Mismatch Repair-Deficient Colorectal Cancer: Building on Checkpoint Blockade

Abstract

Colorectal cancer (CRC) with deficient DNA mismatch repair (dMMR) is characterized by hypermutation leading to abundant neoantigens that activate an antitumor immune response in the tumor microenvironment. Immune checkpoint inhibitors (ICIs) have transformed the treatment of this subset of CRC and other solid tumors with dMMR, by producing frequent and durable responses that extend patient survival. Recently, the anti–programmed death-1 (PD-1) antibody pembrolizumab was shown to produce significantly longer progression-free survival with fewer adverse events compared with chemotherapy as first-line treatment of metastatic CRC (mCRC) with dMMR. Accordingly, single-agent pembrolizumab represents a new standard of care for dMMR mCRCs including patients with Lynch syndrome and the more common sporadic cases. Furthermore, data indicate that the combination of PD-1 and cytotoxic T-cell lymphocyte-4 inhibitors was more effective than single-agent PD-1 inhibition in patients with dMMR mCRCs, suggesting nonredundant mechanisms of action. Although the benefit of ICIs is currently limited to metastatic disease, studies evaluating ICIs as neoadjuvant and adjuvant therapy in earlier-stage dMMR CRC are ongoing. Despite success of ICIs in the treatment of metastatic dMMR cancers, an appreciable proportion of these tumors demonstrate intrinsic or acquired resistance, and biomarkers to identify these patients are needed. Advances in the understanding of immunotherapy resistance mechanisms hold promise for both biomarker identification and development of novel strategies to circumvent treatment resistance. In this review, we present a comprehensive overview of the evidence for the role of immunotherapy in the treatment of dMMR CRC, discuss resistance mechanisms, and outline potential strategies to circumvent primary and secondary resistance with the goal of broadening the benefit of ICIs.

INTRODUCTION

Colorectal cancer (CRC) is the third most common cancer and second leading cause of cancer-related death in the United States with more than 149,500 new cases and 52,980 deaths estimated in 2021.¹ Approximately 15% of CRCs have deficient DNA mismatch repair (dMMR), which results in microsatellite instability (MSI). These tumors are hypermutated with abundant mutation-derived neoantigens that trigger a robust immune response in the tumor microenvironment (TME).^2-4 Phenotypic features of dMMR CRCs include right-sided predominance, tendency for poor differentiation, and better prognosis in the absence of distant metastasis.^2,5-7 The association of dMMR with more favorable prognosis attenuates with advancing stage⁸ and in the setting of metastatic disease, studies indicate a worse prognosis compared with metastatic CRCs (mCRCs) with proficient MMR.⁹

KEY POINTS

• Approximately 15% of colorectal cancers (CRCs) harbor deficient mismatch repair (dMMR) with two thirds of cases being sporadic secondary to MLH1 gene promotor hypermethylation.

• dMMR CRCs are characterized by a high tumor mutation burden that leads to abundant mutation-derived neoantigens that trigger a robust immune response in the tumor microenvironment with tumor-infiltrating lymphocytes.

• Immune checkpoint inhibitors induce frequent and durable responses in dMMR metastatic CRCs. Pembrolizumab is the standard first-line treatment in this population on the basis of a randomized, phase III trial. The combination of nivolumab and ipilimumab showed a higher response rate with a comparable adverse event rate in the same patient population in a phase II study.

• dMMR metastatic CRCs display frequent primary or secondary resistance to immune checkpoint inhibitors, and studies to elucidate mechanisms of resistance and develop strategies for its reversal are ongoing.

CONTEXT

• Key Objective

• Review the current status of the treatment of colorectal cancers with deficient DNA mismatch repair using immunotherapy, including mechanisms of treatment resistance and strategies to overcome such resistance.

• Knowledge Generated

• Discuss data on predictive biomarkers, mechanisms of resistance to immunotherapy, and novel strategies to circumvent such resistance.

• Relevance

• Summarize the current status of immunotherapy for colorectal cancers with deficient mismatch repair including the potential to broaden the use of immunotherapy in this population and to extend benefits to tumors with proficient mismatch repair.

The treatment landscape for dMMR solid tumors changed dramatically in 2015 when an immune checkpoint inhibitor (ICI) targeting programmed death ligand-1 (PD-1) was shown to produce frequent and durable responses and to extend survival in patients with dMMR mCRC.¹⁰ This finding led to the first tumor agnostic approval by the US Food and Drug Administration (FDA) for the treatment of dMMR solid tumors in the metastatic setting, and prompted further studies of immune checkpoint blockade in this tumor subset. This review focuses on the role of immunotherapy for dMMR CRCs and future directions in the therapy of these tumors.

