Deficient mismatch repair/microsatellite instability-high colorectal cancer: current treatment paradigms, limitations and future perspectives

Abstract

The treatment paradigm for deficient mismatch repair (dMMR) colorectal cancer (CRC) has evolved substantially in recent years. Universal mismatch repair testing and broad use of next-generation sequencing have increased the detection of dMMR tumours, which exhibit microsatellite instability-high (MSI-H), hypermutation and abundant neoantigens. The presence of dMMR/MSI-H serves as a robust predictive biomarker for immune checkpoint inhibitor (ICI) therapy, which has demonstrated superior efficacy to cytotoxic chemotherapy in the metastatic setting and led to the first tumour-agnostic Food and Drug Administration approval in 2017 for metastatic dMMR/MSI-H solid tumours. Recent evidence also supports the benefit of ICIs in non-metastatic dMMR CRC. Neoadjuvant immunotherapy has produced high rates of pathological response in both colon and rectal cancers. In locally advanced dMMR rectal cancer, ICI therapy has enabled omission of chemoradiation and surgery in most patients. In resected node-positive dMMR colon cancer, the addition of ICI therapy to chemotherapy has substantially improved disease-free survival, expanding its role earlier in the disease course. Despite these advances, the optimal treatment strategy for non-metastatic dMMR CRC remains undefined due to the lack of direct comparative studies. Importantly, a subset of patients derives limited or no benefit from ICIs despite dMMR/MSI-H status, underscoring the need to further elucidate resistance mechanisms and to develop strategies to overcome them.

Introduction

Colorectal cancer (CRC) is the third most common cancer in the USA and is the second leading cause of cancer-related mortality.¹ Approximately 10–15% of CRCs exhibit deficient mismatch repair (dMMR) resulting in microsatellite instability-high (MSI-H). These tumours are hypermutated and generate abundant neoantigens that elicit a pronounced intratumoral immune response.^{2 3} Across solid tumours, including CRC, dMMR/MSI-H is a predictive biomarker for responsiveness to immune checkpoint inhibitor (ICI) therapy.^{4 5} The dMMR phenotype arises either from germline pathogenic variants in mismatch repair (MMR) genes (MLH1, MSH2, MSH6 or PMS2) conferring the Lynch syndrome36,8 or, more commonly, from sporadic mechanisms such as epigenetic silencing of MLH1 or somatic inactivation of MMR genes. The prevalence of dMMR/MSI-H is higher in localised CRC (10–15%) and declines substantially in metastatic disease (4–5%).9,12

Universal testing of MMR protein expression by immunohistochemistry (IHC) is routine for all patients with newly diagnosed CRC to identify the Lynch syndrome as well as candidates for immunotherapy in the perioperative and metastatic settings.13,18 In metastatic CRC, comprehensive next-generation sequencing (NGS) is increasingly used as it identifies both MSI-H status and actionable genetic alterations. In this review, we summarise the evolving clinical landscape of dMMR/MSI-H CRC and evaluate the evidence and its limitations in guiding current treatment strategies, with an eye towards the future.

Methods

We performed a narrative review of the current medical literature pertaining to dMMR/MSI-H CRC, emphasising diagnostic approaches to MSI detection and systemic therapy strategies across the perioperative and metastatic settings. As this is a narrative rather than a systematic review, study selection was based on clinical relevance and scientific significance, and no predefined inclusion or exclusion criteria were employed.

Identification of dMMR/MSI-H CRC in the era of NGS

Identification of MMR/MSI status is essential in clinical practice for detection of patients with the Lynch syndrome as well as candidates for immunotherapy. Historically, two approaches are used for identification of MMR and MSI status: IHC for MMR protein expression and PCR for amplification of microsatellite markers.^{19 20} Current National Comprehensive Cancer Network (NCCN) guidelines recommend MMR/MSI assessment of all newly diagnosed cases of CRC.²¹ Sensitivity of MMR IHC has been reported to be between 85% and 100% with specificity ranging from 85% to 92%, while sensitivity and specificity of MSI by PCR may be as high as 98.7% and 95.8%, respectively, using the Pentaplex panel.²² An advantage of IHC is that it identifies the specific MMR protein(s) that is lost, which guides germline testing, or additional testing such as MLH1 promoter methylation or BRAF mutation assessment.

