Use of Genomic Profiling to Assess Risk for Cardiovascular Disease and Identify Individualized Prevention Strategies—A Targeted Evidence-Based Review Glenn E

REVIEW

Use of genomic profiling to assess risk for cardiovascular disease and identify individualized prevention strategies—A targeted evidence-based review Glenn E. Palomaki, BS1, Stephanie Melillo, MPH2,3, Louis Neveux, BA1, Michael P. Douglas, MS2,3, W. David Dotson, PhD2, A. Cecile J. W. Janssens, PhD4, Elizabeth A. Balkite, MS, CGC5, and Linda A. Bradley, PhD1,2

Purpose: To address the key question of whether using available “car- INTRODUCTION diogenomic profiles” leads to improved health outcomes (e.g., reduction in CVD is a major contributor to morbidity and mortality in the rates of myocardial infarction and stroke) and whether these profiles help in United States. An estimated 80 million adults have one or more making medical or personal decisions. Methods: A targeted evidence- types of CVD (48% occurring in individuals aged 60 years or based review based on published Evaluation of Genomic Applications in older), and preliminary 2006 mortality data indicate that CVD Practice and Prevention methodologies. Results: No study addressed the accounts for one in every 2.9 deaths.1 The 2005 overall death overarching question directly. Evidence for the analytic validity of genomic rate from CVD was 279 per 100,000, with death rates higher in profiles was inadequate for most genes (scale: convincing, adequate, and men than women, and in blacks than whites. Consequently, the inadequate), but based on gray data, the analytic sensitivity and specificity might be adequate. For the 29 candidate genes (58 separate associations burden of CVD is high, and the cost (direct and indirect) in 2008 2 reviewed), the credibility of evidence for clinical validity was weak (34 is estimated at 448.5 billion dollars. Prevention and manage- associations) to moderate (23 associations), based on limited evidence, ment of CVD, particularly ischemic heart disease and stroke, 3 potential biases, and/or variability between included studies. The associa- present a difficult challenge for health care and public health. tion of 9p21 variants with heart disease had strong credibility with odds Major nonmodifiable risk factors include increasing age, male ratios of 0.80 (95% confidence interval: 0.77–0.82) and 1.25 (95% confi- gender, and heredity, whereas modifiable risk factors include dence interval: 1.21–1.30), respectively, for individuals with no, or two, smoking, hypertension, dyslipidemia, obesity, physical inactiv- 4–6 at-risk alleles versus those with one at-risk allele. Using a multiplicative ity, and diabetes. African Americans have more severe hy- model, we combined information from 24 markers predicting heart disease pertension and increased risk of heart disease compared with 7,8 and from 13 markers for stroke. The areas under the curves (64.7% and whites. In men, the average annual rate of initial cardiovas- 55.2%, respectively), and overall screening performance (detection rates of cular events increases from three per 1000 at 35–44 years to 74 24% and 14% at a 10% false-positive rate, respectively) do not warrant use per 1000 at 85–94 years. Similar increases occur in women but as stand-alone tests. Conclusion: Even if genomic markers were indepen- approximately a decade later in life.2 dent of traditional risk factors, reports indicate that cardiovascular disease The term “CVD” encompasses a broad range of disorders of risk reclassification would be small. Improvement in health could occur the heart and circulatory system, and classification of CVDs is with earlier initiation or higher adherence to medical or behavioral inter- not uniformly applied in the literature. We defined two main ventions, but no prospective studies documented such improvements (clin- groups of outcomes based on the International Statistical Clas- ical utility). Genet Med 2010:12(12):772–784. sification of Diseases and Related Health Problems (I00-I99).9 The grouping of CHD includes coronary artery disease, isch- Key Words: evidence review, genomic profiles, cardiovascular dis- emic heart disease, and MI. The second grouping of stroke ease, analytic validity, clinical validity, clinical utility includes intracerebral and subarachnoid hemorrhage, ischemic stroke, and other diseases (e.g., cerebral infarction and occlu- sion/stenosis of cerebral arteries). In addition to the conven- From the 1Women & Infants Hospital/Alpert Medical School of Brown tional risk factors, biomarkers have also been identified for use 2 University, Providence, Rhode Island; Office of Public Health Genomics, in potentially improving the prediction of CVD risk. In one Centers for Disease Control and Prevention, Atlanta, Georgia; 3McKing Consulting Corporation, Atlanta, Georgia; 4Department of Public Health, study, 10 biomarkers (e.g., C-reactive protein, fibrinogen and Erasmus MC University Medical Center, Rotterdam, The Netherlands; and homocysteine) provided only modest additive contribution to 5ABQ Associates, Inc. Durham, North Carolina. risk prediction compared with the use of conventional risk 10 Glenn E. Palomaki, BS, Women & Infants Hospital, 70 Elm Street, 2nd factors alone. Floor, Providence, RI 02903. E-mail: [email protected]. Over the last decade, candidate genes for CVD susceptibility Disclaimer: The findings and conclusions in this report are those of the have been the focus of many relatively small studies that in- authors and do not necessarily represent the official position of the Centers clude one or a few genes at a time.11 These candidate gene- for Disease Control and Prevention. disease association studies often have lacked power, and the Disclosure: The authors declare no conflict of interest. reproducibility in subsequent confirmatory studies has generally been poor. More recently, genome-wide association (GWA) Supplemental digital content is available for this article. Direct URL citations appear in the printed text and are provided in the HTML and PDF versions of studies have tested tens of thousands of single-nucleotide poly- this article on the journal’s Web site (www.geneticsinmedicine.org). morphisms (SNPs) in an effort to identify associations with 11–13 14 Submitted for publication April 27, 2010. CVD. Despite notable gaps in knowledge, genomic profiling tests for CVD (sometimes referred to as “heart health”) Accepted for publication August 10, 2010. continue to be offered in the health care market and can be Published online ahead of print November 11, 2010. ordered online without the involvement of a physician. The goal DOI: 10.1097/GIM.0b013e3181f8728d of this evidence review is to assess what is known about the

