Genetics of Colorectal Cancer (PDQ®) 4/6 —Health Professional Version - National Cancer Institute

Genetics of Colorectal Cancer (PDQ®)—Health Professional Version - National Cancer Institute

Genetics of Colorectal Cancer (PDQ®)–Health Professional Version

Lynch Syndrome

Introduction

Lynch syndrome is the most common inherited CRC syndrome and accounts for approximately 3% of all newly diagnosed cases of CRC. It is an autosomal dominant condition caused by pathogenic variants in the MMR genes MLH1 (mutL homolog 1), MSH2(mutS homolog 2), MSH6 (mutS homolog 6), and PMS2 (postmeiotic segregation 2), as well as the gene EPCAM (epithelial cellular adhesion molecule, formerly known as TACSTD1), in which deletions in EPCAM cause epigenetic silencing of MSH2. Lynch syndrome is also associated with a predisposition for developing several extracolonic manifestations, including sebaceous adenomas and cancers of the endometrium and ovaries, stomach, small intestine, transitional cell carcinoma of the ureters and renal pelvis, hepatobiliary system, pancreas, and brain. Lynch syndrome–associated cancers exhibit MSI; therefore, tumor testing is a key component in the diagnosis of Lynch syndrome, in addition to family history. Universal tumor testing of all CRCs is now recommended as a strategy to screen for Lynch syndrome and identify those individuals who may subsequently benefit from germline genetic testing. Intensive cancer screening and surveillance strategies, including frequent colonoscopy, along with risk-reducing surgeries, are mainstays in patients with Lynch syndrome.

History of Lynch syndrome

Between 1913 and 1993, numerous case reports of families with apparent increases in CRC were reported. As series of such reports accumulated, certain characteristic clinical features emerged: early age at onset of CRC; high risk of synchronous (and metachronous) colorectal tumors; preferential involvement of the right colon; improved clinical outcome; and a range of associated extracolonic sites including the endometrium, ovaries, other sites in the GI tract, uroepithelium, brain, and skin (sebaceous tumors). Terms such as cancer family syndrome, and hereditary nonpolyposis colorectal cancer (HNPCC) were used to describe this entity.[246]

The term Lynch syndrome replaced HNPCC and is applied to cases in which the genetic basis can be confidently linked to a germline pathogenic variant in a DNA MMR gene. Moreover, HNPCC is misleading as many patients have polyps and many have tumors other than CRC.

With the increased recognition of families that were considered to have a genetic predisposition to the development of CRC, research for a causative etiology led to the development of the Amsterdam criteria in 1990.[247] The Amsterdam criteria were originally used for the identification of high-risk families and included fulfillment of all of the following: three or more cases of CRC over two or more generations, with at least one diagnosed before age 50 years, and no evidence of FAP.

In 1987, a chromosomal deletion of a small segment of 5q led to the detection of a genetic linkage between FAP and this genomic region,[248] from which the APC gene was eventually cloned in 1991.[249] This led to searches for similar linkage in families suspected of having Lynch syndrome who had multiple cases of CRC inherited in an autosomal dominant fashion and young onset of cancer development. The APC gene was one of several genes (along with DCC and MCC) evaluated in families that fulfilled Amsterdam criteria, but no linkage was found among the Lynch kindreds. In 1993, an extended genome-wide search resulted in the recognition of a candidate chromosome 2 susceptibility locus in large families. Once MSH2, the first Lynch syndrome–associated gene, was sequenced, it was evident from the somatic variant patterns in the CRC tumors that the MMR family of genes was likely involved. Additional MMR genes were subsequently linked to Lynch syndrome, including MLH1, MSH6, and PMS2. Lynch syndrome now refers to the genetic disorder caused by a germline variant in one of these DNA MMR genes, distinguishing it from other familial clusters of CRC.

In 2009, a germline deletion in the EPCAM gene was identified as another cause of MSH2inactivation in the absence of a germline pathogenic variant in MSH2. The variant in EPCAMled to hypermethylation of the MSH2 promoter. Thus, EPCAM, which is not a DNA MMR gene, is also implicated in Lynch syndrome and is now routinely tested in at-risk patients along with the DNA MMR genes listed above.

Defining Lynch syndrome families

Families with a preponderance of CRC and a possible genetic predisposition were initially categorized as having Lynch syndrome based on family history criteria, as well as personal history of young-onset CRC. With the advent of molecular tumor diagnostic testing and the discovery of the germline alterations associated with Lynch syndrome, the clinical criteria have currently fallen out of favor due to their underperformance. However, their use, or the risk estimates provided by the Lynch syndrome prediction models, may be applicable among individuals without personal history of cancer but with a family history suggestive of Lynch syndrome, or for those individuals with CRC but without available tumor for molecular diagnostic testing. (Refer to the Universal tumor testing to screen for Lynch syndrome and the Clinical risk assessment models that predict the likelihood of an MMR gene pathogenic variant sections of this summary for more information.)

The first criteria for defining Lynch syndrome families were established by the International Collaborative Group meeting in Amsterdam in 1990 and are known as the Amsterdam criteria.[247] These research criteria were limited to diagnoses of familial CRC. In 1999, the Amsterdam criteria were revised to include some extracolonic cancers, predominantly endometrial cancer.[250] These criteria provide a general approach to identifying Lynch syndrome families, but they are not considered comprehensive; nearly half of families meeting the Amsterdam criteria do not have detectable pathogenic variants.[251]

Amsterdam criteria I (1990):

1. One family member diagnosed with CRC before age 50 years.
2. Two affected generations.
3. Three affected relatives, one of them an FDR of the other two.
4. FAP should be excluded.
5. Tumors should be verified by pathological examination.

Amsterdam criteria II (1999):

• Same as Amsterdam criteria I, but tumors of the endometrium, small bowel, ureter, or renal pelvis can be used to substitute an otherwise qualifying CRC.

These criteria were subsequently used beyond research purposes to identify potential candidates for microsatellite and germline testing. However, the Amsterdam criteria failed to identify a substantial proportion of Lynch syndrome kindreds; families that fulfilled Amsterdam criteria I but did not have evidence of MSI and were without a pathogenic germline variant in a DNA MMR gene, were referred to as familial colorectal cancer type X (FCCX). (Refer to the FCCX section of this summary for more information.)

With the hallmark feature of MSI associated with Lynch syndrome tumors, and the limitations of the Amsterdam criteria related to low sensitivity, the Bethesda guidelines were introduced in 1997. The Bethesda guidelines are a combination of clinical, histopathologic, and family cancer history features that identify cases of CRC that warrant MSI tumor screening. The Bethesda guidelines (with a subsequent revision in 2004) were formulated to target patients in whom evaluation of CRC tumors for MMR deficiency should be considered, and to improve the sensitivity of clinical criteria used to identify individuals who are candidates for mutational DNA analysis.[252,253] (Refer to the Genetic and molecular testing for Lynch syndrome section of this summary for more information about testing for MSI and IHC.)