IHC for MMR proteins is widely used as the initial test for identifying MMR deficiency; however, indeterminate or discordant results should prompt reflex molecular MSI testing. While highly sensitive and specific, real-world data for IHC indicate modestly lower accuracy in community practice for reasons including tissue processing, use of different antibodies and interpretive challenges.²³ Low rates of discordance have been reported at many academic centres. A retrospective analysis of 3228 patients with CRC treated at academic medical centres in France found the rate of discordance between MMR IHC and MSI PCR to be 1.6%, with over half of discordant cases related to IHC misclassifications.²² Another study of 593 colon cancers available via the Northern Ireland Biobank found that 97.8% of tumours found to be dMMR by IHC were MSI-H by PCR.²⁴ A retrospective analysis of 809 CRCs from a US academic centre reported only a 0.4% discordance rate.²⁵

While current American Society of Clinical Oncology (ASCO) guidelines recommend IHC and/or PCR for detection of DNA MMR defects in patients with solid tumours being considered for immunotherapy, use of a validated MSI NGS panel is another option.¹⁹ In the era of NGS, which is routinely obtained in the setting of metastatic disease, IHC may have the most utility in non-metastatic disease as a screening assay.²⁰ In a study of 1417 patients with CRC whose pathological specimens were reviewed by surgical pathologists with subspecialty expertise at an academic centre in the USA, NGS was able to identify an additional 1% of patients who were MSI-H when compared with IHC.²⁶ Further, in a recent immunotherapy trial in dMMR metastatic CRC (CheckMate 8HW), 10% of patients receiving ipilimumab plus nivolumab (21/202 patients) were found to have microsatellite stable (MSS)/proficient MMR (pMMR) tumours on central laboratory confirmatory testing using both IHC and MSI PCR, conflicting with local MMR/MSI reporting, highlighting the need for improvement in our methods to identify dMMR/MSI-H tumours.¹⁶ An additional study among 520 patients with endometrial cancer or CRC, conducted at a large referral centre, reported a concordance rate of only 90% between MMR IHC and MSI by NGS.²⁷ These studies likely underscore the limitations of IHC, which may result in uninterpretable or misclassified results in a subset of patients. In addition to issues outlined above, some tumours with functional MMR deficiency (eg, via missense mutations) may retain protein expression on IHC. Despite such reports, an analysis of 28 105 CRCs found a concordance rate of 99.74% between IHC and MSI assessment via NGS.²⁸

A further benefit of NGS is the ability to identify POLE and POLD1 mutations which result in ultra-hypermutated tumours that show responsiveness to immunotherapy.29,33 POLE and POLD1 genes each perform an essential function in DNA proofreading and can be identified in 2.4–6.1% of CRCs.^{30 34} IHC and PCR are unable to identify POLE/POLD1 mutations, which underscores the importance of NGS to identify patients with these tumours who demonstrate high sensitivity to PD-1 blockade, with response rates often exceeding those observed in MSI-H cancers.³⁵

Performance of NGS for all stages of CRC has been proposed given its ability to identify additional candidates for immunotherapy, in addition to genes with therapeutic significance (KRAS, BRAF, others) in the metastatic setting. In patients with non-metastatic CRC, one must consider the significant cost of screening by NGS. That being said, the benefit of immunotherapy to a patient with a hypermutated tumour can be substantial and is in some cases curative.

Perioperative systemic therapy for dMMR/MSI-H CRC

Adjuvant chemotherapy with a fluoropyrimidine with or without oxaliplatin has been considered standard of care for high-risk stage II or stage III colon cancer.^{36 37} In the case of stage II dMMR/MSI-H colon cancer, adjuvant therapy is generally not recommended given the excellent prognosis in this population, with the exception of T4b tumours whose prognosis appears to be similar to patients with stage III cancers.³⁸ In patients with stage II and III dMMR/MSI-H colon cancers, adjuvant therapy with a fluoropyrimidine alone did not improve survival when compared with no adjuvant therapy.¹¹ In stage III dMMR/MSI-H colon cancer, a benefit of adjuvant oxaliplatin is suggested by data from a retrospective, observational cohort study³⁹ and a pooled analysis.⁴⁰ However, in patients with locally advanced colon cancer who received neoadjuvant fluoropyrimidine plus oxaliplatin in the FoxTROT trial, subgroup analysis showed that tumour regression was much less common in dMMR than in pMMR tumours.⁴¹ Furthermore, no reduction in recurrence was seen in patients with dMMR tumours. As such, recent studies incorporating ICIs into the perioperative setting have significantly altered the treatment paradigm of localised, dMMR/MSI-H CRC.

