Immune checkpoint inhibitor response in mismatch repair-deficient colorectal cancer and other solid tumors: is it truly disease-agnostic?

Mismatch repair deficiency (MMR-D) in solid tumors has been investigated over the decades particularly, due to its unique characteristics such as hypermutability leading to increased neoantigen generation and tumor-reactive immune infiltration [1]. MMR-D occurs as a consequence of loss of function alterations in any MMR genes (MLH1, PMS2, MSH2, MSH6) and this results in an inappropriate response against DNA mismatches, which frequently occur during the DNA replication and homologous recombination [2,3]. The abnormal MMR response accelerates mutagenesis, particularly in microsatellite regions of the DNA, resulting in variable sizes of microsatellites, which is known as microsatellite instability (MSI). MSI, one of the hallmarks of MMR-D tumors, introduces frameshift mutations, which leads to further hypermutability status in the DNA. These frameshift mutations trigger the generation of mutation-associated neoantigen and subsequent to this, mutation-associated neoantigen-reactive immune cells, which are also known as tumor-infiltrating lymphocytes, migrate into tumor microenvironment [4]. Tumor-infiltrating lymphocytes, particularly CD8⁺ cytotoxic T cells orchestrate significant antitumor response to eliminate malignant clones and inhibit disease progression [5].

This foundation of knowledge summarized above initiated clinical trials of immune checkpoint inhibitors in MMR-D/MSI-H patients with different types of cancers, including colorectal cancer [6–8]. In the study by Le et al. [8], patients with MMR-D cancers with different histologies including colorectal, gastric, endometrial and small bowel cancer, as well as ampullary tumor and cholangiocarcinoma, were treated with pembrolizumab. The authors have reported highly promising treatment response across different tumor types included in the trial. Patients with MMR-D colorectal cancer had 40% objective response rate (ORR) and patients with MMR-D non-colorectal tumors had an ORR of 71%. Although the size of the cohort was quite small for each arm (a total of 17 patients in both arms), these promising results led to US FDA (MD, USA) disease-agnostic approval of pembrolizumab in all solid tumors, regardless of tumor histology, in 2017. Follow-up results of this study included a larger cohort with 86 MMR-D cancer patients with 12 types of primary cancers [9]. In this study, patients with MMR-D colorectal cancer (n = 40) had an ORR of 52% and this was 54% for patients with non-colorectal MMR-D tumors, again demonstrating a promising response across the board. More recently, the KEYNOTE 158 study investigated the efficacy of pembrolizumab in a Phase II trial, which enrolled 233 patients with exclusively noncolorectal cancers [10]. Notably, the authors included 27 cancer types in their cohort, where the most common cancers were endometrial (n = 49), gastric (n = 24), cholangiocarcinoma (n = 22), pancreatic (n = 22) and small intestine (n = 19). The authors also included other solid tumors types that are less commonly seen in Lynch syndrome patients such as ovarian cancer and brain tumors. The ORR for the general cohort was 34%, which is comparable with colorectal cancer data from previous reports [8]. The authors also reported the efficacy of pembrolizumab by cancer type. The highest ORR was observed in patients with endometrial cancer including eight (16%) patients with complete response and 20 (41%) with partial response (ORR = 57.1%). The ORR for cholangiocarcinoma, gastric cancer, small intestine cancer, ovarian and pancreatic cancer was 40.9, 45.8, 42.1, 33.3 and 18.2%, respectively. Interestingly, no objective response was seen in patients with primary brain tumors (n = 13) enrolled in this study. The authors also reported 12- and 24-month progression-free survival (PFS) rates of 33.9 and 29.3% indicating durable response among responders, which is consistent with previously reported experiences with immune checkpoint inhibitors. The median PFS for the overall cohort was 4.2 months and most notably, PFS for pancreatic cancer and brain tumors was 2.1 and 1.1 months, respectively. Highly promising median overall survival (OS) outcomes were also noted and that was 23.5 months for the overall cohort. However, the median OS for patients with pancreatic cancer and brain tumors was 4.0 and 5.6 months, respectively.

