Introduction

Whole-genome sequencing (WGS) tests are being offered selectively in clinical care and are expected to become more widely used in the future. The American College of Medical Genetics and Genomics (ACMG) released recommendations that specify which secondary findings (SFs) from WGS should be returned to patients.1 These recommendations identify 56 genes associated with 24 conditions in which mutations are considered clinically actionable (e.g., treatment or behavior modification that leads to improved health outcomes) (see the definitions in Box 1 ).

The ACMG made the recommendation to return certain SFs based on an established benefit of clinical utility, and all of the genes included have clinically available tests.1,2,3,4 However, when the full implications of finding and reporting SFs found with WGS are assessed, the full range of benefits, risks, and costs must be considered. The cost-effectiveness of returning SFs from WGS in the general population or in specific targeted/high-risk population clinical scenarios has not been assessed. Our aim was to identify published economic evaluations of detecting mutations in the 56 genes associated with the 24 conditions that are considered clinically actionable by the ACMG among the general population or in targeted/high-risk populations to identify available economic evidence and gaps in knowledge. Our results inform the costs/benefits of genetic testing for these conditions in both targeted populations and general populations, which may occur as WGS enters clinical care more broadly.

Materials and Methods

Search strategy

We identified economic evaluations for screening for all conditions and respective genes defined on the ACMG’s list of returnable results.1 We conducted a targeted search of PubMed for English-language articles of cost-effectiveness, cost–utility, or cost–benefit analyses that specifically addressed the conditions and genes of interest and published from 1994 through November 2014. We limited the search to the past 20 years because the majority of all economic analysis of genetic testing has been done during this time frame, and we used Medical Subject Headings (MeSHs) and keywords to limit the studies to economic analyses and the disease or condition found in the ACMG’s list of returnable results.1 The search strategy used the following terms:

  • Disease or condition—MeSH terms and key words for each. The MeSH term was limited further using the genetics or diagnosis subheading for selected conditions (see Supplementary Appendix A online for specific terms used).

  • Cost-effectiveness—“Cost–Benefit Analysis” (MeSH) and “Cost-effectiveness” (key word)

For conditions (e.g., Lynch syndrome and hereditary breast ovarian cancer syndrome) that returned a large number of results, we limited the search by adding the specific terms for the relevant gene(s), using MeSH terms, when available, or key words (e.g., MSH1, BRCA1, or BRCA2). We combined the results of these searches and validated our search, as described below. We augmented our search strategies for economic evaluations related to genetic screening for the disease/conditions and genes on the ACMG list by reviewing references in the Tufts Center for the Evaluation of Value and Risk in Health Cost-Effectiveness Analysis Registry (CEAR) database5,6 and in recently published cost-effectiveness and cost–utility studies.7

Article selection

Titles and abstracts were screened by two independent reviewers according to prespecified inclusion and exclusion criteria, and the full text of relevant articles were retrieved. Disagreements were resolved through discussion and referral to a third reviewer, if necessary. The article inclusion criteria were original articles only, focus on one of the conditions and genes listed in the ACMG list, English language only, provide a measure of relative economic value defined as a cost-effectiveness or cost-to-utility ratio or net benefits/benefit-to-cost ratio, and an economic evaluation screening for the gene, not a treatment. Article exclusion criteria were reviews, editorials, or methods articles and studies solely of “costs.” We screened all full-text articles to identify those that included at least one analysis of general population screening (or opportunistic screening of a general population) and those that included only analyses of targeted/high-risk population screening.

Data extraction

For each included study, two reviewers independently extracted relevant data and study details. Basic article details including population demographics and cost-effectiveness data—specific versus general population studied, incremental cost-effectiveness ratios (ICERs), and author conclusions—were abstracted (see Supplementary Appendix C online). Discrepancies were resolved through discussion.

We defined “cost-effective” as an ICER ≤$100,000 per life-year or quality-adjusted life-year (QALY) gained.8,9 Using this definition, we determined whether at least one clinical scenario per targeted population study was “cost-effective” for screening for the condition of interest. We assumed testing in a general population scenario to mean the population tested was not limited by other risk factors (e.g., increased risk of the condition because of family history) and a targeted population scenario to mean the population tested was limited by outside factors (e.g., family history of the condition, other clinical factors suggestive of the condition).

