A systematic review and economic evaluation of diagnostic strategies for Lynch syndrome

Background Lynch syndrome (LS) is an inherited autosomal dominant disorder characterised by an increased risk of colorectal cancer (CRC) and other cancers, and caused by mutations in the deoxyribonucleic acid (DNA) mismatch repair genes. Objective To evaluate the accuracy and cost-effectiveness of strategies to identify LS in newly diagnosed early-onset CRC patients (aged < 50 years). Cascade testing of relatives is employed in all strategies for individuals in whom LS is identified. Data sources and methods Systematic reviews were conducted of the test accuracy of microsatellite instability (MSI) testing or immunohistochemistry (IHC) in individuals with CRC at risk of LS, and of economic evidence relating to diagnostic strategies for LS. Reviews were carried out in April 2012 (test accuracy); and in February 2012, repeated in February 2013 (economic evaluations). Databases searched included MEDLINE (1946 to April week 3, 2012), EMBASE (1980 to week 17, 2012) and Web of Science (inception to 30 April 2012), and risk of bias for test accuracy was assessed using the Quality Assessment of Diagnostic Accuracy Studies-2 (QUADAS-2) quality appraisal tool. A de novo economic model of diagnostic strategies for LS was developed. Results Inconsistencies in study designs precluded pooling of diagnostic test accuracy results from a previous systematic review and nine subsequent primary studies. These were of mixed quality, with significant methodological concerns identified for most. IHC and MSI can both play a part in diagnosing LS but neither is gold standard. No UK studies evaluated the cost-effectiveness of diagnosing and managing LS, although studies from other countries generally found some strategies to be cost-effective compared with no testing. The de novo model demonstrated that all strategies were cost-effective compared with no testing at a threshold of £20,000 per quality-adjusted life-year (QALY), with the most cost-effective strategy utilising MSI and BRAF testing [incremental cost-effectiveness ratio (ICER) = £5491 per QALY]. The maximum health benefit to the population of interest would be obtained using universal germline testing, but this would not be a cost-effective use of NHS resources compared with the next best strategy. When the age limit was raised from 50 to 60 and 70 years, the ICERs compared with no testing increased but remained below £20,000 per QALY (except for universal germline testing with an age limit of 70 years). The total net health benefit increased with the age limit as more individuals with LS were identified. Uncertainty was evaluated through univariate sensitivity analyses, which suggested that the parameters substantially affecting cost-effectiveness: were the risk of CRC for individuals with LS; the average number of relatives identified per index patient; the effectiveness of colonoscopy in preventing metachronous CRC; the cost of colonoscopy; the duration of the psychological impact of genetic testing on health-related quality of life (HRQoL); and the impact of prophylactic hysterectomy and bilateral salpingo-oophorectomy on HRQoL (this had the potential to make all testing strategies more expensive and less effective than no testing). Limitations The absence of high-quality data for the impact of prophylactic gynaecological surgery and the psychological impact of genetic testing on HRQoL is an acknowledged limitation. Conclusions Results suggest that reflex testing for LS in newly diagnosed CRC patients aged < 50 years is cost-effective. Such testing may also be cost-effective in newly diagnosed CRC patients aged < 60 or < 70 years. Results are subject to uncertainty due to a number of parameters, for some of which good estimates were not identified. We recommend future research to estimate the cost-effectiveness of testing for LS in individuals with newly diagnosed endometrial or ovarian cancer, and the inclusion of aspirin chemoprevention. Further research is required to accurately estimate the impact of interventions on HRQoL