Author + information
- Received June 6, 2006
- Revision received October 30, 2006
- Accepted December 4, 2006
- Published online April 17, 2007.
- David T. Miller, MD, PhD⁎,§,1,
- Paul M. Ridker, MD, MPH, FACC†,‡,§,2,
- Peter Libby, MD, FACC‡,§,⁎ ( and )
- David J. Kwiatkowski, MD, PhD⁎,§
- ⁎Division of Hematology, Brigham and Women’s Hospital, Boston, Massachusetts
- †Division of Preventive Medicine and Center for Cardiovascular Disease Prevention, Brigham and Women’s Hospital, Boston, Massachusetts
- ‡Division of Cardiovascular Medicine, Brigham and Women’s Hospital, Boston, Massachusetts
- §Donald W. Reynolds Cardiovascular Clinical Research Center on Atherosclerosis at Brigham and Women’s Hospital and the Harvard Medical School, Boston, Massachusetts
- ↵⁎Reprint requests and correspondence:
Dr. Peter Libby, Donald W. Reynolds Cardiovascular Clinical Research Center, Cardiovascular Medicine, Brigham and Women’s Hospital, Mallinckrodt Professor of Medicine, Harvard Medical School, 77 Avenue Louis Pasteur, NRB 7, Boston, Massachusetts 02115.
Recent rapid advances in genomic tools and techniques hold great promise for transforming the practice of cardiovascular medicine. Resources including the Human Genome Project and the International HapMap project, major technological advances in high-throughput genotyping and methods of statistical analysis, and methods for high-throughput gene expression and small molecule profiling allow researchers to confront issues that will fundamentally change the practice of cardiovascular medicine during the 21st century. Genomic, proteomic, and metabolomic studies of complex cardiovascular diseases such as atherosclerosis will bridge epidemiology and basic biology, and promise increased understanding of cardiovascular disease processes. Genetic approaches applied to atherosclerosis will continue to identify genes and pathways involved in the predisposition to and pathophysiology of atherosclerosis. Gene expression profiling refines our understanding of the dynamic nature of the atherosclerotic vascular wall and promises discovery and validation of targets for therapeutic intervention. Opportunities to translate genetic, genomic, proteomic, and metabolomic information into cardiovascular clinical practice have never been greater, but their fruition requires validation in large independent cohorts, achieved only through collaborative effort. Their continued success will depend on ongoing cooperation within the cardiovascular research community.
This state-of-the-art review focuses on research that applies major technological developments in genomic medicine to atherosclerotic cardiovascular disease. High-throughput technologies facilitate identification of genetic, genomic, proteomic, and metabolomic markers of coronary artery disease (CAD) risk that may find a place in clinically useful prediction algorithms. Such risk models would augment the utility of established clinical tools such as the Framingham risk score. However, putative genomic, proteomic, and metabolomic markers will require the same rigor in application as other epidemiologic markers. Ultimately, the translation of such information to clinical practice should enhance “personalized medicine.” The coupling of information gained through genomic, proteomic, and metabolomic methodologies to traditional tools should sharpen our ability to assess and modify management of cardiovascular disease.
Despite steady progress, atherosclerotic cardiovascular disease remains a growing public health burden in developed countries, and advances in cardiovascular research will not realize their full impact unless translated to care of individual patients (1). Hope for determining new therapeutic targets in atherosclerosis increasingly rests on research progress in genetic studies, expression profiling, and proteomics (2–7). Combining new markers of cardiovascular disease with currently available screening tools promises to promote this translational process.
Before a personalized medicine approach to atherosclerosis based on genomics, proteomics, and metabolomics can become reality, researchers must validate novel markers across different cohorts and in relation to various environmental modifiers. The operation of intricate networks of genes, environmental factors, and gene-by-environment interactions further complicates our understanding of the genetic components of atherosclerosis (8,9). Combined genomic approaches, often called genomic convergence, are necessary in atherosclerosis research (7).
Technological revolutions in genetics and genomics have facilitated 2 major approaches to understanding disease pathogenesis and risk. First, the Human Genome Project and International HapMap Project now allow us to obtain with relative speed vast amounts of deoxyribonucleic acid (DNA)-based information applicable to research subjects with atherosclerotic cardiovascular disease. Second, current technology enables collection of information on expression for thousands of genes in vascular cells and tissues under different conditions. The human DNA sequence laid the groundwork for studies of genetic susceptibility to disease, and expression databases assist definition of disease subtypes and variance related to environmental interactions. Although currently less mature technologies, proteomic and metabolomic studies promise to complement genomic approaches. Table 1compares advantages and disadvantages of methods for marker identification.
