Key points

Introduction

Colorectal cancers (CRCs) may be divided via molecular phenotyping into tumors with normal DNA mismatch repair (MMR) function and those with DNA MMR deficiency (MMR-D), representing ∼15% of CRCs. The hallmark of MMR-D CRC is a distinct type of genomic instability, referred to as high-frequency microsatellite instability (MSI-H). MMR-D/MSI-H is associated with specific clinical and pathologic features, and in some cases, may provide the initial clinical indication suggesting the presence of Lynch syndrome, an inherited form of CRC, characterized by germline genetic defects in one of the DNA MMR genes ( MLH1, MSH2, MSH6, PMS2, EPCAM ). Lynch syndrome remains significantly underdiagnosed, and recent measures to expand screening of patients with Lynch-associated cancers, including those with CRC and endometrial cancer, provides one of the most effective ways to identify at-risk families. The MMR-D/MSI-H phenotype also serves as an important prognostic and predictive marker in CRC. Although MMR-D/MSI-H CRCs tend to be early stage at diagnosis and are associated with a good prognosis, the value of 5-fluorouracil (5-FU) as an effective chemotherapeutic treatment of this type of CRC has been debated. As our understanding of MMR-D/MSI-H CRCs continues to evolve, oncologists remain at the forefront of diagnosing, treating, and managing patients with this unique subtype of CRC and ensuring that, for appropriate patients, referral to clinical cancer genetics for Lynch syndrome evaluation is implemented.

The adenoma-carcinoma model of colorectal cancer

The development of CRCs via the adenoma-carcinoma sequence, initially described by Fearon and Vogelstein in 1990, proposed a multistep genetic model of CRC carcinogenesis. Most CRCs arise along the chromosomal instability pathway, characterized by widespread imbalances in chromosome number (aneuploidy) and loss of heterozygosity, as well as the accumulation of characteristic mutations in tumor suppressor genes and oncogenes critical for CRC initiation and progression. The second pathway, accounting for 15% to 20% of CRCs, results from defective DNA MMR, which leads to the molecular phenomenon of MSI-H shown within the tumor. Microsatellites are short, tandemly repeated DNA sequences that are distributed throughout the human genome and consist of mononucleotide, dinucleotide, or higher-order DNA base repeats. Such repetitive sequences are prone to the accumulation of mutations mostly because of base-base mismatches and insertion-deletion loops. The correction of errors in microsatellites is performed by the MMR proteins, the most notable of which are MLH1, MSH2, MSH6, and PMS2. In tumors with MMR-D, either because of germline, somatic, or epigenetic inactivation, the correction of such mutations is impaired, leading to the accumulation of DNA errors, resulting in the MSI-H phenotype.

In most cases, MMR-D is a result of an epigenetic phenomenon in the tumor, most commonly hypermethylation of the MLH1 promoter, which leads to the development of a sporadic (noninherited) CRC. However, in about 30% of cases, MMR-D is secondary to a germline mutation in one of the MMR genes, leading to a diagnosis of Lynch syndrome, an inherited form of CRC. Deciphering the cause of the MMR-D is of upmost importance, because it helps to differentiate sporadic from inherited forms of colon cancer.

Screening for defective DNA mismatch repair/high-frequency microsatellite instability in patients with colorectal cancer: should a universal approach be applied?

The identification of the MMR-D/MSI-H phenotype in a colorectal tumor may be the first indication that a patient may have a diagnosis of Lynch syndrome. Patients identified to have Lynch syndrome benefit from life-saving cancer surveillance measures. In addition, their at-risk family members would benefit from predictive genetic testing and, if needed, implementation of high-risk cancer surveillance. Although measures aimed at increasing awareness about hereditary CRC, both among individuals as well as among physicians, may be undertaken, one of the most effective ways to identify potential at-risk Lynch families is by evaluating patients who are diagnosed with either CRC or endometrial cancer, the 2 most common Lynch-associated malignancies.

