Guidelines on Genetic Evaluation and Management of Lynch Syndrome

A Consensus Statement by the US Multi-Society Task Force on Colorectal Cancer

Francis M Giardiello MD; John I Allen; Jennifer E Axilbund; C Richard Boland; Carol A Burke; Randall W Burt; James M Church; Jason A Dominitz; David A Johnson; Tonya Kaltenbach; Theodore R Levin; David A Lieberman; Douglas J Robertson; Sapna Syngal; Douglas K Rex


Am J Gastroenterol. 2014;109(8):1159-1179. 

In This Article

Identification of Lynch Syndrome

Several strategies have been developed to identify patients with LS. These include clinical criteria, prediction models, tumor testing, germline testing, and universal testing. The effectiveness of these strategies will be discussed here (Table 7).

Clinical Criteria

Amsterdam Criteria. Utilizing Amsterdam II criteria (Table 5) involves the clinical evaluation of the patient and patient's pedigree for colorectal and other LS cancers. Analysis from several sources reveals that patients and families meeting Amsterdam II criteria have a 22% sensitivity and 98% specificity for diagnosis of LS.[5,6,8,99,100] However, when a large number of families were collected and exhaustive searches performed for germline mutations in DNA MMR genes, fully 40% of families that meet the Amsterdam I criteria do not have LS.[69]

Revised Bethesda Guidelines. These guidelines specify circumstances in which a patient's CRC should be tested for MSI (Table 6). The sensitivity and specificity for LS in those meeting any one of the guidelines is 82 and 77%, respectively.[6,7]

Colorectal Cancer Risk Assessment Tool. Clinical criteria to identify patients at high risk for CRC are complex and difficult to apply in a busy office or endoscopy practice. Kastrinos and colleagues[101] developed and validated a simple 3–question CRC risk assessment tool. When all 3 questions were answered "yes," the tool correctly identified 95% of individuals with germline mutations causing LS. The cumulative sensitivity was 77% to identify patients with characteristics suggestive of hereditary CRC and who should undergo a more extensive risk assessment. This tool can be found in Figure 1.

Figure 1.

Colorectal cancer risk assessment tool. Adapted with permission from Kastrinos et al. (101).

Computational Models

Several clinical prediction models exist to determine an individual's risk for LS, including the MMRpredict, MMRpro, and the PREMM1,2,6 models. All appear to outperform existing clinical criteria, including the revised Bethesda guidelines.[99,100,102,103]

MMRpredict Model. This model uses sex, age at diagnosis of CRC, location of tumor (proximal vs distal), multiple CRCs (synchronous or metachronous), occurrence of EC in any first-degree relative, and age at diagnosis of CRC in first-degree relatives to calculate risk of the patient having an LS gene mutation. Reported sensitivity and specificity for this model is 69 and 90%, respectively.[5,100] This model appears to have the best specificity for LS of other calculators of gene mutation. This model can be accessed online at:

MMRpro Model. This model utilizes personal and family history of colorectal and endometrial cancer, age at diagnosis, and molecular testing results for MMR genes, when available, to determine the risk of a patient having a germline mutation of MLH1, MSH2, or MSH6.[104] This calculator also indicated the risk for future cancer in presymptomatic gene carriers and other unaffected individuals. The sensitivity and specificity of this model is 89 and 85%, respectively, and can be found at:

PREMM 1,2,6 , Model. Variables utilized in this model include proband, sex, personal, and/or family history of colorectal, endometrial, or other LS cancers.[105] This calculator gives a specific estimate of risk for a MLH1, MSH2, and MSH6 mutation. Analysis of the accuracy of this model reveals a sensitivity of 90% and specificity of 67%. PREMM1,2,6 appears to have the best sensitivity but worse specificity compared with the others. The use of this model to determine risk of LS in the general population was a cost-effective approach when a 5% cutoff was used as a criterion for undergoing germline genetic testing.[106] This model can be found at:

Tumor Testing

Testing of tumor tissue can be done on archived formalin-fixed tissue from surgical resection specimens or biopsies from colorectal or endometrial cancer. Some experts would also recommend testing adenomas >1 cm in size in appropriate individuals. Laboratories in the United States are required to save specimens for at least 7 years.

Microsatellite Instability Testing. The sensitivity for diagnosing LS using molecular testing of CRC tissue for MSI is estimated at 85%, with a specificity of 90%.[107]

Immunohistochemistry Testing. IHC testing of tumor tissue for evidence of lack of expression of MMR gene proteins has an overall reported sensitivity and specificity for LS of 83 and 89%, respectively. As discussed here, loss of MLH1 protein is likely secondary to somatic events, and loss of MSH2 protein is likely from a germline mutation.[107] Of note, the specificity of MSI and IHC testing decreases with increasing age due to increased prevalence of somatic MLH1 hypermethylation. In persons older than age 70 years, the use of BRAF testing (as will be discussed) when loss of MLH1 expression is seen, can help distinguish sporadic CRC tumors with somatic loss of MLH1 from those individuals who do require testing for a germline mutation for LS.[108] An advantage of IHC testing is that lack of a specific mismatch gene protein can direct germline testing to that specific gene.

The accuracy of IHC is operator dependent and varies according to the experience and skill of the laboratory performing the testing. Consequently, prudence would suggest that this testing be performed in recognized reference laboratories with high-quality control measures.

Universal Testing

Utilization of clinical criteria and modeling to identify patients with LS has been criticized for less than optimal sensitivity and efficiency. Studies of molecular testing of all CRCs reveal that up to 28% of LS patients would be missed with the most liberal of clinical criteria—the revised Bethesda guidelines.[25,109,110–112] Evaluation of Genomic Application in Practice and Prevention, a project sponsored by the Office of Public Health Genomics at the Center for Disease Control and Prevention, determined that sufficient evidence exits to offer genetic testing for LS to all individuals with newly diagnosed CRC.[113] The rationale was to reduce morbidity and mortality of relatives of patients with LS. Evaluation of Genomic Application in Practice and Prevention concluded that there was insufficient evidence to recommend a specific genetic testing strategy.[113] Universal testing for LS has also been endorsed by the Healthy People 2020 and the National Comprehensive Cancer Network (NCCN). Evaluation of a universal strategy by Ladabaum et al revealed that a systematic application of testing among patients with newly diagnosed CRC at ≤70 years of age could provide substantial clinical benefits at acceptable costs.[114] Other studies have also reported the cost effectiveness of universal CRC testing.[115] Ladabaum et al. concluded that IHC testing of CRCs for MMR gene proteins followed by BRAF mutation testing of the tumors when MLH1 protein expression is absent, emerged as the most cost-effective approach. Patients with absence of BRAF mutation would then have germline testing for a mutation in the presumed altered MMR gene.

Additional reports suggest that universal tumor IHC testing among individuals with CRC had greater sensitivity for identification of LS compared with other strategies, including Bethesda guidelines, or a selective strategy (tumor testing of patients with CRC ≤70 years of age or older patients meeting Bethesda guidelines).[112,116]

Although universal testing of CRC is recommended, development and implementation of such a screening system are complicated. These programs require cooperation and effective communication across multiple disciplines, ensuring that patients at risk for LS are identified, notified of abnormal results, and referred for genetic counseling and genetic testing.[117]

Panel testing for germline mutations in >20 cancer-causing genes (which include the MMR and EPCAM genes) is now available commercially as a single test. Inevitably, advances in technology will decrease the cost of such analysis. In the future, germline testing, rather than tumor evaluation, might be the most cost-effective universal testing approach.