Bethesda guidelines (1997):

1. Cancer in families that meet the Amsterdam criteria.
2. The presence of two Lynch syndrome–related cancers, including synchronous and metachronous CRCs or associated extracolonic cancers. [Note: Endometrial, ovarian, gastric, hepatobiliary, or small-bowel cancer or transitional cell carcinoma of the renal pelvis or ureter.]
3. The presence of CRC and a FDR with CRC and/or Lynch syndrome–related extracolonic cancer and/or a colorectal adenoma; one of the cancers diagnosed before age 45 years, and the adenoma diagnosed before age 40 years.
4. CRC or endometrial cancer diagnosed before age 45 years.
5. Right-sided CRC with an undifferentiated pattern (solid/cribriform) on histopathology diagnosed before age 45 years. [Note: Solid/cribriform defined as poorly differentiated or undifferentiated carcinoma composed of irregular, solid sheets of large eosinophilic cells and containing small gland-like spaces.]
6. Signet-ring–cell CRC diagnosed before age 45 years. [Note: Composed of more than 50% signet ring cells.]
7. Adenomas diagnosed before age 40 years.

Revised Bethesda guidelines (2004)*:

1. CRC diagnosed in an individual younger than 50 years.
2. Presence of synchronous, metachronous colorectal, or other Lynch syndrome–associated tumors.**
3. CRC with MSI-H pathologic associated features diagnosed in an individual younger than 60 years. [Note: Presence of tumor-infiltrating lymphocytes, Crohn-like lymphocytic reaction, mucinous/signet-ring differentiation, or medullary growth pattern.]
4. CRC or Lynch syndrome–associated tumor** diagnosed in at least one FDR younger than 50 years.
5. CRC or Lynch syndrome–associated tumor** diagnosed at any age in two FDRs or second-degree relatives.

*One criterion must be met for the tumor to be considered for MSI testing.

**Lynch syndrome–associated tumors include colorectal, endometrial, stomach, ovarian, pancreatic, ureter and renal pelvis, biliary tract, and brain tumors; sebaceous gland adenomas and keratoacanthomas in Muir-Torre syndrome; and carcinoma of the small bowel.[253,254]

Although the Bethesda guidelines were able to identify a higher proportion of Lynch syndrome carriers than the Amsterdam criteria, they still missed approximately 30% of Lynch syndrome families.[255] Furthermore, the Bethesda guidelines were not consistently used in clinical practice to identify the subset of individuals with CRC who should have MSI tumor testing; the guidelines were deemed cumbersome and difficult to remember by health care providers and the opportunity to refer for genetic evaluation was missed.[256]

With the advent of alternative approaches, including universal testing of all newly diagnosed cases of CRC for MSI (regardless of age at diagnosis or family history of cancer), clinical criteria for Lynch syndrome have been rendered obsolete. While the Bethesda guidelines were intended for individuals with cancer, their performance in individuals unaffected by cancer may still be of use. Given the limited modalities available to assess unaffected individuals for Lynch syndrome, family history and the use of clinical criteria may be appropriate in identifying those who warrant further genetic evaluation and testing.

Clinical risk assessment models that predict the likelihood of an MMR gene pathogenic variant

Because health care providers ineffectively use clinical criteria to select individuals with CRC for genetic referral and evaluation for Lynch syndrome, computer-based clinical prediction models were developed and introduced in 2006 as alternative modalities to provide systematic genetic risk assessment for Lynch syndrome. The risk models include the PREMM (PREdiction Model for gene Mutations) modelsExit Disclaimer, MMRpredictExit Disclaimer, and MMRproExit Disclaimer.[257-260]

Three models (PREMM[1,2,6], MMRpredict, and MMRpro) quantify an individual’s probability of carrying an MMR gene variant in MLH1, MSH2, and MSH6. The PREMM(1,2,6) model was subsequently extended to include prediction of pathogenic PMS2 and EPCAMvariants and is the only model to provide prediction of all five genes associated with Lynch syndrome (PREMM5).[260]

While the models were all created for the same purpose, they differ in the way they were developed and the variables used to predict risk. In addition, the populations in which they were validated reveal each model’s specific characteristics that may impact accuracy.[261-270] Deciding on which model to use in the risk assessment process depends on both the clinical setting in which it is applied and the patient population that is being evaluated. MMRpro’s predictions account for family size and unaffected relatives, the possibility of including molecular tumor data in the risk analysis, and the option of predicting pathogenic variant carrier status following germline testing. The major limitation in the widespread use of MMRpro in routine practice is the need to input data from the entire pedigree (including individuals without cancer), which is relatively time-consuming. Its best use is likely to be as a genetic counseling tool in a specialized high-risk clinic or research setting, as its accessibility is also limited. PREMM’s major advantages include that it is easy to use, available as an online tool, and has been extensively validated, including in a self-administered setting in a gastrointestinal clinic.[271] It includes risk prediction based on personal and family cancer history up to second-degree relatives for a broad spectrum of extracolonic cancers. However, the model does not take into account family size and may overestimate the likelihood of a pathogenic variant in a pedigree that includes multiple elderly family members who are unaffected by CRC or endometrial cancer. Given the ease with which one can use the PREMM model (it has been deemed less time-consuming than MMRpro in validation studies),[266] it may be used by diverse health care providers whose primary aim is to identify patients who should be referred for genetic evaluation, and is likely to be most useful in the pretesting decision-making process. Lastly, MMRpredict’s use may be limited overall because of its less accurate risk estimates [272] when used to evaluate families with Lynch syndrome–associated cancers and older individuals affected by CRC; the model was developed using data from young-onset CRC cases (patients diagnosed at age <55 y) and did not include extracolonic malignancies. Furthermore, the model does not incorporate tumor testing results or provide post-hoc risk estimates based on gene sequencing results.

Overall, there is ample evidence that each of the models has superior performance characteristics of sensitivity, specificity, and positive and negative predictive values that support their use when compared with the existing clinical guidelines for diagnosis and evaluation of Lynch syndrome. Because of the diverse clinical settings in which a health care provider has the opportunity to assess an individual for Lynch syndrome, prediction models offer a potentially feasible and useful strategy to systematically identify at-risk individuals, whether or not they are affected with CRC.

Summary

In conclusion, the presence of tumor MSI in CRCs, along with a compelling personal and family history of cancer, warrants germline genetic testing for Lynch syndrome, and most clinical practice guidelines provide for such an approach. These guidelines combine genetic counseling and testing strategies with clinical screening and treatment measures. Providers and patients alike can use these guidelines to better understand available options and key decisions. (Refer to Table 14 for more information about practice guidelines for diagnosis and colon surveillance in Lynch syndrome.)

Genetics of Lynch syndrome

The genetics of both the tumor and the germline have an important role in the development and diagnosis of Lynch syndrome. Tumor DNA in Lynch syndrome–associated tumors exhibits characteristic MSI, and in these cases, there is typically loss of IHC expression for one or more of the proteins associated with the MMR genes. Molecular testing with MSI and/or IHC has been adopted as a universal screen for diagnosis of Lynch syndrome in newly diagnosed patients with CRC and endometrial cancer. IHC testing results can potentially direct gene-specific germline testing. Many genetic testing laboratories offer multigene (panel) tests that simultaneously test for pathogenic variants in all of the Lynch syndrome–associated genes (and often additional genes associated with inherited cancer susceptibility).