772 Genetics IN Medicine • Volume 12, Number 12, December 2010 Genetics IN Medicine • Volume 12, Number 12, December 2010 Genomic profiling for risk of CVD analytic validity, clinical validity, and clinical utility of these cardiogenomic profiling tests. Table 1 Key questions relating to the analytic framework A recent HuGE Navigator search (Phenopedia, using the 1. Does the use of “cardiogenomic profiling” lead to improved disease term “cardiovascular diseases”) identified 6493 publi- outcomes for the patient/consumer, or is it useful in medical or cations, 3331 genes, 169 meta-analyses, and 60 GWA study personal decision making? (Overarching question) publications. This targeted review addresses only those genes included on genomic profiles aimed at CVD or “heart health” 2. What is known about the analytic validity of tests that identify variations in genes associated with “heart” or cardiovascular that were available when the study was undertaken in mid 2008. health, including the analytic sensitivity and specificity, assay robustness (e.g., failure rates and resistance to changes in variables such as sample quality), and other factors? Clinical scenario 3. What is the clinical validity of cardiogenomic profiles, including The clinical scenario focuses on individuals in the general clinical sensitivity and specificity and positive and negative population being offered “cardiogenomic profiling” or “heart predictive value? health” tests. In general, these individuals will not have an existing diagnosis of a specific CVD event (e.g., stroke and MI) a. What is the strength of association of cardiovascular health but may have intermediate findings (e.g., hyperlipidemia). The outcomes with the presence of specific gene variants (e.g., odds proposed clinical utility for testing is that the knowledge of a ratios)? personal gene variant(s) associated with an increase in risk for b. How well does this testing alone predict specific cardiovascular CVD might improve health outcomes (e.g., morbidity and mor- outcomes (e.g., MI and stroke)? tality) by (1) providing motivation beyond routine health mes- c. How well does this testing in combination with other CVD sages to change health behaviors; (2) bringing potential risk to testing (e.g., cholesterol) predict specific cardiovascular the attention of health care providers for clinical follow-up; or outcomes? (3) making or revising treatment decisions. The evidence review was commissioned and will be used by the Evaluation of d. Do other factors (e.g., race/ethnicity, family history, smoking, Genomic Applications in Practice and Prevention (EGAPP) diet, exercise level, and other conditions) affect the clinical Working Group (EWG) to inform the development of formal validity of the testing? recommendations for clinical practice. This review addresses 4. What are the issues relating to the use of cardiogenomic profiles the overarching question: “Does the use of ‘cardiogenomic in the designated populations and what is the impact on patient/ profiling’ lead to improved outcomes for the patient/consumer, consumer outcomes? or is it useful in medical or personal decision-making?” a. What are the management options for patients/consumers based on cardiogenomic profile results in a medical model vs. a DTC model and how would they differ from routine health messages? METHODS b. How could the results of cardiogenomic profiling in the general The process used to identify, review, analyze, evaluate, and population impact health behaviors or inform decision making summarize the evidence is briefly presented in this review, but by patients and their healthcare providers that affect outcomes? an in-depth presentation of EWG methods has been published.15 c. In what ways could the use of cardiogenomic profiling in the Consultants from Women & Infants Hospital, Department of designated populations impact clinical outcomes (e.g., Pathology and Laboratory Medicine, with experience in evi- morbidity/mortality) dence review of genetic tests were contracted by the Office of d. What is known about other contextual issues, such as cost- Public Health Genomics at the Centers for Disease Control and effectiveness, likelihood of behavioral change, and family Prevention to perform the targeted review with the assistance of history considerations. four EGAPP staff members and other outside consultants. Guid- ance was provided by a Technical Expert Panel. 5. What are the potential harms associated with cardiogenomic An analytic framework and key questions (Table 1, Fig. 1) profiling (e.g., marketing direct to consumers, distress or stigma for a “poor prognosis” result, misinterpretation of results leading were developed and refined by the EWG and Technical Expert to excessive or inadequate treatment, and exploitation of Panel members (see “Acknowledgments”), with support from hypervigilant people). EGAPP staff. The review focuses on clinical validity but ad- MI, myocardial infarction; CVD, cardiovascular disease; DTC, direct-to-con- dresses the limited information available on analytic validity sumer. and considers proposed measures of clinical utility in clinical practice and in direct-to-consumer settings. Standard methods used include systematic search criteria for identification of published and gray literature; application of inclusion/exclusion criteria; abstraction of data; meta-analysis; assessment of the Data sources for analytic validity quality of individual studies; and overall strength of evidence.15 PubMed௡18 searches were performed for the alleles and Limiting key questions and truncating search strategies (e.g., SNPs (e.g., as “AGTR1”or“AGTR1 genotyping”) and specific targeted gray literature searches) are two common methods in terms (e.g., “analytic validity” and “clinical test”). Gray litera- targeted reviews.16 In reviewing the available evidence, ques- ture searches (e.g., company and genetic testing web sites) were tions from the ACCE (Analytic validity, Clinical validity, Clin- conducted to collect information on laboratories offering testing ical utility and Ethical, Legal and Social Implication) review for these markers and the methodologies used. When these framework were often used to identify and organize the specific sources were not sufficient, we contacted the companies information needed to address the key questions.17 The draft offering heart health genomic panels for information, first by report was then revised in response to reviewers’ comments, an e-mail questionnaire and then by phone (for those not and the final manuscript was submitted to the EWG. responding).

Genetics IN Medicine • Volume 12, Number 12, December 2010 773 Palomaki et al. Genetics IN Medicine • Volume 12, Number 12, December 2010

Fig. 1. The analytic framework: Genomic profiles and cardiovascular disease. This schematic shows the analytic framework underlying the current review. The numbers indicate the four key questions contained in Table 1.

Clinical validity literature search association studies.22 These “Venice” criteria focus on amount A comprehensive review of all gene/CVD associations was of evidence, replication of evidence, and protection from bias. determined by the EWG to be outside the scope of the review, Each of these three criteria was assigned a grade of “A,” “B,” or which was limited to (1) addressing two major CVD outcomes “C” (see Appendix, Supplemental Digital Content 2, groupings (i.e., CHD and stroke), (2) examining only those http://links.lww.com/GIM/A125, for more detail). Epidemio- genes/polymorphisms included in cardiogenomic panels avail- logical evidence for significant association was rated as able in April 2008, and (3) using existing high-quality meta- “strong” if the meta-analysis received three A grades, “moder- analyses whenever possible. Literature searches were conducted ate” if it received any B grade but not any C grade; and “weak” using HuGE Navigator v1.319,20 after crosschecking a subset of if it received a C grade in any of the three criteria.22 These genes by a PubMed search.18 Specific search strategies for each grades provide a quick but powerful overview of the likely gene are contained in Appendix, Supplemental Digital Content reliability of the reported summary estimates. 1, http://links.lww.com/GIM/A124. Reference lists of retrieved There is no generally accepted process to revise the effect publications were examined for relevant studies. Searches were size estimate, if credibility is rated as moderate or low due to performed between June 2008 and January 2009. One investi- potential biases (e.g., publication bias). To provide an indication gator (L.M.N., S.M., or G.E.P.) had primary responsibility for of the extent of potential change in effect size, we performed a each gene, and results were reviewed by another (usually cumulative effects analysis23 for combinations having a mini- G.E.P.). Discrepancies were resolved by discussion. mum of six studies (three of which included 500 or more participants). The analysis adds one study at a time, from Criteria for inclusion of studies for clinical validity smallest to largest confidence interval, to create multiple esti- To be included, a publication needed to be in English and mates for the summary OR. If the range of these cumulative include information about whites. The outcome needed to be a ORs is wide, potential for bias exists. We defined the range as primary CVD event, heart disease, or stroke. Sufficient data the difference between the final cumulative OR and the first needed to be present to express the effect size as an OR with stable estimate (the third of three consecutive estimates all confidence intervals. Existing meta-analyses were used prefer- within 10% of each other). Usually, but not always, this oc- entially, but we created structured summaries of original pub- curred within the first three large studies. The ranges Ͻ0.1, lications when no suitable meta-analyses were found. 0.1–0.19, and Ն0.2 were considered small, medium, and large ranges, respectively. Data analyses for clinical validity Summary ORs and corresponding 95% CIs were derived Combining CVD genetic markers using a random-effects model (Comprehensive Meta-analysis, As a way to set a reasonable upper limit on the effect size of Version 2, Englewood, NJ), from the original source (published several genetic markers, we chose a multiplicative model that meta-analysis), from a reanalysis of the reported data, or from a assumes each marker is an independent predictor of CVD risk. new literature summary. The preferred method for determining Before multiplication, the ORs were adjusted, so that there was the summary ORs was to compare wild-type individuals (no no overall impact on CVD risk.24 For example, if an at-risk at-risk alleles) with heterozygotes (one at-risk allele) and then genotype was present in 10% of the population and was asso- separately compare wild type with homozygotes (two at-risk ciated with a summary OR of 1.1, an OR of 1.09 would be used alleles). Other models were also used. For example, when the for the 10% at risk and 0.99 for the remaining 90%. A Monte frequency of the at-risk allele was very low, heterozygotes and Carlo simulation generated 100,000 individuals with, and homozygotes were combined and compared against the wild 100,000 without, heart disease (or stroke) with their associated type. Some existing meta-analyses included formal assessment cumulative ORs. When multiple variants were available for a of heterogeneity, usually a ␹2-based Q statistic. This was con- given gene, the one with the strongest evidence was used. If two verted to the I2 statistic21 for ease of interpretation and com- variants has similar credibility, the one with the largest impact parison between studies. Some also included a formal exami- (effect size time ϫ at risk proportion) was used. nation of publication bias or stratified the summary OR by sample size. If sufficient data were available, we looked for Clinical utility potential publication bias. A systematic review of the literature on the clinical utility of the 29 markers was not undertaken. Rather, PubMed searches18 Evaluating credibility of the cumulative evidence for a focused on identifying potential benefits and harms associated gene-disease association with addition of CVD-associated alleles/SNPs to existing risk In 2007, a consensus group recommended evaluation guide- assessment algorithms based on conventional risk factors and lines to assess the cumulative evidence provided by genetic on the potential for genetic information to motivate behavioral

change. Articles from the clinical validity review were also a X 9P21

reviewed for information related to clinical utility. SNPs

RESULTS Tests for cardiogenomic risk assessment Table 2 lists the eight genomic test panels related to CVD/ heart health available in April 2008, with the relevant genes/ variants included. There is signiﬁcant variability among these tests including the CVD outcomes (e.g., MI, coronary artery disease, and stroke); panels and platforms used for testing; populations for which testing is recommended; personal and family history information collected; pretest information provided on potential beneﬁts and harms of testing; ways in which results are presented and interpreted (e.g., gene by gene, risk algorithm, and general or personalized health messages); and whether physicians, genetic counselors, or other health care providers are involved. A summary of this information for each genomic test or panel is available in Appendix, Supplemental Digital Content 2, http://links.lww.com/GIM/A125. Table 3 contains the health implications for each of the genes, as described by the companies offering the tests.

