Mismatch Repair–Proficient Colorectal Cancer: Finding the Right TiME to Respond

In this issue of Clinical Cancer Research, Llosa and colleagues describe a gene expression–based approach to characterize the tumor immune microenvironment (TiME) of patients with advanced mismatch repair proficient (MMRp) colorectal cancer (1). This approach uncovers evidence that upregulation of intratumoral IL17-mediated signaling may attenuate responses to checkpoint inhibition even in the context of an immunoreactive microenvironment within MMRp tumors. As such, the work by Llosa and colleagues provides a roadmap for significantly expanding the population of patients with advanced colorectal cancer eligible to receive therapy with checkpoint inhibitors.

Currently, the role of immunotherapy in advanced colorectal cancer is limited to the mismatch repair deficient (MMRd) subtype, which represents only a minority (∼5%) of cases (2). As shown in multiple phase II studies, treatment of advanced MMRd colorectal cancer with anti-programmed cell death protein 1 (PD-1) antibody alone or in combination with anti-cytotoxic T lymphocyte–associated antigen 4 antibody is associated with durable clinical benefit (2). The molecular genetic driver for this activity is clearly understood to be adaptive immune recognition of novel expressed peptides (neoantigens) produced from a high burden of somatic frameshift and point mutations in the tumor cell population.

While the benefit of checkpoint blockade in MMRd colorectal cancer is now well-established, there has been considerable effort to understand potential mechanisms of primary and secondary (acquired) resistance in this subtype. Primary resistance appears to be uncommon, although misdiagnosis of MMR status was recently implicated in a substantial proportion of cases (3). Shin and colleagues also identified inactivating JAK1/2 mutations as a potential cell-intrinsic mechanism of primary resistance (4). For secondary resistance, current evidence points toward the acquisition of defects in antigen processing and presentation as playing a major role (e.g., inactivation of β-2-microglobulin or mutation of HLA class I). These insights will require further exploration and validation, but nonetheless will inform future therapeutic strategies for patients with advanced MMRd colorectal cancer.

In contrast, limited progress has been made to date with respect to the utility of checkpoint inhibitors for patients with MMRp colorectal cancer, which is the most common subtype where response rates to anti-PD-1 therapy are poor. In a landmark phase II study (KEYNOTE-016, clinicaltrials.gov ID NCT01876511), Le and colleagues observed a disease control rate of only 11% (2/18) for patients with advanced MMRp colorectal cancer who received anti-PD-1 monotherapy in the second-line or greater (5). Furthermore, unlike the case for other common cancers, PD-L1 expression does not appear to be a predictive biomarker for MMRp colorectal cancer. Consequently, two critical questions have emerged: (i) is there a molecularly defined subset of MMRp colorectal cancer tumors that are immunosensitive; (ii) can MMRp tumors convert from immunoresistant to immunosensitive?

These questions have been the focus of comprehensive efforts to define the tumor cell–intrinsic and TiME-mediated mechanisms of resistance to checkpoint blockade in MMRp colorectal cancer, which is a challenge shared across many tumor types (6). An important perspective emerging from this work is that the somatic landscape of MMRp colorectal cancer clearly plays a significant role in modulating the antitumor immune response. Elegant preclinical studies of murine colorectal cancer models have provided evidence that key oncogenic drivers, such as MAPK and TGFβ activation, may promote the exclusion of CD8⁺ cytotoxic T cells from or recruitment of myeloid-derived suppressor cells to the TiME, respectively. By virtue of demonstrating enhanced efficacy with the combination of targeted small-molecule inhibitors and anti-PD-1 therapy in these models, such studies highlight not only the apparent plasticity of the MMRp colorectal cancer microenvironment, but also potential biomarkers as well. While these insights are promising, successful translation into therapeutic clinical trials has been proven difficult. For example, in a recent negative phase III study (IMblaze370, clinicaltrials.gov ID NCT02788279) testing the efficacy of anti-PD-L1 with or without MEK inhibition in advanced colorectal cancer, RAS/RAF mutation status was not correlated with response rate (7).

