Fax: (713) 792-7586
Cost-effectiveness analysis of prevention strategies for gynecologic cancers in Lynch syndrome
Version of Record online: 27 MAY 2008
Copyright © 2008 American Cancer Society
Volume 113, Issue 2, pages 326–335, 15 July 2008
How to Cite
Kwon, J. S., Sun, C. C., Peterson, S. K., White, K. G., Daniels, M. S., Boyd-Rogers, S. G. and Lu, K. H. (2008), Cost-effectiveness analysis of prevention strategies for gynecologic cancers in Lynch syndrome. Cancer, 113: 326–335. doi: 10.1002/cncr.23554
- Issue online: 8 JUL 2008
- Version of Record online: 27 MAY 2008
- Manuscript Accepted: 29 FEB 2008
- Manuscript Revised: 27 FEB 2008
- Manuscript Received: 13 NOV 2007
- Lynch syndrome;
- gynecologic cancer;
- prevention strategy
Women with Lynch syndrome (hereditary nonpolyposis colorectal cancer) have an increased lifetime risk for endometrial and ovarian cancer. Screening and prophylactic surgery have been recommended as prevention strategies. In this study, the authors estimated the net health benefits and cost-effectiveness of these strategies in a Markov decision-analytic model.
Five strategies were compared for a hypothetical cohort of women with Lynch syndrome: 1) no prevention (‘reference’); 2) prophylactic surgery (hysterectomy and bilateral salpingo-oophorectomy) at age 30 years; 3) prophylactic surgery at age 40 years; 4) annual screening with endometrial biopsy, transvaginal ultrasound, and CA 125 from age 30 years; and 5) annual screening from age 30 years until prophylactic surgery at age 40 years (combined strategy). Net health benefit was measured in quality-adjusted life years (QALYs), and the primary outcome measured was the incremental cost-effectiveness ratio (ICER). Baseline and transition probabilities were obtained from published literature, and costs were from the U.S. Department of Health and Human Services and Agency for Health Care Quality and Research. Sensitivity analyses were performed for uncertainty around various parameters.
The combined strategy provided the highest net health benefit (18.98 QALYs) but had an ICER of $194,650 per QALY relative to the next best strategy (prophylactic surgery at age 40 years). Prophylactic surgery at age 30 years and annual screening were dominated by alternate strategies.
Annual screening followed by prophylactic surgery at age 40 years was the most effective gynecologic cancer prevention strategy, but the incremental benefit over prophylactic surgery alone was attained at substantial cost. The ICER would become favorable by improving the effectiveness and reducing the costs of screening in this population. Cancer 2008. © 2008 American Cancer Society.
Lynch syndrome (hereditary nonpolyposis colorectal cancer [CRC]) is an autosomal-dominant, inherited cancer-susceptibility syndrome characterized by a lifetime risk of 40% to 60% for colorectal and endometrial cancer.1 In the general population, the mean age at diagnosis of endometrial cancer is 63 years; however, among women with Lynch syndrome, the mean age is significantly younger at around 48 years.1-3 These women also are at risk of developing ovarian cancer, with a lifetime risk of 10% to 12%.4 Prophylactic hysterectomy with bilateral salpingo-oophorectomy (HBSO) after the completion of childbearing has been recommended as a prevention strategy.5 Schmeler et al reported that prophylactic surgery prevented 100% of endometrial and ovarian cancers in this population.1-3 A decision analytic model by Chen et al identified prophylactic surgery at age 30 years as the most effective gynecologic cancer prevention strategy in Lynch syndrome.6 However, to our knowledge it is unclear whether prophylactic surgery is an acceptable option for women at this age because of the postoperative consequences, such as loss of fertility, vasomotor symptoms, urogenital atrophy, and osteoporosis.7, 8
Although to our knowledge there is little evidence that screening for endometrial and ovarian cancer is beneficial for women with Lynch syndrome, current consensus statements recommend screening with annual endometrial biopsy and transvaginal ultrasound (TVUS) beginning at ages 30 years to 35 years.5, 9 In the absence of a proven survival difference between prophylactic surgery and screening, patient preferences and costs are an important consideration in determining the optimal strategy. In this report, we present a cost-effectiveness analysis of gynecologic cancer prevention strategies for women with Lynch syndrome.