As discussed below, neoadjuvant immunotherapy has revolutionised the treatment of dMMR/MSI-H locally advanced rectal cancer. Data in a relatively modest number of patients have shown that all treated patients achieved complete clinical response to neoadjuvant anti-PD-1 therapy such that chemoradiation and surgery with their attendant morbidities were avoided. Despite these highly provocative data, further data are needed in a real-world setting to determine if response rates in routine clinical practice are as robust.

Neoadjuvant therapy for dMMR/MSI-H colon cancer

The NICHE-2 trial is a single-arm, phase II trial that enrolled patients with stage II or III, dMMR/MSI-H colon adenocarcinoma who were deemed surgically resectable.¹³ Enrolled patients received two doses of nivolumab (3 mg/kg) on days 1 and 15, and one dose of ipilimumab (1 mg/kg) on day 1, followed by surgical resection within 6 weeks of study enrolment. The coprimary endpoints were timely surgery and 3-year disease-free survival (DFS). Of 115 evaluable patients, 64% had a cT4 tumour and 67% were clinically node positive. The first primary endpoint of timely surgery was met with all patients completing recommended neoadjuvant therapy and proceeding to R0 resection, with only two patients experiencing a delay in surgery due to treatment-related toxicity, namely myositis. A pathological response was observed in 98% of patients, with pathological complete response (pCR) in 68%. There were no disease recurrences reported at a median follow-up of 26 months, and an updated analysis at a median follow-up of 36.5 months showed a 3-year DFS of 100%.⁴² Other smaller studies have suggested the benefit of neoadjuvant immunotherapy for locally advanced dMMR/MSI-H colon cancer. NICHE-3 evaluated the combination of nivolumab and relatlimab, an anti-LAG3 antibody, reporting a major pathological response in 92% and clinical complete response (cCR) in 68% of patients, with 98% of patients remaining disease free at a median follow-up of 9 months.⁴³ Additional phase II studies have shown efficacy of toripalimab (anti-PD-1) alone or combined with celecoxib, pembrolizumab monotherapy and camrelizumab (anti-PD-1) plus apatinib [vascular endothelial growth factor (VEGF)-R2 inhibitor] as neoadjuvant therapy in patients with dMMR/MSI-H CRC.44,49 In a systematic review of 628 patients with non-metastatic dMMR/MSI-H CRC who received neoadjuvant immunotherapy, a pCR rate of 66.6% was reported.⁵⁰ A summary of select trials evaluating neoadjuvant and adjuvant immunotherapy for dMMR/MSI-H CRC is shown in table 1.

Table 1. Select trials evaluating perioperative immunotherapy for dMMR/MSI-H colorectal cancer.

Line of therapy	Trial name	Colon cancer, n (%)	Rectal cancer, n (%)	Disease stage*, n (%)	Intervention/control	Clinical CR, n (%)	Pathological CR, n (%)	3-year DFS
Neoadjuvant (all phase II)	NICHE-213 42	115 (100)	0	I–II: 38 (33) III: 77 (67)	Ipilimumab+nivolumab (15d)	NR	75 (68)	100%
	PICC44 49	30 (83)†	6 (17)†	II: 4 (12) III: 32 (94)	Toripalimab+celecoxib versus toripalimab	NR	15/17 (88) vs 11/17 (65)	NR‡
	NEOCAP46	46 (88)	6 (12)	II: 2 (4) III: 50 (96)	Camrelizumab+apatinib followed by surgery if no clinical CR	28/52 (54)	14/23 (61)	NR§
	NICHE-343	59 (100)	0	II: 22 (37) III: 37 (63)	Nivolumab+relatlimab	NR	40 (68)	NR¶
	NEOPRISM-CRC47	32 (100)		NR (high-risk stage II or stage III)	Pembrolizumab**	NR	17 (53)	NR††
	RESET-C48	85 (100)	0	I–II: 34 (40) III: 51 (60)	Single-dose pembrolizumab	NR	37/84 (44)	NR
	Cercek et al60 61	0	50 (100)	I–II: 8 (16) III: 42 (84)	Dostarlimab	49/49 (100)	N/A	NR‡‡
Adjuvant (phase III)	ATOMIC14	712 (100)	0	III: 712 (100)	FOLFOX+atezolizumab versus FOLFOX	N/A	N/A	86.4% vs 76.6%, p<0.0001

^*Clinical staging was reported for all neoadjuvant trials, whereas pathological staging was reported for the ATOMIC trial.