Genetic and molecular testing for Lynch syndrome

MSI

The presence of MSI in colorectal tumor specimens is a hallmark feature of Lynch syndrome and can be cause for suspicion of a germline pathogenic MMR gene variant. Microsatellites are short, repetitive sequences of DNA (mononucleotides, dinucleotides, trinucleotides, or tetranucleotides) located throughout the genome, primarily in intronic or intergenic sequences.[273,274] The term MSI is used when colorectal, endometrial, or metastatic tumor DNA [275] shows insertions or deletions in microsatellite regions when compared with normal tissue. MSI indicates probable defects in MMR genes, which may be due to somatic variants, germline variants, or epigenetic alterations.[276] In most instances, MSI is associated with absence of protein expression of one or more of the MMR proteins (MSH2, MLH1, MSH6, and PMS2). However, loss of protein expression may not be seen in all tumors with MSI and not all tumors with loss of protein expression on IHC will be microsatellite unstable.

Certain histopathologic features are strongly suggestive of MSI phenotype, including the presence of tumor-infiltrating lymphocytes (refer to Figure 4), Crohn-like reaction, mucinous histology, absence of dirty necrosis, and histologic heterogeneity.[277]

ENLARGE

Figure 4. Tumor-infiltrating lymphocytes are a histopathologic feature suggestive of microsatellite instability.

Initial designation of a colorectal adenocarcinoma as microsatellite unstable was based on the detection of a specified percentage of unstable loci from a panel of three dinucleotide and two mononucleotide repeats that were selected at a National Institutes of Health (NIH) Consensus Conference and referred to as the Bethesda panel. If more than 30% of a tumor's markers were unstable, it was scored as MSI-H; if at least one, but fewer than 30% of markers were unstable, the tumor was designated MSI-low (MSI-L). If no loci were unstable, the tumor was designated microsatellite stable (MSS). Most tumors arising in the setting of Lynch syndrome will be MSI-H.[278] The clinical relevance of MSI-L tumors remains controversial; the probability is very small that these tumors are associated with a germline pathogenic variant in an MMR gene.

The original Bethesda panel has been replaced by a pentaplex panel of five mononucleotide repeats,[278] which has improved the detection of MSI-H tumors.

(Refer to the Prognostic and therapeutic implications of MSI section of this summary for more information about the treatment implications of MSI testing.)

(Refer to the Universal tumor testing to screen for Lynch syndrome section of this summary for information about the utilization of MSI status in the diagnostic workup of a patient with suspected Lynch syndrome.)

IHC

IHC methods are cheaper, easier to understand, and more widely available as a surrogate for MSI and, for these reasons, have replaced polymerase chain reaction (PCR)–based MSI testing in most institutions. IHC is performed in the colorectal or endometrial tumor (or metastatic sites) [275] for protein expression using monoclonal antibodies for the MLH1, MSH2, MSH6, and PMS2 proteins. Isolated loss of expression of any one of these proteins may suggest which specific MMR gene is altered in a particular patient.[279-282] However, certain proteins can form heterodimers (or have other binding partners) and yield loss of two proteins expressed on IHC.

MSI can lead to nucleotide-pairing slippage (looping) in which single nucleotide mispairs are introduced. Heterodimers of MMR proteins are formed to identify the errors and bind the DNA at these sites.[276,283] For example, MSH2 protein complexes with MSH6 protein to form MutSα, which has the main ability to repair single base pair mismatches and single base pair loop-out lesions that can occur during the replication of a mononucleotide repeat sequence. In the absence of MSH6 protein, the MSH2 protein will dimerize with the MSH3 protein forming the MutSβ complex, which has the ability to trigger repair of larger loop-out DNA mismatches, but also has some overlapping activity to repair lesions usually repaired by MutSα.

ENLARGE

Figure 5. Immunohistochemical tumor testing for protein expression of the mismatch repair genes associated with Lynch syndrome, depicted for a single patient with colorectal cancer. Protein expression is preserved for MSH2 and MSH6 (inset) and absent for MLH1 and PMS2 (inset). Absence of MMR protein expression is suggestive of Lynch syndrome and warrants additional evaluation.

As a result, when the germline pathogenic variant is in the MSH2 gene, the tumor IHC may not express both MSH2 and MSH6, as the latter protein requires binding to MSH2 for stability. In this case, if no pathogenic variant is found in either gene, germline pathogenic variant testing for EPCAM should be considered if it was not already included. Approximately 20% of patients with absence of MSH2 and MSH6 protein expression by IHC and no MSH2 or MSH6 pathogenic variant identified will have germline deletions in EPCAM.[284] The latter mechanism accounts for approximately 5% of all Lynch syndrome cases.[284] A deletion in one allele of exon 9 of the EPCAM (TACSTD1) gene, which is immediately upstream of the start site of MSH2 and in the same orientation, can lead to transcriptionalread-through and methylation of the MSH2 promoter, and subsequent silencing of MSH2 in any tissue that expresses EPCAM. The presence of EPCAM pathogenic variants showing similar methylation-mediated MSH2 loss has been reported in numerous families.[285] On the strength of these observations, germline EPCAM testing is performed in patients with loss of MSH2 protein expression on IHC testing of their CRCs but who lack a detectable MSH2 germline pathogenic variant and is included with MSH2 testing in all colon cancer gene panels.

In patients with no variants in any of these genes, tumor sequencing may reveal double somatic MSH2 variants. (Refer to the EPCAM and Lynch-like or HNPCC-like syndromesections of this summary for more information.)

Similarly, the loss of MLH1 (either by germline pathogenic variant or hypermethylation of the MLH1 promoter) results in the absence of expression of both MLH1 and PMS2 proteins in the tumor. The most common abnormal IHC pattern for DNA MMR proteins in colorectal adenocarcinomas is loss of expression of MLH1 and PMS2. PMS2 and MLH1 function as a stable heterodimer known as MutLα. MutLα binds to MutSβ and guides excision repair of the newly synthesized DNA strand.[276] A functional defect in MLH1 results in degradation of both MLH1 and PMS2, while a defect in PMS2 negatively affects only PMS2 expression. Thus, a loss of MLH1 and PMS2 indicates an alteration in MLH1 (promoter hypermethylation or germline variant), while loss of PMS2 expression indicates a germline PMS2 variant. However, among 88 individuals with PMS2-deficient CRC, PMS2 germline pathogenic variant testing followed by MLH1 germline pathogenic variant testing revealed pathogenic PMS2 variants in 49 individuals (74%) and MLH1 pathogenic variants in 8 individuals (12%).[286] Eighty-three percent of the alterations in MLH1 were missense variants, but two relatives carried identical MLH1 variants, and one individual, who developed two tumors with retained MLH1 expression, carried an intronic variant that led to skipping of exon 8.[286] Therefore, in CRCs with solitary loss of PMS2 expression, an MLH1 germline pathogenic variant should be sought if no PMS2 germline variant is found. Tumors with MSI and loss of MSH2 and MSH6 protein expression are generally indicative of an underlying MSH2 germline variant (inferred MSH2 pathogenic variant). Unlike the case with MLH1, MSI with MSH2 loss is rarely associated with somatic hypermethylation of the promoter.