MATERIALS AND METHODS
We developed a Markov decision-analytic model with the best available data to estimate outcomes in a cohort of women with Lynch syndrome at risk for endometrial and ovarian cancer. We compared 5strategies by determining the incremental cost-effectiveness ratio (ICER), which we defined as the additional cost of a specific strategy divided by its health benefit compared with an alternative strategy. The numerator of the ICER was the average lifetime cost (U.S. dollars in the year 2006), and the denominator was the average quality-adjusted life-expectancy gain, which was expressed in quality-adjusted life-years (QALYs). A strategy that was more costly and less effective than an alternative strategy was ‘dominated.’ A strategy that was more costly but more effective than an alternate strategy was considered cost-effective if its ICER was below a willingness-to-pay threshold of $50,000 per QALY gained.10 In keeping with the recommendations of the U.S. Panel on Cost-Effectiveness in Health and Medicine, we adopted a societal perspective and discounted all costs and health benefits at a rate of 3% per year.11 We performed sensitivity analyses to account for uncertainty around various parameters.
The model was programmed by using TreeAge Pro 2007. A hypothetical cohort of women with Lynch syndrome enters the model at age 30 years; and, each year, they face age-dependent risks of endometrial and ovarian cancer up to age 70 years. The model consists of 4 health states that are mutually exclusive, as illustrated in Figure 1: 1) not at risk for gynecologic cancer, 2) at risk for gynecologic cancer, 3) gynecologic cancer, and 4) dead. All women begin in the ‘at risk for gynecologic cancer’ state. They move to the ‘not at risk’ state if they have undergone prophylactic surgery. They transition to the ‘gynecologic cancer’ state if they are diagnosed with endometrial or ovarian cancer. At the end of each 1-year cycle, women return to 1 of the health states according to baseline and transition probabilities defined in the model. If they have endometrial or ovarian cancer but have not developed disease recurrence within 5 years, then they move to the ‘not at risk’ state; however, they are still at risk for CRC. This process continues until all women in the cohort reach the ‘dead’ state because of cancer or other age-related causes according to U.S. life tables.12 We conducted a Monte-Carlo simulation of 10,000 women in each strategy to estimate the number of endometrial and ovarian cancer cases in each cohort as well as CRC. We did not estimate the number of all other Lynch syndrome-related cancers. Quality-of-life weights (utilities) were applied to noncancer states13 and to other health states in the model (prophylactic surgery, endometrial cancer, ovarian cancer).14-22 With an average utility of 0.86 for prophylactic surgery, 1 year is quality-adjusted to an equivalent of 0.86 years without this intervention (because of loss of fertility and other effects of premature menopause). This adjustment is carried through until age 45 years, when quality of life begins to decline among women in the general population.13 To account for differences in patient preference, these utilities were varied over a wide range of estimates.
Five strategies are defined: 1) no prevention (‘reference’ strategy, no screening or prophylactic surgery); 2) prophylactic surgery at age 30 years; 3) prophylactic surgery at age 40 years; 4) annual screening from age 30 years with endometrial biopsy, CA 125, and TVUS; and 5) annual screening from age 30 years until prophylactic surgery at age 40 years. An abnormal screen (elevated CA 125 and an abnormal TVUS or endometrial biopsy) leads to exploratory laparotomy, HBSO, and staging.
For the base case, we assume that 1) all women who enter the model are at risk for both endometrial and ovarian cancer, and these risks are eliminated after prophylactic surgery; 2) the lifetime risks of endometrial and ovarian cancer are 50% and 10%, respectively; 3) compliance with surveillance and subsequent interventions is 100%; 4) if endometrial or ovarian cancer is missed on a screen or test, then the diagnosis is made within the next year; and 5) the lifetime risk of CRC is 42%. Sensitivity analyses were used to evaluate alternate assumptions to those used in the base case.