^†While 34 patients were enrolled in this trial, two had synchronous rectal and ascending colon primaries; therefore, 36 tumours were analysed.

^‡At a median follow-up of 14.9 months, all patients were alive and without disease recurrence.

^§At a median follow-up of 11.2 months from time of complete response among 20 patients with clinical CR, none had evidence of disease recurrence.

^¶At a median follow-up of 8 months, one patient had disease recurrence.

^**31 of 32 patients had TMB-H disease and planned to receive three cycles of pembrolizumab, whereas one TMB-low patient planned to receive one cycle.

^††At a median follow-up of 6 months, no patients have had recurrent disease.

^‡‡96% recurrence-free survival at 2 years.

CR, complete response; DFS, disease-free survival; dMMR, deficient mismatch repair; MSI-H, microsatellite instability-high; N/A, not applicable; NR, not reported; TMB-H, tumor mutation burden- high.

Adjuvant therapy for dMMR/MSI-H colon cancer

ATOMIC is the first phase III trial that evaluated adjuvant chemoimmunotherapy for stage III dMMR/MSI-H colon cancer.¹⁴ This study enrolled 712 patients with stage III dMMR/MSI-H colon adenocarcinoma who underwent R0 resection and were randomised (1:1) to 6 months of mFOLFOX6 plus atezolizumab (840 mg every 2 weeks), followed by atezolizumab monotherapy for an additional 6 months versus mFOLFOX6 alone for 6 months, which was standard of care at time of study enrolment. The primary study endpoint was DFS, and secondary endpoints were overall survival (OS) and adverse event (AE) profile. The study arms were well balanced for clinicopathological features, and approximately 54% of patients had clinical high-risk stage III disease, that is, T4 and/or N2 tumours, in both arms. At a median follow-up of 40.9 months, the 3-year DFS rate was 86.4% in the chemoimmunotherapy arm versus 76.6% in the chemotherapy-only arm (HR 0.50, 95% CI 0.34 to 0.72, p<0.0001). Longer follow-up is needed for OS given relatively few OS events thus far, although standard use of ICIs for recurrent/metastatic disease may make OS difficult to interpret. The AE profile of the chemoimmunotherapy arm did not show a clinically meaningful increase in grade 3 or 4 (G3/4) events compared with the chemotherapy arm. The finding of a 50% reduction in recurrence or death in the atezolizumab plus mFOLFOX6 arm suggests a new standard of care for patients with stage III dMMR/MSI-H colon cancer. Results of the ATOMIC trial have been incorporated into the updated 2025 NCCN guidelines, which also extend these findings to T4bN0 stage II colon cancers.⁵¹

Neoadjuvant studies are not directly comparable to the ATOMIC trial, as ATOMIC is a randomised phase III study, whereas NICHE-2 is a single-arm study that enrolled a selected patient cohort, reported a lower event rate and had inherent challenges with accurate radiographic tumour staging. The NICHE-2 investigators expressed an opinion that patients with bulky disease can benefit from a neoadjuvant approach, as some of these patients may not be candidates for upfront R0 resection. In contrast, all patients in ATOMIC had upfront surgery which enabled pathological tumour staging. It is important to note that 33% of patients in NICHE-2 had stage II disease, whereas ATOMIC exclusively enrolled patients with stage III disease. Challenges exist for the preoperative staging of CRC, and patients with dMMR/MSI-H tumours are more likely to have an increased number and size of regional lymph nodes which can result in overstaging and overtreatment.^{52 53} In this regard, a prior analysis of CT staging in 14 455 patients with surgically resected colon cancer, including 3351 patients with dMMR/MSI-H tumours, found that patients with dMMR/MSI-H disease had overstaging of nodal status in 33% of cases.⁵⁴

