Tumors with mismatch repair deficiency (MMRd) for DNA replication machinery exhibit high microsatellite instability (MSI-H) and high tumor mutational burden (TMB) in their genotype. Such tumors present with poor prognosis in the advanced setting and benefit less from conventional treatments compared with MMR-proficient tumors (MMRp). On the other hand, MMRd and MSI-H status have been associated with increased sensitivity to immunotherapy — attributed to elevated burden of immunogenic frameshift mutations and mutation-associated neoantigens (Fig. 1). Accordingly, immune checkpoint blockade (ICB) has revolutionized the outcome of these cancers, with recent studies demonstrating remarkable and durable responses in the neoadjuvant and advanced settings. Here we discuss recent clinical advances that provide substantial hope for patients with MMRd or proofreading-deficient diseases, and that will guide future investigations of combination therapies and approaches to refine patient selection in those histology-agnostic cancer indications.

Fig. 1: MMRd and POLE-mutated tumors display high TMB and sensitivity to immunotherapy.
figure 1

MMRd and POLE-mutant tumors contain DNA repair defects responsible for the accumulation of mutations and are therefore characterized as having a high TMB. As a consequence, they exhibit elevated numbers of mutation-associated neoantigens. Presentation of these neoantigens by MHC-I molecules can facilitate tumor recognition by CD8+ T cells and immune-mediated cancer cell killing. These tumors might escape from the immune system through the upregulation of immune checkpoints, including PD-L1, CTLA4 and LAG3 (lymphocyte activation gene 3). ICB can restore anti-tumor immunity and generate durable tumor responses in patients with either somatic or germinal MMRd or POLE mutations.

Full size image

Phase 1/2 clinical trials

Early phase immunotherapy trials have shown remarkable results in MMRd tumors in recent years. The ongoing CheckMate 142 phase 2 trial is a multicohort non-randomized study evaluating nivolumab (monoclonal antibody to programmed death 1 (PD-1)) monotherapy or in combination with ipilimumab (monoclonal antibody to cytotoxic T lymphocyte–associated antigen 4 (CTLA4)) in metastatic MMRd colorectal cancer (CRC), either in the first-line setting or in later lines, for which five-year follow-up data were presented at the 2022 ASCO Annual Meeting1. The median overall survival (OS) was not reached for patients in the combination arm, and survival curves were similar whether the therapy was given in first or later lines. Indirect comparisons suggest that the survival of patients treated with the combination therapy is better than treatment with nivolumab alone; however, head-to-head comparisons are needed for confirmation. Additionally, as combination ICB also comes with increased risk of immune-related adverse events; predictive biomarkers are needed to identify patients for whom combination therapy is essential, versus patients who would receive benefit from nivolumab alone. Results of the randomized CheckMate 8HW study, which compares nivolumab alone versus nivolumab in combination with ipilimumab versus chemotherapy in the first-line setting (NCT04008030), are therefore eagerly awaited.

Given the success of ICB in MMRd patients with advanced cancer, there is now interest in using these drugs in the neoadjuvant setting. NICHE is an open-label proof-of-concept study with an adaptive design in which patients with non-metastatic stage II and III, resectable and untreated MMRd or MMRp colon cancers received one low dose of ipilimumab (1 mg kg–1) on day 1 and one full dose of nivolumab on days 1 and 15 (ref. 2). Surgery took place within six weeks of enrollment. The study showed a pathological response in 100% of patients with MMRd, including a pathological complete response (pCR) in 69% of patients, compared with a pathological response in only 29% of MMRp patients and no pCR (20 and 15 patients per arm, respectively). The results of the NICHE 2 trial, presented at the 2022 ESMO Congress, confirmed previous finding from an expanded cohort of 112 patients with MMRd with 95% major pathological responses, including a pCR in 67% of patients3. No patient experienced disease recurrence after a median follow up of 13 months (range 1–57 months). The results are also interesting because ICB was not provided as an adjuvant treatment and there was low toxicity, with grade 3–4 immune-related adverse events in only 13% of patients, probably because of the short treatment exposure and low single dose of ipilimumab. Another phase 2 trial with neoadjuvant anti-PD-1 (toripalimab) with or without celocoxib in patients with non-metastatic stage II and III MMRd CRC has also recently reported an impressive pCR rate (NCT03926338).