Unlike MLH1 and MSH2 (which both dimerize with other proteins or have other binding partners), germline pathogenic variants in MSH6 and PMS2 result in the isolated loss of those specific proteins by IHC. However, tumors from MSH6 pathogenic variant carriers may not display the MSI phenotype at a frequency as high as MLH1 and MSH2 carriers (despite an inactive DNA MMR system), as there are pathogenic missense variants that do not completely abrogate protein expression yielding false negative results by IHC testing.[265,287] In a study that reported tumor testing results among MMR germline carriers enrolled through the Colon Cancer Family Registry, 7 of 24 carriers (28%) with MSH6pathogenic variants had tumors that displayed normal protein expression on IHC staining. IHC tumor testing was more informative for MLH1 and MSH2 pathogenic variant carriers in which 93% of MLH1 carriers had correlating loss of MLH1 protein expression and 96% of MSH2 carriers had loss of MSH2 protein expression.[265]

In some cases, tumors manifest MSI and/or IHC shows loss of DNA MMR protein expression, but no germline pathogenic variant is identified. This condition is known as Lynch-like (or HNPCC-like) syndrome and the tumor phenotype is predominantly due to biallelic somatic inactivation of DNA MMR genes and not a pathogenic germline alteration. (Refer to the Lynch syndrome–related syndromes section of this summary for more information.)

Table 10. Protein Loss and Potential Germline Defect(s)

Loss of Protein Expression	Germline MMR Defect Predicted by IHC Protein Expression Loss
	MLH1	MSH2	MSH6	PMS2	EPCAM
IHC = immunohistochemistry; MMR = mismatch repair.
MLH1/PMS2	X
MSH2/MSH6		X			X
MSH6			X
PMS2	X			X
MLH1	X
MSH2		X

Somatic MLH1 hypermethylation

It is important to recognize that hypermethylation of the MLH1 promoter, a somatic event confined to the tumor, can lead to abnormal protein expression of MLH1 on IHC. Approximately 10% to 15% of sporadic CRC cases have a microsatellite unstable tumor phenotype due to MLH1 hypermethylation and are not heritable. These sporadic MSI colon cancers [288] have a generalized excess of DNA methylation referred to as CIMP.[289] (Refer to the CIMP and the serrated polyposis pathway section in the Introduction section of this summary for more information.) Because loss of MLH1 protein expression on IHC occurs in both Lynch syndrome and sporadic tumors, its specificity for predicting germline MMR gene variants is lower than for the other MMR proteins, and additional molecular testing is often necessary to clarify the etiology of MLH1 absence.

BRAF pathogenic variants have been detected in 68% of CRC tumors with MLH1 promoter hypermethylation and very rarely, if ever, in CRC from patients with Lynch syndrome.[290-293] This suggests that detection of somatic BRAF V600E pathogenic variant detection in CRC may be useful in excluding individuals from germline variant testing. As a result, BRAFV600 testing and/or MLH1 hypermethylation assays are increasingly utilized in universal Lynch syndrome–testing algorithms in an attempt to distinguish between an absence of MLH1 protein expression caused by hypermethylation and germline MLH1 pathogenic variants. Making such a distinction is also a more cost-effective approach in excluding individuals from germline testing.

Biallelic mismatch repair deficiency (BMMRD)

Rarely, patients with MMR gene variants carry such variants in both parental alleles. When two variant alleles are identified, whether homozygous or compound heterozygous, this is termed biallelic mismatch repair deficiency (BMMRD) or constitutional mismatch repair deficiency (CMMRD). The likelihood of BMMRD involving homozygous MMR gene pathogenic variants will inevitably be higher among consanguineous unions. The incidence of consanguinity may be higher in rural and otherwise geographically and/or culturally isolated populations.[294]

Tumor studies yield characteristic abnormalities. In a series of 28 patients with BMMRD,[295] 17 brain tumors showed loss of staining for the MMR protein in the normal stromal cells in addition to neoplastic cells, showing a contradistinction from tumors in patients with Lynch syndrome in which normal staining is retained in nontumor cells. In contrast to this characteristic feature seen with IHC, PCR-based MSI analysis was not reliable, as 20 of 28 tumors were MSS. Of the tumors that were MSI-H, essentially all were colon cancers.

The PMS2 gene is markedly overrepresented in cases of BMMRD. It has been suggested that the presence of homozygosity of variants in the other MMR genes is a prenatally lethal state, while the otherwise milder expression of PMS2 is consistent with survival when present in both parental alleles.

(Refer to the BMMRD section in the Prevalence, clinical manifestations, and cancer risks associated with Lynch syndrome section for more information about the clinical phenotype of BMMRD.)

Table 11. Hereditary Colorectal Cancer (CRC) Syndromes and Associated Tumor Phenotypea

Clinical Phenotype	Pathogenic Germline Variant in DNA MMR	Somatic Inactivation of DNA MMR	Tumor Phenotype
BMMRD = biallelic mismatch repair deficiency; FCCX = familial colorectal cancer type X; MMR = mismatch repair; MSI = microsatellite instability; MSS = microsatellite stable.
aAdapted from Carethers et al.[296]
Lynch syndrome	Present in one allele	Present in one allele	MSI
Sporadic CRC with hypermethylation of MLH1 promoter	Absent	+BRAF	MSI
BMMRD	Present in two alleles	Absent	MSI (tumor and normal tissue)
Lynch-like	Absent	Present in two alleles	MSI
FCCX	Absent	Absent	MSS

Constitutional epimutation

While somatic hypermethylation of the MLH1 promoter is acquired and not uncommon, examples of MLH1 promoter hypermethylation have been described in the germline and are generally not associated with a stable Mendelian inheritance. This constitutional methylation of MMR genes occurs most often in MLH1 and, to a lesser extent, MSH2 and is termed constitutional epimutation .[297] A constitutional epimutation (also referred to as a primary epimutation) is an acquired alteration in normal tissue that silences an active gene or activates an inactive gene.[298] Such epimutations occur most often in maternal alleles. In some cases all somatic cells appear involved, while in others there is evidence of mosaicism. Tumors in patients with primary epimutations are generally indistinguishable from those otherwise typical of Lynch syndrome germline variant carriers, including age at onset, tumor spectrum, and presence of abnormal MSI and IHC. Since these are not inherited in a Mendelian fashion, antecedent family history of tumors is minimal, and risk to offspring somewhat unpredictable. Epimutations present in a de novo case seem to typically be "erased" in the process of gametogenesis and to not be passed to the next generation. Very rare cases of inherited MLH1 epimutations have been reported.[299,300]

Interpreting molecular alterations in tumors and distinguishing the likely primary epimutation cases from those of sporadic MSI poses significant challenges. Most instances of absence of MLH1 expression are caused by the sporadic hypermethylation of the MLH1promoter. Rare instances of a de novo constitutional epimutation in MLH1 [301] or an inherited germline MLH1 methylation [302] add some complexity to the interpretation of MSI associated with absence of MLH1 expression. Akin to sporadic MSI, primary epimutation tumors show methylation of the MLH1 promotor and may show BRAF variants as well. As noted above, family history of cancer in such cases tends to be minimal or absent, as in true sporadic MSI. Distinguishing such cases from sporadic cases may call for assaying normal tissue (e.g., blood or normal colon mucosa) for evidence of MLH1methylation, which will be absent from true sporadic cases and absent from carriers of conventional Lynch syndrome MMR pathogenic variants.

