Colorectal cancer (CRC) is a genetically heterogeneous disease, and patients with CRC show widely disparate outcomes. Recently, the significant immunotherapeutic benefit of immune checkpoint inhibitors (ICIs) in CRCs has been reported in patients with dMMR (mismatch repair-deficient) CRC (1–3). dMMR tumors may occur via several mechanisms. In Lynch syndrome CRC, the underlying mechanism is usually a germline mutation of one of the four (MLH1, MSH2, MSH6, and PMS2) MMR genes, leading to microsatellite instability (MSI). In contrast, in sporadic CRC, dMMR is primarily caused by epigenetic silencing through CpG methylation of the MLH1 gene promoter, among other causes (4). dMMR tumors are often accompanied by a higher density of tumor-infiltrating lymphocytes (TILs) and more cytotoxic T cells than mismatch repair-proficient (pMMR) tumors (5, 6). The most likely explanation for the vital benefit of immunotherapy is the high levels of neoantigens, which trigger T-cell response to tumor cells in patients with dMMR (7, 8).

The dMMR status of CRCs varies across different stages of the disease. Overall, approximately 15% of all CRCs have dMMR status, but dMMR status becomes less common in later stages: approximately 20% in stage II, 12% in stage III, and only 2% to 4% in stage IV (9, 10). Previous studies have proposed that this phenomenon is associated with a decrease in immune surveillance (11). Patients with early-stage CRC with dMMR have a significantly better prognosis and longer survival than pMMR patients; however, in contrast with early-stage CRCs, dMMR predicts a significantly worse prognosis in advanced-stage CRCs (12). Considering the relationship between T cells and the dMMR status, it is hypothesized that the heterogeneity of T cells contributes significantly to the transformation of the immune microenvironment from the early stage to the advanced stage in dMMR CRCs.

The response rate to immunotherapy differs in early- and advanced-stage dMMR tumors. Immunotherapy may be highly effective in patients with early-stage dMMR CRC. In an exploratory NICHE study (NCT03026140) of neoadjuvant treatment with nivolumab (anti–PD-1 antibody) and ipilimumab (anti–CTLA-4 antibody) in patients with dMMR, the pathological response rate was 100% in 20 patients with dMMR (13). However, in advanced-stage dMMR tumors, a meta-analysis of 939 dMMR/MSI-H (MSI-high) patients pretreated with ICIs from 14 studies indicated a pooled overall response rate of 41.5% (14). In addition, certain biomarkers can predict the efficacy of ICIs, such as the Immunoscore, which measures TILs in the core and invasive margins of a tumor (15), the tumor mutation burden (16), and the T-cell phenotype (17). Although the mechanisms of immunotherapy resistance are unclear, some possible mechanisms have been proposed, such as the infiltration of CD8+ T cells or FOXP3+ cells (18, 19), the limited repertoire of cytotoxic T cells (20), and the absence of memory T cells (21). T-cell immunotherapies have shown great promise in patients with advanced cancer and have revolutionized treatment (22). Therefore, we presumed that the heterogeneity of T cells played a crucial role in ICI response at different stages of dMMR tumors. Among all the immune checkpoints, we only explored PD-1/PD-L1 because this checkpoint has been well studied, and a considerable amount of literature on it has been published. The PD-1/PD-L1 inhibitors are the most widely used ICIs. Current research is investigating opportunities for their use in every stage of CRC (23). Although this area has been the focus of much research, knowledge of TILs between early-stage and advanced-stage dMMR CRC is limited.

In the present study, we sought to study the distribution of individual immune cells subsets, especially T cells, in early- and advanced-stage dMMR CRCs and to reveal their effect on patient survival. We designed two multiplex immunohistochemistry (mIHC) panels consisting of immune cell markers. The following T-cell markers were considered: CD3, which is highly expressed by T cells; CD4, which is a marker of T helper cells; CD8, which is highly expressed by cytotoxic T cells; CD8+CD45RO+, which is a memory cytotoxic T-cell marker; CD4+FOXP3+, which is a marker of CD4+ regulatory cells; CD68, which is a pan-macrophage marker; CD68+CD163−, which represents M1-like macrophages; CD68+CD163+, which is a marker of M2-like macrophages; and PD-1 and PD-L1, which are immune checkpoint markers (24–28). The simultaneous quantification of various markers in both epithelial regions and stromal regions profiles the immune characteristics of dMMR CRCs across tumor stages and may assist in the identification of potential prognostic markers.