Literature searching To determine whether HuGE Navigator was sufficient to use as the platform for literature searching, we chose three genes (GPX1, SOD2, and MTR) and searched for gene-disease association studies using both HuGE Navigator v1.3 and PubMed. For GPX1, the HuGE Navigator search identified three studies, one of which satisfied the inclusion criteria. A similar PubMed XX search identified an additional 59 articles, none of which met inclusion criteria. For SOD2 and MTR, PubMed searches identified 161 and 43 additional articles, respectively. None of these satisfied the inclusion criteria. Given the ease, speed of use, and

apparent completeness, we chose to rely mainly on HuGE Genes included on test panels for CVD Navigator searches for the identiﬁcation of publications for the clinical validity analyses. In a separate analysis of the association of 9p21 SNP markers and heart disease,25 we discovered that HuGE Navigator search was less sensitive in identifying articles relating to 9p21 (a text search) than for searches involv- ing genes (missed articles used terms such as “9p” and “9p21.3,” which were not indexed in HuGE Navigator).25

Analytic validity Identification of data sources PubMed searches for each of the 29 genes (e.g., as “AGTR1” genes. or “AGTR1 genotyping”) and specific terms (e.g., “analytic validity” and “clinical test”) identified no relevant articles, whereas searches for individual genes using less specific search terms yielded too many articles to review (e.g., Ͼ700 citations for [“9p21” and “assay”]). Testing methods used in the research XXXXXXXXX studies included in clinical validity were not relevant, as they CDKN2A/CDKN2B did not include method comparisons and were not representa- tive of clinical practice. Websites from the eight companies (Table 2) were reviewed for information on analytic validity, but no data were found. XXXXXX X XX X XX X XXX XX XXX XXX X X X X X X X X X X GeneTests26 queries found no US laboratories offering testing for 20 of the 29 genes (Table 2). Testing for three of the nine remaining genes, F5 (Factor V Leiden), F2 (prothrombin ACE AGT AGTR1 APOB APOC3 APOE CBS CETP CYBA CYP11B2 F2 F5 GNB3 GPX1 IL1B IL6 ITGB3 LPL MTHFR MTR MTRR NOS3 PAI-1 PON1 SELE SOD2 SOD3 TNF G20210A), and MTHFR, is widely available to assess throm- bophilia risk. An American College of Medical Genetics/Col- Currently available genetic tests for cardiovascular disease (CVD) along with the genes included on each panel lege of American Pathology proficiency testing program is also (plus profile) (premium) (profile) genomics available.27 Two systematic reviews address F5/F2 analytic genetics DeCODE MI GenelexScionaGenovations XGenosolutions XGenovations Integrative X X X X X X X X X X X X X X X X X X X X X X Interleukin Sometimes referred to as SNPs near the Testing laboratory (distributor) Clinical claims Lifestyle changes validity and found adequate evidence (scale: convincing, ade- Table 2 a SNP, single-nucleotide polymorphism.

Genetics IN Medicine • Volume 12, Number 12, December 2010 775 Palomaki et al. Genetics IN Medicine • Volume 12, Number 12, December 2010

quate, and inadequate) that the analytic sensitivity, analytic Table 3 Genes and associated health implications based specificity, and reproducibility of commonly used methods were on individual company website literature high, although generalizability was limited due to lack of strat- 28 Genes Health implicationsa ification by method/platform. Publications providing information on the analytic validity of testing for MTHFR and the ACE Blood flow (a) remaining six genes (i.e., ACE, AGT, AGTR1, APOB, APOE, AGT Increased levels of circulating angiotensin; and CBS) were not identified. hypertension (a) Relevant methodologies AGTR1 Hypertension, coronary artery disease and kidney The literature on allele/SNP genotyping described laboratory disease (a) developed tests (LDTs) and a large number of commercially APOB Lipid metabolism (d) available reagents and platforms.29,30 The methods used for the nine genes listed on GeneTests were reported as “sequencing APOC3 Triglyceride metabolism (b) and/or mutation analysis.” In general, genotyping methods have APOE Atherosclerosis, MI, stroke, and osteoporosis, and involved discrimination of alleles by primer extension, hybrid- higher antioxidant activity (d) ization, ligation or enzymatic cleavage, and detection using fluorescence, mass, gel electrophoresis, or chemilumines- CBS Metabolism of vitamin B6 and removal of 29 homocysteine (b) cence. Mistaken alleles, allelic dropout (i.e., amplification of only one of two alleles in a heterozygous individual), and other CETP Lower CETP activity, lower LDL-cholesterol, higher genotyping errors can result from a number of causes. These HDL cholesterol; CV in women or high triglycerides; have included interaction with flanking DNA sequences, low D442G not defined (a); Cholesterol metabolism (b) quality/quantity of the DNA in samples, laboratory problems CYBA Reduced vascular NADPH oxidase activity and related to reagents/protocols/equipment, and human error. coronary atherosclerosis (a) Questions remain about causes of errors in many newer technologies (e.g., multiplex assays, chips, and SNP arrays) used in CYP11B2 Blood pressure via kidney and fluid balance (d) routine clinical practice and their potential impact on patient F2 Increased levels of prothrombin and venous results.31 thrombosis (a) A survey of companies offering genomic panels/tests F5 Abnormal inactivation of Factor V and venous thromboembolism (a) Standardized e-mails were sent to the Chief Medical Officer (or equivalent) of each of the eight companies. One company GNB3 Enhanced G-protein activation, hypertension, responded (deCODE, Inc., Reykjavik, Iceland). Client Services atherosclerosis, myocardial infarction and LVH (a) (or equivalent) from the seven remaining companies were con- GPX1 Oxidative stress (d) tacted by telephone; both information about this review and a request to be contacted were left. One additional company IL1B Over-expression of inflammation and risk for responded (Interleukin Genetics, Inc., Waltham, MA). cardiovascular disease (e) Information on analytic validation of the deCODE MI™ IL6 Inflammatory response (b) LDT was provided. The test is based on Centaurus™ Assay (Nanogen, Inc., San Diego, CA), a methodology using DNA ITGB3 Increased platelet aggregation and risk of clot formation hybridization technology, allele-specific fluorescence detection (a) of polymerase chain reaction products by spectrometry, and LPL Cholesterol metabolism (b) allele calling by automatic algorithm. A published reference for the real-time polymerase chain reaction methodology this test MTR Removal of homocysteine (b) was based on was also provided.32 The information included MTRR Metabolism of Vitamin B12 (b); vascular Integrity via results from comparisons with a gold standard in 100 unselected homocysteine management (d) patients and 12 positive controls (100% concordance). No fail- ures were reported. Patients/controls were run in quadruplicate, NOS3 Blood flow (b) and all results agreed. Long-term quality control measures were PAI-1 Higher PAI-1 levels and thrombosis (a) also in place. The laboratory, located in Iceland, reported hold- ing CLIA and College of American Pathologists accreditations. PON1 Vascular Integrity Information on the Inherent Health™ Heart Health LDT SELE E-selectin adhesion leading to atherosclerosis/restenosis; (also provided under the name Gensona™) was provided by enhanced thrombin production, risk of coagulation (c) Interleukin Genetics (Waltham, MA). A single-base extension SOD2 Antioxidant defense (b); Oxidative stress (d) technology was used in a tagged fluorescent assay, and the GenomeLab™ SNPStream௡ Genotyping System (Beckman SOD3 Antioxidant defense (b) Coulter, Fullerton, CA) was used to identify the relevant SNPs. TNF Inflammatory response (b); Vascular integrity (d) Genotype results using this method for 15 control DNA samples (Coriell Cell Repositories, Camden, NJ) agreed with the known 9p21 SNPs Unknown mechanisms (c) genotypes (100% concordance). Results for 20 samples were aHealth implications as provided on the company website: (a) Genovations/ compared with sequencing results from an outside laboratory Genosolutions, (b) Sciona, (c) deCODE, (d) Integrative Genomics, and (e) Inter- (100% concordance). Replicate samples (2–9 buccal swabs) leukin Genetics. from 19 of the 20 volunteers were successfully tested for three SNPs in 98.4% of samples, and concordance was 100%. In another validation study of DNA from 104 buccal swabs and 136 control samples, 98.2% were successfully genotyped. The

