OBJECTIVE: To estimate the cost-effectiveness of screening strategies for Lynch syndrome among newly diagnosed endometrial cancer patients.
METHODS: A decision analysis compared four strategies to screen women with newly diagnosed endometrial cancer for Lynch syndrome: 1) Amsterdam criteria strategy, where full gene sequencing was performed for women who meet Amsterdam criteria; 2) Sequence-all strategy, where full gene sequencing was performed for all women with endometrial cancer; 3) Sequence aged younger than 60 years strategy, where full gene sequencing was performed for women aged younger than 60 years with endometrial cancer; and 4) immunohistochemistry/single gene strategy, where immunohistochemistry was performed for the DNA mismatch repair genes for all women after single gene sequencing for specific women lacking protein expression. Prevalence rates, probabilities of immunohistochemistry staining loss, and gene mutation rates were calculated from published data. Costs were estimated from Medicare reimbursement rates. Cost-effectiveness ratios and incremental cost-effectiveness ratios were estimated for each strategy.
RESULTS: For the estimated 40,000 women diagnosed annually with endometrial cancer, the sequence-all strategy detects 920 patients with Lynch syndrome at a cost of $105 million. The Amsterdam criteria give the least-expensive strategy ($7 million), but detect the fewest patients (n=83) with Lynch syndrome. The immunohistochemistry/single gene sequencing strategy detects 858 patients at a cost of $17 million; this strategy has an incremental cost-effectiveness ratio of $13,812. The sequence aged younger than 60 years strategy was less effective and more costly than other strategies.
CONCLUSION: Of the strategies studied, immunohistochemical evaluation of tumor specimens for mismatch repair protein expression after single gene sequencing for patients with endometrial cancer is a cost-effective strategy for detecting Lynch syndrome.
LEVEL OF EVIDENCE: III
Immunohistochemical evaluation of tumor specimens after single gene sequencing for patients with endometrial cancer is a cost-effective strategy for detecting Lynch syndrome.
From the 1Division of Gynecologic Oncology and 3Division of Human Genetics, Department of Internal Medicine, Ohio State University College of Medicine, Columbus, Ohio; and 2Division of Gynecologic Oncology, University of Alabama at Birmingham, Birmingham, Alabama.
Corresponding author: David E. Cohn, MD, 320 West 10th Avenue, M-210 Starling Loving Hall, Columbus, OH 43210; e-mail: David.Cohn@osumc.edu.
Financial Disclosure Ms. Hampel received honoraria from Myriad Genetic Laboratories, Inc (Salt Lake City, UT) for serving on an advisory group on Lynch syndrome. The other authors did not report any potential conflicts of interest.
Lynch syndrome or hereditary nonpolyposis colorectal cancer (HNPCC) is the most common hereditary cancer syndrome in the United States and Europe. Greater than 50% of women diagnosed with Lynch syndrome will present with a gynecologic malignancy as their sentinel cancer; the majority of these cancers will be endometrial cancer. Lynch syndrome is caused by a germline mutation in one of the DNA mismatch repair genes MLH1, MSH2, MSH6, and PMS2. The prevalence of Lynch syndrome among all endometrial cancer patients (just over 2%) is identical to that in patients with colorectal cancer.1,2 Although most women with endometrial cancer present with early stage disease, a diagnosis of colon cancer has a worse prognosis. The identification of probands with Lynch syndrome will lead to a reduction in morbidity and mortality from colorectal cancer, given the opportunity for screening and prevention strategies; thus, endometrial cancer patients are a rich population in which to screen for Lynch syndrome. A number of guidelines, including the modified Bethesda guidelines and Amsterdam criteria, recommend screening for Lynch syndrome based on a family history of colorectal cancer. The current Bethesda guidelines are the best tool we have for identifying those individuals whose tumors require further genetic testing, but these may not be sensitive for individuals with small family pedigrees or those with a predominance of familial endometrial cancer. In newly diagnosed colorectal cancer patients, the sensitivity of the Bethesda guidelines for detecting individuals with MLH1 or MSH2 mutations approaches 100%.3 Although we do have to extrapolate some to apply these statistical tests to our gynecologic population, Hampel et al1,2 found that 8 in 13 endometrial cancer patients with Lynch syndrome did not meet any published clinical criteria for Lynch syndrome. Thus, if personal or family history or both were used as justification for referral for genetic testing 61.5% of patients would have been missed in the patients presenting with endometrial cancer. It is clear that a screening modality that is independent of personal or family history is necessary to identify endometrial cancer patients at risk for Lynch.