Most recently, a precision oncology trial, NCI-MATCH, investigated the biomarker-based treatments for various cancer types and reported the results for immune checkpoint inhibitor therapy in MMR-D noncolorectal cancers [11]. The study screened approximately 4902 patients and identified 99 (2%) patients who had MMR-D disease. Among these, 42 patients were eligible for the trial and received nivolumab 480 mg every 4 weeks. The most common cancers included were endometrial cancer (n = 13; 30%) and prostate cancer (n = 5; 12%). The study also included three (7%) cholangiocarcinoma patients and one (2%) pancreatic cancer patient. The authors reported an ORR of 36% for the overall cohort. The median PFS and median OS were 6.3 and 17.3 months, respectively [11]. Similar to pembrolizumab, durable responses were noted (12- and 18-months PFS rates; 46.2 and 31.4%, respectively). In this study, the authors did not report disease-specific outcomes rates given the limited number of patients enrolled for each disease type.

Overall, the studies above clearly suggest that MMR-D in solid tumors is a predictor of response to immune checkpoint inhibitor therapy. Moreover, the ORR and durable responses were also similar to experiences with immune checkpoint inhibitors in patients with MMR-D colorectal cancer. However, cancer type-specific responses reported in the KEYNOTE 158 study are indicative of variable clinical outcomes. For example, lack of any objective response in brain tumors and relatively lower response rates in pancreatic cancer suggest that the disease biology may have an independent impact on immune checkpoint inhibitor response, regardless of MMR-D status. Perhaps, multiple disease-specific biologic factors such as tumor microenvironment and heterogeneity in founder mutations that initiate and drive carcinogenesis can at least play a partial role. For example, the pancreatic cancer microenvironment has distinct characteristics with the infiltration of myeloid-derived suppressor cells along with T-regulatory cells in tumor stroma [12]. It has been also suggested that dense stroma in pancreatic cancer with high desmoplastic reaction, may function as a physical barrier for the migration of effector T cells into tumor microenvironment [13]. Notably, brain tumors, which also had a relatively poor response to pembrolizumab, have also distinct microenvironmental characteristics including the blood–brain barrier which reduces the translocation of the inflammatory cell to the brain parenchyma [14]. Although immune checkpoint inhibitors have the potential to be effective in metastatic brain tumors where the blood–brain barrier is disrupted by invading metastatic tumors, their role in primary CNS tumors remains unclear.

Another plausible reason that may lead to distinct responses to immune checkpoint inhibitor therapy could be founder mutations that determine the molecular behavior of cancer. For example, a recent landmark study evaluating Lynch Syndrome in patients with solid tumors showed that approximately 36% of Lynch syndrome patients had microsatellite stable disease (MSS) [15]. Notably, most of the MSS cancers were noncolorectal/nonendometrial cancer. This important finding suggests that the existence of germline mutations in MMR genes may not necessarily dictate carcinogenesis to be triggered by mismatch repair dysfunction and perhaps, alternative independent oncogenic pathways may also trigger oncogenesis leading to MMS tumors in the background of co-existing MMR-D. Notably, that may be more prevalent in those cancers that are not associated with Lynch syndrome [15]. These discoveries indicate that MSI and high tumor mutation burden may be a better predictor of immune checkpoint response, as these markers are highly associated with MMR-D-driven carcinogenesis while the loss of function alterations in MMR genes may sometimes be a passenger mutation/alteration [16]. Markedly, a study investigating the impact of BRCA gene alterations in solid tumor reported BRCA mutations in non-BRCA associated cancers may not have any clinical implication [17] suggesting highly actionable molecular alterations may have a distinct impact (or potentially no impact) on different types of cancer.

These findings collectively suggest that although MMR-D may be surrogate of immune checkpoint inhibitors in most of the solid tumors, this response is not independent of disease biology, which appears to lead to distinct outcomes across different types of cancers. From tumor microenvironment to the mutational landscape of the individual tumor type, several other factors may operate in this complex cancer cell and immune system interaction. Perhaps, the characterization of immune infiltration to the tumor microenvironment may further define the effectivity of antitumor response, hereby immune checkpoint inhibitor responsiveness. Recently, immunoscore – which is a scoring system based on the CD3 and CD8-positive T-cell infiltration into tumor and tumor margin – provided promising outcomes as a prognostic tool, particularly in early-stage MSS colorectal cancer [18]. Further prospective studies are needed to evaluate the predictor role of characterization of T-cell infiltrate on immune checkpoint inhibitor response/resistance in cancer patients with MMR-D tumors. Currently, ongoing basket trials enrolling different types of cancers carrying a unique molecular alteration (such as BRAF, FGFR, etc) may also shed further light on the actionability of a single molecular target across different tumors with distinct biological characteristics.