Direct costs of screening were obtained from the Medicare payment schedule of the U.S. Department of Health and Human Services using Health Care Financing Administration Common Procedure Codes.23 Direct costs of prophylactic surgery (eg, clinic visit, professional fees, inpatient hospital charges) were from the Health Care Utilization Project summary on statistics for U.S. hospitalizations for principal diagnoses and procedures.24 Direct costs of surgery for presumptive or confirmed endometrial or ovarian cancer were from the Agency for Health Care Research and Quality report on Hospital and Ambulatory Surgery Care for Women's Cancers.24, 25 We did not include the costs of adjuvant radiotherapy or chemotherapy or the costs of treatment for recurrent disease. Indirect healthcare costs included time costs according to the Bureau of Labor Statistics26 based on an estimated 0.5 days for each screen and 6 weeks for surgery and also included travel costs based on automobile expenditures for urban travel using a national average fuel consumption index and gasoline costs in the year 2006.27 Selected data for the base case are presented in Tables 1 through 5.
|Variable||Reference(s)||Base case, %||Range, %|
|Lifetime risk||Aarnio 1999,50 Dunlop 1997,51 Lynch&de la Chapelle 2003,52 Vasen 1996,53 Watson&Lynch 199454||50||30-61|
|Incidental diagnosis at prophylactic surgery||Schmeler 20063||5||2-8|
|Mean age at diagnosis, y||Schmeler 20063||48||46-54|
|Probability of abnormal bleeding||Berek&Hacker 1994,55 Soliman 200456||80||46-90|
|Sensitivity of endometrial biopsy||Dijkhuizen 200057||91||80-95|
|Specificity of endometrial biopsy||Dijkhuizen 200057||98||85-99|
|Stage distribution, reference strategy||Schmeler 2006,3 Broaddus 2006,46 Boks 200258|
|Stage distribution at prophylactic surgery||Schmeler 2006,3 Renkonen-Sinisalo 200733|
|Stage distribution at screening||Renkonen-Sinisalo 200733|
|Stage-specific survival||Boks 2002,58 Cohn 2007,59 Eltabbakh&Moore 1999,60 Kosary 2007,61 Cook 199962|
|Variable||Reference(s)||Base case, %||Range, %|
|Lifetime risk||Watson 2001,4 Vasen 1996,53 Aarino 199563||10||4-18|
|Incidental diagnosis at prophylactic surgery||Schmeler 2006,3 Renkonen-Sinisalo 200733||1.1||0-2|
|Mean age at diagnosis, y||Schmeler 2006,3 Watson 2001,4 Hampel 200564||45||42-54|
|Probability of symptoms*||Attanucci 2004,65 Goff 2007,66 Goff 200467||70||57-80|
|Sensitivity of screening (TVUS, CA 125)||Buys 2005,68 Jacobs 1999,69 Lacey 2006,70 Menon 2005,71 Van Nagell 2000,72 van Nagell 2007,73 Olivier 200674||50||40-85|
|Specificity of screening (TVUS, CA 125)||Menon 200571||99.8||95-99.9|
|Stage distribution, reference strategy||Schmeler 2006,3 Watson 20014|
|Stage distribution at prophylactic surgery||Schmeler 2006,3 Watson 2001,4 Rebbeck 200275|
|Stage distribution at screening||Schmeler 2006,3 Watson 20014|
|Stage-specific survival||Watson&Lynch 200176|
|Lifetime risk of peritoneal carcinoma after prophylactic surgery||Schmeler 2006,3 Finch 2006,77 Piver 199378||0||0-2|
|Perioperative mortality within 30 days||HCUP, 200224||0.1||0.1-1.0|
|Colorectal cancer risks||Reference(s)||Base case, %||Range, %|
|Lifetime risk||Aarnio 1999,50 Dunlop 199751||42||30-54|
|Mean age at diagnosis, y*||Schmeler 2006,3 Lynch&de la Chapelle 2003,52 Hampel 200564||50||45-60|
|Stage distribution||Bertario 1999,79 Watson 1998,80, Jarvien 200081|
|Stage-specific survival||Bertario 1999,79 Barnetson 2006,82 de Vos tot Nederveen Cappel 200483|
|Screening costs||Reference(s)||Cost, U.S. dollars|
|Initial consultation (HCPC 99245)||$212||$100-400|
|Annual office visits (HCPC 99205)||$144||$70-290|
|TVUS (HCPC 76830)*||U.S. DHHS 200723||$103||$180-350|
|CA 125 (HCPC 86304)||MDACC 200784||$22||$20-30|
|Endometrial biopsy (HCPC 58100 and 88305)*||U.S. DHHS 200723||$188||$60-100|