Such MLH1-predominant primary epimutations are to be distinguished from secondary epimutations such as those occurring when MSH2 is methylated as a consequence of inherited variants in the upstream EPCAM gene. (Refer to the EPCAM section of this summary for more information.)

Molecular diagnostic tumor testing to screen for Lynch syndrome in clinical practice

While many molecular pathology laboratories can assess both MSI and IHC, an approach that uses IHC testing as the initial screen for defective MMR activity has been favored because it is less labor intensive and more cost-effective.[303,304] Part of this rationale is that the information provided by IHC may target germline genetic testing toward one specific MMR gene (with the exception of loss of MLH1 expression) as opposed to a comprehensive testing strategy of all Lynch syndrome–related MMR genes that would be directed by the use of MSI alone.[255,303,305-308] While MSI testing was originally favored in the oncologic evaluation of individuals with CRC for its prognostic and therapeutic implications, screening for Lynch syndrome can be more effectively directed by IHC testing.

Universal tumor testing to screen for Lynch syndrome

Use of MSI and/or IHC testing in all newly diagnosed cases of CRC, regardless of the age at diagnosis or family history of cancer, increases the sensitivity of the initial screen for Lynch syndrome, especially for carriers of MSH6 and PMS2 pathogenic variants. This approach is more sensitive than existing clinical criteria, as many individuals with Lynch syndrome are diagnosed at older ages (>50 y) and have less striking family histories of CRC than previously appreciated. This universal testing of colorectal (and endometrial) tumors using either MSI or IHC testing has been recommended by many professional organizations and is being widely adopted.[309,95,310-312]

Genetic risk assessment and MMR gene variant testing in individuals with newly diagnosed CRC can lead to improved outcomes for the patient and at-risk family members. Dating back to 2009, the Evaluation of Genomic Applications in Practice and Prevention (EGAPP), a project developed by the Office of Public Health Genomics at the Centers for Disease Control and Prevention (CDC), reported that there was sufficient evidence to recommend offering tumor screening for Lynch syndrome to individuals with newly diagnosed CRC to reduce morbidity and mortality in relatives.[313,314] At that time, there was insufficient evidence to recommend a specific testing strategy between MSI and IHC.

Several studies have demonstrated the feasibility of universal screening for Lynch syndrome. Initial experience from one institution found that among 1,566 patients screened using MSI and IHC, 44 patients (2.8%) had Lynch syndrome. For each proband, an average of three additional family members were subsequently diagnosed with Lynch syndrome.[255] A subsequent pooled analysis of 10,206 incident CRC patients tested with MSI/IHC as part of four large studies revealed a pathogenic variant detection rate of 3.1%.[315] This study compared four strategies for tumor testing for the diagnosis of Lynch syndrome: (1) testing all individuals meeting at least one criterion of the Bethesda guidelines; (2) testing all individuals meeting Jerusalem recommendations;[316] (3) testing all individuals with CRC aged 70 years or younger, or older than 70 and meeting at least one criterion of the Bethesda guidelines; and (4) universal testing of all individuals with CRC.[315] Tumor testing with MSI involved panels individualized at each institution and IHC involved testing all four of the DNA MMR genes involved with Lynch syndrome, across all institutions. The strategy of tumor testing in all individuals diagnosed with CRC at age 70 years or younger and testing individuals over age 70 who met one of the revised Bethesda guidelines yielded a sensitivity of 95.1%, a specificity of 95.5%, and a diagnostic yield of 2.1%. This strategy missed 4.9% of Lynch syndrome cases, but 34.8% fewer cases required IHC/MSI testing, and 28.6% fewer cases underwent germline testing than in the universal approach.

The consideration to further stratify the recommendation for molecular tumor testing by age (i.e., 70 y) warrants attention as it influences the cost-effectiveness of universal screening strategy.

Loss of MLH1 and PMS2 due to somatic hypermethylation is not uncommon, and is more frequently detected with increasing age at CRC diagnosis.[317] Therefore, additional molecular tumor testing including BRAF and MLH1 hypermethylation testing is recommended in cases in which there is loss of MLH1 and PMS2 expression on IHC, thereby decreasing the number of individuals referred for unnecessary germline genetic testing. A testing strategy including MLH1 hypermethylation analyses in individuals aged 70 years or younger with CRC who had loss of MLH1 on IHC was shown to be cost-effective in a population-based study of 1,117 individuals.[318]

Screening individuals with CRC for Lynch syndrome is most often performed in a stepwise fashion based on IHC tumor testing results that evaluate protein expression for the four MMR genes related to Lynch syndrome. One proposed strategy is summarized in Figure 6. This framework does not incorporate a germline testing approach that simultaneously evaluates multiple cancer susceptibility genes (multigene [panel] testing), which may be useful in select patient populations. (Refer to the Multigene [panel] testing section of this summary for more information.)

ENLARGE

Figure 6. A proposed strategy to evaluate individuals with colorectal cancer for Lynch syndrome based on immunohistochemical tumor testing results. Adapted from Geiersbach KB, Samowitz WS. Microsatellite instability and cancer. Arch Pathol Lab Med 135(10):1269-77, 2011.

Cost-effectiveness of universal tumor screening for Lynch syndrome

Results are available from a Markov model that incorporated the risks of colorectal, endometrial, and ovarian cancers to estimate the effectiveness and cost-effectiveness of strategies to identify Lynch syndrome among persons aged 70 years or younger with newly diagnosed CRC .[304] The strategies incorporated in the model were based on clinical criteria, prediction algorithms, and tumor testing or up-front germline pathogenic variant testing followed by directed screening and risk-reducing surgery. IHC followed by BRAFpathogenic variant testing was the preferred strategy in this study. An incremental cost-effectiveness ratio of $36,200 per life-year gained resulted from this strategy. In this model, the number of relatives tested (3–4) per proband was a critical determinant of both effectiveness and cost-effectiveness. These results were similar to earlier analyses conducted by EGAPP which found that the most cost-effective approach was to test all tumors for absence of protein expression of MSH2, MLH1, MSH6, and PMS2 followed by targeted germline testing of MSH2, MLH1, or MSH6 offered depending on which protein was absent. If there was absence of MLH1, testing was offered for BRAF variant-negative tumors.[314]

NCCN 2018 guidelines support universal screening of all CRCs with IHC and/or MSI.[95] Universal screening in all individuals irrespective of age was associated with a doubling of incremental cost per life-year saved compared with screening only those younger than 70 years.[304] The authors of this analysis conclude that screening individuals younger than 70 years appears reasonable, while screening all individuals regardless of age might also be acceptable, depending on willingness to pay.

However, it is important to note that the conclusions from this study were contingent upon the number of at-risk relatives who underwent germline testing (through a process known as cascade screening ) based on the identification of a germline MMR gene variant in the index case of CRC in the family. In their model, to meet the accepted $50,000 cost-effective threshold, testing a minimum of three to four relatives was necessary.[304] This emphasizes the importance of provider-to-patient communication, family communication, and the need to ensure improved uptake of germline testing in Lynch syndrome families with a known causative gene. (Refer to the Psychosocial Issues in Hereditary Colon Cancer Syndromes section of this summary for more information about family communication and uptake of genetic testing in families with Lynch syndrome.)

