|Home | About | Journals | Submit | Contact Us | Français|
Lynch syndrome is the most common cause of inherited colorectal and endometrial cancers yet it is under-recognized in clinical practice. The relative merits of screening for Lynch syndrome among healthy adults without cancer versus among adults with colorectal or endometrial cancer are discussed in this Perspectives article. Newly diagnosed colorectal cancer patients are a much easier target population for screening and leads to more informative genetic test results, at a lower cost in most cases.
Lynch syndrome (LS), the most common cause of inherited colorectal and endometrial cancer, has not received the same amount of attention as hereditary breast-ovarian cancer syndrome. As a result, individuals with LS are grossly underdiagnosed. Indeed, one can estimate that the population incidence of LS is approximately 1 in 370. This estimate is based on the 2.8% incidence of Lynch syndrome among newly diagnosed colorectal cancer patients (1, 2) and the 5% lifetime risk for colorectal cancer in the U.S. (3). The penetrance of a mutation in the mismatch repair genes for colorectal cancer is about 50% (4), however, and so the incidence of LS in the general population is about double the incidence among colorectal cancer patients, or 0.028 × 0.05 × 2 = 0.0028, which is 1 in 370 individuals. Therefore, as many as 829,747 of the approximately 307,006,550 people in the U.S. today could have LS (of course, this is somewhat inflated since some LS cancer patients will die of their disease). Although the precise number of LS diagnoses in the U.S. is difficult to ascertain, it is safe to say that it is probably fewer than 10,000 cases, meaning that no more than 1.2% (10,000 / 829,747) of all individuals with LS are aware of their diagnosis at present. This state of underdiagnosis is especially troubling since ample data (5–7) indicate that an early diagnosis of LS followed by intense cancer surveillance and/or prophylactic surgery can prevent morbidity and mortality from LS cancers.
Any efforts aimed at increasing the numbers of individuals with LS who are diagnosed and thus can follow appropriate management guidelines, such as the efforts reported by Dinh et al. in this issue of the journal (8), are to be applauded. This perspective article will examine the challenge of the Dinh et al. approach for busy primary-care physicians who need to obtain family-history information from all of their patients and provide risk assessment for numerous conditions, not just LS. We also will examine the currently accepted clinical practice in the genetics community when evaluating an unaffected (not yet with a cancer known to be part of the LS-tumor spectrum), at-risk individual who has a family history suggestive of LS, which is to begin genetic testing with an affected (already had or has cancer known to be part of the LS-tumor spectrum) family member. We will also explore the advantages of the alternative approach we have proposed, which is to screen all newly diagnosed colorectal and endometrial cancer patients for LS and then test the at-risk relatives of those found to have an LS gene mutation.
Although primary care physicians obtain a family medical history from all patients as standard of care, their time for collecting the history and performing a risk assessment is limited. Studies have shown that the average family-history discussion lasts fewer than three minutes (9–11). Furthermore, the accuracy of self-reported family history is questionable. A 2007 evidence review by the Agency for Healthcare Research and Quality (AHRQ) found that the accuracy in reporting the presence of colorectal cancer in first-degree relatives ranged from 57% to 65% in studies using personal interview and from 86% to 90% in studies using telephone interview and self report (12). The follow-up evidence review by AHRQ in 2009 found that there is insufficient evidence on how to collect family-history information accurately in the primary care setting and on how taking a family history affects patient outcomes (13, 14). These family histories need to be assessed for a multitude of adult-onset conditions, not just LS. Running a separate computer risk model (e.g., PREMM1,2,6 or BRCApro) for every adult-onset condition with a hereditary component is not user-friendly for busy clinicians, and recommendations to do so are not likely to promote adherence. There are several family-history tools (on-line and paper) available to physicians, and they are reported to improve data recording by 46%–78% over data recording via family history in patient charts (15). Developing a standardized tool for collecting family history, performing an automated risk assessment, and incorporating this information into the electronic medical record appear to be the optimal way to make it easier for clinicians to quickly and consistently determine which patients need a genetics evaluation and consideration for genetic testing. Even then, physicians can only perform this assessment for the patients who come for a preventive physical exam, and studies have found that only about 21% of the population do so annually.