|Patient (per screen)†||U.S. DoL BoLS 200726||$40||$20-60|
|Prophylactic surgery costs|
|Initial consultation (HCPC 99245)||$212||$100-400|
|Professional§||U.S. DHHS 200723||$1035||$1000-5000|
|Patient†||U.S. DoL BoLS 200726||$1442||$1000-2500|
|Cancer surgery costs|
|Endometrial-professional§||U.S. DHHS 200723||$1345||$700-3000|
|Ovarian-professional§||U.S. DHHS 200723||$1520||$750-5000|
|Patient†||U.S. DoL BoLS 200726||$1442||$1000-2500|
|Prophylactic HBSO (premenopausal)||Grann 2002,15 van Roosmalen 200222||0.86||0.82-0.99|
|Quality weights associated with health states|
|Endometrial cancer||Carter 1998,14 Grann 2000,17 Hillner 1986,18 Weinstein&Schiff 1983,20 Sun 200621||0.83||0.68-0.95|
|Ovarian cancer||Grann 2002,15 Grann 1998,16 Ortega 199719||0.75||0.65-0.82|
The average unadjusted life expectancy, quality-adjusted life expectancy, and total discounted costs for each strategy are presented in Table 6. Prophylactic surgery at age 30 years provided the highest unadjusted life expectancy (21.59 years); however, the combined strategy provided the highest quality-adjusted life expectancy (18.98 QALYs). Annual screening alone was dominated, because it was more costly and less effective than prophylactic surgery at age 30 years. Prophylactic surgery at age 40 years was a cost-effective alternative to prophylactic surgery at age 30 years because of its favorable ICER of $5025 per QALY gained. However, the combined strategy was not a cost-effective alternative to prophylactic surgery at age 40 years because of its ICER of $194,650 per QALY gained.
|Strategy||Cost, U.S. dollars||Unadjusted life expectancy, y||Quality-adjusted life expectancy, QALYs||ICER, U.S. dollars/QALY*|
|No prevention (reference)||$13,620||19.5910||18.4582||—|
|PS at age 30 y||$18,523||21.5873||18.8115||$13,877|
|PS at age 40 y||$19,184||20.6444||18.9430||$5025|
|Annual screen starting at age 30 y||$30,912||20.0405||18.6627||Dominated†|
Figure 2 summarizes the results of a sensitivity analysis in which the age at prophylactic surgery in the combined strategy is varied from ages 40 to 50years. The highest net health benefit is achieved when prophylactic surgery occurs before age 42 years. Quality-adjusted life expectancy decreases as long as prophylactic surgery is postponed. Then, we varied the age at the start of screening in the combined strategy to determine whether screening at a later age would decrease the lifetime cost and provide a more favorable ICER relative to prophylactic surgery alone at age 40 years. Although the average lifetime cost decreases when screening is started at a later age (Table 7), the ICER remains well above $50,000 per QALY gained, as illustrated in Figure 3.
|Age at screen, y||Combined strategy (PS at age 40 years)|
|Cost, U.S. dollars||Effectiveness, QALYs|
The results were highly sensitive to preference ratings or utilities for prophylactic surgery. When the utility of prophylactic surgery is decreased from 1.0 (ie, no effect on quality of life) to 0.82, the quality-adjusted life expectancy also decreases, as shown in Table 8. At a utility threshold of 0.88, prophylactic surgery at age 40 years (with or without prior screening) yields a higher net benefit than prophylactic surgery at age 30 years.
|Utility||PS at age 30 years||PS at age 40 years||Combined strategy (PS at age 40 years)|
|0.86 (Base case)||18.8115||18.9430||18.9766|
Results of the Monte-Carlo simulation are presented in Table 9. Because fewer women are dying with endometrial or ovarian cancer after prophylactic surgery at age 30 years, they have the highest number of CRC cases. Women undergoing prophylactic surgery at age 40 years have a higher number of endometrial and ovarian cancer cases than women who are screened for 10 years before prophylactic surgery (in the combined strategy).