IMMUNOTHERAPY FOR dMMR mCRC

Immunotherapy in Chemotherapy-Refractory Disease

The first in-human phase I clinical trial of pembrolizumab (BMS-936558), a fully human monoclonal anti–PD-1 antibody, included 39 patients with treatment-refractory solid tumors. One patient with mCRC had a prolonged complete response (CR), and his tumor was found to have dMMR.¹¹ This observation was hypothesis-generating, and suggested that dMMR CRCs may respond to immunotherapy because of their increased tumor mutational burden (TMB), as seen in other tumor types such as melanoma and non–small-cell lung cancer. A subsequent phase II study (KEYNOTE 016) demonstrated clinical activity of pembrolizumab in patients with dMMR mCRC (objective response rate [ORR] of 40%) and led to the first FDA tissue/site agnostic approval of this antibody for unresectable or metastatic treatment-refractory dMMR solid tumors.¹⁰

A phase II study (CHECKMATE 142) demonstrated clinical activity of nivolumab (an anti–PD-1 antibody) as monotherapy and in combination with ipilimumab (an anticytotoxic T-cell lymphocyte-4 [CTLA-4] antibody) in patients with dMMR mCRC refractory to a fluoropyrimidine, oxaliplatin, and irinotecan.^12,13 These data led to FDA approval of nivolumab in 2017 and its combination with ipilimumab in 2018 in this patient population. More recently, the anti–PD-1 antibody dostarlimab was granted FDA approval for its use in the treatment of refractory dMMR solid cancers on the basis of the results from the phase I GARNET study.¹⁴ The chronology of the evaluation of ICIs in patients with dMMR mCRCs is outlined in Table 1 and is discussed in detail in the Data Supplement (online only).

TABLE 1.

Clinical Studies of Immunotherapy for Metastatic Mismatch Repair-Deficient Colorectal Cancer

Immunotherapy as First-Line Treatment

In another cohort of the CHECKMATE 142 study, 45 patients with treatment-naïve dMMR mCRC were treated with nivolumab (3 mg/kg once every 2 weeks) plus ipilimumab (1 mg/kg once every 6 weeks) until disease progression or treatment discontinuation. The primary end point was investigator-assessed ORR, which was 69% and was 62% by blinded independent central review (BICR) at a median follow-up of 29.0 months. Six patients (13%) had CR and 25 patients (56%) had partial response (PR). BICR considered 11 patients (24%) to have CR and 17 patients (38%) to have PR. Of note, 71% of the responses were durable and lasted at least 12 months. Responses were seen in all subgroups. Patients with BRAF^V600E tumors had an ORR of 76%, whereas KRAS-mutated patients had an ORR of 80%. Median PFS and OS were not reached, but the 24-month progression-free survival (PFS) and overall survival (OS) rates were 74% and 79%, respectively. Grade 3-4 treatment-related adverse events (AEs) were seen in nine patients (20%). Health-related quality of life was maintained throughout treatment.¹⁹

KEYNOTE 177 was a phase III, open-label trial that enrolled 307 patients with previously untreated mCRC with dMMR. Patients were randomly assigned to receive pembrolizumab (200 mg once every 3 weeks) monotherapy or chemotherapy with fluorouracil-based treatment with or without bevacizumab (antivascular endothelial growth factor antibody) or cetuximab (antiepithelial growth factor receptor antibody). The study primary end points were PFS and OS. At a median follow-up of 32.4 months, pembrolizumab showed superior median PFS (16.5 months v 8.2 months, hazard ratio [HR] 0.60; 95% CI, 0.45 to 0.80; P = .0002) compared with chemotherapy. Grade 3-4 treatment-related AEs were observed in 22% of patients in the pembrolizumab arm versus 66% in the chemotherapy arm.²⁰ On the basis of these data, the FDA approved pembrolizumab monotherapy as first-line treatment for patients with unresectable or metastatic dMMR CRC on June 29, 2020.²¹ At further follow-up, analysis of PFS2 (time from random assignment to progression on the next line of therapy or death from any cause) showed that patients treated with pembrolizumab had longer PFS2 (median not reached v 23.5 months; HR, 0.63; 95% CI, 0.45 to 0.88) versus the chemotherapy arm.²² Although the final OS analysis at median follow-up of 44.5 months did not reach prespecified statistical significance (P = .0246), there was a 60% crossover rate (36% on study; 24% off study) and median PFS was unchanged at final analysis.²³ Pembrolizumab monotherapy remains the preferred first-line treatment option for patients with dMMR mCRC.

