Molecular Diagnosis of Hereditary Nonpolyposis Colorectal Cancer

When a new strand of DNA is synthesized during the process of replication, errors that are not immediately corrected by the 3′ to 5′ exonuclease activity of DNA polymerase are corrected by a DNA mismatch repair (MR) system. Mutations in gene coding for proteins that participate in the MR system allow persistence of replication errors that would otherwise be repaired. Some of these errors take the form of a phenomenon called microsatellite instability—insertions or deletions of simple repetitive elements within microsatellite sequences that occur throughout the genome. Approximately 0.1% to 0.5% of the population carries a germline mutation in 1 of 6 MR genes, hMSH2, hMSH6, hMLH1, hPMS1, hPMS2, or hTGFBR2 (and perhaps others), that results in a substantial increase in the probability of developing cancer of the colon and rectum, endometrium, kidney, and other sites.¹ Up to 80% of people who harbor a mutation in 1 of the MR genes will develop colorectal cancer (accounting for approximately 5000 new colorectal cancers in the United States yearly)¹ ; up to 60% of women will develop endometrial cancer.²

Prior to identification of the MR gene defects, clinicians recognized that some families are unusually predisposed to develop colorectal cancers and described an autosomal dominant disorder predisposing to cancer of the colon and other sites. This recognition led to the formation of a working group that suggested the Amsterdam criteria (which do not include MR gene sequence information) for diagnosis of hereditary nonpolyposis colorectal cancer (HNPCC).³ The Amsterdam criteria are used both to facilitate research and to identify individuals who may benefit from colonoscopic examination carried out every 1 to 3 years. Such intensive surveillance is of proven value in reducing morbidity and mortality in HNPCC family members.^{4 - 5} Since HNPCC is an autosomal dominant disorder, on average only half of family members harbor a genetic defect that elevates their cancer risk. Thus, if clinical criteria alone were used to screen for HNPCC, twice as many people would undergo the expense and inconvenience of intensive surveillance as would receive full benefit.

The identification of mutations within MR genes of some HNPCC families created both a molecular basis for understanding HNPCC and a basis for molecular diagnosis.¹ Many mutations cause protein truncation. The inheritance of these mutations closely parallels the distribution of cancers in HNPCC families. Nevertheless, MR gene mutations cannot be identified in all families meeting even the strictest criteria for HNPCC.¹ In some cases, the hereditary defect lies in a gene that was not sequenced or even identified. In a few cases, families with "chance clusters" of "sporadic" colorectal cancer are probably identified using clinical criteria, largely are based on family history. This suggests that while the Amsterdam criteria and other proposed screening strategies use distinctive clinical characteristics of HNPCC, these criteria do not provide a completely satisfactory mechanism by which to select high-risk individuals for increased surveillance. It would be useful to have a laboratory test to identify those people most likely to benefit from frequent colonoscopy.

There are currently 2 principal molecular diagnostic approaches that may be used in HNPCC testing—assessment of microsatellite instability in tumor specimens and MR gene sequence analysis of nontumor DNA. Approximately 90% of colorectal cancers arising in individuals with HNPCC demonstrate microsatellite instability. If colorectal tumors from a family suspected of harboring HNPCC do not show microsatellite instability, it is unlikely that known repair gene defects are involved. Hence, the absence of microsatellite instability may be used as a criterion to argue against (but not exclude) a diagnosis of HNPCC. Identification of microsatellite instability does little to support a diagnosis of HNPCC because about 85% of colorectal cancers demonstrating microsatellite instability arise in patients who do not show MR enzyme defects.⁶

There are currently at least 2 practical obstacles to routine assessment of microsatellite instability. First, few (if any) Clinical Laboratory Improvement Act–certified laboratories in the United States offer microsatellite instability assays as a routine clinical laboratory test, and routine microsatellite instability testing seems unlikely to be widely available in the near future. Second, tumors from family members with colorectal cancer are likely to have been removed days, months, or years before genetic counseling and testing are considered. These tumor samples are likely to be formalin-fixed, paraffin-embedded specimens, and fixation and embedding often cause technical artifacts that complicate the performance and interpretation of microsatellite instability assays.