|Cancer type||Estimated no. of cancer cases|
|Reference||PS at age 30 years||PS at age 40 years||Annual screening||Combined strategy|
The benefit of a life-saving preventive strategy (ie, prophylactic surgery in Lynch syndrome) is the gain in life expectancy averaged across the target population receiving the intervention, not the gain actually realized by a particular individual. The average gain may appear to be insignificant, but it represents a very large gain for those few individuals who may have died prematurely without the intervention. In the current study, prophylactic surgery yielded an average life-expectancy gain of 2 years compared with the reference strategy. To put this into perspective, the average gain in life expectancy with triennial cervical cancer screening is estimated to be between 0.1 years to 0.25 years (1-3 months) for the population at risk compared with no screening.28, 29 However, the women whose cancers are detected preclinically actually gain an additional 25 years of life expectancy.28 In a decision-analytic model of prophylactic oophorectomy for BRCA mutation carriers, Schrag et al reported a life-expectancy gain of between 0.3 years to 1.7 years (4-20 months) for women ages 30 years to 40 years.30 The magnitude of benefit for prophylactic surgery in a woman age 30 years with Lynch syndrome, thus, is comparable to prophylactic oophorectomy in a woman aged 30 years with a BRCA mutation.
Although a significant gain in life expectancy is expected with prophylactic surgery at age 30 years, the tradeoff is the effect on quality of life caused by loss of fertility and premature menopause. A delay in prophylactic surgery until age 40 years, when childbearing may be complete for most women, yields a greater quality-adjusted life expectancy than prophylactic surgery at age 30 years, but this outcome is highly sensitive to the utility of prophylactic surgery. To our knowledge, there are limited data regarding the utility of prophylactic surgery at age 30 years, but it will vary according to individual preferences for either preserving reproductive function or reducing cancer risk at this age. The utility may be as low as 0.5 for some women in this population.21 The lower the utility of prophylactic surgery at age 30 years, the lower the net health benefit of that intervention.
Annual screening may appear preferable to prophylactic surgery for premenopausal women,21 but there are conflicting results on the effectiveness of this strategy. Both Rijcken et al and Dove-Edwin et al reported that TVUS failed to identify any cases of endometrial cancer that were only diagnosed at the onset of symptoms.31, 32 However, according to a Finnish population-based study, screening with endometrial biopsy identified more cases of early endometrial cancer or hyperplasia compared with symptoms alone.33 Assuming that screening has imperfect specificity, it will lead to surgery on the basis of false-positive results (ie, atypical endometrial hyperplasia in the absence of endometrial cancer on final pathology or elevated CA 125 and pelvic mass in the absence of ovarian cancer) and thereby decrease the total number of cancer cases. Consequently, the highest net health benefit in this population is achieved with a combination of annual screening from age 30 years until prophylactic surgery at age 40 years. Although this already has been recommended as a risk-reducing strategy,5, 9 it is not a cost-effective alternative to prophylactic surgery alone at age 40 years. The incremental benefit of screening for 10 years before prophylactic surgery is attained at significant cost, and, to decrease the ICER, the sensitivity of screening must be improved while its lifetime costs are reduced. Identification of novel molecular markers and decreased frequency of screening likely are needed to achieve these outcomes.
To our knowledge, there are no other data available concerning other nonsurgical strategies to reduce gynecologic cancer risks in this population, such as oral contraceptives (OCPs) or progestin intrauterine devices. OCPs decrease endometrial and ovarian cancer risks in the general population by approximately 50%,34-43 and progestin intrauterine devices may reverse hyperplasia and early endometrial cancer.44, 45 Because Lynch syndrome-associated gynecologic cancers appear to be similar to sporadic cases,4, 46 these chemoprevention strategies may be beneficial with less adverse effect on quality of life than prophylactic surgery. Our group recently completed accrual to an endometrial cancer chemoprevention study comparing OCPs and Depo-Provera for women with Lynch syndrome, and data analysis currently is ongoing. The Prevention of Endometrial Tumors (POET) study recently was initiated through the Medical Research Council in the U.K. to compare the Mirena Intrauterine System with screening to reduce endometrial hyperplasia and cancer risks in Lynch syndrome.47 Once data are available on chemoprevention strategies in this high-risk population, it will be important to refine the model parameters and update the analysis.