Immune cells play important roles in inflammation and tumorigenesis, changing normal colorectal tissues to tumor tissues, which, in turn, increase the immune cell imbalance in tumors. In previous research, the infiltration of 22 immune cell types in 404 CRCs and 40 adjacent non-tumorous tissues was analyzed using CIBERSORT, a deconvolution algorithm. Macrophages and CD4 memory-activated T cells were found to be infiltrated significantly more in CRCs than in normal tissues in both The Cancer Genome Atlas (TCGA) and Gene Expression Omnibus (GEO) cohorts (30). Inflammation could affect CRC tumorigenesis and modulate the polarization of cells in the TME and the corresponding cytokine milieu, causing tumor-elicited inflammation (31). In inflammation-associated tumorigenesis, chronic inflammation resulting from infections, dysregulated immune responses, or environmental factors can initiate and promote tumorigenesis through the induction of DNA damage or epigenetic changes (32–34). Oncogene activation and tumor suppressor loss sustain tumor-elicited inflammation by indirectly inducing the production of pro-inflammatory cytokines, growth factors, and chemokines that recruit inflammatory immune cells to the tumor site (35, 36).

Little is known about the immune cells that promote dMMR tumor progression. Therefore, further exploration of the immune microenvironment in dMMR tumor progression is needed. Recent advances in mIHC have enabled the simultaneous analysis of several markers within a single FFPE tissue section (29). Our present study used this novel mIHC technique with two panels to show a comparable immune microenvironment across different stages of dMMR tumors, with accurate cell discrimination and spatial information (Supplementary Figure 1). The immune microenvironment at different stages was plotted on the basis of the intensity of several markers of immune cells, providing evidence for a distinct spectrum of immune populations in the early and advanced stages. Researchers have shown that CD8+ cytotoxic T cells and CD8+ memory T cells are crucial to antitumor immunity (22, 37). However, we observed an unusually high proportion of CD8+ cytotoxic T cells and CD8+ memory T cells in patients with advanced dMMR. Researchers have demonstrated that when infiltrating cancer tissues, CD8+ T cells are generally in dysfunctional states (38). Our findings support two possible explanations for these dysfunctional states.

First, some researchers have outlined that the binding of PD-L1 to its receptor PD-1 on activated T cells inhibits antitumor immunity by counteracting T-cell–activating signals (39, 40). Ahn et al. showed that PD-1 plays a major role in CD8+ T-cell exhaustion during cancer and chronic infections (41). Therefore, we hypothesized that PD-L1 colocalization with CD8+ T cells would activate PD-1 and cause cell exhaustion. Our results show that the colocalization of PD-L1 on CD3+ T cells and CD8+ T cells is more strongly enhanced in advanced-stage tumors than in early-stage tumors (Figure 3).

Second, according to Kim et al., there is a growing consensus that CD4+ regulatory T cells play a crucial role in the suppression of the antitumor immune response and contribute to tumor escape from the host immune system (42). Our study also shows that CD4+ regulatory T cells are significantly more prevalent in advanced-stage dMMR CRCs than in early-stage dMMR CRCs (Figure 2). Considering the lack of validation and limited sample size, further investigation and experimentation with a larger sample size are strongly recommended. In conclusion, our study results highlight the possibility of heterogeneous distribution of T cells and PD-L1 colocalization in early-stage and advanced-stage dMMR CRCs.

PD-L1–positive cells were more frequently observed in advanced-stage tumors than in early-stage tumors in both epithelial and stromal regions (Figure 1), although this difference was not statistically significant. As mentioned in the literature, PD-L1 may counterbalance a vigorous immune microenvironment and promote dMMR tumor progression (39, 40). Therefore, our observation of a high number of PD-L1 colocalizing T cells, such as CD8+PD-L1+ cytotoxic T cells and CD8+PD-L1+ memory T cells, in advanced-stage CRC supports the use of ICIs, such as anti–PD-L1 drugs, in advanced-stage CRC. Further research should be conducted to investigate the possibility of the use of ICIs as a first-line therapy (3). PD-1–positive cells were more frequently seen in early-stage tumors than in advanced-stage tumors in epithelial regions. However, researchers have demonstrated that PD-1 expression level is not associated with the response rate of ICIs (1). In the case of macrophages, the proportion of PD-L1+ macrophages was increased in epithelial regions of advanced-stage dMMR tumors. Furthermore, among all immune cell phenotypes presented in Table 2, CD3+PD-1+ T cells were the only promising prognostic factor (Figure 4).