Originally, the Amsterdam criteria and subsequently, the Revised Bethesda guidelines, have served as criteria to select patients with CRC or endometrial cancer in whom further evaluation for MMR-D/MSI-H was indicated. However, in the clinical setting these criteria have proved to be cumbersome to implement and lack specificity and sensitivity. Some groups have recommended expansion of the criteria for MMR/MSI testing to either universal testing, with testing of all patients with CRC, or to include all individuals with CRC diagnosed younger than 70 years. Using universal testing, Lynch syndrome was diagnosed in 2.4% to 3.7% of all unselected patients with CRC. Implementing universal screening for endometrial cancer yielded similar results, with about 1.8% to 3.9% of patients being diagnosed with Lynch syndrome. In the CRC cases, a Lynch syndrome diagnosis would have been missed in ∼12% to 28% of patients if the Revised Bethesda guidelines were used for selection criteria. More expansive testing of CRC and endometrial cases has also proved to be cost effective.

Based on these data, the Mallorca group, consisting of European experts in Lynch syndrome, the Evaluation of Genomic Applications in Practice and Prevention Working Group, and the National Comprehensive Cancer Network have recently updated their recommendations to suggest that either all patients with CRC or all patients with CRC younger than 70 years at diagnosis, and those 70 years or older who meet Bethesda guidelines, should undergo testing by either immunohistochemistry (IHC) or MSI for Lynch syndrome. Implementation of such reflex testing is already occurring across large medical centers, with a recent study suggesting that the specific implementation procedures influenced patient follow-through with genetic testing after an initial positive screening test. For example, institutions with a high level of involvement of genetic counselors in the tracking and communicating of screening results improved patient follow-through and reduced barriers to patient contact.

Identification of defective DNA mismatch repair/high-frequency microsatellite instability tumors

Two accepted methods for the detection of MMR-D are available. MSI testing relies on polymerase chain reaction for amplification of specific microsatellites repeats. The original accepted panel of 5 microsatellite markers, referred to as the Bethesda panel, includes 2 mononucleotides (BAT-25, BAT-26) and 3 dinucleotides (D5S346, D2S123, D17S250). The presence of instability is determined based on comparison of the length of a specific microsatellite marker in the tumor versus the normal DNA, such as adjacent colonic mucosa or a blood sample. Instability in 2 or more markers is defined as an MSI-high tumor, whereas those with 1 unstable marker are designated MSI low. For a sample to be microsatellite stable (MSS), no instability in any of the markers should be present. Given concerns over limited sensitivity of dinucleotide repeats, the 2002 National Cancer Institute workshop made further revisions, with recommendations to include a secondary panel of mononucleotide markers, such as BAT-40, to exclude MSI-low cases in which only a dinucleotide repeat is mutated.

In 1996, monoclonal antibodies against MMR proteins became available, rendering IHC detection of MMR proteins possible. A lack of expression of 1 or more of these proteins is diagnostic of MMR-D, with the specific pattern of expression also helping to pinpoint which gene is most likely to harbor a mutation or may be inactivated by another mechanism ( Table 1 ). Initial concerns over lower sensitivity of IHC as opposed to MSI screening have largely been overcome with the introduction of the 4-antibody panel (MLH1, MSH2, MSH6, PMS2). In the clinical setting, both IHC and MSI testing are used broadly and are essentially interchangeable techniques with greater than 90% concordance and similar sensitivity of around 85% to 92%. At our institution, the preference to use IHC as opposed to MSI testing for initial screening was largely based on the ability of IHC to pinpoint which genes should be targeted for subsequent germline analysis in MMR-D cases, thereby simplifying the genetic testing process.

Defective DNA mismatch repair/high-frequency microsatellite instability colorectal cancer: making the diagnosis of Lynch syndrome

An MSI-H or an abnormal IHC result does not distinguish between a somatic (sporadic) versus a germline (inherited) defect in the MMR system. Without further analysis, the cause of the defect remains elusive. In approximately two-thirds of cases, MSI-H indicates a sporadic CRC, with the molecular phenomenon being caused by an epigenetic event, most commonly MLH1 promoter hypermethylation, leading to MLH1 gene inactivation. In about 70% of such promoter hypermethylated MSI-H cases, the presence of a BRAF V600E mutation can be identified. The presence of a BRAF V600E mutation in a colorectal tumor would make the diagnosis of Lynch syndrome less likely, and together with an absence of suggestive family history, may reassure clinicians of the sporadic cause of the MSI-H CRC. Cases of BRAF mutated tumors in Lynch syndrome–associated CRCs have been reported, and therefore a careful review of the clinical and family history should be undertaken to determine whether further genetic counseling/testing is indicated. In many centers, an MSI-H CRC or a CRC showing loss of MLH1/PMS2 protein expression is immediately tested via direct MLH1 hypermethylation analysis or BRAF V600E mutation analysis. These are generally not considered genetic tests but are used for risk stratification to determine which patients may benefit from subsequent genetic testing.