The Society of Gynecologic Oncologists recently published a committee statement on the triage of patients for genetic counseling based upon the clinician's perception of a patient's risk.4 The Society of Gynecologic Oncologists guidelines, however, lack concrete recommendations regarding the form of screening that should be undertaken. Recent literature suggests that immunohistochemistry for the mismatch repair proteins may be used as primary triage. Two recent retrospective studies have identified defective mismatch repair based on immunohistochemistry in 18–34% of newly diagnosed endometrial cancer patients aged younger than 50 years.5 Although these studies lacked follow-up with genetic sequencing, Lu et al6 recently provided evidence that prospective screening with immunohistochemistry, followed by sequencing in all endometrial cancer patients aged younger than 50 years, is a feasible, sensitive option. We set out to determine whether screening for Lynch syndrome is feasible in probands with endometrial cancer and to estimate the cost-effectiveness of these methods.
MATERIALS AND METHODS
A hypothetical cohort of 40,000 patients with newly diagnosed endometrial cancer serves as the basis for this decision analysis (Fig. 1). Patients with endometrial cancer were evaluated with four different screening strategies to determine if they had Lynch syndrome: 1) Amsterdam criteria strategy: full gene sequencing for women with endometrial cancer who meet revised Amsterdam criteria7; 2) Sequence all strategy: full gene sequencing for all women with endometrial cancer; 3) Sequence aged younger than 60 years strategy: full gene sequencing for women aged less 60 with endometrial cancer; 4) immunohistochemistry/single gene strategy: immunohistochemistry for all women with endometrial cancer after single gene sequencing (based on the lack of protein expression at immunohistochemistry).
Using data from both published and unpublished sources, we estimated the probabilities of a variety of outcomes, including prevalence rates, probabilities of immunohistochemistry staining loss, and gene mutation rates (Table 1). When published data were not available to estimate these probabilities, clinical experience was used to estimate these outcomes. A decision tree was generated (Fig. 1) using the estimates of clinical outcomes as described in Table 1.
We conducted our analysis from the perspective of a third-party payer, and specifically used Medicare reimbursement to provide conservative cost estimates (Table 2). We used only 2008 direct costs and did not use charges billed or indirect costs. Baseline costs were varied across clinically reasonable ranges in sensitivity analyses. The costs of cancer surveillance, cancer therapy, surgical complications, and morbidity from adjuvant therapy were not included in the model because the objective of this decision analysis was to examine costs associated with screening for Lynch syndrome and not those related to the management of patients diagnosed. Costs related to full gene sequencing were $2,474. The cost of genetic consultation was $150 and immunohistochemistry was $261. Single gene mutation testing ranged from $683 to $983, depending on the specific mutation. Screening for MSH2 included testing for large genomic deletions (Table 2).