Ongoing studies aim to determine whether combining immunotherapy with chemotherapy will be more efficacious versus either alone. COMMIT is a phase III randomized open-label study in the first-line setting that was designed with three arms including modified fluorouracil, leucovorin, and oxaliplatin (mFOLFOX6) plus bevacizumab with or without atezolizumab (a programmed death ligand-1 [PD-L1] inhibitor), or atezolizumab alone in patients with treatment-naïve dMMR mCRC. The primary study end point is PFS. On the basis of the results of KEYNOTE 177, the protocol was modified to close the chemotherapy-only arm.²⁴ CHECKMATE 8HW is an international, randomized, phase III study (on the basis of Checkmate 142) in patients with dMMR mCRC. The study has three arms including chemotherapy of investigator's choice, nivolumab alone, or nivolumab with ipilimumab. The primary end point is PFS assessed by BICR.²⁵

NEOADJUVANT AND ADJUVANT IMMUNOTHERAPY

Neoadjuvant Immunotherapy

The only randomized phase III study of neoadjuvant chemotherapy for patients with nonmetastatic colon cancer (FOxTROT) failed to demonstrate a statistically significant improvement in recurrence rate in the overall population and in the subset of dMMR tumors.^26,27 A similar study of neoadjuvant chemotherapy in patients with locally advanced dMMR rectal cancer was also negative.²⁸ In a single-center, open-label study evaluating neoadjuvant immunotherapy in patients with dMMR and proficient mismatch repair (pMMR) early-stage colon cancers, all patients received one dose of ipilimumab (1 mg/kg) on day 1 and two doses of nivolumab (3 mg/kg) on days 1 and 15 (NICHE; ClinicalTrial.gov identifier: NCT03026140). The primary study end point was safety and feasibility. Treatment was well tolerated, and pathologic responses were seen in all dMMR patients with 19/20 (95%) showing major pathologic response (≤ 10% residual viable tumors) and 12 (60%) demonstrating pathologic complete response (pCR).²⁹ In the phase II VOLTAGE-A study (ClinicalTrials.gov identifier: NCT02948348) that enrolled T_3-4N_anyM₀ rectal cancer patients (37 pMMR in cohort A1 and five dMMR in cohort A2), all received chemoradiation followed by nivolumab (240 mg once every 2 weeks) for five cycles and then underwent surgical resection. A pCR rate of 60% (3/5) was seen in dMMR patients.³⁰

A single center, open-label phase II study (ClinicalTrials.gov identifier: NCT04082572) enrolled 35 dMMR nonmetastatic solid tumor patients with localized unresectable or high-risk resectable (≥ 20% recurrent risk) disease. All patients received pembrolizumab (200 mg once every 3 weeks for eight cycles) followed by surgery with the option to continue pembrolizumab treatment for a total of 18 cycles followed by observation. The primary end points were safety and pCR rate. Among patients with CRC, 69% (21/27) had a radiographic response and of the 12 patients who underwent surgery, 10 had a pCR.³¹

Although existing data are considered exploratory, they support a potential role for immunotherapy in the neoadjuvant treatment of dMMR CRC. Ongoing studies include a phase II trial (EA2201; ClinicalTrials.gov identifier: NCT04751370) in patients with dMMR stage II-III rectal cancer treated with two cycles of nivolumab (480 mg) and ipilimumab (1 mg/kg) once every 4 weeks, followed by short-course radiation (5 Gy for 5 days). Patients then receive an additional two cycles of nivolumab plus ipilimumab. Total mesorectal excision is planned 8-12 weeks after completion of cycle 4 of immunotherapy. The primary end point is tumor pCR rate.

Adjuvant Immunotherapy

Since patients with stage II dMMR CRC generally have an excellent prognosis without a clear survival benefit from adjuvant fluoropyrimidine-based adjuvant therapy,³² current guidelines do not recommend adjuvant treatment for these patients. However, the combination of a fluoropyrimidine and oxaliplatin remains the standard of care for adjuvant treatment of stage III colon cancers, including dMMR tumors.⁷ A recent analysis of pooled data in stage III dMMR colon cancers from the ACCENT database demonstrated a survival benefit for patients treated with a fluoropyrimidine plus oxaliplatin as adjuvant therapy, and this benefit was limited to low-risk (T_1-3N₁) tumors.^33,34