In one-third of MSI-H tumors, the molecular defect is secondary to a germline mutation in one of the MMR genes, diagnostic of Lynch syndrome. In patients with CRC in whom Lynch syndrome is suspected based on early age at diagnosis or family history of Lynch-associated malignancies, an abnormal IHC or MSI-H result should prompt referral for genetic counseling and genetic testing. If IHC was performed, then the specific protein expression loss helps guide genetic testing (see Table 1 ). For example, in a tumor with absence of MSH2 and MSH6 expression, genetic testing for mutations in the MSH2 gene should be undertaken initially. Identification of a genetic mutation confirms the diagnosis of Lynch syndrome. In such patients, appropriate cancer surveillance and options for cancer risk reduction need to be reviewed and implemented. Moreover, expansion of genetic counseling and testing for family members should be undertaken.

A difficult clinical situation that arises not infrequently is that workup of a patient with an MSI-H or MMR-D CRC results in ambiguous or uninformative results with respect to the origin of the MMR defect. For example, in a young patient with CRC with absence of MLH1/PMS2 protein expression, genetic testing of the MLH1 and PMS2 genes may be unrevealing, with no mutation identified. Coupled with the absence of MLH1 promoter hypermethylation and absence of a BRAF mutation, in such a clinical circumstance, Lynch syndrome cannot be ruled out. In such situations, a careful evaluation of the family history may help to identify other cancer-affected family members whose tumors could be evaluated for the MMR-D/MSI-H molecular signature and, if positive, would confirm the presence of an occult mutation. These cases require the careful input of genetic counselors and cancer geneticists for recommendations for appropriate cancer surveillance for both the patient and at-risk family members.

Clinical and pathologic features of defective DNA mismatch repair/high-frequency microsatellite instability colorectal cancers

Regardless of the cause of the MMR deficiency, MMR-D/MSI-H CRCs seem to be associated with distinct clinical and pathologic features, which often serve as the initial clue that a particular tumor may harbor an MMR defect ( Table 2 ). As opposed to the 25% of MSS colorectal tumors, 85% of MMR-D/MSI-H tumors are proximal to the splenic flexure. Moreover, the age distribution of MMR-D/MSI-H tumors seems to be bimodal, with about 24% of patients with CRC younger than 40 years and 19% older than 70 years having MSI-H tumors, as opposed to only 8% of patients with CRC aged between 50 and 59 years. This U-shaped age distribution reflects the higher prevalence of Lynch-associated MMR-D/MSI-H CRCs in young patients and the similarly higher prevalence of sporadic MMR-D/MSI-H phenotype observed in older patients.

Despite the improved prognosis (see later discussion) associated with the MMR-D/MSI-H phenotype, such tumors tend to show a higher histologic grade and are often mucinous tumors with presence of signet-ring cells. A recently recognized feature is the presence of the distinct medullary subtype of colorectal adenocarcinoma, associated with a unique histologic appearance and a more favorable prognosis compared with standard poorly differentiated colonic carcinomas. Medullary carcinomas are generally associated with a right-sided predominance, older female patients, and a lower incidence of lymph node metastases. Morphologic features of such tumors are characterized by a syncytial growth pattern and large vesicular nuclei, with conspicuous nucleoli and prominent peritumoral lymphocytic infiltrates. Because there is significant morphologic overlap between medullary carcinomas and poorly differentiated carcinomas, these may be difficult to differentiate. However, the presence of medullary carcinoma subtype, even if categorized under a poorly differentiated carcinoma, is associated with a favorable prognosis. In 1 study, MLH1 antibody staining was absent in nearly 80% of medullary carcinomas, indicating the close association of MMR-D/MSI-H with this subtype of CRC. Other common features include prominent tumor-infiltrating lymphocytes, lack of necrotic debris, and a Crohn-like lymphocytic host reaction. These distinct histologic features evoke the possible presence of MMR-D/MSI-H, prompting the pathologist to perform subsequent MSI or IHC testing.