776 © 2010 Lippincott Williams & Wilkins Genetics IN Medicine • Volume 12, Number 12, December 2010 Genomic profiling for risk of CVD laboratory reported having accreditation from CLIA and four erable overlap, the cORs are higher among cases. Overall, 10% states (CA, MD, NY, and RI). of controls and 23% of cases are assigned a cOR of 1.38 or higher. Limitations The cOR distributions can also be displayed as a receiver Method descriptions and in-house analytic validation data operator characteristic curve (Fig. 3). The AUC is 64.7% and, were not provided by the other six companies. This data are by itself, would not be considered a useful test for heart disease considered Level 4 (the lowest quality).15 The overall quality of risk stratification.33,34 evidence for analytic validity of these testing panels was considered inadequate. Specific limitations included no information Association of individual gene variants with stroke about the platform used for the majority of genomic tests, The last two columns in Table 4 provide similar data for the limited information on the impact of DNA degradation from 28 genes in association with stroke. Less data were available, sample collection/processing and shipment using real sam- with only 19 genes having identified publications that include ples, and for some genes, limited information about the 20 combinations of variants/models. Six of the 20 combinations actual variant tested. For some of the genes, the prevalence (30%) were associated with a statistically significant OR. None of the least common genotype is low, so the positive predic- received a strong credibility rating. Overall, 50% (10 of 20) had tive value is highly dependent on analytic sensitivity. The moderate credibility, with the remaining having weak credibil- in-house information from two companies was encouraging, ity. Figure 2B shows that the distributions of cORs generated by and platforms exist that could allow high sensitivity and a separate Monte Carlo simulation have only modest separation specificity. However, estimating analytic sensitivity and between cases and controls. Overall, 15% of cases and 11% of specificity for the majority of genomic panels for CVD/heart controls had cORs of 1.38 or higher. Figure 3 also shows the health was not possible. receiver operator characteristic curve for the stroke model, with an AUC of 55.2%. Clinical validity Assessing the potential for biases in the effect size Association of individual gene variants with coronary A cumulate effects analysis was performed for 17 of the 38 heart disease combinations for heart disease. Of these, 2 (12%) had a large Table 4 provides a summary of the evidence. The gene range, 5 (29%) had a moderately wide range, and the remaining acronym (and common aliases) is shown in the first column, 10 (59%) had a relatively small range. In all instances of a large along with the variant(s) reviewed, the proportion of the popu- or moderate range, the final consensus effect size was larger lation at risk, and the model(s) used in the literature in the next than the first consistent estimate. Only 7 of the 13 combinations three columns. For example, the first row summarizes the in- for stroke had sufficient data. Of these, one (14%) had a large formation for the ACE gene insertion/deletion variant. The range, three (43%) had a moderately wide range, and three at-risk genotype (present in 27% of whites) is the presence of (43%) had a relatively small range. Among the four combina- two at-risk deletion alleles (DD). Also provided is the Venice tions with moderate or large ranges, one (MTHFR) had a first assessment of cumulative credibility. None of the grades are stable estimate (1.35) that was larger than the consensus (1.1). based on effect size. For the ACE I/D variant, the cumulative Overall, this suggests that the findings shown in Table 4, and credibility is classified as “weak.” Although the amount of Figures 2 and 3 are upper limits, with an “unbiased” effect size for evidence is large (indicated by the “A”), the extent of replica- many of the gene marker combinations likely being closer to 1.00. tion is poor (indicated by the “C”). The protection from bias is not evaluated, because if any of the three grades was a C, the Risk prediction: Combining traditional risk factors and credibility is considered weak regardless of the grade assigned genomic markers for bias. The consensus OR is 1.22, with the “s” indicating Limited information is available on whether combining significance (P Ͻ 0.05). In the example of ACE, several meta- genomic markers with traditional risk factors (TRFs) improves analyses showed a strong association between sample size and prediction of heart disease. Only three studies35–38 report the effect size (smaller studies having larger effects), but no attempt accuracy of 10-year risk prediction using traditional risk factors to change the effect size was undertaken. A description of the with, and without, measurement of the 9p21 SNPs, and a literature search, how the at-risk genotypes were defined, summary of these results has been published.25 The net reclas- sources for the at-risk frequencies, included confidence inter- sification index, a measure of improvement39, ranged from vals, tests of heterogeneity, and evaluation of publication bias Ϫ0.2% (a nonsignificant decrease in prediction by adding 9p21 are all included in Appendix, Supplemental Digital Content 1, SNP information) to 4.9% (a small, but significant improvement http://links.lww.com/GIM/A124. Overall, 38 different variants/ in risk prediction). However, the mechanisms of 9p21 markers models were reported for 28 genes (no data were found for in modifying risk for heart disease are not yet fully under- SOD3), and 17 (45%) had at least one statistically significant stood.35–37,40 OR. Only one marker (2.4%) had strong credibility (A,A,A)— The change in the heart disease risk is relatively small, when the 9p21 SNPs. Among the remaining combinations, 15 (37%) compared with change in risk estimates possible using conven- and 25 (61%) have moderate and weak credibility, respectively. tional risk factors, and other tests (e.g., electrocardiogram) and Without considering the 25 combinations with weak credibility, biomarkers.41,42 How to analyze the potential improvement over the ORs range from a high of 1.25 to a low of 0.80 (both risk prediction using traditional risk factor (e.g., AUC, event- estimates are found using the 9p21 SNPs). specific reclassification, integrated discrimination improvement test, and other new approaches) remains an active area of Cumulative effect of multiple genes tested current and future research.39,43–45 The bolding of the OR (Table 4) indicates which of the effects was chosen for modeling for each gene. The distribution Limitations of cumulative ORs (cORs) resulting from the Monte Carlo Most analyses have been restricted to whites, usually of simulation are shown in Figure 2A. Although there is consid- European descent; results in other race/ethnic groups might

Genetics IN Medicine • Volume 12, Number 12, December 2010 777 Palomaki et al. Genetics IN Medicine • Volume 12, Number 12, December 2010

