Herein, we identify two distinct clusters of advanced colorectal carcinomas with unique immune microenvironment signatures. Our findings suggest that this distinction is based predominantly on activation states of the immune cell composition, and largely independent of the CMS subtype or of driving oncogenic mutations. This study adds to the expanding literature which focuses on the importance of the tumor microenvironment in disease heterogeneity. Using currently available advanced technology, we demonstrate, similar to the findings by Galon et al. in this cancer subtype, that an activated immune environment (including CD8 T cells but also dendritic cells, tertiary lymphoid structures, among others) may be associated with improved patient prognosis. Further studies utilizing these techniques to better characterize the activation state of various immune cell subpopulations will help to uncover whether tumors with activated immune environments may be more sensitive to checkpoint inhibition, offering additional therapeutic implications.
The tumor microenvironment comprises a complex network of cell types such as lymphocytes, dendritic cells, myeloid cells, fibroblasts, endothelial cells and extracellular components (cytokines, chemokines, etc.), all of which heavily influence the ability of a tumor to grow and disseminate [23–25]. The characteristics of this environment are thought to have significant prognostic and even therapeutic implications in numerous cancer subtypes including ovarian, colorectal, lung and breast, and several studies have shown that the inflammatory network in particular plays a pivotal role in the evolution of cancer [26–28]. However, it is not just the density and cell type but also a cell’s specific phenotype which remains imperative in directing the immune response to be either immunostimulatory or immunotolerant, a feature that is becoming more evident as our ability to characterize these components advances. Although we demonstrate that an activated CRC TIME is associated with improved survival, it is not clear at this time if this improved survival is due to improved benefit with cytotoxic therapy, or rather a feature of a less aggressive CRC phenotype.
Cancer cells have evolved multiple mechanisms to escape immune surveillance, such as defects in antigen presentation machinery, upregulation of negative regulatory pathways, and the recruitment of immunosuppressive cell populations which ultimately result in inefficient cytotoxic response. Our ability to characterize and define the TIME has significantly evolved. Earlier studies relying on manual inspection of glass slides and single surface marker analysis by immunohistochemistry were limited in their ability to assess the dynamic nature of the TIME. This lack of specificity in defining subpopulations of cells may account for some of the inconsistencies in the literature between cell type and prognosis. Due to the complexity of the immune microenvironment that we now recognize, computational algorithms using bulk transcriptome data have become a popular approach to better understanding this interaction between tumor and host. The CMS classification scheme, for example, was one of the first subclassification studies using RNA-Seq data in colorectal carcinomas to incorporate features of the immune microenvironment in addition to molecular alterations within the tumor, in order to derive subclassifications [6]. This is pertinent in colorectal cancer, as multiple studies have demonstrated a correlation between the presence of mature T cells and dendritic cells and a more favorable prognosis, the former being the basis for the Immunoscore [9,12,13,29,30]. Similarly, the presence of B-cell and T-cell aggregates, also known as tertiary lymphoid structures (TLS) in the TIME, is considered a favorable prognostic feature as they promote T cell activation and antitumor effects[8,31]. Dendritic cells also play a prominent role in tumor response, as they vary significantly in their immunostimulatory or immunosuppressive activity at different stages of cancer progression and are incredibly dependent on local signaling [32,33]. In general, immature dendritic cells are peripherally located and non-mobile but are highly phagocytic and can be stimulated by various factors (including tumor antigens) to undergo maturation and activation, which then elicits an immune response by presenting antigens to T cells, inducing cytotoxic T lymphocyte immune response [34]. They also secrete chemokines that cause NK cells and T cell migration into the tumor and cytokines that maintain cytotoxic functions [35,36]. Not surprisingly, studies in a variety of solid tumors have found that increased dendritic cells, particularly those in the mature state with higher expression of costimulatory molecules including CD40, CD83, and CD86, are associated with improved prognosis [37–40]. On the other hand, tumor-associated dendritic cells have been found to often exhibit impaired function, including decreased uptake and presentation of antigens, reduced expression of costimulatory surface molecules, and inefficient migration, all of which may in reduced immune response and tumor evasion [41]. Studies have shown that tumor cells can release cytokines such as IL-6, IL-10 and TGF-B that cause dendritic cells to remain in an immature or immunosuppressive stage [33,42–44]. There are also metabolic inhibitors including IDO and arginase, which are produced by tumor associated macrophages (particularly the M2 phenotype) and upregulation of receptors such as CTLA4 and PD-1, all of which may be contributing to tumor immune escape [45,46].
The findings of the current study are unique in that we were able to evaluate a specific cohort of patients with advanced colorectal carcinoma, mostly Stage IV, who had already failed first line therapy. Within this cohort, we first identified two distinct groups through unsupervised clustering that differed predominantly in the activation state of the dendritic cells and other immune cells. Our findings are in keeping with other studies which have found an association between increased Th1 and activated dendritic cells and improved prognosis, whereas increased inactivated dendritic cells are generally associated with progression of disease [7,11,47]. Prior research has also characterized KRAS mutant CRC as having limited cytotoxic T cell infiltration, reduced T helper 1 responses, and reduced INF-gamma signaling, generating an overall immunosuppressive tumor microenvironment phenotype, which correlates with our finding of increased KRAS mutations in the subgroup with a suppressed TIME [48,49]. Another unique finding of the current study is that the reactivity state of the TIME did not depend on site of disease, either primary versus metastasis or location of metastasis. This is an important point, as prior studies have suggested that liver metastases are associated with immunosuppression, and patients with liver metastases have been shown to have minimal response to systemic immunotherapy [50]. However, we identified metastatic liver tumors with both activated and suppressed TIME, suggesting that a subset of patients with hepatic involvement may benefit from immunotherapy.
There are several recognized limitations of the current study. The cohort of patients with advanced colorectal carcinoma in which sufficient material available for RNA-Seq was relatively small. In addition, while review of the histomorphology and IMC™ analysis were performed in order to confirm our findings in situ, this comparison was somewhat limited by the fact that the FFPE tissue areas used for histopathology, IHC and IMC™ analysis may not exactly correspond to matched frozen tissue areas used for bulk RNA-Seq. Second, the IMC™ analysis was limited to discrete regions of interest that may not be representative of the entire tissue sample. Lastly, we were unable to fully characterize the immune population through IMC™ given the lack of additional immune markers in the available panel. Additional studies that can isolate dendritic cells and activations states are warranted.
In conclusion, this study supports the pertinent role of the immune microenvironment in tumor progression and patient prognosis, even within a group of patients with advanced stage disease who have failed multiple lines of therapy. We characterized the immune cell composition utilizing novel, more advanced technologies such as RNA-Seq and IMC™ in order to better understand the complexity and vital role that activity states play in determining response to tumor infiltration, thereby affecting response to therapy and overall patient outcome.