Immunodivergence in Metastatic Colorectal Cancer

Abstract

Van den Eynde et al. publish in this issue of Cancer Cell that metastatic colorectal cancer shows marked heterogeneity in T cell infiltration among different lesions and patients. Measurements of T cell infiltration in metastases by immunoscore offer some prognostic information and support immune editing by coevolving adaptive immune responses.

Coevolution of cancer and the antitumor immune response is a terrible, somber, and fascinating tug-of-war. Similar to most malignancies, colorectal cancer is linked to genomic instability and epigenetic changes that determine how malignant cells interact with the stroma. There have been many studies on clonal evolution of tumor cells at the genetic level (McGranahan and Swanton, 2017). However, stromal, and specifically immune cell, adaptations to such tumor cell changes are considerably less defined.

The group of Jerome Galon has produced over the years extensive evidence that the density of CD8⁺ T cells infiltrating primary colorectal carcinomas is associated with better relapse-free survival with high accuracy that outperforms other pathology-based approaches and the TNM staging system (Fridman et al., 2012, Galon et al., 2006). This evidence has been validated prospectively in an international, multicenter effort that used a simple and elegant combination of digital pathology and two immunohistochemical stainings (CD3 and CD8) per tumor to predict risk of relapse (Pagès et al., 2018).

Although some clonal heterogeneity may exist within primary tumors, dramatic clonal heterogeneity has been discovered when studying metastases in a variety of malignancies. Van den Eynde et al. now report studies on the B and T cell infiltration in a retrospective series of colorectal cancer patients with metastasis at the initial diagnosis (synchronous) or following surgical resection of the primary tumor (metachronous) (Van den Eynde et al., 2018). The prognostic value of CD8⁺ T cell infiltration regarding disease-free survival and overall survival in metastases is clearly lower than in primary tumors but is still significant. Several parameters such as the area of the metastases and the variation of the infiltration level, considering mainly CD3⁺, CD8⁺, CD4⁺, CD20⁺, and FOXP3⁺ lymphocytes, in each metastasis are studied. The most striking reported feature in patients bearing more than one metastatic lesion is the heterogeneity of such lesions in both lymphocyte infiltrates and the pattern of analyzed hotspot mutations. Even if only T cell density without functional parameters was considered in the study, this clearly supports tumor and immune diversity among metastases.

Divergent adaptive immune responses across tumor lesions could be due to variable antigenic content of tumor cells, changes in antigen processing and presentation machineries, metabolic or ionic adaptations limiting T or B cell expansion/survival, and enrichment of potent regulatory signals or mechanical barriers for T cells in the tumor microenvironment. However, the contribution of the immune properties and propensity of the underlying non-tumor metastatic tissue site to support immune infiltration cannot be excluded. In this regard, the liver and brain are common metastatic sites that have been considered as immune privileged due to their relatively low baseline immune infiltration and resistance to immune responses (Lang et al., 2006, Louveau et al., 2015). Variation of immune responses within different tissue or organ regions is also expected due to distinct local cell composition, vascular structure, functional properties, and antigenic exposure.

Notably, the study identified association between elevated T cell infiltration (T cell hot) and lower number of metastases, most likely reflecting sustained adaptive immune pressure on tumor development and spreading. Possibly due to the same phenomenon, T cell cold metastases are more numerous and associated with worse prognosis (Figure 1). However, the directionality of the association between tumor burden and adaptive immune cell infiltration in colorectal tumors cannot be established by a simple correlation and will require mechanistic studies to functionally link the processes. In addition, until similar findings are experimentally documented, caution is advised in extrapolating these conclusions to other malignancies. An important methodological note for future biomarker development is that in spite of intralesional heterogeneity, informatics simulations of needle trajectory of biopsies that could have been performed in removed metastatic lesions very accurately predict lymphocyte infiltration of the lesion as a whole.

Figure 1.

Multiverse of Colon Cancer Metastases with Different Degrees of Immune Cell Infiltration

It is well established that T cell infiltration in the primary lesions of colon cancer determines reduced probability of metastasis following surgery. Successful synchronous or metachronous metastasis must have overcome the control by the adaptive immune system. Estimating T cell infiltrates in the metastatic lesions of colorectal cancer also provides prognostic information, albeit less accurate than in the case of primary tumors. Co-evolution of the cytotoxic immune response and cancer probably explains much of the phenomenology conducive to the number, size, and rate of progression of distant metastases. Density of CD8+ T cell infiltrates in colon cancer can be assessed by the so-called immunoscore (IS), grading 0 to 4 the presence of T cells in the center and invasive margins of the tumor masses. The temperature analogies of hot (IS ≥ 3), lukewarm (IS = 2), and cold (IS < 2) lesions with respect to the intensity of T cell infiltration in primary and metastatic lesions probably speak of a history of immunosurveillance, equilibrium, and escape in the relationship of anticancer immunity and the malignant cells, which ultimately results in heterogeneity of lesions and cancer patients.