Patients in the Amsterdam criteria strategy were those who fulfilled the Amsterdam criteria and were assigned the expense of referral for genetic consultation and full gene sequencing from peripheral blood. All patients in the sequence-all strategy underwent full gene sequencing. In the sequence aged younger than 60 years strategy, patients who were aged younger than 60 years at the diagnosis of their endometrial cancer underwent full gene sequencing. In the immunohistochemistry/single gene strategy, all patients had their tumor specimens evaluated for defective mismatch repair by immunohistochemistry for MLH1, MSH2, and MSH6. Those with normal expression of all proteins received no further evaluation. Patients who had absence of MLH1 staining were triaged based on their age: those who were aged 60 years or older did not undergo any further evaluation, whereas patients who were aged younger than 60 years underwent genetic consultation followed by mutation analysis of MLH1. This strategy is based on the fact that the majority of patients aged 60 years or older (without a personal or family history of a Lynch syndrome associated cancer) with absent MLH1 staining have this defect due to methylation of the MLH1 promoter (an epigenetic change) rather than due to Lynch syndrome (a genetic alteration in MLH1). Patients who had absence of staining for MSH6 or MSH2 had stepwise genetic testing starting with MSH6 testing followed by MSH2 testing in those patients with a wild-type MSH6 analysis.
Cost-effectiveness ratios, defined as the cost to detect one case of Lynch syndrome, were calculated for each strategy. Incremental cost-effectiveness ratio, defined as the cost to detect one additional case of Lynch syndrome, was also estimated for each strategy. Of note, the denominator of the incremental cost-effectiveness ratio is cases of Lynch syndrome detected and does not estimate quality-adjusted life years as an endpoint. It is for this reason that we do not offer $50,000 as the cutoff cost of a viable incremental cost-effectiveness ratio. Although incremental cost-effectiveness ratio is typically used with quality-adjusted life expectancy as the denominator, we merely offer the incremental cost-effectiveness ratio as a guide for which screening modality is cost-effective. Sensitivity analyses were performed to evaluate the effects of certain parameters on the model. All modeling and calculations were performed using a decision analysis program (DATA 3.5, TreeAge Pro Software, Williamston, MA).
Under our baseline assumptions, using 40,000 hypothetical patients with newly diagnosed endometrial cancer, we compared the four screening strategies (Table 3). The sequence-all strategy detects 920 patients with Lynch syndrome at a cost of $105 million. This results in the highest (least favorable) cost-effectiveness ratio of $114,087 and an incremental cost-effectiveness ratio of greater than $1 million. The Amsterdam criteria strategy is the least expensive ($7 million) but detects the fewest patients with Lynch syndrome (83 patients). The immunohistochemistry/single gene sequencing strategy detects 858 patients at a cost of $17 million. This strategy has the lowest cost-effectiveness ratio ($20,270) and an incremental cost-effectiveness ratio of $13,812. Using this strategy, each additional case of Lynch syndrome can be detected for less than $14,000. The sequence aged younger than 60 years strategy was less effective (800 patients) and more costly ($52 million) than other strategies, and therefore is dominated by other strategies, and would not be recommended as a screening strategy.
A series of one-way sensitivity analyses were performed to evaluate the effect of single parameters while all other estimates remained the same. We specifically evaluated the cost of full gene sequencing. Our baseline cost for full gene sequencing was $2,474; we varied this cost from $500 to $5,000 to evaluate a plausible range of costs. As the cost of full gene sequencing decreases, the total cost and cost-effectiveness ratio for the sequence-all strategy and sequence aged younger than 60 years strategy decreases. When the cost reaches $600, the sequence aged younger than 60 years strategy becomes the most cost-effective strategy, with a cost-effectiveness ratio of $18,750 and an incremental cost-effectiveness ratio of $12,318. This confirms that the cost of full sequencing effects the model; specifically, as sequencing becomes less expensive, additional strategies become cost-effective.
We also wanted to evaluate the age cutoff for screening; specifically, aged younger than 50 years compared with aged younger than 60 years. In a two-way sensitivity analysis, we estimated that 12% of patients with endometrial cancer are aged younger than 50 years, and 9% have Lynch syndrome. Using these estimates, this strategy would detect only 432 patients (compared with 800 patients with an aged younger than 60 years cutoff). Given the decreased number of patients being tested, the total costs ($13 million compared with $52 million with an aged younger than 60 years cutoff) decrease; thus the cost-effectiveness ratio improves to $29,156, and this strategy is no longer dominated (incremental cost-effectiveness ratio $16,931). However, the immunohistochemistry/single gene strategy is still preferred, with a cost-effectiveness ratio of $20,270 and an incremental cost-effectiveness ratio of $13,812.