Conversely, Cercek et al. investigated single-agent anti-PD-1 dostarlimab, administered every 3 weeks in patients with MMRd stage II or III rectal adenocarcinoma, in a phase 2 study4. Patients underwent close clinical follow up and reassessment after 6 months. Patients that achieved a clinical complete response (cCR) could forgo additional treatments in favor of a non-operative follow up. Remarkably, all 14 patients enrolled had a cCR, as well as a pCR on serial biopsies. These results are particularly encouraging in rectal cancer, where both surgery and chemoradiation come with frequent functional complications. In addition, for MMRd gastric or oesophageal junctional adenocarcinoma, neoadjuvant ICB may allow for an organ preservation strategy. A major challenge for organ-sparing strategies is how to reliably assess the treatment response, as the NICHE study reported poor concordance between clinical and pathological responses. The timing of surgery is also a key consideration; for example, treatment in the NICHE study was given for 6 weeks2, whereas Cercek et al. reported that the median time from the start of treatment to a cCR was >3 months4. Additional cycles of ICB and longer observation times for responders may lead to the identification of patients reaching a cCR who could then avoid surgery. The advent of monitoring circulating tumor DNA as a pre-emptive surrogate for immunotherapy efficacy will be highly valuable in guiding decision making in this context. The long-term preventive impact for patients with Lynch syndrome is still unknown, raising further the need for longer follow ups and potential assessment of longer treatments. Although major benefits are expected in terms of patient survival (potentially cured), quality of life (avoidance of surgery) and societal costs (shorter treatment durations and less relapses), those treatments are not currently available in the neoadjuvant setting for most patients. Larger prospective studies with longer follow ups will likely be required prior to formal approvals, even though the safety profile of neoadjuvant ICB is not predicted to greatly differ from that in the adjuvant setting.

Phase 3 clinical trials

KEYNOTE-177 is the first phase 3 study to randomize patients with metastatic MMRd CRC in the first-line setting to receive pembrolizumab or standard of care5. Patients who were randomized to chemotherapy had the option to cross over to pembrolizumab in the event of disease progression. The study showed doubled progression-free survival for patients who received pembrolizumab (16.5 versus 8.2 months). Despite an improvement with a hazard ratio of 0.74, OS did not yet reach its endpoint — a key factor being that 60% of patients crossed over. This study definitively established ICB as a standard of care in this patient population, and pembrolizumab finally obtained approval as a first-line treatment for patients with unresectable or metastatic MMRd CRC in Europe, the USA and many other countries. Nonetheless, nearly 30% of patients in this study experienced early progression with ICB, compared to 12.3% with chemotherapy with or without bevacizumab or cetuximab, raising considerations for the design of future clinical studies. First, the KEYNOTE-177 trial did not include central assessment of MMRd status, raising the possibility that some patients in the study were inaccurately diagnosed as MMRd, which has been identified as a cause of primary resistance in MMRd CRC6. Additionally, response evaluation using RECIST1.1 criteria may have underestimated true benefits as pseudoprogressions can occur in this patient population. Lastly, there is certainly a number of patients for whom a single ICB treatment is not sufficient, and ICB combination strategies are needed. ICB in combination with chemotherapy is also under investigation for patients with MMRd CRC in the first-line setting in the COMMIT study: atezolizumab versus mFOLFOX in combination with bevacizumab and atezolizumab (NCT02997228). Of note, the number of progressive patients is much higher in the metastatic setting than in the localized setting, suggesting differences in immunotherapy sensitivity and therefore biological differences between primary and metastatic MMRd tumors. This underscores the importance of early diagnosis of MMRd status to allow for early ICB treatment, for which the success rate is higher.