Given the benefit of immunotherapy seen in the metastatic setting, there are currently two phase III randomized trials evaluating the role of ICI as adjuvant therapy in patients with resected dMMR stage III tumors. The ongoing ATOMIC study (ClinicalTrials.gov identifier: NCT02912559) plans to enroll 700 patients who are randomly assigned to receive fluorouracil, leucovorin, and oxaliplatin (FOLFOX) for 12 cycles alone or combined with the anti–PD-L1 antibody, atezolizumab, which is given along with FOLFOX and then continued as monotherapy for an additional 6 months in the experimental arm. The primary end point of this study is disease-free survival.³⁵ Another phase III study named POLEM (ClinicalTrials.gov identifier: NCT03827044) is currently enrolling stage III patients with dMMR or POLE exonuclease domain mutations, and has a target accrual of 402 patients. In POLEM, patients are randomly assigned to receive fluoropyrimidine-based chemotherapy (CAPOX for 12 weeks or capecitabine for 24 weeks) alone or followed by the PD-L1 antibody, avelumab (10 mg/kg once every 2 weeks for 24 weeks). Disease-free survival is also the primary study end point.³⁶ The results of these adjuvant studies are awaited with keen interest since they have the potential to establish a new standard of care for the treatment of stage III dMMR colon cancers. Should these studies be positive, the next question will be whether an ICI as adjuvant monotherapy can achieve similar results in the absence of systemic chemotherapy in dMMR colon cancers.

BIOMARKERS AND RESPONSIVENESS TO ICIs

The ability to identify patients likely to exhibit intrinsic resistance to ICIs can avoid ineffective treatment with potential side effects and considerable cost. In patients with mCRC, neither the level of PD-L1 expression nor RAS/BRAF mutation status is predictive of ICI outcomes.^10,12,13,19 Furthermore, similar outcomes from ICIs are seen in patients with dMMR mCRCs independent of whether MMR deficiency is due to a hereditary versus sporadic etiology.^12,13,19,20 ICIs harness naturally occurring antitumor T-cell responses, and the efficacy of ICIs depends on the pre-existing immune response. In a study that quantified CD3⁺ and CD8⁺ T-cell densities in 278 dMMR and 283 pMMR colon cancers, higher T-cell densities were observed in dMMR versus pMMR tumors, yet greater intertumoral heterogeneity was also found among dMMR cancers.³⁷ The presence of T cells within the TME is the simplest indicator of a pre-existing antitumor immune response and has been associated with responsiveness to ICIs in patients with melanoma.³⁸ However, this finding has not been universal, and local immunosuppressive factors may limit clonal expansion of T cells or other processes. Accordingly, tumor-infiltrating lymphocytes (TILs) are not a particularly sensitive surrogate for the presence of a de novo antitumor immune response and their predictive utility for immunotherapy, including Immunoscore, in dMMR tumors has not been demonstrated.³⁹ Responses require not only the presence of T cells, but their activation as indicated by production of interferon-γ (IFN-γ) as was reported in patients with dMMR CRCs.¹⁰

Both TMB and an inflamed TME were more strongly associated with response across multiple prospective studies of anti–PD-1 therapy than was either biomarker alone.⁴⁰ TMB has limitations as predictor of ICI response that may be due, in part, to tumor heterogeneity regarding the clonality of mutations. Clonal mutations are shared across all tumor cells in a given patient, and are critical to generation of an effective antitumor response.⁴¹ Importantly, mutations must be effectively presented to the immune system. Among dMMR mCRCs, median TMB was 46.1 mutations/Mb compared with 3.5 mutations/Mb in pMMR tumors on the basis of the Foundation Medicine (FM) database of 18,140 mCRC cases of which 821 were dMMR. In patients with dMMR mCRCs (N = 22) treated with an ICI, TMB predicted ORR and PFS with an optimal cutpoint estimated to be between 37 and 41 mutations/Mb. This cutoff resulted in 13 patients being classified as TMB-high who responded, whereas six of nine TMB-low patients had disease progression.⁴² However, a recent analysis of dMMR mCRCs (N = 29) from the KEYNOTE 177 study found that TMB was not predictive of response to anti–PD-1 therapy while high clonality of immunogenic mutations and clonally expanded T cells correlated with disease response.⁴³ Responsive tumors were also rich in PD-1+ CD8 T cells interacting with PD-L1+ CD74+ presenting macrophages. Of note, FDA approved pembrolizumab for the treatment of TMB-high unresectable or metastatic solid tumors defined as a TMB cutoff of ≥ 10 mutations/Mb on the basis of a retrospective analysis of 10 patient cohorts (N = 102) from the KEYNOTE 158 study (ClinicalTrials.gov identifier: NCT02628067). In this analysis, pembrolizumab produced a 29% ORR with 4% CR in TMB-high tumors,⁴⁴ yet there were no CRCs included.^45,46 In the KEYNOTE 016 study, 12 dMMR primary refractory cases were identified, of which three had exome sequencing with an average 1,413 nonsynonymous mutations, which did not differ significantly in number from that of responders.⁴⁷ This finding suggests that the type of mutations in addition to their quantity is also an important factor. Of note, the MSIsensor score, a validated algorithm quantifying the number of unstable microsatellites against the reference genome, was shown to correlate with frameshifting indel mutations and demonstrated greater predictive utility than did MSI status for responsiveness to immunotherapy.⁴⁷