Another study addressed the cost-effectiveness of testing for pathogenic variants in the Lynch syndrome–associated genes and evaluated 21 screening strategies, including clinical criteria, use of clinical Lynch syndrome prediction models, and molecular tumor testing.[319] The model included two steps: (1) measurement of the newly identified number of Lynch syndrome diagnoses; and (2) measurement of the life-years gained as a result of confirming Lynch syndrome in a healthy carrier. Among all of the strategies modeled, screening the proband with a predictive model such as PREMM(1,2,6) followed by IHC for MMR protein expression and germline genetic testing was the best approach, with an incremental cost-effectiveness ratio of $35,143 per life-year gained. Germline genetic testing on all probands was the most effective approach, but at a cost of $996,878 per life-year gained. The authors concluded that the initial step of Lynch syndrome screening should utilize a predictive model in the proband, and that both universal testing and general population screening strategies were not cost-effective screening strategies for Lynch syndrome.

Establishment of an upper age limit for universal tumor testing remains controversial. Some experts have endorsed testing only individuals with CRC who are younger than 70 years (reserving testing in individuals ≥70 y for only those meeting the revised Bethesda criteria; with this strategy, 5% of carriers would be missed).[320] However, others have advocated against an upper age limit for testing given the potential benefit to younger generations via cascade screening and the opportunity for increased surveillance and other prophylactic interventions in individuals found to carry a known familial pathogenic variant.

Another cost-effectiveness analysis was performed using data from 179 consecutive endometrial cancer patients diagnosed at or before age 70 years and screened with MMR IHC and reflex MLH1 promoter hypermethylation, among whom seven Lynch syndrome carriers (3.9%) were identified.[321] Only one of the seven Lynch syndrome probands was age 50 years or younger at endometrial cancer diagnosis. The authors calculated that screening women diagnosed with endometrial cancer at age 51 to 70 years resulted in an additional 29.3 life-years gained (on top of the 45.4 life-years gained by screening women diagnosed at age ≤50 y), and the incremental cost-effectiveness ratio for screening all diagnoses at age 70 years or younger versus diagnoses at age 50 years or younger was 5,252 euro per life-year gained. Universal tumor-based screening of all women age 70 years or younger was also cost-effective, compared with strategies using the Bethesda guidelines to guide MMR and MSI testing with an incremental cost-effectiveness ratio of 6,668 euro per life-year gained.

The cost-effectiveness of universal tumor testing in both CRC and endometrial cancer is largely driven by the assumption of cascade screening through which other at-risk family members will be identified, tested, and subsequently pursue their own cancer risk reduction.[304]

The cost of germline genetic testing continues to decrease with advancements in DNA mutational analyses, including simultaneous testing of multiple germline variants associated with malignancy, through multigene (panel) tests. As a result, additional cost-effective analyses using more updated data related to germline testing will need to be conducted. Multigene (panel) testing may become a more favorable and cost-effective approach in the future.

Considerations and limitations related to universal tumor testing for Lynch syndrome

While universal screening continues to be adopted nationally, there is significant variability in the uptake and approach to molecular testing. A 2011 survey of the National Society of Genetic Counselors revealed that more than 25% of respondents had some form of universal screening implemented at their center. Tumor screening methods varied; 34 (64.2%) of 53 centers started with IHC, 11 (20.8%) of 53 centers started with MSI testing, and 8 (15.1%) of 53 centers performed both tests on newly diagnosed colorectal tumors.[322] A 2012 survey suggested that some form of universal screening was being routinely performed at 71% of the National Cancer Institute (NCI) Comprehensive Cancer Centers, but utilization dropped to 15% among a random sample of community hospital cancer programs.[323]

Because adherence to universal screening for Lynch syndrome may be poor (many patients are not referred for genetic evaluation and testing), a prospective quality improvement study utilizing the Six Sigma conceptual framework was conducted to improve the implementation of universal genetic screening among young patients with CRC.[324] The main aim of the study was to increase the proportion of tumor studies for deficient MMR among patients with early-onset CRC (aged 18–50 y). The intervention involved patient and provider education, in addition to visual cues provided at point of care. The study demonstrated an improvement of 21.5% in the rate of IHC testing in young adults with CRC over the 12-month postintervention period compared with the preintervention period.

Studies reporting uptake of genetic testing for Lynch syndrome have largely focused on individuals and families who were selected for potential risk of Lynch syndrome based on family history or clinical characteristics. While universal tumor screening is increasingly being adopted to identify newly diagnosed patients who may have a germline variant, few studies have examined the uptake of genetic testing after universal tumor testing. An important implication of universal screening for Lynch syndrome is that it does not result in automatic germline testing in appropriate individuals. In the clinical setting, more follow-up by health care teams to facilitate referral to genetic counseling for patients with abnormal tumor screening results may improve completion of genetic testing.[325] Higher levels of patient completion of genetic testing after abnormal tumor screening may be associated with having genetic counselors involved in this process to disclose screen-positive results, provide counseling after tumor testing, or facilitate referrals.[326]

Subsequent genetic counseling requires coordination between the pathologist, the referring surgeon or oncologist, and a cancer genetics service. As an illustration, a population-based screening study found that only 54% of patients with an IHC-deficient tumor (that was BRAF pathogenic variant–negative) ultimately consented to and proceeded with germline MMR testing.[327] One institution found 21 pathogenic variants among 1,100 patients who underwent routine MSI and IHC testing after a diagnosis of CRC. This study found markedly increased uptake of genetic counseling and germline MMR gene testing when both the surgeon and a genetic counselor received a copy of abnormal MSI/IHC results, especially when the genetic counselor played an active role in patient follow-up.[325]

In contrast to tumor testing, which is commonly performed without a patient's prior knowledge, germline genetic testing, such as germline testing for MMR pathogenic variants, generally includes genetic counseling and requires patient permission before it is performed. A cross-sectional survey of U.S. cancer programs (20 NCI–designated Comprehensive Cancer Centers and 49 community hospital cancer programs) found that, of those that performed MSI and/or IHC testing as part of standard pathologic evaluation at the time of colon cancer diagnosis in all or select cases, none required written informed consent before tumor testing.[323]

Diagnostic strategies for all individuals diagnosed with endometrial cancer

Given the increased prevalence of endometrial cancer among carriers of MMR pathogenic variants, there is a growing consensus to screen patients with endometrial cancer for Lynch syndrome.

In a study that examined the feasibility and desirability of performing tumor screening of all endometrial cancers, regardless of age at diagnosis or family history of cancer, at least 2.3% (95% CI, 1.3%–4.0%) of newly diagnosed patients had Lynch syndrome.[328,329] Eight of thirteen cases diagnosed with Lynch syndrome were aged 50 years or older, eight did not meet published family history criteria for Lynch syndrome, and two would have been missed by MSI testing. Because of the increased prevalence of endometrial cancer and the results of this study, the authors support universal screening of endometrial cancers for Lynch syndrome. (Refer to the IHC section of this summary for more information about performing IHC for MMR protein expression.)