The ATOMIC trial was designed before results of the IDEA (International Duration Evaluation of Adjuvant therapy) collaboration⁵⁵ were published such that all participants received 6 months of adjuvant FOLFOX chemotherapy. The study did not have an immunotherapy-only arm since chemotherapy was the standard of care for patients with stage III dMMR/MSI-H colon cancer at the time of trial initiation, and omission of chemotherapy was considered unethical. An ongoing phase III trial (AZUR-2) is randomising patients with T4N0 stage II or stage III dMMR/MSI-H resectable colon adenocarcinoma to perioperative dostarlimab or surgery, followed by standard of care chemotherapy (defined as FOLFOX or CAPEOX for 3–6 months or watch and wait at the treating provider’s discretion).⁵⁶

Biologically, there are benefits to having an intact immune system at the time of immunotherapy treatment. In a phase II trial (S1801) in patients with stage IIIB–IVC melanoma who were randomised to neoadjuvant plus adjuvant pembrolizumab versus adjuvant pembrolizumab alone, a significantly longer event-free survival (EFS) was observed in the group that had received neoadjuvant therapy.⁵⁷ Furthermore, the phase III NADINA trial randomised patients with stage III melanoma to neoadjuvant ipilimumab/nivolumab with response-driven adjuvant nivolumab and found that those who had received neoadjuvant therapy had a longer EFS.⁵⁸ These trials demonstrate the enhanced efficacy of immunotherapy when given with the primary tumour remaining in situ, likely due to abundant antigen and immune activation signals, enabling robust priming and expansion of effector and memory T cells (figure 1).⁵⁹ After resection, these antigens and immune stimulatory effects disappear, potentially leading to a weaker systemic immunity and limited immune surveillance against micrometastases.

Figure 1. Rationale for neoadjuvant approach for immunotherapy in dMMR/MSI-H colorectal cancer. dMMR, deficient mismatch repair; MSI-H, microsatellite instability-high. Image created using biorender.com.

Neoadjuvant immunotherapy for dMMR/MSI-H rectal cancer

In 2022, publication of a small, phase II study evaluating dostarlimab (anti-PD-1) monotherapy in patients with locally advanced dMMR/MSI-H rectal adenocarcinoma changed the treatment landscape for this disease.⁶⁰ Among 12 patients who received neoadjuvant dostarlimab for 6 months, all had a cCR, and a subsequent expansion of this single-institution trial to 49 total patients showed that all experienced cCR with a reported 2-year recurrence-free survival (RFS) of 96%.⁶¹ Furthermore, a retrospective study among 12 Italian institutions reported a cCR in 16 of 17 patients treated with ICI monotherapy.⁶² However, it will be important to have additional real-world experience for neoadjuvant immunotherapy in patients with dMMR/MSI-H locally advanced rectal cancer where response rates are likely to be lower. Numerous studies have noted limitations with pelvic MRI-based and endoscopic-based restaging following neoadjuvant therapy in rectal cancer.63,65 An interpretation of complete clinical response can enable a watch and wait approach, whereas evidence of likely or apparent residual tumour will typically result in a recommendation for chemoradiation and/or surgical resection. Also, sufficient treatment duration is likely important to achieve a cCR, and how long to treat the patient to achieve this goal is an unanswered question. In rectal cancer, organ preservation is a desired and achievable outcome given the significant morbidity that can be associated with chemoradiation and surgical resection.

In contrast to rectal cancer, organ preservation in patients with colon cancer is much more challenging with regard to determination of residual or recurrent disease using existing imaging modalities. The defined and relatively fixed anatomical location of the rectum enables its evaluation by pelvic MRI, which is not the case with primary colon cancers.

Immunotherapy for dMMR/MSI-H metastatic CRC

Current treatment landscape

The benefit of immunotherapy in dMMR/MSI-H CRC was first demonstrated in a phase I trial of nivolumab which included three patients with metastatic CRC of which one patient had a complete response ongoing at 3 years.⁶⁶ On further review, this patient’s tumour was dMMR/MSI-H, suggesting a link between MSI status and response to immune checkpoint inhibition. This was later followed by KEYNOTE-164, which was a phase II trial assessing pembrolizumab in patients with metastatic dMMR/MSI-H CRC with progression on prior chemotherapy and demonstrated an objective response rate of 33%, suggesting efficacy in this population.⁶⁷ KEYNOTE-177 was a phase III trial that enrolled 307 patients with previously untreated dMMR/MSI-H metastatic CRC and randomised them to receive pembrolizumab monotherapy or standard of care chemotherapy, with primary endpoints of progression-free survival (PFS) and OS.¹⁵ At final analysis, investigators reported a median PFS of 16.5 months with pembrolizumab versus 8.2 months with standard therapy (HR 0.59, 95% CI 0.45 to 0.79).⁶⁸ Median OS was not reached in the pembrolizumab arm and was 36.7 months in the chemotherapy arm. Although this difference did not meet a prespecified alpha of 0.025, crossover was allowed, with 60% of patients in the chemotherapy arm later receiving anti-PD1 or anti-PD-L1 therapy. Furthermore, 22% of patients in the pembrolizumab arm experienced a grade 3 or greater toxicity compared with 66% in the chemotherapy arm. This study established ICI therapy as standard of care for metastatic dMMR/MSI-H CRC in the front-line setting.