Evaluation of methodological quality

We used the “quality rating” assigned by trained coders from the Tufts CEAR to examine the methodological quality of our included studies that were found in Tufts CEAR.5

Results

Studies identified

Our search yielded 607 candidate articles, of which 56 remained after title/abstract review, and 32 remained after full-text review10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41 ( Figure 1 ) (a complete list of studies can be found in Supplementary Appendix B online).

Figure 1
figure 1

Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) diagram of included and excluded studies.

Full size image

Populations, conditions, and genes examined

We found cost-effectiveness analyses (CEAs) for only a fraction (7 of 24; 29%) of all conditions and genes on the ACMG list ( Table 1 ). Only 4 of 32 studies contained analyses for genetic screening in a general population: familial hypercholesterolemia and Lynch syndrome (2 of 24 conditions on the ACMG list, or 8%) ( Table 1 ), whereas all studies included analyses of high-risk or targeted populations ( Table 1 ). Most articles (n = 14) focused on Lynch syndrome (genes MLH1, MSH2, MSH6, and PMS2), hereditary breast ovarian cancer syndrome (genes BRCA1 and BRCA2; n = 6), and familial hypercholesterolemia (genes LDLR, APOB, and PCSK9; (n = 6) ( Table 1 ). The remaining studies (n = 6) addressed MYH-associated polyposis (gene MUTYH) multiple endocrine neoplasia type 2 (gene RET), hypertrophic cardiomyopathy and dilated cardiomyopathy (genes MYBPC3, MYH7, TNNT2, TNNI3, TPM1, MYL3, ACTC1, PRKAG2, GLA, MYL2, and LMNA), and Romano-Ward long QT syndrome types 1, 2, and 3 and Brugada syndrome (genes KCNQ1, KCNH2, and SCN5A) ( Table 1 ).

Table 1 Summary of articles found and cost-effectiveness analyses (CEAs) identified for each condition defined on the American College of Medical Genetics and Genomics list of secondary findings
Full size table

Cost-effectiveness of screening in the general population

The four CEAs that examined general population screening included several different analyses for familial hypercholesterolemia (six analyses) and Lynch syndrome (three analyses examining no risk stratification before genetic testing)11,22,23,40 ( Table 2 ).

Table 2 Details of economic evaluations of the general population
Full size table

Two analyses by Marks et al.22,23 found screening for familial hypercholesterolemia at 16 years of age to be cost-effective in the general population (i.e., mutation prevalence was assumed to reflect the prevalence among the general population) ( Table 2 ). Furthermore, these same studies found screening for familial hypercholesterolemia between the ages of 16 and 55 in the general population, or opportunistically in primary care, borderline cost-effective ($104,502–$125,200/QALY gained) ( Table 2 ).22,23

Both Dinh et al.11 and Snowsill et al.40 found that general population screening with up-front germ-line testing for Lynch syndrome was not cost-effective (ICER values ranged from nearly $129,852 to >$7 million per life-year or QALY gained) ( Table 2 ). However, Dinh et al. found that when the risk thresholds for genetic testing were set to 5 and 10% (a targeted/high-risk population evaluation), the ICERs fell below $50,000 per QALY gained.

Cost-effectiveness of screening in targeted/high-risk populations

For seven conditions, the cost-effectiveness of genetic testing was examined in at least one high-risk or targeted clinical scenario. All of these CEAs found genetic testing to be cost-effective in at least one high-risk or targeted clinical scenario ( Table 1 ; Supplementary Appendix B online).

Evaluation of methodological quality

Although sample sizes were too small to examine in detail the quality of the methods used, we were able to estimate the methodological quality of seven of the included studies using data from the Tufts CEAR. We calculated a mean quality score of 4.5 for the included studies (range: 3.5–6; scale of 1–7), which was equivalent to the mean score (4.5) for all studies in the CEAR (N = 4,007).

Discussion

We found very few (7 of 24; 29%) published health economic evaluations of screening for clinically actionable gene variants among the conditions and genes for which the ACMG recommends returning secondary findings to patients. For the conditions and genes that have been formally evaluated, most studies have focused on targeted/high-risk populations or populations with a high prevalence (i.e., <1:1,500) ( Table 1 ). Our findings suggest that substantial additional data and analyses are required to evaluate fully the benefits, risks, and costs of return of these secondary findings to patients.