The paucity of phase 3 studies reflects the complexity of conducting trials in these rare tumors. Given the consistent unprecedented response rate with ICB across different trials and tumors, together with a known safety profile for those drugs, absolute requirements for phase 3 randomized trials to obtain approval for use as standard of care warrant reconsideration. A good example is the first tumor-agnostic approval by the US Food and Drug Administration for anti-PD1 therapy for MMRd/MSI-H tumors in 2017, based on several phase 2 trials showing frequent and durable responses in patients with previously treated advanced cancers. However, approval was only given in Europe in 2021 and limited to first-line unresected or metastatic CRC after the KEYNOTE-177 phase 3 results were published. Such delay represents a true loss of opportunity for all the patients who could not access those treatments through clinical trials and is ethically challenging for oncologists who cannot administer such life-saving treatments. In addition, most clinical trials have been conducted in CRC where MMRd has a high incidence (up to 15% in early stages; around 5% in advanced stages). However, MMRd and MSI-H status can be found in many tumor types where ICB has been shown to be as effective as in CRC cohorts but, given the low incidence, there is as yet no formal recommendation to broadly test for MMRd status and it is unlikely that sufficiently large datasets will be able to confirm ICB activity in every cancer. Circulating tumor DNA again offers an easy and non-invasive screening tool, not only to test for MMR status but also to predict sensitivity to immunotherapy. Chakrabarti et al., for example, detected 83% concordance between liquid biopsy and tumor-based detection of MSI-H status in pancreatic cancer, and found an overall response rate (ORR) of 77% with durable responses following ICB therapy in this hard-to-treat indication7.

Robust ancillary studies are needed for clinical trials in this rare setting in order to identify predictive biomarkers, understand mechanisms of primary and secondary resistance, and guide future studies.

Preclinical developments

As covered above, a substantial proportion of MMRd patients present with primary resistance in ICB clinical trials, which may depend on the treatment previously received, type of ICB, histotype, stage of the disease or treatment line. This indicates that neither MMRd status nor high TMB alone is sufficient to predict response and points out the heterogeneity even within this subset of tumors. Identifying predictive markers and overcoming resistance to immunotherapy are major challenges being addressed in preclinical studies. By studying the genomic profile of MMRd gastrointestinal tumors, Wang et al. identified AKT serine/threonine kinase-1 (AKT1) and cadherin-1 mutations as independent predictors of poor progression-free survival and primary resistance to ICB8. The combination of these two genes as an immuno-oncology therapy predictor (IOpred) demonstrated good performance in predicting primary resistance (area under the curve 0.751, with 98% specificity), which was validated in an independent cohort. However, the mechanisms underlying the predictive role of the two genes have not been explained and, although the tool may be simple, it does not identify all patients (52% of sensitivity) and may exclude other patients who could potentially respond. Another study identified that the number of activating mutations in the PI3K–AKT–mTOR pathway was negatively correlated with both immune cell infiltration and ICB efficacy in patients with resected gastroesophageal MMRd cancers9. There is growing appreciation that inhibiting PI3K–AKT–mTOR not only targets oncogenic signaling in tumor cells, but also has immuno-modulatory potential by increasing the number of tumor-infiltrating lymphocytes. This study provides rationale to evaluate the combination of ICB and such inhibitors for patients with a high number of mutations. So far, no correlation has been found between TMB, programmed death ligand-1 (PD-L1) expression and the density of T cell infiltration in tumors, which suggests that TMB might behave as an independent biomarker of response to ICB.