The phase III CheckMate 8HW trial enrolled patients with metastatic dMMR/MSI-H CRC and randomised them to receive nivolumab plus ipilimumab, nivolumab monotherapy or chemotherapy±targeted therapies, with coprimary endpoints of PFS of ipilimumab plus nivolumab compared with chemotherapy in the front-line setting and PFS of ipilimumab plus nivolumab compared with nivolumab regardless of prior anticancer therapy.¹⁸ Patients in the ipilimumab plus nivolumab arm received nivolumab 240 mg and ipilimumab 1 mg/kg every 3 weeks for 12 weeks, followed by nivolumab 480 mg every 4 weeks. Those in the nivolumab monotherapy arm received nivolumab 240 mg every 2 weeks for 12 weeks, at which point they transitioned to 480 mg every 4 weeks. In a prespecified interim analysis of 303 patients randomised to ipilimumab plus nivolumab versus chemotherapy, a 24-month PFS rate of 72% was observed in the ipilimumab plus nivolumab arm versus 14% in the standard chemotherapy arm (p<0.001). Furthermore, G3/4 AEs were reported in 23% of the ipilimumab plus nivolumab arm versus 48% in the standard chemotherapy arm. In a subsequent interim analysis comparing ipilimumab plus nivolumab to nivolumab monotherapy, regardless of prior anticancer therapy, including 707 patients, median PFS (mPFS) was not reached in the combination arm compared with 39.3 months with nivolumab (HR 0.62, 95% CI 0.48 to 0.81, p=0.0003).¹⁶ However, ipilimumab plus nivolumab was associated with greater toxicity, with G3/4 events reported in 22% compared with 14% in the nivolumab monotherapy arm. An update presented at the European Society of Medical Oncology (ESMO) 2025 annual meeting reported outcomes of 341 patients with centrally confirmed dMMR/MSI-H status who received ipilimumab plus nivolumab versus nivolumab in the front-line setting and found improved mPFS with ipilimumab plus nivolumab over nivolumab monotherapy (not reached vs 60.8 months; HR 0.69, 95% CI 0.48 to 0.99, p=0.0413).⁶⁹ Based on the results of this trial, ipilimumab plus nivolumab has become a new standard regimen in the front-line treatment of dMMR/MSI-H metastatic CRC.

Despite the benefit of combination immune checkpoint inhibition over single-agent immunotherapy shown in CheckMate 8HW, single-agent immunotherapy may be more appropriate for some patients. In CheckMate 8HW, combination immunotherapy resulted in G3/4 treatment-related AEs in 22%, treatment-related serious AEs in 16% and death in two patients (one due to myocarditis and another due to pneumonitis).¹⁶ G3/4 immune-related AEs included diarrhoea/colitis (5%), adrenal insufficiency (4%), hepatitis, hypophysitis (3% each), pneumonitis, rash (2% each), hypothyroidism (1%), thyroiditis and nephritis (<1% each). Those receiving single-agent nivolumab experienced G3/4 AEs at a lower incidence of 14%, and one treatment-related death was reported (pneumonitis). In KEYNOTE-177, 22% of single-agent pembrolizumab recipients experienced a G3/4 treatment-related AE, although no cases of pneumonitis, myositis or myocarditis were reported. As such, providers may consider single-agent pembrolizumab or nivolumab in place of ipilimumab plus nivolumab in elderly, frail and poor-performance status patients who may not be able to tolerate a significant immune-related AE. A summary of select phase III trials evaluating immunotherapy in the front-line setting for metastatic dMMR/MSI-H CRC is shown in table 2, including the ongoing COMMIT trial that evaluates the role of chemoimmunotherapy in this population.⁷⁰ Aggregate data demonstrate a plateau in long-term survival which suggests that some patients with metastatic dMMR CRC can achieve long-term disease eradication or cure, whereas these tumours were previously found to be incurable with chemotherapy alone.