The second principal molecular diagnostic approach that may be considered for HNPCC diagnosis is MR gene sequence analysis. Clinical laboratories currently use either the protein truncation test or direct sequencing to find mutations in hMSH2 or hMLH1. The protein truncation test identifies mutations that lead to production of an incomplete protein product but does not identify missense mutations that degrade function of a full-length protein product. Direct sequencing may identify all mutations in the regions sequenced at some point in the future.

In this issue of THE JOURNAL, Syngal and colleagues⁷ present results of a study designed to find out how useful direct sequencing is for guiding the clinical care of persons with a family history of colorectal cancer. The authors found that direct sequencing demonstrated a clearly pathogenic mutation of either the hMSH2 or hMLH1 gene in only 25% of families tested. Approximately 40% of the families that fulfilled the most stringent criteria for HNPCC harbored an identifiable MR gene mutation. Only about 16% of families fulfilling the least restrictive criteria harbored such a mutation. These results are similar to those reported in other studies.⁸ Importantly, in about 10% of cases, direct sequencing demonstrated a mutation or polymorphism that, in the authors' judgment, was not necessarily pathogenic.⁷

How then, should one interpret direct sequencing of hMSH2 or hMLH1 genes? First, if a clearly pathogenic mutation is identified within a family, then family members who do not harbor this mutation are unlikely to develop HNPCC or to derive substantially more benefit from intensive colorectal cancer surveillance than from routine screening. Second, family members who have a clearly pathogenic mutation within hMSH2 or hMLH1 are likely to benefit from screening colonoscopy every 1 to 3 years and possibly from increased surveillance for other HNPCC-associated cancers. Third, members of families who do not have an identifiable MR gene sequence abnormality may have mutations in genes that have not been sequenced or identified. It may be nearly as cost-effective to provide intensive colorectal cancer surveillance for all members of potential HNPCC families for which hMSH2 or hMLH1 mutations cannot be identified as it is to provide such surveillance for persons with known MR gene defects. Sequencing of other known MR genes is likely to be unrewarding because mutations of these genes are infrequent. Members of families who have hMSH2 or hMLH1 sequence abnormalities that are not clearly pathogenic should probably be treated as if no sequence abnormality were identified until a better understanding of how to interpret these mutations emerges.

Syngal and colleagues⁷ show that only a minority of families for whom MR gene sequence analysis is performed will have a laboratory result that can be used to reduce the frequency of colorectal cancer screening.⁶ Nevertheless, the proportion of families for which diagnostic testing identifies a pathogenic mutation should increase as clinical criteria are refined and other genes involved in the pathogenesis of HNPCC are identified. Studies of phenotype-genotype correlation may be expected to reduce the number of "ambiguous" nucleic acid changes.

The costs and benefits of HNPCC testing must not be considered in purely economic terms. Laboratory testing for HPNCC has a significant psychological impact on some people.⁹ Based on studies of persons undergoing or considering BRCA1/BRCA2 (familial breast cancer susceptibility) and RET (familial medullary thyroid cancer susceptibility) gene testing,^{10 - 12} it is clear that many people who test "negative" for a familial cancer mutation will still worry about their cancer risk. Those persons told that they definitely harbor a cancer susceptibility gene mutation will often perceive themselves as having a lower quality of life. People who decline cancer susceptibility gene analysis appear to be at an increased risk for depression,¹² whereas people who are depressed are more likely to decline testing.⁹ These potentially adverse mental health effects must be weighed against the likelihood that persons found to have a genetic susceptibility to cancer will adhere more rigidly to surveillance programs, thereby decreasing their cancer-related morbidity and mortality.

At present, molecular testing for HNPCC gene mutations is not appropriate for population screening. As shown by Syngal and colleagues,⁷ the overall clinical utility of MR gene sequencing is modest even for persons whose family histories suggest the possibility of HNPCC. A small percentage of people tested can benefit from a reduction in the frequency of colonoscopic surveillance; others, armed with knowledge of a "positive" test result, will be more likely to comply with an intensive surveillance program. Nevertheless, full benefit is likely to be attained only if these tests are performed in an environment that provides adequate psychological support, genetic counseling, and cancer surveillance and treatment. These resources must be readily available when the possibility of cancer susceptibility gene testing is suggested. Finally, it is clear that analysis of MR gene mutations in the research setting will provide further insight into the pathogenesis of HNPCC, lead to more effective molecular testing strategies, and ultimately improve patient care.