Recent work also suggests that resistance mechanisms to ICB for MMRd disease may differ from other solid tumors. Unlike in melanoma, pathogenic mutations in the B2M gene (which encodes an important component of major histocompatibility complex class I (MHC-I) and is involved in antigen presentation) and in JAK1 and JAK2 (which encode key proteins in cytokine receptor downstream signaling) were not associated with primary resistance to ICB in two independent cohorts of patients with MMRd CRC10. It has been suggested that patients with MHC-I-deficient tumors (which cannot be recognized by CD8+ T cells) experience a compensatory increase in CD4+ T cell tumor infiltration that can potentially activate other cells such as natural killer cells and macrophages to generate anti-tumor immune responses. Specific cell populations in the tumor microenvironment could be of particular clinical relevance — for example, some PD-L1 antigen-presenting macrophages interacting with CD8+ T cells, as suggested recently in ancillary studies of the KEYNOTE-177 trial (NCT02563002).

Chida et al. hypothesized that aside from specific mutations in key genes, gene signatures detected via mRNA sequencing may also be potential biomarkers for predicting sensitivity to immunotherapy11. By analyzing patients with a range of gastrointestinal tumor types, they found that upregulation of angiogenesis pathways was associated with significantly reduced numbers of tumor-infiltrating CD8+ T cells. This suggests a rationale for combining ICB with antiangiogenics in resistant MMRd and MSI-H patients, as proven effective in other tumors such as renal cancer or hepatocellular cancer. In murine tumor models of MMRd, Nebot-Bral et al. also showed that primary resistance to anti-PD-L1 was associated with increased tumor-induced circulating neutrophils and accumulation of intratumoral regulatory T cells. The combination of anti-PD-L1 with anti-CTLA4 therapy was able to partially overcome such resistance, similar to the increased efficacy seen in patients12.

Perspectives

These recent developments collectively reinforce the relevance of MMRd status for gaining benefit from immunotherapy, and provide directions for future preclinical and clinical studies to widen and enhance responses in this patient population. Of note, other interesting work aims to apply the principles underlying these advances to make similar progress in non-MMRd cancers. Some MMRp and microsatellite-stable tumors might present as high TMB; namely, those harboring pathogenic mutations in the gene encoding polymerase epsilon (POLE), which is involved in DNA replication and repair, display ultramutated genotypes (>100 mutations per megabase). Rousseau et al. recently reported the first prospective trial assessing the efficacy of ICB in patients with advanced solid tumors that harbor predefined pathogenic POLE mutations13. This phase 2 study met its primary endpoint with 38% ORR at 12 weeks, and a median OS still not reached. Interestingly, responses were limited to tumors with POLE mutations clustered in DNA-binding or catalytic sites that confer proofreading defects and in turn high TMB, unlike POLE mutations that do not affect proofreading. MMRd and POLE-mutated tumors therefore both show robust responses with immunotherapy, highlighting that the high TMB that they share in common, and/or intrinsic DNA repair defects that could activate the cGAS–STING pathway, are important for these effects. A recent example has been published in metastatic castration-resistant prostate cancer known to derive poor benefit from ICB14. Patients with high TMB had significantly better outcomes with ICB than with chemotherapy (that is, a median OS of 19.9 months, versus 4.2 months). However, TMB also encounters its own challenges, one being the absence of a clear cut-off threshold to predict ICB efficacy across cancers. Also, specific patterns of mutations, as opposed to absolute mutation burden alone, might be associated to better sensitivity to ICB (for example, accumulation of indels).

Most tumors present as low TMB and one idea for triggering their immune response could be to mimic MMRd and POLE-mutated tumors. The drug temozolomide induces the appearance of multiple mutations, including mutations in the MMR gene MSH6. Interestingly, Crisafulli et al. leveraged this in a proof-of-concept phase 2 study enrolling heavily pre-treated patients with O6-methylguanine DNA methyltransferase-methylated MMRp mCRC15. Patients were first treated with temozolomide and then received pembrolizumab if the TMB was ≥20 mutations per megabase after treatment. Preliminary results showed that tumors carrying the characteristic temozolomide signature all developed mutations in MSH6 with increased subclonal or clonal TMB levels and a disease-control rate of 67% in patients who received pembrolizumab.

Along with research into the clinical and biological features of MMRd cancers, rapid uptake of next-generation sequencing for biomarker discovery and patient selection should hopefully provide sustained therapeutic advances for patients in the coming years.