Exploring the clinicopathological features described in the literature, we observed that all dMMR tumors were associated with advanced age, location in the right colon, and poor differentiation, which is consistent with the results of previous studies. In advanced-stage dMMR tumors, a trend toward a sex ratio imbalance was observed (1:2 in favor of males). Among the 15% of CRCs with dMMR, approximately 3% were associated with Lynch syndrome, commonly with MLH1 and MSH2 mutations, and the other 12% were sporadic cases, predominantly caused by epigenetic silencing of MLH1 due to promoter hypermethylation (43, 44). The patients in our cohort were randomly chosen from resected cases; therefore, the proportions of Lynch syndrome cases and sporadic cases in our samples should be similar to those in patients with CRC in the general population. In early-stage dMMR CRC, we found that 3 of 24 patients had a BRAF mutation, and in advanced-stage dMMR CRC, 2 of 19 patients had a BRAF mutation, corresponding to normal percentages of BRAF mutation (Supplementary Table 2). However, considering a higher percentage of MSH2 and MSH6 expression loss in advanced-stage dMMR CRC, we hypothesized that Lynch syndrome would be found more often at the advanced stage. However, further exploration by means of the analysis of hypermethylation of the MLH1 promoter is needed. The mucinous histological subtype is the characteristic morphology of dMMR CRCs. It occurs in up to 40% of patients, according to previous reports (45); in our study, this subtype accounted for 29% of the advanced-stage tumors. Further mechanistic studies are needed to explain the variation in histological subtype across stages in dMMR CRCs.

Although our study was considerably strengthened by the synchronous analysis of multiple immune markers in the early and advanced stages of dMMR CRC, it also has limitations. The primary limitation is the finite number of samples in our patient cohort, which was because of the low proportion of patients with dMMR with advanced-stage cancers and few surgical opportunities to obtain specimens. We did not analyze detailed treatment patterns or all the classical features that were previously noted as potential prognostic factors because of the lack of related data. Furthermore, we hypothesized that PD-L1 colocalization with T cells would activate PD-1 and cause T-cell exhaustion, but we lacked functional studies or investigations of functional markers. Finally, we were not able to collect any information on the immunotherapeutic response, which leaves the predicted value of immune molecules unclear at this point.

In conclusion, we conducted a comprehensive evaluation of the immune microenvironment in the early and advanced stages of dMMR CRCs. Our results demonstrate that high numbers of CD8+ cytotoxic T cells and CD8+ memory T cells, which usually represent a cytotoxic function of the adaptive immune system, are associated with the transformation of the immune microenvironment from early-stage to advanced-stage dMMR CRCs. Furthermore, possibly enhanced inhibition factors, such as CD4+ regulatory T cells and upregulated PD-L1 colocalization of CD3+ T cells and CD8+ T cells, which were hypothesized to be exhausted, were found to be associated with the transformation from early-stage to advanced-stage dMMR CRCs. Furthermore, the immune cell population, such as CD3+PD-1+ cells, can be used to identify patients with dMMR with poor prognoses.

The original contributions presented in the study are included in the article/Supplementary Material. Further inquiries can be directed to the corresponding author.

Written informed consent was obtained from the individual(s) for the publication of any potentially identifiable images or data included in this article.

Conception and design: DH. Development of the methodology: HY and YL. Analysis and interpretation: HY, XW, JQ, and MX. Writing and review: DH, HY, and JP. Study supervision: DH. All authors contributed to the article and approved the submitted version.

This study is supported by grants from Shanghai Natural Science Foundation (19ZR1410200 to DH), National Natural Science Foundation of China (U1932145 to JP and 82002946 to YL), Science and Technology Commission of Shanghai Municipality (18401933402 to JP), Shanghai Sailing Program (19YF1409500 to YL), and “Chenguang Program” supported by Shanghai Education Development Foundation and Shanghai Municipal Education Commission (20CG08 to YL). The sponsors had no involvement in the study design, collection, analysis, interpretation of data, or writing of the report, nor in the decision to submit the paper for publication.

JQ is employed by Genecast Biotechnology Co., Ltd.

The remaining authors declare that the research was conducted in the absence of any commercial or financial relationship that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.