Another smaller study of 242 consecutive endometrial cases demonstrated a 4.5% (11/242) prevalence of MMR-deficient cases lacking somatic MLH1 promoter hypermethylation, including four cases (1.7%) with germline MMR mutations, four cases (1.7%) with two somatic MMR alterations on next-generation sequencing, and two cases (0.8%) with otherwise unexplained MMR-deficiency.[330] Such findings demonstrate that universal MMR tumor screening of endometrial cancers will identify individuals with underlying Lynch syndrome and a spectrum of non-Lynch syndrome cases with various forms of MMR-deficiency.

Another study prospectively evaluated universal IHC-based screening of both CRC and endometrial cancer cases, irrespective of age at diagnosis.[331] In both the tertiary and community settings, 1,290 CRC and 484 endometrial cancer cases were screened between 2011 and 2013. The study additionally calculated PREMM(1,2,6) and PREMM5 scores for all patients in whom a germline pathogenic variant was detected. Abnormal staining was observed in 22% of endometrial cancers and 18.8% of CRCs. After excluding those cases felt to be sporadic because of the presence of BRAF and/or hypermethylation of MLH1, 10.8 % of patients with CRC and 6.6% of patients with endometrial cancer were referred for genetic counselling. Lynch syndrome was diagnosed in 24 individuals (1.4%), 66% of whom had CRC. The overall detection rate of Lynch syndrome was 1.7% in endometrial cancer cases and 1.2% in CRC cases. Among Amsterdam criteria, Bethesda guidelines, PREMM(1,2,6), and PREMM5, the best performing model was PREMM5, which would have detected 82% of cases identified by universal screening.

The cost-effectiveness of tumor testing of women diagnosed with endometrial cancer was examined in a model-based simulation study and included IHC testing in the following scenarios: (1) diagnosis before age 50 years; (2) diagnosis before age 60 years; (3) any age at diagnosis with the presence of an FDR with any Lynch syndrome–associated cancer; and (4) all cases irrespective of diagnosis age and family history. Women fulfilling Amsterdam II criteria or those diagnosed before age 50 years with at least one FDR with any Lynch syndrome–associated cancer were directly referred for genetic counseling and genetic testing without IHC testing. A strategy of IHC testing for MMR protein expression in all patients with endometrial cancer and an FDR with any Lynch syndrome–associated cancer was reported to be cost-effective in the detection of Lynch syndrome.[332] This strategy had an incremental cost ratio of $9,126 per life-year gained relative to the least-costly strategy, which was genetic testing on all women diagnosed with endometrial cancer before age 50 years with at least one FDR with a Lynch syndrome–related cancer. Life expectancy was highest with the most inclusive testing strategy of IHC testing of all women with endometrial cancer irrespective of age at diagnosis or family history, but had the least favorable incremental cost ratio of $648,494 per life-year gained. NCCN recommends tumor testing with IHC and/or MSI, Lynch syndrome–specific genetic testing for MMR genes and EPCAM, or multigene (panel) testing of all endometrial cancers.[95] Despite these recommendations, the uptake of universal screening in women newly diagnosed with endometrial cancer is unclear.

(Refer to the PDQ summary on Genetics of Breast and Gynecologic Cancers for more information about endometrial cancer as a component of Lynch syndrome.)

Germline genetic testing

Genetic testing for germline pathogenic variants in MLH1, MSH2, MSH6, PMS2, and EPCAMcan help formulate appropriate intervention strategies for the affected variant-positive individual and at-risk family members, many of whom may be unaffected by cancer.

If a pathogenic variant is identified in an affected person, then testing for that same pathogenic variant should be offered to all at-risk family members. At-risk relatives who test negative for the identified pathogenic variant in the family are not at increased risk of CRC or other Lynch syndrome–associated malignancies and can follow surveillance recommendations applicable to the general population. Family members who carry the familial pathogenic variant are referred to surveillance and management guidelines for Lynch syndrome. (Refer to the Management of Lynch syndrome section of this summary for more information.)

If no pathogenic variant is identified in the affected family member, then testing is considered negative for Lynch syndrome in that individual. With advances made in DNA sequencing technologies, it is unlikely that current gene testing is not sensitive enough to detect a pathogenic variant in the genes tested. Advances in testing, including the common use of next-generation sequencing (NGS) by most commercial testing laboratories have improved upon the detection of certain alterations such as large deletions or genomic rearrangements as well as the presence of a pseudogene PMSCL in PMS2.

Possible reasons why a pathogenic variant may not be detected include the following:

• The family could have a variant in a yet-unidentified gene that causes Lynch syndrome or a predisposition to colon cancer.
• The individual tested in the family may have developed colon cancer through a nongenetic mechanism (i.e., it is a sporadic case also known as a phenocopy), while the other cases in the family are really the result of a germline variant. If this scenario is suspected, testing another affected individual who has had a Lynch syndrome–associated cancer is recommended.
• In cases in which a CRC tumor displayed MSI and/or abnormal IHC but no germline pathogenic variant was detected, biallelic somatic variants may be the etiology. These cases have been coined Lynch-like and are not considered familial.

Failure to detect a pathogenic variant could mean that the family truly is not at genetic risk despite a clinical presentation that suggests a genetic basis (e.g., the patient may have double somatic variants in an MMR gene). If no variant can be identified in an affected family member, testing should not be offered to at-risk members because results would be uninformative for the relatives. They would remain at increased risk of CRC by virtue of their family history and should continue with recommended intensive screening.

(Refer to the Management of Lynch syndrome section of this summary for more information.)

Multigene (panel) testing

Germline mutation analysis of MLH1, MSH2 (including EPCAM), MSH6, and PMS2 may be considered in instances in which tumor tissue is not available from individuals to test for MSI and/or MMR protein IHC. This approach has become less expensive with the advent of multigene (panel) testing, which is now offered by several clinical laboratories at a cost that may be comparable to single-gene testing. The cost of multigene testing may also approach the cost of tumor screening and may prove to be a cost-effective approach in individuals affected by CRC. At present, multigene tests are not routinely recommended for universal screening for Lynch syndrome among all newly diagnosed CRC patients, but they may be very useful in select populations, such as those with early-onset CRC [333] or from familial, high-risk clinic-based populations. It is also important to note that pathogenic variants may be detected in other cancer-associated genes beyond Lynch syndrome. In a study of 1,112 individuals who met NCCN criteria for Lynch syndrome testing and who underwent multigene testing with a 25-gene panel, as expected, 114 individuals (9.0%) were found to have pathogenic variants in MMR genes; however, 71 individuals (5.6%) were found to have a pathogenic variant in non-Lynch syndrome cancer predisposition genes, such as BRCA1, BRCA2, APC, MUTYH (biallelic), and STK11. Lastly, multigene tests yield a high proportion of VUS. In the aforementioned study, a total of 479 patients (38%) had one or more VUS.[334]

Individuals with early-onset CRC have been shown to have a high frequency and wide spectrum of germline pathogenic variants, indicating that panel testing in this population may be beneficial. In a study of 450 patients with early-onset CRC (mean age at diagnosis, 42.5 y) and a family history including at least one FDR with colon, endometrial, breast, ovarian, and/or pancreatic cancer, 75 germline pathogenic or likely pathogenic variants were identified in 72 patients (16%).[333] The spectrum of variants identified included Lynch syndrome and non-Lynch syndrome–associated genes, including several genes that have not traditionally been associated with CRC (e.g., BRCA1/BRCA2, ATM, CHEK2, PALB2, and CDKN2A). Given the high frequency and variety of hereditary cancer syndromes identified, the authors suggested that multigene testing in this population may be warranted.