Formal family-history criteria to assist in diagnosing LS (previously called hereditary nonpolyposis colorectal cancer syndrome) were developed almost 20 years ago (16). Even though these family history criteria are not used consistently by primary care and other physicians to make appropriate referrals to cancer genetic testing, unaffected individuals who are at-risk for LS because of a family cancer history are already often referred for cancer genetic evaluations. A trained genetic counselor would advise these unaffected at-risk patients that the most informative way to evaluate their family for LS is to start by testing a family member who has or had a cancer which is known to be in the LS tumor spectrum (17). This approach does not delay the evaluation of individuals at risk for LS because one or more relatives already have a potential LS-associated cancer when the at-risk person presents for counseling. Of course, if the affected family members are deceased or unwilling to be tested, then the genetic counselor would offer testing to the unaffected family member. The reason to begin testing with an affected family member is because there is only a 50% chance that a gene mutation will be found in any unaffected first-degree relative of an affected member of an LS family, and this chance decreases with relational distance from the affected family member. Consequently, if an unaffected individual is the first family member to undergo genetic testing and this individual tests negative for an LS gene mutation, the result is uninformative because it has two possible explanations, as follows: a) The affected relative had an LS gene mutation but the consultand did not inherit it, which would be a “true negative” result and the individual could follow general population cancer screening guidelines; or b) the affected relative also is negative for Lynch syndrome gene mutations because the family has some other form of hereditary colon cancer (e.g., attenuated familial adenomatous polyposis, MUTYH-associated polyposis, or, as shown in Fig. 1, familial colorectal cancer of an undetermined genetic origin), meaning that the unaffected individual still has a 50% risk for a causative gene mutation and still needs increased cancer surveillance.
The Dinh et al. study (8) suggests offering genetic testing to all unaffected individuals who exceed a 5% likelihood for having an LS gene mutation, which, as just discussed, would yield an uninformative, negative result in the vast majority of unaffected individuals. In the cases of a negative test, follow-up testing (usually with less-expensive tumor-screening tests discussed below, but with full genetic testing for the sake of this discussion) would need to be offered to an affected relative anyway. Estimated from the costs used by Dinh et al. (8), it would cost $3495 to test the unaffected relative for mutations in all four genes, followed by another $3495 to test the affected relative for mutations in all four genes ($6990 total). If the affected relative also tests negative, the unaffected relative never needed to be tested in the first place and one $3495 test was wasted. If the affected relative tests positive for LS, then the unaffected relative could have single-mutation testing for the known mutation found in their relative, which only costs $298. Therefore, even if the affected relative is found to have LS, testing the unaffected relative first still wasted $3197 ($3495 minus $298). Insurers are becoming increasingly aware of the high costs of genetic testing and of the benefits of involving genetic counselors in the process to ensure that testing proceeds in the most logical and cost-effective manner. Unless the unaffected family member tests positive (which will occur only 5% of the time), it is always more cost-effective to start testing with an affected family member.
Furthermore, unaffected individuals could have LS even if they test negative, depending on how they were tested. LS gene testing is expensive and often challenging, sometimes requiring multiple blood samples and multiple laboratories since very few laboratories in the country offer full sequencing and large-rearrangement testing for all four LS genes. As a result, many patients receive incomplete LS gene testing of only two or three of the four LS genes and some of the genes are only analyzed with sequencing tests which can miss large-rearrangement mutations in and around the genes that can also cause LS.
Last, about 7% of the at-risk individuals undergoing genetic testing for LS will have a variant of uncertain significance in one of the LS genes (1, 2). These results are difficult to interpret in the best of situations and usually require additional testing among affected family members to determine whether or not the mutation is segregating with disease. So, considering these variants and the other confounding factors discussed above, one could argue that the only unaffected individuals for whom the approach recommended by Dinh et al. will provide a clear-cut result which can inform their medical management are the ~5% who test positive for a deleterious mutation in an LS gene. The remaining patients will have uninformative, negative results at best and false-negative results or ambiguous positive results for variants of uncertain significance that may lead to anxiety and confusion regarding appropriate cancer surveillance at worst.