Defective DNA mismatch repair/high-frequency microsatellite instability as a prognostic marker in colorectal cancer

In addition to the distinct clinicopathologic features, numerous studies have also suggested a favorable prognosis of MMR-D/MSI-H colon tumors. Patients with MMR-D/MSI-H tumors tend to have a lower tumor stage at diagnosis and rarely develop metastatic disease. When assessed stage by stage, the presence of MMR-D/MSI-H is noted in ∼20% of stage I/II, ∼12% of stage III, but only ∼4% of stage IV patients, again suggesting an association with earlier stage of disease. Substantial evidence has accumulated from retrospective studies, population-based studies, as well as a meta-analysis to suggest a more favorable prognosis in MMR-D/MSI-H colorectal tumors.

In 2000, Gryfe and colleagues initially reported an association of MSI-H status, with a survival advantage that was independent of all standard prognostic features. This analysis was limited to CRCs diagnosed at age 50 years or younger, limiting the generalizability of this result to all patients with CRC. In the seminal 2003 publication by Ribic and colleagues, of patients with stage II and III colon cancer who did not receive adjuvant chemotherapy, those with MMR-D/MSI-H had significantly improved overall survival compared with MMR-P/MSS tumors with a hazard ratio (HR) for death of 0.31 (95% confidence interval [CI], 0.14–0.72, P = .004).The improved prognosis associated with MMR-D/MSI-H was despite the presence of a higher percentage of poorly differentiated tumors in the MMR-D/MSI-H group (26% vs 9%). A subsequent meta-analyses of 32 studies, including a total of 7642 patients with CRC, including all stages of disease, also found a favorable overall survival in the 1277 MSI-H tumors with an HR of 0.65 (95% CI, 0.59–0.71). When the analysis was restricted to clinical trials only, the benefit was maintained with an HR of 0.69. In this meta-analysis, both treated (5-FU–based adjuvant treatment) and untreated patients were included from phase 3 randomized trials. In the Quick And Simple And Reliable (QUASAR) trial assessing stage II CRCs, MMR-D was an independent prognostic factor, with an improvement in survival with an HR of 0.31. Nonetheless, several studies do not corroborate the survival advantage associated with MMR-D/MSI-H tumors. Such discrepant results may be caused by limited power to detect prognostic differences because of small sample size, different methodologies used for the assessment of defective MMR (ie, 4 vs 2 antibody IHC testing; MSI analysis with different panel and numbers of markers), different treatment and stage of disease included in the analysis, and selection bias. Because patients with MMR-D/MSI-H caused by Lynch syndrome comprise an even smaller group (∼2–3% of all CRCs), power to detect prognostic differences specifically for patients with Lynch syndrome is even more limited. Nonetheless, based on the preponderance of evidence, it is generally well accepted that MMR-D/MSI-H tumors are associated with an improved prognosis, with the more contentious issue being the importance of MMR-D/MSI-H status as a predictive marker in the treatment of CRC.

Defective DNA mismatch repair/high-frequency microsatellite instability status as a predictive marker in colorectal cancer

Summary

The identification of the MMR-D/MSI-H subtype of CRC has important clinical implications for the prognosis of CRC, prediction of response to chemotherapeutic agents, specifically efficacy of 5-FU, and for the need for genetic assessment for Lynch syndrome. Recent expert consensus recommendations suggest that either all patients with CRC or those with CRC at younger than 70 years, should undergo testing by either IHC or MSI, thereby leading to more patients with CRC being recognized as having this subtype of tumor. There is a pressing need for unambiguous answers with respect to the prognostic and predictive impact of MMR-D/MSI-H. Provocative, but preliminary, data suggesting a stage-specific impact and possible inherent biological differences between sporadic versus germline MMR-D/MSI-H tumors remain important questions awaiting validation. With advances in gene expression profiling, next-generation sequencing technologies, and microRNA assessments, further biologically relevant classifications of CRC are anticipated, with MMR-D/MSI-H being just one of many relevant subgroups. Drug development and clinical trial design centered on specific molecular subtypes of tumors is a rapidly emerging approach to novel oncological treatments, and the MMR-D/MSI-H subtype may prove to be a useful marker for such targeted approaches.