This cost-utility analysis reports on the effect of quality of life on the value of screening all new patients with colorectal cancer for Lynch Syndrome.
Published Online: May 09, 2012
Grace Wang, PhD, MPH; Miriam Kuppermann, PhD, MPH; Benjamin Kim, MD, PhD; Kathryn A. Phillips, PhD; and Uri Ladabaum, MD, MS
Purpose: Patients and relatives have varying preferences for genetic testing and interventions related to hereditary cancer syndromes. We examined how the impact of these services on quality of life (QoL) affects the cost-effectiveness of screening for Lynch syndrome among probands newly diagnosed with colorectal cancer and their relatives.
Methods: We constructed a state-transition model comparing screening strategies (clinical criteria, prediction algorithms, tumor testing, and upfront germline testing) with no screening to identify Lynch syndrome. The model incorporated individuals’ health state utilities after screening, germline testing, and risk-reducing surgeries,
with utilities persisting for 12 months in the base case. Outcomes consisted of quality-adjusted lifeyears (QALYs), costs, and cost per QALY gained. Sensitivity analyses assessed how the duration and magnitude of changes in QoL influenced results.
Results: Multiple screening strategies yielded gains in QALYs at acceptable costs compared with no screening. The preferred strategy—immunohistochemistry
of tumors followed by BRAF mutation testing (IHC/BRAF)—cost $59,700 per QALY gained in the base case. The duration and magnitude of decreases in QoL after decisions
related to germline testing and surgeries were key determinants of the cost-effectiveness of screening. IHC/BRAF cost >$100,000 per QALY gained when decrements to QoL persisted for 21 months.
Conclusion: Screening for Lynch syndrome in the population is likely to yield long-term gains in life expectancy that outweigh any short-term decreases in QoL, at acceptable costs. Counseling for individuals should aim to mitigate potential negative impact of genetic testing and risk-reducing interventions on QoL.
(Am J Manag Care. 2012;18(5 Spec No. 2):e179-e185)
Lynch syndrome increases the risk of developing colorectal, endometrial, ovarian, and other cancers. Inherited mutations in DNA mismatch repair genes MLH1, MSH2, MSH6, and PMS2 constitute the molecular basis of Lynch syndrome. Stakeholders have recommended screening for Lynch syndrome using clinical criteria, mutation risk-prediction algorithms, and tumor testing.1,2
Screening is likely to decrease cancer incidence and improve life expectancy in affected families at acceptable costs to payers.3-6 The favorable impact is attributed to cancer risk stratification based on germline testing, followed by preventive interventions including colonoscopy and risk-reducing total abdominal hysterectomy/oophorectomy.7-9
Economic evaluations have not considered the potential effects of these services on quality of life (QoL) among patients with colorectal cancer with Lynch syndrome (probands) and their relatives. The interplay of preferences for screening, germline testing, and preventive care for probands and relatives at risk for developing cancer likely determines the impact of these services on QoL. For example, some may not value germline testing because identifying increased cancer risk may cause distress.10 Others may value total abdominal hysterectomy/oophorectomy highly because it decreases cancer risk and possibly worry.11 Probands’desires and decisions about germline testing affect relatives, but preferences among relatives may differ from those of probands.
Our aim was to examine how changes in individuals’ QoL resulting from medical services for Lynch syndrome affect the cost-effectiveness of screening for Lynch syndrome among persons newly diagnosed with colorectal cancer and their relatives. We performed a cost-utility analysis (CUA) with a decision analytic model using health state utilities from our recent study of patient preferences related to Lynch syndrome to estimate gains in quality-adjusted life-years (QALYs) and incremental cost per QALY gained with screening.4
CUA evaluates the effects of healthcare services in terms of both costs and QALYs gained (not just crude number of life-years gained). We refer readers to the Data Supplement for additional information about CUA and our study methods. The statistical code and data set are available by written agreement.
We adapted a state-transition Markov model of screening for Lynch syndrome, described in detail previously, to incorporate health state utilities related to germline testing and risk-reducing interventions.4 The model compares 16 screening strategies to a referent strategy of no active effort to identify Lynch syndrome. Persons newly diagnosed with colorectal cancer enter the simulation at a mean age of 48 years in the base case, and relatives enter at a mean age of 25 years.