For these and other reasons, we have long advocated that all newly diagnosed colorectal and endometrial cancer patients should be screened for LS (1, 2, 18, 19). This approach starts the testing with an affected individual and thus circumvents many of the issues raised above. A recently commissioned evidence review of this topic by the Evaluation of Genomic Applications in Prevention and Practice (EGAPP) working group (6) led the working group to recommend that all newly diagnosed colorectal cancer patients should be screened for LS to reduce the morbidity and mortality from colorectal cancer in their at-risk unaffected relatives (20). There are two available tumor tests, microsatellite instability (MSI) testing and immunohistochemistry (IHC) staining, that are highly predictive of LS in colorectal or endometrial cancer patients, and one of the tests (IHC) also indicates which of the four mismatch repair genes is likely to harbor the mutation. Both of these tests cost significantly less than genetic testing. Indeed, the most cost-effective approach for screening all newly diagnosed colorectal cancer patients for LS is to test with IHC followed by genetic testing in patients in whom any protein is absent after ruling out epigenetic causes of protein absence (21). The incremental cost-effectiveness ratio (ICER) of this approach is $22,522, which is well below the often-quoted $50,000 threshold at which a screening test is considered cost-effective (21) and is vastly reduced from the $737,025 ICER that applies to genetic testing for all four LS genes in colorectal cancer patients. These calculations suggest that offering tumor testing to the affected relatives of unaffected individuals with a > 5% risk for LS would be significantly more cost-effective than offering genetic testing for all four genes to the unaffected individuals.
Over time, this approach should substantially increase the diagnosis of LS among unaffected individuals and thus address the currently severe underdiagnosis of LS in the population. For example, the Columbus-area LS study utilized tumor testing on affected individuals followed by genetic testing in those with tumor tests suspicious for LS to identify which probands had LS. Then, genetic counseling and single-mutation genetic testing were offered to 306 at-risk relatives of the 58 colorectal and endometrial cancer patients found to have LS (approximately 5 relatives were tested per proband; refs. 1, 2, 18, 19). Fig. 2 is a map showing the impact of this cascade testing among the at-risk family members. Genetic testing among the 306 relatives found 132 with LS, who received intensive cancer surveillance recommendations, and 174 with a “true negative” result, who could follow general population cancer-screening guidelines. The majority of the relatives diagnosed with LS (102/132; 77%) were unaffected at the time of genetic testing.
Population-wide screening of healthy individuals for risk of adult-onset genetic conditions has not been widely accepted. For example, there are three recurrent BRCA1 and -2 gene mutations which cause hereditary breast-ovarian cancer syndrome and are found in 1 of every 40 Ashkenazi Jewish individuals (significantly more frequent than the 1 in 370 incidence of LS in the general population). However, testing for these three mutations is not currently offered routinely to unaffected Ashkenazi Jewish individuals. On the other hand, the National Comprehensive Cancer Network does recommend that all Ashkenazi Jewish women with breast cancer or ovarian cancer at any age should be offered genetic counseling and testing for the three common BRCA mutations (22). It is similarly logical to begin screening for LS with colorectal and probably endometrial cancer patients.
In conclusion, we agree that LS is underdiagnosed and that we need to encourage efforts to increase the numbers of individuals with LS who are diagnosed so that they can benefit from life-saving intensive-cancer surveillance. Although we would also like to encourage physicians to obtain family medical histories and to assess them for all genetic conditions (not only LS), this care will require the development of better, standardized tools that are integrated into the electronic medical record. We do not believe that offering full genetic testing to unaffected individuals with a > 5% risk for LS is the best approach to identifying more individuals with LS. The results will be uninformative in most cases and will be false negatives or will disclose variants of uncertain significance which are difficult to manage in some cases. In the case of an unaffected individual with no living affected relative or with affected relatives who are unwilling to be tested, however, it is good to know that offering testing to the unaffected individual is a cost-effective approach, as described by Dinh et al. (ref. 8; although many genetic counselors would still encourage beginning the process with testing of the deceased relative’s tumor since tumor tissue is often available for years after surgery). It is important to note that just because a testing approach is cost-effective does not mean that it is the most-informative or most cost-effective approach.
Dr. de la Chapelle and Ms. Hampel are supported by grants CA67941 and CA16058 from the National Cancer Institute.
Disclosure of Potential Conflicts of Interest
Ms. Hampel has received honoraria from Myriad Genetic Laboratories, Inc., for serving on a Lynch syndrome Advisory Panel.