Progress in Genetic Testing, Classification, and Identification of Lynch Syndrome

The genetic basis of familial colorectal cancer (CRC) has been substantially clarified over the past 14 years, but the fine points are still emerging. The first breakthrough occurred in 1991, when the adenomatous polyposis coli gene was cloned and found to be the locus of germline mutations causing familial adenomatous polyposis.^{1 - 2} This occurred because the disease has a dramatic and recognizable phenotype, and families were available for study.

With the APC gene identified, attention turned to Lynch syndrome—also called hereditary nonpolyposis colorectal cancer. This disease is more common by an order of magnitude but complicated by the fact that there are usually no physical manifestations of the disease until the affected individual develops cancer.³ As efforts to find the genes causing Lynch syndrome emerged, Vasen et al⁴ established the Amsterdam criteria in 1991 to identify familial clusters of CRC likely to represent this disease. The Amsterdam criteria required that at least 3 members of a kindred have a CRC (later expanded to include other tumors), that 1 member was a first-degree relative of the other 2, that at least 2 generations be involved, that familial adenomatous polyposis be excluded, and that at least 1 cancer be diagnosed in a patient younger than 50 years. This permitted the identification of families in which to find the gene(s) that caused this disease.

In early 1993, the mutational signature in tumors that is now called microsatellite instability (MSI) was discovered.^{3 ,5} It was quickly recognized that MSI is present in most of the CRCs in Lynch syndrome families.^{6 - 8} The recognition that MSI is caused by loss of DNA mismatch repair (MMR) activity led to discovery of the genes that cause Lynch syndrome.^{9 - 10} Although this is an area that continues to evolve, the principal genes that cause Lynch syndrome are the DNA MMR genes, ie, hMSH2, hMLH1, hMSH6, and hPMS2. The first 2 of these (hMSH2 and hMLH1) account for more than 90% of the germline mutations that can be found using conventional DNA analyses and are the only genes for which commercial testing is routinely available. A powerful consequence of the Human Genome Project was that it permitted the conclusion that all of the genes in the human DNA MMR family had been identified.¹¹

With genetic diagnosis available, and using MSI in CRC for case-finding, Lynch syndrome was soon found in families that did not meet the Amsterdam criteria.¹² Most of the familial clusters had CRCs and other tumors with MSI, but some families did not have MSI in their tumors. To make matters more interesting, many families met the Amsterdam criteria and had CRCs with MSI, but no germline mutation could be found in the 2 major Lynch syndrome genes.¹² This indicated that familial colorectal cancer was a bit more complicated. There appeared to be germline mutations in hMSH2 and hMLH1 that were not detectable by DNA sequencing for technical reasons, and there were familial clusters of CRC that were not due to germline mutations in DNA MMR genes.

In a recent study, Casey et al¹³ used the Colon Cancer Family Registry—an international consortium developed to identify familial clusters of colon cancer—as well as state-of-the-art measures to identify mutations in hMSH2 and hMLH1. The investigation included a laboratory technique that permits the separation of paternal and maternal alleles to detect mutations that are masked when the 2 are analyzed together.¹⁴ This substantially increased the diagnostic yield for mutations, particularly for hMSH2. This work demonstrated that many DNA MMR gene mutations—especially large deletions—could be found using additional, more powerful techniques.

A problem with analysis of the MMR genes is that most mutations identified in Lynch syndrome families are unique and are scattered throughout the genes.¹⁵ Thus, all of these genes must be screened to detect a mutation in a new family. However, a few mutations recur in Lynch syndrome families from all over the world. Some of these mutations are associated with shared haplotypes between different families, suggesting a common ancestral origin (ie, founder mutations). Founder mutations are characteristic of isolated populations, such as Finns and Ashkenazi Jews. In another recent report, Lynch et al¹⁶ identified a founder mutation (deletion exon 1-6 MSH2) identified in 9 families in a large outbred population in the United States. An important consequence of this study is that routine testing of US families should include an assay for this mutation.