Despite the increased awareness of Lynch syndrome in women with endometrial cancer, a considerable amount of confusion regarding screening for this disease still exists. It is clear that several management strategies are effective for screening for Lynch syndrome; however, both cost and efficacy must be evaluated. During the past decade, several studies have evaluated the usefulness of screening for Lynch syndrome, yet none have provided a cost analysis.
From our data, we believe that the combination of immunohistochemistry for the mismatch repair proteins followed by directed genetic testing serves as an efficacious strategy to identify patients with Lynch syndrome. This screening strategy is also cost-effective on a societal level. Although primary triage with the Amsterdam criteria is the least costly of the four strategies, it detects the fewest patients. Family history has not proven to be an adequate predictor of Lynch syndrome in our patient population. Hampel et al1,2 found that 9 of 13 patients with endometrial cancer and Lynch syndrome did not meet either the Amsterdam or Bethesda criteria. In a report on MSH6 mutations, Buttin et al11 reported that only 32% of family members of 6 endometrial cancer probands with MSH6 mutations had histories of concern for malignancy. It is apparent that the sensitivity of family history in detecting probands with Lynch syndrome is inadequate as a primary screening strategy.
Although the strategy of sequencing all patients with newly diagnosed endometrial cancer is effective, the exorbitant costs make it an undesirable primary triage strategy. Furthermore, the results of sequencing may be challenging to interpret. At our institution, approximately 50% of MLH1 and MSH6 exonic alterations are missense mutations of uncertain clinical significance (personal communication, Heather Hampel). Additionally, genomic mutations such as large deletions and rearrangements may not always be detected by standard methods of analysis. Casey et al12 reported a 33% improvement in large deletion detection when additional methods of detection were used compared with simple PCR and sequencing. Given that large deletions comprise a significant proportion of MSH2 mutations (approximately 20%), this has implications on the most effective screening strategies for Lynch syndrome.
In the sequence aged younger than 60 years strategy, we found that a substantial number of patients with Lynch syndrome would be identified, but at a relatively high cost. The decreased performance of this strategy combined with its associated higher costs cause this strategy to be dominated by others and thus not recommended as a screening strategy. When our model was adjusted to lower the screening age to patients aged younger than 50 years, we estimated that a substantial number of cases of Lynch syndrome would be missed. Recently, many have argued to revise the Bethesda criteria to include microsatellite instability testing in women presenting with endometrial cancer aged younger than 50 years; however, recent data suggests that the age of 50 to justify referral for genetic consultation may be too low.1,2,13 Results from our model suggest that age alone is not an effective way to screen for Lynch syndrome.
In this decision analysis, we demonstrate that immunohistochemistry followed by single gene sequencing is the least costly and most effective strategy to detect Lynch syndrome in women with newly diagnosed endometrial cancer. We acknowledge that immunohistochemistry followed by single gene sequencing may not be universally accepted; however, immunohistochemistry is already used clinically as a primary screen for Lynch syndrome in patients with newly diagnosed colorectal cancer at many cancer centers. Given the fact that 2% of patients with both endometrial cancer and colorectal cancer will have Lynch syndrome, it is reasonable to recommend that primary screening with immunohistochemistry should be applied to both of these populations.