PD-L1 expression within the TME has shown utility as a predictor of response to ICIs in some tumor types, but not in CRCs including those with dMMR.^10,19 A subgroup analysis of the Checkmate 142 first-line cohort study did not show a significant difference in response rate for dMMR CRCs with PD-L1 ≥ 1% (75%) versus those with PD-L1 < 1% (65%).¹⁹ PD-L1 expression does not necessarily indicate a pre-existing antitumor immune response as some PD-L1–negative tumors can derive benefit from ICIs. Expression of PD-L1 is primarily regulated by INF signaling, and IFN-γ has been shown to stimulate PD-L1 expression, although PD-L1 is also modulated through various other mechanisms.⁴⁸

Mutations in β2 microglobulin (B2M) are enriched in dMMR CRCs.^49,50 β2M is an extracellular component of major histocompatibility complex (MHC) class I and plays a critical role in antigen presentation.⁵¹ Early studies in patients with metastatic melanoma suggested that mutations, deletions, or loss of heterozygosity in B2M was associated with resistance to ICI treatment.⁵² However, conflicting data exist as a study in CRC found that among 44 dMMR CRCs that harbored β2M mutations, 85% (11/13) of patients received clinical benefit from ICI treatment.⁵⁰

INTRINSIC RESISTANCE TO IMMUNOTHERAPY

Although immunotherapy leads to durable responses in nearly 50% of dMMR mCRCs, many patients with dMMR mCRC do not respond to initial ICI (named primary refractory) or develop resistance during the ICI treatment (termed secondary refractory). ICIs are therapeutic antibodies that disrupt negative immune checkpoint regulation and unleash pre-existing antitumor immune responses. To date, the ability to predict which patients are likely to respond to ICIs is limited, and ongoing efforts aim to elucidate the molecular mechanisms of intrinsic resistance.

Tumors that activate the antitumor immune response because of increased TMB and antigenicity are most likely to benefit from ICIs. Even with sufficient antigenicity, however, sensitivity to ICIs can be antagonized by defects in factors regulating antitumor immunity including the mutational landscape, IFN signaling pathway function, expression of antigen-presenting molecules, and immune-evasive oncogenic signaling. These factors influence the priming, activation, and recruitment of T cells to the TME, which are necessary for an immune response in context of ICI blockade. Oncogenic signaling pathways, such as mitogen-activated protein kinase and WNT-β-catenin, govern the recruitment of cells needed for initiation and effectiveness of antitumor immunity. These signaling pathways can affect IFN-γ and antigen presentation, or they can induce immunosuppressive factors in the TME that confer resistance.⁵³ The increased cytotoxic CD8+ T-cell infiltration in hypermutated CRCs is likely enabled by reduced activation of the WNT pathway. Accordingly, tailored approaches that can circumvent defects in IFN-γ signaling and antigen presentation or inhibit immune suppressive pathways may enable more tumors to benefit from ICIs.

The fact that dMMR CRCs have abundant mutation-specific neoantigens and have high response rates to ICIs supports the role of neoantigens in triggering the antitumor immune response. T-cell–specific neoantigens in the TME were shown to expand in response to treatment with an anti–CTLA-4 antibody. Conversely, insufficient antigenicity is an important contributor to resistance to ICIs. A therapeutic strategy is the adoptive transfer of mutation-specific T cells to confer an effective antitumor response.⁵⁴ A T-cell response against a tumor antigen results in expression of IFN-γ in the TME that induces PD-L1 expression via JAK-STAT signaling. It is known that disruption of tumor cell responses to IFN-γ signaling is a key resistance mechanism to ICIs that allows the tumor to evade immune effector function. Mutations in IFN-γ–related genes were observed in patients lacking response to anti–CTLA-4 treatment. Loss-of-function mutations in JAK1 and JAK2 were detected in patients with melanoma who developed late relapses after responding to anti–PD-1 therapy.⁵³ However, clinical studies combining JAK inhibitors with anti–PD-1 antibodies are ongoing, yet early results have not appeared promising.⁵⁶