Table 4 Summary of observed gene/disease associations for cardiovascular disease in whites Heart diseaseb Strokec Gene (aliases) Variant % at risk Comparisona Credibilityd ORe Credibilityd ORe ACE (ACED) D (del), I (ins) 27 DD vs. (II & ID) Weak (A,C,—) 1.22s Moderate (A,A,B) 1.21s AGT M235T 50 MT vs. MM Weak (A,C,—) 1.02n 18 TT vs. MM 1.15s AGTR1 A1166C 48 (AC&CC) vs. AA Weak (A,C,—) 1.11s 8 CC vs. (AC &AA) 1.32s APOB XbaI 58 (TC&TT) vs. CC Weak (A,B,C) 1.14n 12 TT vs. (TC & CC) 1.19s InsDel 53 (DD&DI) vs. II Moderate (B,A,B) 1.15s 10 DD vs. (DI & II) 1.19s EcoRI 30 (AA&AG) vs. GG Weak (C,A,—) 1.32s 3 AA vs. (AG & GG) 1.73s APOC3 (APOC-III) Sst-1 1 S1S1 vs. (S1S2 & S2S2) Moderate (B,A,B) 1.00n T455C 54 TC vs. TT Weak (B,C,—) 1.07n 12 CC vs. TT 1.32n C482T 38 TC vs. CC Moderate (B,A,B) 0.99n 6 TT vs. CC 0.91n ⑀ ⑀ ϭ ⑀ ؉ ⑀ APOE n (n 2,3,4) 21 4 vs. 3 3 Weak (A,C,—) 1.06n Weak (A,C,—) 1.11s 2؉ vs. ⑀3⑀3 0.80n 0.99n⑀ 12 CBS (HIP4) c.844ins68 3 MM vs. (WW & WM) Weak (C,C,—) 0.55n CETP (HDLCQ10) TaqlB (C629A) 50 B1/B2 vs. B1/B1 Moderate (A,B,B) 0.96n 18 B2/B2 vs. B1/B1 0.82n 18 B2/B2 vs. B2- Weak (C,C,—) 0.53s CYBA (CYBA8) C242T 56 (TT&TC) vs. CC Moderate (A,B,B) 0.94n 12 TT vs. (TC & CC) 1.17n CYP11B2 T344C 49 (CC&TC) vs. TT Weak (C,C,—) 1.25n 20 (CC) vs. TT 0.80n F2 (Factor II) G20210A 2 (AA&AG) vs. GG Moderate (B,B,B) 1.21n Moderate (B,B,B) 1.30n F5 (Factor V) G1691A 10 (AA&AG) vs. GG Moderate (A,B,B) 1.10n Moderate (B,B,B) 1.31s GNB3 C825T 41 CT vs. CC Moderate (B,A,B) 1.05n 9 TT vs. CC 1.13n 50 (CT&TT) vs. CC Weak (C,C,—) 1.45s ϭ ؉ GPX1 ALAn (n 5,6,7) 40 ALA6 vs. other Weak (C,C,—) 1.30n Weak (C,C,—) 0.92n IL1B C511T 59 (TT&TC) vs. CC Weak (C,C,—) 0.69n 46 TC vs. CC 0.65n Weak (C,C,—) 1.07n 13 TT vs. CC 0.90n 1.52n IL6 (HGF, IFNB2) G174C 64 (CC؉GC) vs. (GG) Moderate (A,B,B) 1.12n Moderate (B,A,B) 1.07n (Continued)

Table 4 Continued Heart diseaseb Strokec Gene (aliases) Variant % at risk Comparisona Credibilityd ORe Credibilityd ORe LPL (KID, HDLCQ11) S447X 19 (()(SX&XX) vs. SS Moderate (A,A,B) 0.82s 19 (&)(SX&XX) vs. SS Moderate (B,A,B) 0.97n 19 (SX&XX) vs. SS Moderate (B,B,B) 1.33n A291S 11 (AS&SS) vs. AA Weak (C,B,—) 1.48s Pvull 47 P؉Ϫ vs. PϪϪ Weak (B,C,—) 1.17n P؉؉ vs. PϪϪ 0.93n 18 ITGB3 (GP3A) C1565T 28 (TT&TC) vs. CC Weak (A,B,C) 1.13s Moderate (B,A,B) 0.99n MTHFR C677T 45 CT vs. CC Weak (A,C,—) 1.04n Weak (B,C,—) 1.03n 12 TT vs. CC 1.14s 1.19n MTR (MS) A2756G 26 AG vs. AA Weak (C,A,—) 1.02n Weak (C,C,—) 1.18n 3 GG vs. AA 1.25n 4.00n MTRR (MSR) A66G 22 (GG&AG) vs. AA Weak (B,C,—) 1.06n Weak (C,C,—) 0.79n NOS3 (eNOS) G894T 50, 25 TG vs. GG Weak (A,C,—) 1.17s (allele model) 25 TT vs. (GGϩGA) Moderate (B,C,—) 0.98n Intron4 2 AA vs. (BB & BA) Weak (A,C,—) 1.12s T786C 49, 18 CT vs. TT Weak (A,C,—) 1.17s (allele model) PAI-1 G455A 77 4G؉ vs. 5G5G Weak (B,C,—) Weak (A,C,—) 0.98n PON1 Q192R 42 QR vs. QQ Weak (A,C,—) 1.32s Moderate (B,B,B) 1.00n 11 RR vs. QQ 1.38s 1.35s L55M 40 LM vs. LL Moderate (A,B,B) 0.99n Moderate (B,A,B) 1.04n 10 MM vs. LL 0.99n 0.92n SELE (E-Sel) S128R 9 R؉ vs. SS Weak (B,A,C) 1.51s SOD2 (Mn-SOD) C47T 50 CT vs. CC Weak (B,C,—) 1.04n 23 TT vs. CC 0.86n SOD3 (EC-SOD) — TNF G308A 32 (GA&AA) vs. GG Moderate (A,B,B) 1.07n Weak (B,C,—) 1.23n G238A 12 (GA&AA) vs. GG Moderate (B,B,B) 0.95n 9p21 (CDKN2A, CDKN2B) Several 27 BB vs. AB Strong (A,A,A) 0.80s Moderate (B,B,B) 1.02n A is at risk 23 AA vs. AB 1.25s (Additive model) 1.04n aThe at-risk genotype is bolded. bHeart disease includes myocardial infarction, coronary artery disease, coronary heart disease, ischemic heart disease, and ischemic heart failure. cStroke includes intracerebral hemorrhage, subarachnoid hemorrhage, and ischemic stroke. dBased on criteria22 that use three letter grades (A, B, and C) to determine credibility of a meta-analysis. eThe odds ratio for heart disease (or stroke) between the bolded and not bolded comparison groups.

● Amount of evidence: A ϭ over 1000 cases ϩ controls in least common genotype (assuming 1:1), B ϭ 100-1000, C ϭϽ100. ● Extent of replication: A ϭ I2 Ͻ25%, B ϭ I2 25–50%, C ϭ I2 Ͼ50% (or single study); I2 ϭ (Q Ϫ df)/df ϫ 100. ● Protection from bias: A ϭ If present, bias might only affect magnitude, B ϭ no obvious bias, missing information, C ϭ considerable bias/potential for bias that could impact direction, “—” not evaluated due to an earlier C grade, OR ϭ odds ratio. The “s” indicates the results is signiﬁcantly different from 1.0, whereas “n” indicates nonsigniﬁcance. Bolded entries indicate that the OR is used in modeling a combined effect size.

differ in effect size, prevalence of at-risk genotypes, variant of be underpowered or limited in SNPs coverage of the genome to interest, or some combination of these. Some of the gene- reliably conﬁrm or reject real associations in the genes we disease associations under review involve a small effect size for reviewed. The ﬁnding of weak associations between gene vari- relatively uncommon genotypes. The current GWA studies may ants and CVD outcomes does not exclude possible roles for

Genetics IN Medicine • Volume 12, Number 12, December 2010 779 Palomaki et al. Genetics IN Medicine • Volume 12, Number 12, December 2010

Fig. 2. Theoretical overlapping distributions of cumulative odds ratios (cORs) in individuals with and without disease. A, The cOR using markers from 24 genes associated with heart disease from Table 4. The dashed line (with open circles) indicates the distribution for individuals with CHD/MI, and the solid line (with filled circles) indicates the distribution in control individuals. B, A similar analysis with stroke as the outcome of interest. The cOR for stroke is based on markers from 13 genes (Table 4). At a false-positive rate (FPR) of 10% and 11%, these markers have detection rates (DR) of 23% and 15% for heart disease and stroke, respectively.

tivities. Publication bias seems to be an important problem in the literature regarding genomic tests for CVD. Finally, the genes included in this study were, on average, described more than five years ago, with the newest one (9p21) first reported in 2007.46,47 Since that time, one GWA study48 has reported new markers that may also have strong credibility, once confirmatory studies have been reported.