The major criticism of our decision model is that we do not include downstream costs associated with the identification of Lynch syndrome in this population. However, the objective of this study was to evaluate the screening strategies for patients who have been diagnosed with endometrial cancer. This is a common clinical scenario that gynecologists and gynecologic oncologists encounter. We believe that this decision tree adequately analyzes several screening strategies for Lynch syndrome among endometrial cancer patients. We agree that once a patient has been identified as having Lynch syndrome, there are societal costs associated with screening colonoscopies and potential surgical interventions. However, the larger issue is finding a strategy to identify these patients independent of family history or clinicopathologic factors. Lu et al6 offer insight into the cost-effectiveness of the prevention of gynecologic cancers in a population of female Lynch patients. Given that the patients in our model have all undergone hysterectomy and bilateral salpingo-oophorectomy, their findings, although important, may not be applicable to our population.
We have attempted to use accurate but conservative estimates in the model. However, we have included all patients diagnosed with endometrial cancer and have not stratified screening strategies according to clinical scenario, stage, or tumor histology. We recognize that this may not completely represent clinical practice. Furthermore, we did not evaluate the usefulness of genotyping for the screening of mismatch repair deficiency. Hampel et al published the results of prospective Lynch screening in 500 newly diagnosed colon cancer patients. The sensitivity of immunohistochemistry in this population was 94%. The positive predictive value of an abnormal immunohistochemistry for detecting Lynch syndrome was 24%; significantly higher than using either age <50 (10%) or first degree relative with colon cancer or endometrial cancer (9%) as a predictor.14 When Cohn et al8 performed immunohistochemistry on 336 endometrial cancers, positive staining for MLH1 or MSH2 predicted an intact mismatch repair system in 95% of cases.12 These data, as well as others, suggest that mismatch repair screening by either genotyping or immunohistochemistry are both reasonable approaches.
Additionally, there are considerations unique to the endometrial cancer population that must be taken into account when considering screening for mismatch repair deficiency. When using genotyping for microsatellite instability, the positive predictive value of a microsatellite instability-low result is higher in the colorectal population, given the lower prevalence of MSH6 mutations in this population.15 Hampel et al1,2 determined that in endometrial cancer, 50% of the MSH6 mutations were found in tumors that were microsatellite stable or microsatellite instability-L. Thus, immunohistochemistry appears to be more sensitive than microsatellite instability in screening for MSH6 mutation in endometrial cancer.
Moreover, genotyping for mismatch repair deficiency will not discriminate between its presence as a result of epigenetic silencing of MLH1 or mutation of a mismatch repair gene. Up to 30% of endometrial cancer may demonstrate defective mismatch repair secondary to methylation of the MLH1 promoter15; the majority of microsatellite instability seen in endometrial cancer is secondary to MLH1 promoter methylation. Hampel et al1,2 in a study of 543 consecutive cases of endometrial cancer, identified 118 (22%) cases of microsatellite instability. Of these cases, 84 (75%) demonstrated a lack of MLH1 expression by immunohistochemistry, with 79 of 84 (94%) being due to methylation of MLH1. Thus, primary screening with genotyping rather than immunohistochemistry would require the additional step and added cost of methylation studies to ascertain which patients with microsatellite instability require genetic counseling and potential sequencing.
As with any cost-effectiveness model, the results should be used as an adjunct to clinical judgment, because these analyses are unable to address all relevant factors that are used to formulate treatment plans. However, we believe that these data suggest that screening for Lynch syndrome in patients with endometrial cancer is feasible and that a cost-effective strategy exists for the detection of patients with Lynch syndrome. It is our hope that continued research into screening for Lynch syndrome will lead to improved outcomes for both patients and their families.
1. Hampel H, Frankel W, Panescu J, Lockman J, Sotamaa K, Fix D, et al. Screening for Lynch syndrome (hereditary nonpolyposis colorectal cancer) among endometrial cancer patients. Cancer Res 2006;66:7810–7.
2. Hampel H, Panescu J, Lockman J, Sotamaa K, Fix D, Comeras I, et al. Comment on: screening for Lynch syndrome (hereditary nonpolyposis colorectal cancer) among endometrial cancer patients [letter]. Cancer Res 2007;67:9603.