We compared different strategies for identifying Lynch syndrome among persons with new diagnoses of colorectal cancer. The strategies included clinical criteria, such as Amsterdam II and revised Bethesda guidelines,12,13 which require an extensive family history of Lynch-associated cancer (colorectal, endometrial, or other associated cancer) and consider age of cancer onset. We included risk-prediction algorithms, such as MMRpro, PREMM126, and MMRpredict,14-17 which estimate the probability of having a genetic mutation based on factors like family history of cancer, tumor characteristics, histology, and age at diagnosis among patients and relatives. We also examined several tumor-testing strategies (eg, microsatellite instability testing, immunohistochemistry [IHC], microsatellite instability testing combined with IHC, and IHC followed by BRAF testing).18-22 Testing a tumor sample for microsatellite instability can identify impairment in the DNA replication and repair system, which may result from mutations in mismatch repair (MMR) genes. IHC determines MMR protein expression in tumor samples by applying antibodies against the MMR proteins. Loss of IHC expression of specific MMR proteins guides germline testing. For example, IHC showing loss of expression of MSH2 but not MLH1, MSH6, or PMS2 would suggest germline testing of only the MSH2 gene. Combining BRAF testing with IHC identifies sporadic loss of MLH1 protein expression that is not related to mutations in the MMR genes but may result from hypermethylation of the MLH1 promoter. Finally, germline testing of a blood sample can identify inherited mutations in MMR genes through sequencing, deletion or duplication analysis, or rearrangementanalysis. When a specific mutation is identified in a proband, subsequent germline testing can focus on identifying the same mutation in relatives. We modeled different clinical management programs and acceptance rates among probands and relatives until age 75 years based on their germline testing results and cancer risk.23
Major events were first colorectal, endometrial, or ovarian cancer; metachronous colorectal cancer; cancer treatment complications; and death resulting from cancer or other causes. We accounted for the more favorable cancer prognosis associated with Lynch syndrome versus sporadic cancers.24 Persons were observed until death or age 100 years.
Health State Utilities
Health state utilities (herein referred to as utilities) represent the strength of an individual’s preferences for specific health-related outcomes and can be used as preference weights to make QoL adjustments to the number of life-years gained after a health intervention. Our model includes utilities that reflect how views about germline testing and management of Lynch syndrome affect QoL among probands and relatives (Table 1).25 Utilities related to Lynch syndrome and living with cancer were measured in our recent study using preferenceelicitation exercises and the time tradeoff metric (Appendix).
We applied utilities related to Lynch syndrome for 12 months in the base case and assumed that decrements to QoL resulting from Lynch syndrome-related services were transient. We chose 12 months based on literature and to avoid bias in favor of screening.10,26-34 Furthermore, studies of hereditary cancer syndromes have similarly applied utilities in a timelimited fashion.35,36 To test how our assumption affected costeffectiveness, we varied time in health states related to Lynch syndrome, testing from 0 to 36 months in sensitivity analyses.
After living for 12 months in health states related to Lynch syndrome, utilities for both probands and relatives reverted to those of the general population, as measured and adjusted for age and sex starting at age 45 years by Fryback et al.37 We assigned a utility of 1 (representing perfect health) to relatives entering the simulation at age 25 years, as suggested by Kwon et al,36 and interpolated values through age 44 years based on values reported for age 45 years by Fryback et al.
We accounted for changes in QoL for individuals living with colorectal, endometrial, and ovarian cancers based on findings from our utilities study (Kuppermann et al, manuscript submitted for publication). Utilities for persons living with cancer were applied for 5 years; they then reverted to general population utilities.
Clinical and Cost Inputs
We derived clinical inputs from an Evaluation of Genomic Applications in Practice and Prevention meta-analysis.38 Surveillance, Epidemiology, and End Results data informed cancer risk estimates.4,39
We adjusted costs from published sources and Medicare schedules to 2010 US dollars using the medical component of the Consumer Price Index.40 Costs reflected all direct expenses associated with screening, germline testing and genetic counseling, preventive interventions, complications, and cancer care (Appendix).
Outcomes and Cost-Utility Analyses
Applying a third-party payer perspective and an annual 3% discount rate, we calculated primary outcomes of mean QALYs per person, mean cost per person, incremental QALYs gained, and incremental cost per QALY gained (ie, incremental cost-effectiveness ratio [ICER]). Analyses focused on families with a representative number of at-risk relatives, and results reflect weighted averages for probands and relatives with and without Lynch syndrome.4
To examine effects of changes in QoL associated with germline testing and management decisions on outcomes, we focused on the tumor testing strategy of IHC followed by BRAF mutation testing. Both our previous cost-effectiveness analysis and this current cost-utility analysis found this to be the preferred strategy.4 This strategy also reflects the current movement toward reflexive screening of all colon cancer samples for Lynch syndrome.2,5,41
PDF is available on the last page.