Clinical utility Health outcomes and clinical scenarios A test has clinical utility when the results change clinical management, which, in turn, leads to measurable improvements in primary health outcomes (i.e., reduced morbidity and mortality related to heart disease). Secondary outcomes of interest include the impact on health care providers’ decision making and the individuals’ personal motivation. The potential clinical utility of testing is likely to differ depending on the clinical Fig. 3. CHD: The receiver operator characteristic (ROC) scenario, most importantly the setting in which the testing is curves for the cumulative odds ratio (cOR). The genomic offered (e.g., clinical and direct-to-consumer) and the popula- test panel detection rate (sensitivity) is on the vertical axis, tion targeted to be offered testing (e.g., high risk or general whereas the false-positive rate (1-specificity) is on the hor- population). Three likely clinical scenarios are (1) genomic izontal axis. The 45[degree] dashed reference line indi- testing is not performed, and actions continue to be based on cates a “useless” test where detection and false-positive risk estimated using TRFs; (2) testing is ordered by a physician rates are equal. The ROC curve for the 24 CHD/MI markers or other health care provider and the results considered in the is shown with open circles, whereas the corresponding context of CRFs; or (3) testing is ordered by a consumer (who curve for the 13 stroke markers is shown with filled circles. may or may not have a CRF) who may or may not seek advice The areas under the curve for the CHD/MI and stroke and/or interpretation from a health care provider. markers are 64.7% and 55.2%, respectively. Potential interventions and related benefits and harms No studies were identified on long-term CVD outcomes after these variants in the pathway of lipid metabolism or in other personalized clinical interventions based on genomic test re- pathways associated with CVD. Many studies have confirmed sults. In the clinical setting, physician changes to clinical man- both gene/lipid associations and that the variants identified are agement based on increased heart disease risk might include functional. However, the function of the marker with the highest earlier initiation or higher rates of pharmacotherapy (e.g., statins credibility, 9p21, is not well described. Understanding its role in and antihypertensives), increased use of other medical tests or disease etiology might result in an improved understanding of interventions (e.g., health plan prevention and management the process, leading to improved treatments or prevention ac- programs), and provision of targeted recommendations for life-

780 © 2010 Lippincott Williams & Wilkins Genetics IN Medicine • Volume 12, Number 12, December 2010 Genomic profiling for risk of CVD style changes (e.g., smoking cessation, diet, and exercise). In- system (clinical utility). For genomic tests to add substantially creased use of diagnostic or prognostic tests, such as exercise to existing predictive models for heart disease, it would likely stress testing, myocardial infusion imaging, and echocardiogra- require many markers such as 9p21 to be included.54,55 phy, could improve clinical outcomes when targeted to those with the highest risk (potential benefit) or increase costs and patient anxiety without improving health outcomes (potential DISCUSSION risk). There is preliminary evidence that physician knowledge of CHD risk scores may translate into short-term benefits, such Quality of evidence as increased prescription of drugs and modest reductions in risk Analytic validity factors (e.g., patient blood pressure), without identified clinical The quality of evidence for analytic validity is inadequate. 49 harms. However, further studies are needed to determine For most of the offerings (Table 2), it was not possible to whether such findings are replicable, apply across different identify the specific variant tested within a specific gene. We practice settings, and lead to improved long-term CHD out- found no published literature on the specific testing platforms, 49 comes. analytic sensitivity/specificity, or predictive values for the eight The challenge of motivating long-term behavioral change is genomic panels relating to heart health from the eight compa- key in addressing modifiable risk factors (e.g., compliance with nies advertising these tests. Two of the companies provided 50 treatment, diet/exercise, weight loss, and smoking cessation), in-house data and referrals to a publication. Testing for the gene whether in the clinical or direct-to-consumer setting. However, variants described in the review can be done reliably with measuring behavioral change is uniquely important in demon- currently available technologies (e.g., F2 and F5), but there is strating the utility of CVD genomic profiling test result reported inadequate evidence that this is so for all the companies iden- directly to consumers. No studies were identified that assessed tified in this report. the clinical utility of providing genetic risk information to consumers outside the clinical setting. In fact, very little is currently known about all aspects of the direct-to-consumer Clinical validity scenario, including how many tests are ordered, who is ordering The quality of evidence for clinical validity varied widely the tests, what their reasons and motivation are for having the among the 29 genes/variants. For this reason, we used the tests, and how the information is being used. Venice grading system22 to rate the credibility of gene/disease Individuals at risk based on one or more CRFs and a positive association studies. The most credible evidence is for the 9p21 genetic test result seem most likely to receive the potential SNP markers and heart disease (but not stroke). This effect is benefits. However, more studies are needed to determine (1) the highly reproducible in multiple large studies and unlikely to be balance between potential benefits, potential risks (e.g., false influenced by major biases. On the other extreme are several reassurance and perceived genetic determinism), and implemen- gene/variant combinations that are based on only a few heter- tation issues (e.g., costs, available, and effective interventions); ogeneous small studies, often with an effect size that is suspect (2) whether tests should be routinely offered for risk screening due to important potential biases. Some of the strongest reported or targeted to refining risk in patients with identified CRF; (3) effect sizes are associated with weak credibility (credibility what level of increased risk warrants a management change; and grades of C, C, -). This review focused on the outcomes of heart (4) whether such changes will improve long-term heart disease disease and stroke. The identification of gene/variants has outcomes. Arguments about the inherent value of information helped greatly in understanding the relevant pathways. Finding are difficult to support if no action is taken (e.g., clinical or little or no association with a given health outcome does not personal decision making) or if interventions do not improve mean the gene does not play an important part in, for example, health outcomes. lipid metabolism. The literature is not consistent in how genotypes are grouped Summary when reporting gene/disease association studies. When the at- A test with demonstrated clinical validity does not necessar- risk allele frequency is low, those rare homozygotes are often ily result in improved health outcomes (clinical utility). A recent combined with heterozygotes to improve the power of the study. survey found that only approximately 37% of US physicians These ad hoc decisions make combining information from reported regular use of any heart disease risk score (93% of studies more difficult and may mask larger effect sizes in the whom used the Framingham system).51 Patient adherence to smaller groups. By contrast, an allele-specific OR fits the 9p21 lipid-lowering medications is also an important factor in attain- data very well (i.e., the change in risk from no to one at-risk ing improved heart disease outcomes, but preliminary data allele is equivalent to the change from one to two at-risk suggest that only approximately half of patients reach target alleles).25 The reliability of this finding is partly due to the lipid levels, and only one in four continue drug treatment long relatively high frequencies of the three 9p21 genotypes and to term.52 Other complicating factors may be access to medical the large study sizes. care and medications. Therefore, a positive impact of reclassi- Although we rated the credibility of gene/disease associa- fication based on addition of genetic markers to CRF cannot be tions, it was not possible to determine the extent to which biases assumed, even if risk assessment were to be improved. In fact, might account for some (or all) of the effect size when the it has been proposed that heart disease may be more effectively credibility is not rated as strong. Lohmueller et al.56 addressed prevented by implementing an inexpensive standardized multi- this issue and concluded that up to 25% of reported gene- drug intervention (i.e., the polypill) in an easily defined popu- disease associations might represent real associations. Ntzani et lation (i.e., all people aged 55 years or older), regardless of al.57 examined the relationship between AT1R and MI and found individual risk levels.53 a highly significant per-allele OR of 1.13. However, there was Clinical trials are needed to demonstrate that use of genetic high heterogeneity, with the larger studies showing no effect. tests is associated with changes in physician management de- Thus, the point estimates for the effect size for genes/variants cisions, patient motivation and long-term behavioral changes, graded as weak (or even moderate) must be viewed skeptically, improved outcomes, and/or reduced costs to the health care until sufficient information is available to determine the extent