Tumor cells evade destruction by downregulating cell surface MHC class I expression that is critical for tumor antigen presentation. The importance of IFN-γ signaling in antitumor immunity may be related to its ability to induce/enhance MHC class I antigen presentation, which requires genes including B2M.⁵⁷ Patients with melanoma who received immunotherapy were shown to lose functional expression of B2M and thereby MHC class I expression.^58,59 Furthermore, acquired resistance to ICIs have been reported in tumors with mutations in genes encoding antigen processing machinery, including B2M,^52,55 as shown in dMMR CRCs.⁶⁰

Circumventing Immunotherapy Resistance

For tumors that fail to respond to single or combination ICI, the goal is to trigger antitumor immune responses by enhancing antigen presentation and immune responses against existing antigens. Such strategies include (1) inducing a proinflammatory state that overwhelms the mechanism of immune suppression in the TME, (2) inducing immunogenic cell death, or (3) recruiting antigen-presenting cells for efficient priming against tumor antigens. Figures 1A and 1B summarize different approaches that are under investigation to circumvent immunotherapy resistance.

FIG 1.

(A) Schematic overview of potential approaches to circumvent immunotherapy resistance: (1) Rechallenge with anti–PD-1 or anti–PD-L1 inhibitor, (2) combine with other immune checkpoint inhibitors, (3) combine with chemotherapy or radiation, (4) tumor vaccine and other agents that may increase neoantigen presentation, and (5) ACT. (B) Schematic overview of ACT approaches including TILs, TCR therapy, and CAR-T cell therapy. For TIL treatment, TILs are isolated from tumor tissue followed by REP before adoptive infusion into the patient. In TCR and CAR therapy, peripheral blood T cells are isolated by leukophoresis and then transduced using viral vectors to express either a TCR or CAR, respectively. Patients require a lymphodepleting treatment before ACT infusion. ACT, adoptive cellular therapy; CAR, chimeric antigen receptor; MHC, major histocompatibility complex; PD-1, programmed death-1; PD-L1, programmed death ligand-1; REP, rapid expansion protocol; TCR, T cell receptor; TIL, tumor-infiltrating lymphocyte. Figures were created with BioRender.⁸¹

Combination ICIs

In patients with chemo-refractory mCRC with dMMR treated in the phase II Checkmate 142 study, response rates for nivolumab and nivolumab plus ipilimumab were 31.3% and 55%, respectively.^12,13 Of note, CHECKMATE 142 enrolled 45 patients and the OS result is not yet mature. In the first-line setting, nivolumab plus ipilimumab led to an ORR of 69% compared with 43.8% for pembrolizumab as shown in the KEYNOTE 177 study.^19,20 In this cross-trial comparison among dMMR mCRCs, the CR rate increased to 13% from 7% for nivolumab plus ipilimumab versus pembrolizumab.^19,20 Although a study that directly compares dual ICIs with single-agent PD-1 inhibition is lacking, the observed higher response rate of dual ICIs indicates nonredundant mechanisms of action. A planned comparison of nivolumab versus its combination with ipilimumab in patients with treatment-naïve mCRC with dMMR will occur in the CHECKMATE 8HW (ClinicalTrials.gov identifier: NCT04008030) study.²⁵ These data are needed to guide the use of dual ICIs and treatment sequencing in the first-line setting, and studies are also needed in patients who fail initial pembrolizumab monotherapy.

ICI Rechallenge

Data suggest that retreatment with PD-1 blockade can be effective in some patients with prior resistance to anti–PD-1 treatment. In the final analysis of KEYNOTE 164, nine patients with dMMR tumors were retreated with pembrolizumab after prior progression on pembrolizumab, and six of nine had a PR.⁶¹ Data in other solid tumors suggest that adding a CTLA-4 inhibitor to a PD-1 (or PD-L1) inhibitor may reverse immunotherapy resistance. In a retrospective multicenter study that included 84 patients with advanced melanoma whose disease progressed on anti–PD-1 therapy, 47 patients received ipilimumab and 37 received ipilimumab plus nivolumab with response rates of 16% and 21%, respectively.⁶² In a retrospective study with 45 patients with metastatic renal cell carcinoma who had disease progression on a PD-1 inhibitor, ipilimumab and nivolumab treatment was associated with an ORR of 20% and a 4-month median PFS, suggesting that this combination can salvage patients refractory to PD-1 inhibition.⁶³ These data and the observation of increased response to dual PD-1 and CTLA-4 inhibition in dMMR mCRCs suggest that ipilimumab may play an important role in the treatment of PD-1 or PD-L1 refractory tumors.