Current management of metastatic colorectal cancer patients for whom surgery is prescribed frequently involves presurgical regimens of chemotherapy with anti-VEGF or anti-EGFR monoclonal antibodies. This has permitted Van den Eynde et al. to explore differences in the immune infiltrate of lesions depending on the type of neoadjuvant regimen received. While chemotherapy together with an anti-EGFR IgG1 monoclonal antibody leads to increased T cell infiltration particularly in invasive margin of the metastasis, anti-VEGF in combination with chemotherapy did not increase T lymphocyte infiltrates but increased the expression of B cell and NK cellgenes. The differences could be attributed to the mechanism of action of the EGFR monoclonal antibody that coats tumor cells and activates Fcγ receptors on NK and myeloid cells, whereas anti-VEGF neutralizes a cytokine. Of interest, RAS mutation status does not seem to be a driver of baseline immunoscore but determines lower T cell infiltration in response to anti-EGFR plus chemotherapy. Over the recent years, it has been established that almost all forms of cancer therapy affect antitumor immunity. This includes anti-VEGF, anti-EGFR, and conventional chemotherapy that reshapes the composition of the microenvironment and may induce immunogenic cell death.

The study by Van den Eynde and colleagues only considers lymphocyte infiltration density and location, which might be an oversimplification of a much more complex process. In fact, numerous immune inhibitory receptors and potent regulatory mechanisms are probably at play and more likely so in T cell hot lesions. There are two aspects of metastatic colorectal cancer that could be crucial in determining hot and cold metastasis in terms of the density of the T cell infiltrate and its functions not covered by this study: the expression of active TGF-β (Tauriello et al., 2018) and the presence of immunosuppressive myeloid leukocytes. TGF-β is a powerful immunosuppressive factor deeply involved in colon cancerimmunobiology in addition to promoting extracellular matrix deposition that thwarts lymphocyte infiltration. The abundance and composition of myeloid infiltrates is also likely important in antigen presentation, production of pro- or anti-inflammatory signals, and local metabolic adaptations. Finally, and based on emerging data, the gut microbiome could have a prominent effect in the immune conditioning and evolution of colorectal malignancies (Zitvogel et al., 2018).

Technical aspects of the in situ immune analysis of tumor tissue samples also pose prominent challenges. These include the careful validation of selected antibodies for detection of markers, optimization and standardization of assays, clear and data-driven definition of specific tumor areas and compartments for selective quantification of targets, the reproducibility of analyses over time and across laboratories, and the clinical implications of findings.

Another aspect not addressed by the study is the behavior of genetically unstable, mismatch repair deficient colorectal cancers associated with fewer metastases and commonly sensitive to anti-PD1 monoclonal antibodies (Le et al., 2015). This has been attributed to higher neoantigen burdens resulting from defects in DNA repair. In these cases, even more clonal heterogeneity occurs as a result of more frequent unrepaired mutations, and there is reciprocally more opportunity for eventual adaptation and escape. Along these lines, we do not know yet whether the proposed immunoscore approach will be a useful biomarker to predict response to immunotherapies either in microsatellite stable (MSS) or microsatellite instability-high colon cancer, which needs to be studied. It must be stressed that so far, MSS metastatic colorectal cancer remains essentially resistant to immune checkpoint blockers that are showing partial efficacy in other digestive tract malignancies.

Immunodiversion makes colon cancer a formidable enemy for immunotherapy, even if the immune system has inherent abilities to adapt. Coevolution of the immune response and the malignant cells fitting the three E model (elimination, equilibrium, and escape) (Mittal et al., 2014) probably explains these observations. When studying primary tumors, we deal, to some extent, with equilibrium, but when considering metastases, we are more likely dealing with escape. Such immune escape is likely to be the heterogeneous Darwinian result of serial successful escape variants explaining immunodivergence of the metastases in the snapshots that we see upon surgery or necropsy.