Table 2. Select phase III trials evaluating immunotherapy for metastatic dMMR/MSI-H colorectal cancer.

Trial name	Patient (n)	Study arms	ORR (%)	mPFS (months)	mPFS P value	mOS (months)	mOS P value	G3/4 immune-related adverse events, n (%)
KEYNOTE-17715 68	307	Pembrolizumab	43.8	16.5	0.0002	Not reached	0.036*	14 (9)
		Chemotherapy	33.1	8.2		36.7		3 (2)
CheckMate 8HW† 16 18 69	303	Ipilimumab+nivolumab	NR	54.1‡	<0.001	NR	NR	37 (18.5)
		Chemotherapy		5.9‡				1 (1)
	707 (any line of therapy)	Ipilimumab+nivolumab	71	Not reached	0.0003	NR	NR	55 (15.6)§
		Nivolumab	58	39.3				27 (7.7)§
	341 (first-line therapy)	Ipilimumab+nivolumab	73	Not reached	0.0413¶	Not reached	NR	24%
		Nivolumab	61	61.8		Not reached		17%
COMMIT70	120 (goal accrual)	Atezolizumab	Ongoing
		mFOLFOX6/bevacizumab+atezolizumab

^*Overall survival did not achieve prespecified alpha of 0.025; therefore, this difference was not significant.

^†Response rate and progression-free survival are reported among patients who were deemed to have dMMR/MSI-H tumours on central laboratory confirmation.

^‡mPFS as reported in an updated analysis.

^§Two patients in the ipilimumab+nivolumab arm and one patient in the nivolumab arm experienced a grade 5 immune-related adverse event.

^¶Boundary for statistical significance was p<0.0383.

dMMR, deficient mismatch repair ; G3/4, grade 3/4; mOS, median overall survival; mPFS, median progression-free survival; MSI-H, microsatellite instability-high; NR, not reported; ORR, objective response rate.

Limitations to current approach

Primary resistance to immunotherapy is a well-recognised phenomenon across solid malignancies.^{71 72} In stage III patients enrolled in the ATOMIC trial, 14% of those receiving atezolizumab and mFOLFOX6 experienced disease recurrence or death.¹⁴ In patients with metastatic CRC enrolled in KEYNOTE 177, 29.4% of patients receiving pembrolizumab had progression of disease as their best response, which was higher than among patients receiving chemotherapy in the control arm.¹⁵ In CheckMate 8HW, among those who received ipilimumab plus nivolumab or nivolumab monotherapy in any line of therapy, 10% and 19% had progressive disease as their best response.¹⁶ There is an unmet need to identify additional predictive biomarkers beyond MMR or MSI-H status for immunotherapy outcomes. An analysis of tumours from KEYNOTE-177 found that response to pembrolizumab was associated with high clonality of immunogenic mutations, clonally expanded T cells, low activation of Wnt signalling, deregulation of the interferon gamma pathway and active immune escape mechanisms. Responsive hypermutated CRCs were also rich in cytotoxic and proliferating PD1⁺CD8 T cells interacting with PDL1⁺ antigen-presenting macrophages.⁷³ Additional studies have assessed for predictors of poor response/resistance to immunotherapy, with mechanisms including a tumor growth factor-β (TGFβ)-rich tumour microenvironment,⁷⁴ alterations in the Wnt and JAK-STAT signalling pathways and mutations in MHC or beta-2 microglobulin impacting antigen presentation.⁷⁵ Novel drug combinations are needed to overcome these resistance mechanisms to improve the outcomes of patients with dMMR/MSI-H mCRC.⁷⁶

Studies in patients with dMMR/MSI-H CRCs have shown that those with liver metastases may respond poorly to immunotherapy with poorer PFS when compared with other sites of metastatic disease.77,79 The poor response to immunotherapy in liver metastases has been seen in other tumour types treated with immunotherapy, including melanoma and non-small cell lung cancer.⁸⁰ Preclinical studies in murine models have shown that liver metastases act as a ‘sink’ for activated CD8 T cells from circulation, where they get trapped in the liver and undergo apoptosis, resulting in immunotherapy resistance through T cell depletion.⁸⁰ This phenomenon is not limited to dMMR/MSI-H tumours, as a phase I study in MSS metastatic CRC found that patients with liver metastases had worse objective response rates and poorer PFS compared with those without liver metastases.⁸¹