ICI Plus Chemotherapy and/or Radiation

Chemotherapy and radiation can induce immunogenic cell death via multiple proposed mechanisms.⁶⁴ Both treatment modalities are dependent on T cells, and both can augment the impact of ICIs. However, chemotherapy and radiation also have immunosuppressive functions as well that can induce extrinsic mechanisms of resistance to immunotherapy.⁶⁵ Therefore, these standard therapies are unlikely to represent a primary approach to overcoming intrinsic resistance, yet their ability to control disease burden and trigger immunogenic cell death may be useful with emerging combination immunotherapies.

On the basis of promising data from a phase Ib study (ClinicalTrials.gov identifier: NCT01633970) of atezolizumab combined with FOLFOX and bevacizumab,⁶⁶ a phase II study (AtezoTRIBE) was conducted whereby 218 patients with mCRC were randomly assigned (1:2 ratio) to receive up to eight cycles of FOLFOXIRI and bevacizumab alone (TRIBE arm; n = 73) or combined with atezolizumab (AtezoTRIBE arm; n = 145). Treatment was then followed by maintenance fluorouracil/leucovorin and bevacizumab with or without atezolizumab until disease progression. The TRIBE and AtezoTRIBE study arms included 7% and 6% dMMR cases, respectively. A statistically significant improvement in median PFS to 13.1 months was observed in the AtezoTRIBE arm versus 11.5 months in the TRIBE arm (HR, 0.69; 80% CI, 0.56 to 0.985; P = .012). Patients with dMMR tumors in the AtezoTRIBE arm showed significantly increased median PFS (not reached at a median follow-up of 20.6 months). An interaction analysis showed that the magnitude of PFS benefit varied according to MMR status (P-interaction .010).⁶⁷

The ongoing COMMIT study in dMMR mCRCs will determine whether immunotherapy plus chemotherapy is superior to single-agent anti–PD-L1 therapy. Other studies exploring the role of ICIs in combination with chemotherapy and biologics are summarized in Table 2.

TABLE 2.

Studies Evaluating the Role of Immune Checkpoint Inhibitor With Chemotherapy in mCRCs (including both dMMR and pMMR cases)

ICI With Targeted Agents

Regorafenib, a multitarget tyrosine inhibitor approved for treatment of mCRC, is a potent vascular endothelial growth factor inhibitor that reduces tumor-associated macrophages and decreases PD-L1 expression leading to a more inflamed TME. Synergy with regorafenib and an ICI was reported in preclinical CRC models and led to the REGONIVO study. This study enrolled 50 chemorefractory patients including 25 with mCRC, where treatment with regorafenib plus nivolumab achieved an ORR of 36% in patients with mCRC including one dMMR case.⁷³ However, only five PRs (7.1%) were found in a phase II study evaluating the same regimen also in chemorefractory mCRC patients.⁷³

To address resistance to ICIs driven by oncogenic signaling pathways, strategies to boost antitumor immunity include inhibitors of mitogen-activated protein kinase, PI3K, and WNT signaling combined with ICIs. Binimetinib is an MEK inhibitor that increases MHC class I expression, increases TILs, and downregulates multiple immunosuppressive cytokines in preclinical models. In a phase III study known as IMblaze 370, however, the combination of binimetinib and atezolizumab failed to show an OS benefit versus regorafenib in chemorefractory mCRC patients.

Combination of ICIs With Novel Immunotherapy

Multiple antibodies and small molecules targeting other putative immune checkpoints (ie, LAG3, TIM3, TIGIT, etc) can disrupt negative regulation between tumor cells and T cells or myeloid cells and T cells, and are undergoing clinical development. Lymphocyte-activating gene 3 (LAG3, CD223) is a member of the immunoglobulin superfamily that inhibits T-cell function and promotes tumor immune escape. LAG3 overexpression has been detected in human CRCs. Relatlimab is a human immunoglobulin G4 LAG3-blocking antibody with the potential to restore the effector function of exhausted T cells. In an international, double-blinded, and randomized phase III study (RELATIVITY-047), nivolumab combined with relatlimab significantly prolonged mPFS from 4.6 to 10.1 months (HR, 0.75; 95% CI, 0.6 to 0.9; P = .0055) versus nivolumab monotherapy in 714 treatment-naïve patients with advanced melanoma. Toxicity was manageable with a grade 3/4 treatment-related AE rate of 18.9%.⁷⁵ Other LAG3 inhibitors are under investigation including LBL-007, which showed significant antitumor activity alone or combined with an anti–PD-1 antibody in CRC xenografted mice.⁷⁶ Clinical trials with TIM3 (T-cell immunoglobulin and mucin domain-containing protein 3) and TIGIT (T-cell immunoreceptor with Ig and ITIM domains) in CRC are ongoing.