Genetics IN Medicine • Volume 12, Number 12, December 2010 781 Palomaki et al. Genetics IN Medicine • Volume 12, Number 12, December 2010 to which potential biases might be influencing the size or even ● What methodology should be used to determine the extent the presence of a reported effect. to which genomic (or nongenomic) markers add useful Combining multiple genomic makers together is likely to information to an existing risk model. increase the predictive power over that possible with any one ● How information about genomic tests for CVD can be kept marker. However, little or no evidence exists as to how these current, given the rapid increase in knowledge and tech- combinations should be made. The most common method, and nology. the one we used in our modeling, is to assume independence of ● the markers and multiple the effect sizes. Using only the com- Alternative strategies for prevention of heart disease and bination of 24 gene/variants to predict heart disease, it was how genomic markers might impact these strategies. possible to identify 24% of individuals with disease along with ● How genomic markers that modify CVD risk derived from 10% of those without disease, at a cut-off OR of 1.38. Poorer traditional risk factors will change the pattern of clinical performance was found for stroke. This modeling might over- practice. estimate the effect if some of the markers are related (e.g., in the ● Are there behavioral changes related to providing the re- same biological pathway), but insufficient data are available to sults of genomic testing, and would these changes plausi- create more sophisticated models. It is also possible, but un- bly lead to improved health, and what factors might influ- likely, that some markers interact with each other to provide ence these changes (e.g., setting, method of delivery, and effect sizes that are larger than the product of the two. Thus, the change in risk). results of modeling the effect size for genomic panels should not be considered definitive. Research agenda Laboratories performing analytic validation studies for Clinical utility genomic panels should consider publishing their detailed results This review did not rate quality of evidence for clinical in peer-reviewed journals, as a way to build the evidence base utility. Rather, it focused on an overview of issues related to for reliable testing. A consensus method of handling data with clinical utility as part of the framework for evaluating genomic poor credibility and/or the existence of possible bias that could testing in clinical practice. No study was identified that actually have a nontrivial impact on the effect size should be developed. demonstrated an improvement in health outcomes based on use This would allow more consistent and reliable modeling to of genomic markers. Assessment of risks and benefits associ- occur. Further work on standardizing genotype models, sum- ated with genomic testing for CVD risk must consider whether marizing/evaluating the literature, combinations of genomic CRFs are included and whether the testing scenario is through markers, and combinations of genomic and nongenomic mark- a health care provider or direct to consumer. Preliminary work ers should be continued.22,44 Such work should capitalize on the on the interest, barriers, intent and behavioral changes with existing efforts occurring in various complex disease-specific respect to genomic testing is underway. An important collabo- areas such as breast cancer and diabetes. A recent scientific rative effort between the National Institutes of Health and two statement from the American Heart Association44 proposed six health plans (Multiplex Initiative) is underway to investigate phases in the evaluation of novel risk makers for CVD. The “… the interest level of healthy young adults in receiving 9p21 markers are the most developed of the novel genomic genetic testing for eight common conditions …,” as well as who markers and would be considered to be in Phase 3 (show decides to take the testing, how they interpret the results, and incremental value over standard risk markers) or Phase 4 how the results impact their future health care decisions.58 (change risk sufficiently to modify therapy). It has already passed through proof of concept (Phase 1) and prospective validation (Phase 2). Assuming that clinical utility can be dem- Gaps in knowledge onstrated by a change in recommended therapy, Phase 5 would require a randomized trial to demonstrate that a health outcome ● Little or no available information on the analytic validity is improved, whereas Phase 6 would determine whether the of genomic panels, either in the published literature, or on marker’s use is cost-effective. the company websites. Often, it was not possible to even determine the testing platform or methodology being used. ● The specific genes and variant(s) included on the genomic Update on the availability and content of genomic panel. panels for heart disease risk ● How genotypes should be grouped/modeled for risk (e.g., This targeted review addresses only those genes included on allele-specific OR, recessive model, and dominant model) genomic profiles aimed at CVD or “heart health” that were and what evidence is needed to decide. available when the study was undertaken in mid 2008. Because ● Which of the gene/variant associations identified might it is possible that additional or updated panels may now exist, benefit from further validation and/or analysis to improve we repeated our search strategy in June 2010. Of the genomic their credibility. panels included in Table 2, one had added new markers (de- ● How information gained from GWAS might be helpful in CODE MI), two panels no longer seem to be available (Geno- determining the effect size and credibility of existing gene/ vations and Genosolutions), and the remaining five were un- changed. No new genomic panels for heart health were disease associations. identified. The updated offering from deCODE was released in ● Which, if any, of the gene/disease associations identified May 2009 after this review had been conducted. That “heart with moderate or weak credibility might be overestimated attack” panel currently includes markers in nine genes not included due to potential biases (e.g., publication bias). in the present review (BRAP, CELSR2/PSRC1, CXCLI2, MIA3, ● How multiple genomic markers for CVD should be com- MRAS, PHACTR1, SH2B3, SLC5A3/MRPS6/KCNE2, and bined, and the types of data needed to inform these models. WDR12).