There is clinical uncertainty with regard to the optimal duration of immunotherapy treatment. In CheckMate 8HW and KEYNOTE-177, immunotherapy was continued for 2 years, assuming patients did not experience progression or toxicity leading to discontinuation.^{15 18} In a retrospective analysis of patients with dMMR/MSI-H metastatic CRC who received an ICI, investigators compared those who discontinued treatment prior to 395 days for an indication other than progressive disease versus those who continued treatment beyond 395 days and found no significant difference in PFS and OS.⁸²

The role of artificial intelligence in the management of dMMR/MSI-H CRC

A number of studies have demonstrated the ability of deep learning (DL) to predict MMR/MSI status in digitised whole slide images from CRCs, with a high degree of accuracy.83,85 This has the potential to expedite and enhance the detection of dMMR/MSI-H tumours. Furthermore, recent studies have used DL to extract clinically relevant prognostic and predictive information from digitised H&E-stained whole slide images of human CRCs and other malignancies. A digital pathology segmentation algorithm was developed to identify and quantify tumour morphological features and was used to build prognostic models in CRC, thus enhancing interpretive accuracy.⁸⁶ Data have been generated in large, independent patient populations whereby an unsupervised algorithm was shown to correlate with and outperform established morphological and molecular prognostic features.⁸⁷ Consistent results have been obtained across tumour and nodal stage, and results have the potential to guide the selection of adjuvant treatment to avoid overtreatment in low-risk patients and to identify those who may benefit from more intensive treatment. In other malignancies, such as melanoma, machine learning (ML) models have been developed to predict response to immunotherapy based on digitised pathology, with potential application to other tumours such as CRC.⁸⁸ Furthermore, ML has been used to assess both clinical and transcriptomic data to predict immunotherapy response in CRC.⁸⁹ These examples highlight the evolving role for DL/ML to further advance prognosis and prediction in CRC, including patients with dMMR/MSI-H tumours, with the potential to guide treatment decisions.⁹⁰

Conclusions

Outcomes for patients with dMMR/MSI-H colon and rectal cancers continue to improve with the use of immunotherapy across the neoadjuvant, adjuvant and metastatic settings. In non-metastatic disease, the optimal treatment strategy remains uncertain. Short-duration neoadjuvant immunotherapy can produce high rates of pCR, making it an appealing option.¹³ It is important to note that these data are derived from single-arm studies with selected patient populations, and there is difficulty in precise radiological tumour staging preoperatively. This is made more challenging by the fact that dMMR/MSI-H tumours tend to have more numerous and larger peritumoral nodes compared with patients with pMMR/MSS tumours, which can lead to overtreatment. Adjuvant chemoimmunotherapy is another effective strategy, and the phase III ATOMIC trial showed that the addition of the anti-PD-L1 antibody, atezolizumab, to mFOLFOX6 significantly reduced the risk of recurrence or death compared with mFOLFOX6 in stage III dMMR/MSI-H colon cancer, indicating a new standard of care.¹⁴ Nevertheless, important questions remain, including the contribution of cytotoxic chemotherapy to immunotherapy and its duration. Ongoing results from the phase III AZUR-2 trial—which compares neoadjuvant dostarlimab followed by surgery plus adjuvant dostarlimab versus upfront surgery followed by fluoropyrimidine-oxaliplatin chemotherapy—may help clarify an optimal approach.⁵⁶

In the metastatic setting, the CheckMate 8HW trial demonstrated superior efficacy of nivolumab plus ipilimumab compared with nivolumab monotherapy, with only a modest increase in toxicity.^{16 18 69} Consequently, dual checkpoint blockade and anti-PD-1 monotherapy represent standard options, with combination therapy being a preferred strategy for patients with favourable performance status. Despite high response rates and durable treatment benefit, it is important to note that many patients will display primary resistance or develop acquired resistance to immunotherapy, which underscores the need for continued investigation into the mechanisms of resistance and strategies to overcome them.

Data availability statement

Data sharing not applicable as no datasets generated and/or analysed for this study.