Tumor Vaccines

Tumor-specific antigens or neoantigens resulting from spontaneous mutations in tumor cells are considered excellent targets for immunotherapy using a tumor vaccine approach. In a phase I/II study (ClinicalTrials.gov identifier: NCT03639714) that enrolled 22 patients with various cancers (10 pMMR CRCs, 10 gastroesophageal cancers [GEA], and two non–small-cell lung cancer), all were treated with a patient-specific and neoantigen-directed heterologous prime/boost vaccine. This vaccine leveraged 20 neoantigens that were identified/selected and was administered together with nivolumab and ipilimumab. Among 18 evaluable patients, there was one CR (GEA), four stable disease (three CRCs and one GEA), and 11 with disease progression. Of the nine patients with pMMR CRCs, five were progression-free beyond 6 months and of these, four of five had circulating tumor DNA that decreased ≥ 50% from baseline.⁶⁵ Potent immunogenicity, as evidenced by CD8⁺ neoantigen-specific T cells, was detected in all patients. On the basis of these data, phase II/III clinical trials are planned to further evaluate the activity of a personalized neoantigen vaccine in CRC. Instead of a personalized neoantigen-directed tumor vaccine, off-the-shelf shared neoantigen vaccines are also undergoing evaluation.

Adoptive Cellular Therapy

Adoptive cellular therapy (ACT) is a personalized immunotherapy approach that uses the patients' own immune cells after ex vivo expansion, modification, and then reinfusion. The clinical success of ACT has been shown in patients with metastatic melanoma.⁷⁷ The majority of ACT uses T cells, but @other immune cell subsets such as natural killer cells and dendritic cells have also been studied. There are three main classes of ACT including TILs, engineered T-cell receptors (TCRs), and chimeric antigen receptor that are in the process of development for solid tumors (Fig 1B and Data Supplement). Table 3 lists selected current and ongoing immunotherapy trials in patients with dMMR metastatic CRC.

TABLE 3.

Selected Studies Evaluating ICI Treatment in dMMR CRCs

In conclusion, tumors with an increased mutational burden and neoantigenicity activate an antitumor immune response and, accordingly, have the greatest likelihood of benefit from ICIs. ICIs are highly effective in patients with dMMR mCRCs. Treatment of CRCs and other solid tumors with dMMR using ICIs has produced remarkable and durable tumor responses. Studies have shown that even higher response rates can be achieved with the combination of PD-1 inhibitors and anti–CTLA-4 antibodies, suggesting nonredundant mechanisms. Although progress in CRC is currently limited to the metastatic disease, ongoing studies will determine whether these advances can be applied to earlier-stage disease. Despite substantial progress made, an appreciable proportion of dMMR cancers display intrinsic resistance to immunotherapy. Resistance mechanisms include tumor-intrinsic genetic defects in IFN-γ signaling and antigen presentation, as well as oncogenic signaling pathways that alter these processes or induce immunosuppressive factors in the TME. An increased understanding of these processes and their role in resistance is expected to advance predictive biomarker identification as well as facilitate development of targeted strategies to bypass such defects and broaden the benefit of ICIs. Current strategies to circumvent resistance include combination immunotherapies and rechallenge with ICIs. Novel approaches being tested include use of antibodies and small molecules targeting other immune checkpoints including LAG3, TIM3, and TIGIT as well antitumor vaccines and adoptive cellular therapies.

AUTHOR CONTRIBUTIONS

Conception and design: All authors

Administrative support: Frank A. Sinicrope

Collection and assembly of data: All authors

Data analysis and interpretation: All authors

Manuscript writing: All authors

Final approval of manuscript: All authors

Accountable for all aspects of the work: All authors

AUTHORS' DISCLOSURES OF POTENTIAL CONFLICTS OF INTEREST

Mismatch Repair-Deficient Colorectal Cancer: Building on Checkpoint Blockade

The following represents disclosure information provided by authors of this manuscript. All relationships are considered compensated unless otherwise noted. Relationships are self-held unless noted. I = Immediate Family Member, Inst = My Institution. Relationships may not relate to the subject matter of this manuscript. For more information about ASCO's conflict of interest policy, please refer to www.asco.org/rwc or ascopubs.org/jco/authors/author-center.

Open Payments is a public database containing information reported by companies about payments made to US-licensed physicians (Open Payments).