In this context, this study by Llosa and colleagues provides a significant step forward in our understanding of other clinically meaningful TiME factors (1). The investigators adeptly leveraged a repository of primary tumor samples obtained from treatment naïve patients with advanced MMRp colorectal cancer who had received anti-PD-1 monotherapy through the aforementioned KEYNOTE-016 trial. By systematically comparing the expression of 44 immune-related genes between MMRd tumors that responded (partial or complete) versus MMRp tumors that did not (stable or progressive disease), the investigators discovered a concise 10-gene signature that predicts immunosensitivity. Upon further analysis and validation, approximately three (21%) of 14 unselected MMRp colorectal cancer tumors harbored the immunosensitivity signature, suggesting that such tumors may response to checkpoint inhibition (however this was not formally tested). In addition, CD8⁺ tumor-infiltrating lymphocytes (TIL) from these predicted immunosensitive tumors displayed an exhaustion phenotype, characterized by upregulation of PD-1/PD-L1 and CXCL13, and downregulation of IFNγ. Most strikingly, four (36%) of the remaining 11 MMRp colorectal cancer tumors that were putatively classified as nonimmunosensitive were found to be enriched for an IL17⁺ TIL population coexisting in the background of an exhausted phenotype.

The gene expression–based signature presented by Llosa and colleagues holds interesting potential as a novel predictive biomarker. However, a major confounding factor of their approach is that the immunosensitive signature was derived by comparing MMRd responders to MMRp nonresponders. A direct comparison of MMRp responders to nonresponders might capture other important elements in the TiME that were not seen in this study. Yet, due to the rarity of MMRp responders, such an analysis is not feasible without pooled outcomes data and gene expression profiles from multiple therapeutic trials involving checkpoint blockade and MMRp colorectal cancer.

Nonetheless, as reported, the prospect of identifying 10%–20% of patients with MMRp colorectal cancer who might benefit from checkpoint inhibition would represent a major advancement in the standard-of-care. Furthermore, the observation that IL17⁺ T cells are enriched in a subset of putatively immune-resistant tumors is especially interesting considering the multitude of roles that IL17 signaling may play in maintaining gut and microbiota homeostasis (8), and the clinical availability of IL17 inhibitors. This finding warrants further investigation, as it may point toward a novel combinatorial strategy (e.g., anti-IL17 and anti-PD) for overcoming resistance to checkpoint blockade in MMRp.

The work by Llosa and colleagues also raises key questions: (i) how does the immunosensitive TiME signature correlate with relevant patient demographics and other well-established biomarkers for MMRp colorectal cancer (e.g., RAS/RAF mutation status, tumor sidedness, and consensus molecular subtype); (ii) how does exposure to cytotoxic chemotherapy or targeted therapy change the TiME, if at all; and (iii) does the signature capture biology that is unique to the microenvironment of the primary tumor or biology that is relevant across different metastatic sites (e.g., liver, lung, and peritoneum) as well?

Moving forward, independent validation of the immunosensitive TiME signature as a predictive biomarker may be challenging, especially given the limited number of treatment naïve MMRp colorectal cancer patients who have received anti-PD-1 monotherapy through clinical trial. Nonetheless, this work provides a useful framework for understanding the heterogeneous nature of the MMRp colorectal cancer immune microenvironment (Fig. 1). It also exemplifies the critical need to embed correlative multi-omic analyses within therapeutic trials for advanced MMRp colorectal cancer.

M.J. Overman is a consultant/advisory board member for Merck and Bristol-Myers Squibb. E. Vilar is a consultant/advisory board member for Janssen Research and Development. No potential conflicts of interest were disclosed by the other author.

This work was supported by grant Research Training in Academic Medical Oncology T32 Award CA009666-23 (U.S. NIH/NCI to J.A. Willis); R01 CA219463 (US NIH/NCI) and a gift from the Feinberg Family to E. Vilar; and P30 CA016672 (US NIH/NCI) to the University of Texas Anderson Cancer Center Core Support Grant.