The current study contributes to existing literature indicating that prophylactic surgery is an effective gynecologic cancer risk-reducing strategy for women with Lynch syndrome. With respect to CRC risk-reducing strategies, Syngal et al reported that colonoscopic surveillance provided the greatest quality-adjusted life expectancy compared with prophylactic proctocolectomy or subtotal colectomy,48 and Vasen et al also reported that it was cost-effective for this population.49 Colonoscopic surveillance and a combined strategy of gynecologic cancer screening followed by prophylactic HBSO appear to provide the highest net benefit compared with alternate CRC and gynecologic cancer prevention strategies, respectively. Intuitively, the combination of these interventions should be the optimal prevention strategy for women with Lynch syndrome. An important caveat of the current study is the limited availability of empiric data on screening and patient preferences in this population, which imposes various assumptions on the model. Prophylactic surgery at age 30 years prolongs life expectancy but at the expense of quality of life. The decision to proceed with prophylactic surgery likely will be influenced by an individual's perception of risk and preference for various health states, including premature menopause and cancer. Screening before prophylactic surgery provides additional benefit although at substantial cost, which underscores the need to improve the effectiveness and decrease the lifetime costs of this intervention.
- 12United States life tables, 2003. Natl Vital Stat Rep. 2006; 54: 1–40..
- 21Preferences for cancer prevention strategies (CPS) in women with hereditary nonpolyposis colorectal cancer (HNPCC) [abstract]. J Clin Oncol. 2006; 24(18S suppl June 20 ): 1018. Abstract 1018..
- 23U.S. Department of Health and Human Services Physician Fee Schedule. Baltimore, Md: Centers for Medicare and Medicaid Services. Available at: http://www.cms.hhs.gov/PhysicianFeeSched/01_overview.asp. Accessed June 10, 2007.
- 24Health Care Utilization Project (HCUP). Statistics on Stays in U.S. Hospitals, Principal Diagnosis. Washington, DC: Agency for Healthcare Research and Quality, Department of Health and Human Services; 2002.
- 25Health Care Utilization Project (HCUP). Hospital and Ambulatory Surgery Care for Women's Cancers. Washington, DC: Agency for Healthcare Research and Quality, U.S. Department of Health and Human Services. Available at: http://www.ahrq.gov/data/hcup/highlight2/highlight2.pdf. Accessed June 10, 2007.
- 26U.S. Department of Labor, Bureau of Labor Statistics. Overview of BLS Statistics on Wages, Earnings, and Benefits. Washington, DC: U.S. Bureau of Labor Statistics; 2007.
- 27Energy Information Administration. A Primer on Gasoline Prices. Washington, DC: Energy Information Administration U.S. Department of Energy; 2005.
- 47National Cancer Research Institute/Medical Research Council. POET study (Prevention of Endometrial Tumors). 2006. Available at URL: http://www.londonideas.org/internet/professionals/documents/ShirleyHodgson.pdf
- 55Practical Gynecologic Oncology. 2nd ed. Philadelphia, Pa: Lippincott, Williams & Wilkins; 1994., .
- 61Cancer of the corpus uteri. In: RiesLAG,YoungJL,KeelGE,EisnerMP,LinYD,HornerM-J, eds. SEER Survival Monograph: Cancer Survival Among Adults: U.S. SEER Program, 1998–2001, Patient and Tumor Characteristics. NIH Pub. No. 07-6215. Bethesda, Md: National Cancer Institute, SEER Program; 2007: 123–132..
- 84M. D. Anderson Cancer Center. Charges and Costs Associated With Gynecologic Cancer Screening. Houston, Tex: M. D. Anderson Cancer Center; 2007.
- 85Agency for Healthcare Research and Quality. Statistics forU.S. community hospital stays, principal procedure based on CCS (Clinical Classifications Software), 2004. Available at: http://hcupnet.ahrq.gov/HCUPnet.jsp?Id=A9364FA8859E427&Form=SelDXR&JS=Y&Action=%3E%3ENext%3E%3E&_DXPR=PreRunPCCHPR1. Accessed June 10, 2007.