3. Rodríguez-Moranta F, Castells A, Andreu M, Piñol V, Castellví-Bel S, Alenda C, et al. Clinical performance of original and revised Bethesda guidelines for the identification of MSH2/MLH1 gene carriers in patients with newly diagnosed colon cancer: a proposal of a newer and simpler set of recommendations. Am J Gastroenterol 2006;101:1104–11.
4. Lancaster JM, Powell CB, Kauff ND, Cass I, Chen LM, Lu KH, et al. Society of Gynecologic Oncologists Education Committee statement on risk assessment for inherited gynecologic cancer dispositions. Gynecol Oncol 2007;107:159–62.
5. Matthews KS, Estes JM, Conner MG, Manne U, Whitworth JM, Huh WK, et al. Lynch syndrome in women less than 50 years of age with endometrial cancer. Obstet Gynecol 2008;111:1161–6.
6. Lu KH, Schorge JO, Rodabaugh KJ, Daniels MS, Sun CC, Soliman PT, et al. Prospective determination of lynch syndrome in young women with endometrial cancer. J Clin Oncol 2007;25:5158–64.
7. Vasen HF, Watson P, Mecklin JP, Lynch HT. New clinical criteria for hereditary non-polyposis colorectal cancer (HNPCC, Lynch syndrome) proposed by the International Collaborative Group on HNPCC. Gastroenterology 1999;116:1453–6.
8. Cohn DE, Frankel WL, Resnick KE, Zanagnolo V, Copeland LJ, Hampel H, et al. Improved outcomes with an intact DNA mismatch repair system in endometrial cancer. Obstet Gynecol 2006;108:1208–15.
9. Esteller M, Levine R, Baylin SB, Ellenson LH, Herman JG. MLH1 promoter hypermethylation is associated with the microsatellite instability phenotype in sporadic endometrial carcinomas. Oncogene 1998;17:2413–7.
10. Ries LAG, Melbert D, Krapcho M, Stinchcomb DG, Howlader N, Horner MJ, et al. SEER cancer statistics review, 1975-2005. Based on Nov 2007 SEER data submission, posted 2008. Bethesda (MD): National Cancer Institute. Available at: http://seer.cancer.gov/csr/1975_2005/
. Retrieved June 8, 2009.
11. Buttin BM, Powell MA, Mutch DG, Babb SA, Huettner PC, Edmonston TB, et al. Penetrance and expressivity of MSH6 germline mutations in seven kindreds not ascertained by family history. Am J Hum Genet 2004;74:1262–9.
12. Casey G, Lindor NM, Papadopoulos N, Thibodeau SN, Moskow J, Steelman S, et al. Conversion analysis for mutation detection in MLH1 and MSH2 in patients with colorectal cancer. JAMA 2005;293:799–809.
13. Goodfellow PJ, Buttin BM, Herzog TJ, Rader JS, Gibb RK, Swisher E, et al. Prevalence of defective DNA mismatch repair and MSH6 mutation in an unselected series of endometrial cancers. Proc Natl Acad Sci U S A 2003;100:5908–13.
14. Hampel H, Frankel WL, Martin E, Arnold M, Khanduja K, Kuebler P, et al. Feasibility of screening for Lynch syndrome among patients with colon cancer. J Clin Oncol 2008;26:5783–6.
15. Modica I, Soslow RA, Black D, Tornos C, Kauff N, Shia J. Utility of immunohistochemistry in predicting microsatellite instability in endometrial carcinoma. Am J Surg Pathol 2007;31:744–51.
16. Garg K, Leitao MM Jr, Kauff ND, Hansen J, Kosarin K, Shia J, et al. Selection of endometrial carcinomas for DNA mismatch repair protein immunohistochemistry using patient age and tumor morphology enhances detection of mismatch repair abnormalities. Am J Surg Pathol 2009; 33:925–33.
Figure. No caption available.