ACKNOWLEDGMENTS 18. U.S. National Library of Medicine. National Institutes of Health. PubMed. gov. Available at: http://preview.ncbi.nlm.nih.gov/pubmed, 2009. Accessed This report was supported by the Office of Public Health October 1, 2010. Genomics, Centers for Disease Control and Prevention through 19. HuGE Literature Finder. Available at: http://www.hugenavigator.net/ a contract (200-2003-01396-0128) with McKing Consulting HuGENavigator/startPagePubLit.do, 2009. Accessed October 1, 2010. Corporation, Inc. The authors thank the EGAPP Working 20. Yu W, Wulf A, Yesupriya A, Clyne M, Khoury MJ, Gwinn M. HuGE Watch: tracking trends and patterns of published studies of genetic associ- Group members of the Technical Evaluation Panel including ation and human genome epidemiology in near-real time. Eur J Hum Genet Ned Calonge, MD, MPH (Colorado Department of Public 2008;16:1155–1158. Health and Environment); Celia Kaye, MD, PhD (University of 21. Higgins JP, Thompson SG, Deeks JJ, Altman DG. Measuring inconsistency Colorado School of Medicine); Carolyn Sue Richards, PhD, in meta-analyses. BMJ 2003;327:557–560. 22. Ioannidis JP, Boffetta P, Little J, et al. Assessment of cumulative evidence FACMG (Oregon Health & Science University); and Joan A. on genetic associations: interim guidelines. Int J Epidemiol 2008;37:120– Scott, MS, CGC (Johns Hopkins University). The authors also 132. thank unpaid consultants Christopher J. O’Donnell, MD, MPH 23. Borenstein M, Hedges LV, Higgins JPT, Rothstein HR. Introduction to (National Heart, Lung and Blood Institute/NHLBI Framingham Meta-Analysis. New York: John Wiley & Sons, Ltd., 2009. 24. Janssens AC, van Duijn CM. Five genetic variants associated with prostate Heart Study) and Colleen McBride, PhD (Chief, Social and cancer. N Engl J Med 2008;358:2739; author reply 2741. Behavioral Research Branch, National Human Genome Re- 25. Palomaki GE, Melillo S, Bradley LA. Association between 9p21 genomic search Institute [NHGRI]) who provided guidance and com- markers and heart disease: a meta-analysis. JAMA 2010;303:648–656. ments on drafts of this manuscript. The authors also thank the 26. GeneTests. Available at: http://www.ncbi.nlm.nih.gov/sites/GeneTests, 2009. reviewers of this manuscript and evidence report, including 27. CAP 2010 Surveys. Available at: http://viewer.zmags.com/publication/ Donna K. Arnett, PhD, MSPH (University of Alabama at Bir- f475b8bc#/f475b8bc/1, 2009. Accessed October 27, 2010. mingham) and Yuling Hong, MD, PhD, Associate Director for 28. Segal JB, Brotman DJ, Emadi A, et al. Outcomes of genetic testing in adults Science, Division for Heart Disease and Stroke Prevention, with a history of venous thromboembolism, 2009. Evidence report/technology assessment no. 180 (Prepared by Johns Hopkins University Evidence Centers for Disease Control and Prevention, Atlanta, GA. Based Practice Center under contract no. HHSA 290-2007-10061–1). AHRQ Publication No. 09-E011. Rockville, MD: Agency for Healthcare Research and Quality. Accessed October 1, 2010. REFERENCES 29. Kim S, Misra A. SNP genotyping: technologies and biomedical applications. 1. Lloyd-Jones D, Adams R, Carnethon M, et al. Heart disease and stroke Annu Rev Biomed Eng 2007;9:289–320. statistics—2009 update: a report from the American Heart Association 30. Ragoussis J. Genotyping technologies for genetic research. Annu Rev Statistics Committee and Stroke Statistics Subcommittee. Circulation 2009; Genomics Hum Genet 2009;10:117–133. 119:480–486. 31. Pompanon F, Bonin A, Bellemain E, Taberlet P. Genotyping errors: causes, 2. Rosamond W, Flegal K, Furie K, et al. Heart disease and stroke statistics— consequences and solutions. Nat Rev Genet 2005;6:847–859. 2008 update: a report from the American Heart Association Statistics Com- 32. Kutyavin IV, Milesi D, Belousov Y, et al. A novel endonuclease IV post- mittee and Stroke Statistics Subcommittee. Circulation 2008;117:e25–e146. PCR genotyping system. Nucleic Acids Res 2006;34:e128. 3. Mensah GA, Ryan US, Hooper WC, et al. Vascular endothelium summary 33. Wald NJ, Hackshaw AK, Frost CD. When can a risk factor be used as a statement II: cardiovascular disease prevention and control. Vascul Phar- worthwhile screening test? BMJ 1999;319:1562–1565. macol 2007;46:318–320. 34. Wald NJ, Morris JK, Rish S. The efficacy of combining several risk factors 4. Hobbs FD. Cardiovascular disease: different strategies for primary and as a screening test. J Med Screen 2005;12:197–201. secondary prevention? Heart 2004;90:1217–1223. 35. Anderson JL, Horne BD, Kolek MJ, et al. Genetic variation at the 9p21 locus 5. World Heart Association. Cardiovascular disease risk factors, 2009. Avail- predicts angiographic coronary artery disease prevalence but not extent and able at: http://www.worldheart.org/cardiovascular-health/cardiovascular- has clinical utility. Am Heart J 2008;156:1155.e2–1162.e2. disease-risk-factors. Accessed June 12, 2009. 36. Talmud PJ, Cooper JA, Palmen J, et al. Chromosome 9p21.3 coronary heart 6. van Wyk JT, van Wijk MA, Sturkenboom MC, Moorman PW, van der Lei disease locus genotype and prospective risk of CHD in healthy middle-aged J. Identification of the four conventional cardiovascular disease risk factors men. Clin Chem 2008;54:467–474. by Dutch general practitioners. Chest 2005;128:2521–2527. 37. Brautbar A, Ballantyne CM, Lawson K, et al. Impact of adding a single 7. Clark LT, Ferdinand KC, Flack JM, et al. Coronary heart disease in African allele in the 9p21 locus to traditional risk factors on reclassification of Americans. Heart Dis 2001;3:97–108. coronary heart disease risk and implications for lipid-modifying therapy in 8. American Heart Association. Heart disease & stroke statistics, 2009. 2009 the Atherosclerosis Risk in Communities Study. Circ Cardiovasc Genet update-at-a-glance. Available at: http://www.americanheart.org/downloadable/ 2009;2:279–285. heart/1240250946756LS-1982%20Heart%20and%20Stroke%20Update.042009. 38. Paynter NP, Chasman DI, Buring JE, Shiffman D, Cook NR, Ridker PM. pdf. Accessed July 8, 2010. Cardiovascular disease risk prediction with and without knowledge of ge- 9. World Health Organization. International statistical classification of diseases netic variation at chromosome 9p21.3. Ann Intern Med 2009;150:65–72. and related health problems, 10th Revision. Version for 2007. Available at: 39. Pencina MJ, D’Agostino RB Sr, D’Agostino RB Jr, Vasan RS. Evaluating http://www.who.int/classifications/icd/en/. Accessed January 12, 2009. the added predictive ability of a new marker: from area under the ROC curve 10. Wang TJ, Gona P, Larson MG, et al. Multiple biomarkers for the prediction of to reclassification and beyond. Stat Med 2008;27:157–172; discussion 207– first major cardiovascular events and death. N Engl J Med 2006;355:2631–2639. 212. 11. Casas JP, Cooper J, Miller GJ, Hingorani AD, Humphries SE. Investigating 40. Visel A, Zhu Y, May D, et al. Targeted deletion of the 9p21 non-coding the genetic determinants of cardiovascular disease using candidate genes and coronary artery disease risk interval in mice. Nature 2010;464:409–412. meta-analysis of association studies. Ann Hum Genet 2006;70:145–169. 41. Wilson PW, D’Agostino RB, Levy D, Belanger AM, Silbershatz H, Kannel 12. Pearson TA, Manolio TA. How to interpret a genome-wide association WB. Prediction of coronary heart disease using risk factor categories. study. JAMA 2008;299:1335–1344. Circulation 1998;97:1837–1847. 13. Humphries SE, Yiannakouris N, Talmud PJ. Cardiovascular disease risk 42. Wang TJ. New cardiovascular risk factors exist, but are they clinically prediction using genetic information (gene scores): is it really informative? useful? Eur Heart J 2008;29:441–444. Curr Opin Lipidol 2008;19:128–132. 43. Cook NR. Statistical evaluation of prognostic versus diagnostic models: 14. Ioannidis JP. Personalized genetic prediction: too limited, too expensive, or beyond the ROC curve. Clin Chem 2008;54:17–23. too soon? Ann Intern Med 2009;150:139–141. 44. Hlatky MA, Greenland P, Arnett DK, et al. Criteria for evaluation of novel 15. Teutsch SM, Bradley LA, Palomaki GE, et al. The Evaluation of Genomic markers of cardiovascular risk: a scientific statement from the American Applications in Practice and Prevention (EGAPP) initiative: methods of the Heart Association. Circulation 2009;119:2408–2416. EGAPP Working Group. Genet Med 2009;11:3–14. 45. Janssens AC, van Duijn CM. Genome-based prediction of common diseases: 16. Watt A, Cameron A, Sturm L, et al. Rapid reviews versus full systematic methodological considerations for future research. Genome Med 2009;1:20. reviews: an inventory of current methods and practice in health technology 46. Helgadottir A, Thorleifsson G, Manolescu A, et al. A common variant on assessment. Int J Technol Assess Health Care 2008;24:133–139. chromosome 9p21 affects the risk of myocardial infarction. Science 2007; 17. Haddow JE, Palomaki GE. ACCE: a model process for evaluating data on 316:1491–1493. emerging genetic tests. In: Khoury M, Little J, Burke W, editors. Human 47. McPherson R, Pertsemlidis A, Kavaslar N, et al. A common allele on Genome Epidemiology: a scientific foundation for using genetic information chromosome 9 associated with coronary heart disease. Science 2007;316: to improve health and prevent disease. Oxford: Oxford University Press, 1488–1491. 2003:217–233. 48. Kathiresan S, Voight BF, Purcell S, et al. Genome-wide association of

Genetics IN Medicine • Volume 12, Number 12, December 2010 783 Palomaki et al. Genetics IN Medicine • Volume 12, Number 12, December 2010

early-onset myocardial infarction with single nucleotide polymorphisms and profiling for the prediction of coronary heart disease. Am Heart J 2009;158: copy number variants. Nat Genet 2009;41:334–341. 105–110. 49. Sheridan SL, Crespo E. Does the routine use of global coronary heart disease 55. Janssens AC, Gwinn M, Bradley LA, Oostra BA, van Duijn CM, Khoury risk scores translate into clinical benefits or harms? A systematic review of MJ. A critical appraisal of the scientific basis of commercial genomic the literature. BMC Health Serv Res 2008;8:60. profiles used to assess health risks and personalize health interventions. Am J 50. King DE, Mainous AG 3rd, Carnemolla M, Everett CJ. Adherence to healthy Hum Genet 2008;82:593–599. lifestyle habits in US adults, 1988–2006. Am J Med 2009;122:528–534. 56. Lohmueller KE, Pearce CL, Pike M, Lander ES, Hirschhorn JN. Meta- 51. Sposito AC, Ramires JA, Jukema JW, et al. Physicians’ attitudes and analysis of genetic association studies supports a contribution of common adherence to use of risk scores for primary prevention of cardiovascular variants to susceptibility to common disease. Nat Genet 2003;33:177– disease: cross-sectional survey in three world regions. Curr Med Res Opin 2009;25:1171–1178. 182. 52. Schedlbauer A, Schroeder K, Fahey T. How can adherence to lipid-lowering 57. Ntzani EE, Rizos EC, Ioannidis JP. Genetic effects versus bias for candidate medication be improved? A systematic review of randomized controlled polymorphisms in myocardial infarction: case study and overview of large- trials. Fam Pract 2007;24:380–387. scale evidence. Am J Epidemiol 2007;165:973–984. 53. Wald NJ, Law MR. A strategy to reduce cardiovascular disease by more than 58. NIH News, National Human Genome Research Institute. Study to probe how 80%. BMJ 2003;326:1419. healthy younger adults make use of genetic tests, 2007. Available at: 54. van der Net JB, Janssens AC, Sijbrands EJ, Steyerberg EW. Value of genetic http://www.genome.gov/25521052. Accessed July 6, 2010.