Multigene testing has also been examined in a larger study of 1,058 individuals with CRC who were unselected for age at diagnosis, personal or family history, or MSI/MMR test results.[335] Germline pathogenic variants in cancer susceptibility genes were identified in 105 individuals (9.9%). While 33 individuals (3.1%) carried pathogenic variants in Lynch syndrome genes, 74 (7.0%) had pathogenic variants in non-Lynch syndrome–associated genes, including APC, MUTYH, BRCA1/BRCA2, PALB2, CDKN2A, TP53, and CHEK2. These data illustrate the breadth of variants that may be identified in unselected CRC patients; thus, use of a comprehensive multigene test may be warranted.

A 2017 study examined the frequency of pathogenic Lynch syndrome–associated gene variants in individuals undergoing multigene testing at a single commercial United States laboratory between 2012 and 2015, and reported on the characteristics of those carriers identified with Lynch syndrome.[336] The study reports on the largest cohort of individuals tested through multigene testing to date; data was reported on 34,980 individuals who had undergone various multigene panel tests that included the MMR and EPCAM genes, where the indication for testing was not limited to Lynch syndrome. A total of 618 pathogenic variants were identified in 612 individuals (1.7%) and analyses were conducted on 579 subjects (after exclusion of 33 individuals who had a Lynch syndrome–associated variant and a second MMR variant or other pathogenic alteration in another cancer predisposition gene). The majority of carriers were affected by cancer, including non-Lynch syndrome–associated malignancies, where breast cancer was most frequently reported (124/423, 23.5%). MSH6 variants were most prevalent (29.3%), followed by PMS2 (24.2%), MSH2(23.7%), MLH1 (21.6%), and EPCAM (1.2%). This finding differs from previous data where MSH2 and MLH1 variants were more prevalent, as individuals were more often selected for Lynch syndrome–specific testing due to a personal and/or family history of CRC.

The study reports on genotype-phenotype correlations on 528 Lynch syndrome carriers, the majority of whom had CRC (186, 35.2%) and endometrial cancer (136, 25.8%), followed by breast cancer (124, 23.5%) and ovarian cancer (74, 14%).[336] One hundred forty-five carriers presented with breast or ovarian cancer as their sentinel tumor and did not carry a prior diagnosis of CRC or endometrial cancer prior to the time of multigene testing. When examining MMR gene variant distribution among tumor-specific subgroups, a higher frequency of MSH6 and PMS2 variants were detected in carriers with breast cancer only than MLH1 and MSH2, where the latter pathogenic variants were more frequent in subjects with CRC only. For patients with breast cancer only, the frequency of PMS2 gene variants was significantly higher than population estimates, which was not the case for MLH1, MSH2, or MSH6. A comparable retrospective study reported similar findings. Standardized incidence ratios (SIRs) of breast cancer were calculated by comparing observed breast cancer frequencies in a population of 423 women with pathogenic or likely pathogenic variants in MMR genes with those in the general population. The authors reported a statistically significant age-standardized risk of breast cancer for MSH6 carriers (SIR = 2.11; 95% CI, 1.56–2.86) and PMS2 carriers (SIR = 2.92; 95% CI, 2.17–3.92).[337] A critical limitation of both of these studies was the excess of breast cancer cases in the overall referral population as well as the known high background population prevalence of MSH6 and PMS2 germline pathogenic variants.

Clinical criteria for the identification of Lynch syndrome, including the Amsterdam criteria, revised Bethesda guidelines, or the PREMM(1,2,6) risk prediction model, would have failed to identify 27.3% of Lynch syndrome carriers in this study.[336] Given the increased prevalence of breast and ovarian cancers, 58.9% met the NCCN guidelines for BRCA1/BRCA2testing and of these, 36.7% also met NCCN guidelines for Lynch syndrome testing. Lastly, there were limited data on tumor testing results, available only on 18.8% of pathogenic variant carriers, where results were often discordant with the altered gene, which was most often reported in MSH6 and PMS2 carriers. Results of this study support the use of multigene testing for Lynch syndrome and further study of the respective cancer risks, as current testing strategies limit identification of Lynch syndrome carriers and associated malignancies.

Lastly, germline MMR genes have been detected unexpectedly among individuals undergoing multigene testing for cancers not commonly associated with Lynch syndrome, such as breast and prostate cancer. As a result, the cancer spectrum associated with Lynch syndrome may be wider than previously appreciated. (Refer to the Breast cancer and Prostate cancer sections of this summary and the Genetics of Prostate Cancer summary for more information.)

(Refer to the Multigene [panel] testing section in the PDQ summary on Cancer Genetics Risk Assessment and Counseling for more information about multigene testing, including genetic education and counseling considerations, and research examining the use of multigene testing.)

Cost-effectiveness of multigene (panel) testing

As genetic testing becomes routine rather than the exception, questions regarding the cost of testing are inevitable. Historically, a cost-effectiveness ratio of $50,000 per quality-adjusted life-year (QALY) has been utilized as the benchmark for good value for care.[338] Over time it has been suggested that this threshold is too low and that other thresholds such as $100,000 or $150,000 be utilized.[338]

A 2015 study evaluated the cost-effectiveness of multigene testing for CRC and polyposis syndromes in patients referred to a cancer genetics clinic.[339] These authors developed a decision model to estimate the immediate and downstream costs for patients referred for evaluation and of CRC surveillance in family members identified as carriers of pathogenic variants. The costs were estimated on the basis of published models from the CDC and from an academic molecular genetics laboratory. They classified the syndromes on the basis of inheritance pattern and penetrance of CRC. Four custom panels were compared with the standard of care. The four panels tested for (1) Lynch syndrome–associated genes only (MLH1, MSH2, MSH6, PMS2, and EPCAM); (2) genes in panel 1 and additional genes associated with autosomal dominant inheritance and high CRC penetrance (APC, BMPR1A, SMAD4, and STK11); (3) genes in panels 1 and 2 and those associated with autosomal recessive inheritance with high CRC penetrance (MUTYH); or (4) all genes in the first three panels and those associated with autosomal dominant conditions with low penetrance (PTEN, TP53, CDH1, GALNT12, POLE, POLD1, GREM1, AKT1, and PIK3CA). The respective costs were as follows: panel 1, $144,235 per QALY; panel 2, $37,467 per QALY; panel 3, $36,500 per QALY; and panel 4, $77,300 per QALY when compared with panel 3. The authors concluded that the use of an NGS multigene test that includes highly penetrant CRC and polyposis syndromes and Lynch syndrome cancer genes was the approach most likely to provide clinically meaningful results in a cost-effective fashion.

The cost of germline genetic testing continues to decrease with advancements in technology since the time this model analysis was conducted; additional studies are needed to continue to assess the cost-effectiveness of this testing approach.