration in right-side colorectal cancer (CRC). Correlations between (A) VEGFA and T-cell infiltration (right side: r = −0.267, p = 0.0001; left side: r = −0.115, p = 0.029), (B) VEGFA and Th1-cells (right side: r = −0.239, p = 0.0007; left side: r = −0.111, p = 0.034), and (C) VEGFA and perforin (PRF1) expression (right side: r = −0.187, p = 0.007; left side: r = −0.02, p = 0.702). (D) Infiltration of CD8 cells and expression of VEGFA detected in CRC tissue microarrays (20×). (E) The number of CD8 alpha-positive cells in the VEGFA low and high group in right-side tumors. The group was divided into two subgroups based on the median value (*p = 0.0386). (F) Correlation between VEGFA and CD8 alpha-positive cells in right-side tumors (r value represents Pearson correlation coefficient). Two-tailed p-values are presented for significance (<0.05; r = −0.3903, p = 0.0441; n = 28). (G) Kaplan–Meier curves for cancer-specific survival based on CRC immune infiltration classes. The CD8^hiVEGFA^lo class had the best survival, whereas other classes were associated with poor outcome (log-rank test, p = 0.0082). Two-tailed Student’s t-tests and Kaplan–Meier survival log-rank test were used for those analyses. Error bars represent the mean ± SEM. *p < 0.05; **p < 0.01; ***p < 0.001; p ≥ 0.05, not significant.

Figure 5

The colorectal cancer (CRC) immune landscape and implications for immunotherapy. Right-side colon (cecum to transverse colon) cancer and left colorectal (splenic flexure to rectum) cancer have different mutation loads, immune infiltration, and blood supplies; furthermore, different immune cell infiltration profiles, signaling pathways, and effector molecule in right-side colon cancer and left-side CRC might lead to different treatment responses. All of these differences might lead to different prognoses with immunotherapy.