These studies did not address the issue of whether there would be Lynch syndrome–like families with no mutation in any DNA MMR gene. For a long time, families have been recognized that are characterized by the clustering of CRCs at an advanced age and by the absence of other Lynch syndrome–associated cancers.¹⁷ The ability to characterize familial clusters using precise genetic criteria led to an interesting conundrum: how to characterize a family that met the traditional Amsterdam criteria for Lynch syndrome but that had no detectable germline mutation in a DNA MMR gene? Furthermore, what if the tumors did not have MSI? This could represent a different disease and might require different management approaches.

In this issue of JAMA, Lindor and colleagues¹⁸ have applied rigorous diagnostic measures to 161 familial clusters of CRC. All of these families met the Amsterdam criteria for Lynch syndrome. The investigators divided the families into those in which MSI was present in the CRCs (group A) and those in which MSI was not present (group B). It was reasonable to hypothesize that these 2 groups might have different diseases, and Lindor et al demonstrate that this is apparently the case. The families in group A had all the classic clinical features of Lynch syndrome, and the average age to develop cancer was 48.7 years. There was a 6-fold increase in risk for CRC and significantly increased risks for cancers of the uterus, stomach, kidney, ovary, small intestine, and ureter. This is classic Lynch syndrome.

However, families in group B (which met the Amsterdam criteria but did not have DNA MMR abnormalities) had a significant 2-fold increase in risk for CRC but not for any of the other tumor. Additionally, the average age of developing cancer in group B relatives was 60.7 years, significantly older than in group A. By genetic and clinical criteria, Lindor et al conclude that this is a different disease, which they term “familial colorectal cancer type X.” This is not Lynch syndrome, and the genetic basis of this disease remains completely unknown. The authors stressed that the surveillance protocol of frequent colonoscopies and endometrium surveillance usually advised for Lynch families are not recommended for these families. Instead, they advise a less stringent protocol including colonoscopy every 5 years, starting at a more advanced age.

Identification of Lynch syndrome families is important because it makes it possible to target preventive measures. The study by Piñol and colleagues,¹⁹ also in this issue of JAMA, evaluated the best strategy to identify Lynch families. In 1996, a consortium of investigators developed the Bethesda guidelines to determine which tumors should be tested for MSI³ ; these have been subsequently updated.²⁰ To assess the effectiveness of these guidelines, Piñol et al performed MSI testing and immunohistochemical analysis of the MMR gene proteins in tumors from approximately 1200 patients, followed by mutation analysis (for MSH and MLH1) in those with an abnormal test result. One quarter of the patients fulfilled 1 of the revised Bethesda guidelines and one third of the tumors showed MMR deficiency; mutations were identified in 11 patients (0.9%). The authors concluded that the new Bethesda guidelines are very effective for the identification of Lynch syndrome families.

There is still more to learn about the causes of hereditary colorectal cancer. For example, the mutational analyses undertaken by Piñol et al did not include a search for MSH6 mutations, which may constitute 7% of Lynch syndrome mutations.¹⁵ Second, in their immunohistochemical analyses, only antibodies against MLH1 and MSH2 proteins were used, whereas a recent study suggests that adding antibodies against MSH6 and PMS2 will increase the diagnostic yield.²¹ Third, the authors used only the BAT26 microsatellite marker instead of the full panel of markers suggested by the National Cancer Institute.²² Perhaps up to 5% of MSI-H CRCs were missed using this approach. And fourth, is immunohistochemical analysis as effective as MSI analysis in identifying MMR-deficient CRCs, as the authors conclude? Some studies suggest that MSI and IHC analyses are equally effective,²³ while others suggest that MSI analysis is more sensitive.²⁴ The best approach might be to perform IHC and MSI analyses in a different order, depending on the probability of finding a relevant germline mutation. In families meeting the Amsterdam criteria, there is a mutation detection rate of 40% to 50%.¹² Immunohistochemical analysis might be the first step, followed by MSI analysis in cases with normal immunohistochemical results. The advantage of immunohistochemistry is that it may direct the search for mutations to the specific gene. In patients who meet the Bethesda guidelines, the first step might be MSI analysis followed by immunohistochemical analysis of all tumors classified as MSI-high.²⁵

Even though MSI and immunohistochemical analyses have been available for several years, they are being applied on a relatively small scale at this time. Therefore, efforts should be undertaken to make these molecular diagnostic tools available for all clinicians. Finally, to make optimal use of them, clinicians should be familiar with the Bethesda guidelines to know in which patients the tests should be performed.