The few CEAs that addressed testing for specific genes in the general population suggest that such testing in low-prevalence populations may not be cost-effective ( Table 2 ). With WGS, however, it must be appreciated that a relatively low testing cost may be incurred to obtain information for multiple conditions. If most of the cost incurred from screening is from the test itself rather than the follow-up treatment and health-care utilization, then WGS-enabled screening may improve cost-effectiveness. To illustrate, several models have found that screening for familial hypercholesterolemia in some general populations (i.e., at 16 years), but not in others, may be cost-effective.22,23 Testing for this condition in high-risk populations (family members of persons with clinical diagnosis or genetic mutations) using a case-finding/cascade screening approach usually represents the most cost-effective (lowest ICER) application of the screening technology.22,23,25,34,37,41 Furthermore, in the case of general population screening for familial hypercholesterolemia, the clinical utility of the genetic test compared with a standard cholesterol test to evaluate an individual patient’s possible need for lipid-lowering drugs (e.g., statins) may be the more cost-effective approach considering the clinical care pathway would certainly require a cholesterol test before statin use anyway. In Lynch syndrome models demonstrate that general population screening of non-risk-stratified individuals with traditional gene sequencing methods is not cost-effective.11,40 When a risk threshold for genetic testing was set to 5 and 10% (nonuniversal screening), the ICER values fell below the common benchmark threshold of $50,000 per QALY gained.11 We found no CEAs of BRCA1/2 screening in the general population.

In all of the CEAs included we found at least one high-risk or targeted population screening scenario in which screening was suggested to be cost-effective.10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41 Whether substituting WGS (with its yield of SFs) instead of condition-specific targeted testing in these scenarios would improve clinical effectiveness and/or cost-effectiveness remains unclear. However, it is important to consider that lowering WGS costs may affect the cost-effectiveness of using single-gene tests. For example, three of the four general population studies included in our study included sensitivity analyses of the test cost, and in all three cases the cost-effectiveness was relatively more sensitive to test cost compared with other variables.11,22,40 In one case a 50% decrease in test cost resulted in a 35% reduction of the ICER value for universal screening ($14,400 vs. $22,154 cost/life year gained).22 Furthermore, the proportion of the total cost per life year gained that was attributable to the test cost ranged from 0.7 to 6.7% ($3,495/$490,31511; $2,500/$129,85240; $1,492/$22,15422; and $1,603/23,805 (ref. 23)).

Limitations

We searched only PubMed and thus we may have missed articles not indexed in PubMed, but we expect that this would be rare, and we supplemented our search findings with other databases and key articles. Because we found few relevant CEAs addressing general population screening, these results should be considered only suggestive of what might be found with a larger sample. The individual analyses that we identified do not directly address the cost-effectiveness of WGS, and sequencing of multiple genes at once will have a different range of effects and a different ICER than screening for only one condition.

Conclusions and future research

It would be useful to have more economic analyses of the conditions and genes on the ACMG list among the general population or high-risk populations. CEAs that examine WGS in populations at high risk for specific conditions, in which the primary aim is to examine the genes associated with this condition but where SFs in other genes also are reported to patients, would be highly informative. Comparisons of WGS versus existing methods, including the use of gene panels, would be of great interest. Furthermore, one must consider how to integrate the economic evaluation of multiple or all of these conditions and the impact of jointly considering these genes/conditions on the cost-effectiveness of WGS.42,43 For example, the use of existing CEA results (e.g., QALYs, ICERs) in the development of future CEA models for multiple conditions may not be appropriate data inputs. It is possible that WGS may be more cost-effective than screening for individual conditions/genes because one is screening for multiple diseases. However, WGS may result in a higher number of false-positives that generate further tests/costs, thus negatively affecting the cost-effectiveness.

In summary, the cost-effectiveness of screening for most of the conditions and genes on the ACGM actionable list remains undefined. Furthermore, the clinical and economic consequences that may follow ascertainment of findings for conditions and genes on the ACGM list have not been explored. When considering the full range of effects that WGS may have in the general population, it will be necessary to integrate the clinical and economic effects stemming from identifying clinically actionable findings, findings with unclear implications for management, and findings of unknown significance across a large number of clinical conditions and associated genes.

Disclosure

K.A.P. has received honoraria for serving on a scientific advisory panel for Illumina and for a speaking engagement for Life Technologies. Disclosures have been reviewed by the University of California, San Francisco. D.A.M. receives support as an ad hoc consultant in health economics for Optum Insight. The other authors declare no conflict of interest.