A patient-specific risk profile for cardiovascular disease generated by knowledge of the genetic underpinnings of human disease risk would likely assist the clinician in providing targeted interventions, e.g., drugs that act on only a subset of the population or avoiding a particular environmental exposure known to interact with a genetic variant. Both examples illustrate personalized medicine. Although current technology enables researchers to amass huge amounts of genomic data in a relatively short time frame, translation to the clinic will require demonstrated effects on outcomes. Indeed, the bottleneck for translation of genomic medicine to everyday practice lies at that intersection of knowledge.
Genetic Studies of Atherosclerotic Cardiovascular Disease
Genetic linkage studies
Using families and anonymous DNA markers, genetic linkage studies correlate inherited genomic regions with inherited familial characteristics. Although the many genes and environmental factors influencing atherosclerosis make direct linkage difficult, numerous examples link phenotypes with atherosclerotic cardiovascular disease. The work of Goldstein and Brown (10) on low-density lipoprotein (LDL)-receptor mutations in familial hypercholesterolemia furnished a foundation for understanding the mechanism of LDL lowering by 5-hydroxy-3-methylglutaryl-coenzyme A (HMG-CoA) reductase inhibitors. However, single-gene defects in lipid metabolic pathways account only for a very small portion of familial and sporadic atherosclerosis (11). In most people, multiple genes and environmental factors determine elevated lipid levels and atherosclerotic events. Among families in which a single gene plays a stronger role, evidence of linkage frequently identifies a chromosomal locus but not single genes for atherosclerotic cardiovascular disease (12–16).
Notably, linkage analysis has successfully identified specific genes that may contribute to cardiovascular event risk, for example, myocyte enhancer factor-2 (MEF2A) with myocardial infarction (MI) risk (3) and arachidonate 5-lipoxygenase activating protein gene (ALOX5AP) with MI and stroke risk (17). Helgadottir et al. (17) showed linkage between the ALOX5APgene region and MI in 296 Icelandic families, including 713 subjects. The ALOX5APgene encodes the enzyme 5-lipoxygenase activating protein (FLAP), which participates in leukotriene synthesis. Such leukotrienes, especially leukotriene B4 (LTB4), mediate inflammation within the vasculature and participate in murine atherosclerosis (18). Helgadottir et al. (17) followed up their suspicions about ALOX5APwith a genetic association study approach (see the following text).
Human families with CAD are not the only resource for linkage analysis. Almost 2 decades ago, linkage analysis among inbred mouse strains identified an atherosclerosis susceptibility region on mouse chromosome 1 (Ath1). This region contains the tumor necrosis factor (ligand) superfamily member 4 (Tnfsf4) gene, also termed Ox40l(19). Atherogenic in mice, OX40L stimulates T cells. Other human studies recently linked the TNFSF4region and the region containing the gene for OX40 (i.e., the OX40L receptor) to CAD (16,20), but did not identify the underlying gene in either case. Comparison of mouse and human genetics ultimately led to TNFSF4, and further studies showed an association between a TNFSF4single nucleotide polymorphism (SNP) and MI risk in women from 2 human cohorts (19). Most recently, another study associated an OX40SNP with MI risk (21), suggesting that this ligand-receptor interaction may provide a therapeutic target.
Genetic association studies
Many genetic and environmental factors likely influence complex diseases such as atherosclerosis. Viewed separately, each factor exerts a relatively small effect. Combined, they might explain increased disease risk. Compared with linkage studies, genetic association studies provide greater statistical power and more detailed knowledge of a genetic region, enhancing our attempt to understand the genetics of complex diseases such as atherosclerosis. In general, these studies seek to identify differences in the inheritance of particular SNP alleles among subjects with a differing clinical phenotype, such as MI (cases vs. control subjects) or differing levels of a biomarker such as C-reactive protein (CRP) or cholesterol (high vs. low levels).
Candidate gene association studies
Association studies may take a candidate gene or genome-wide approach. Such studies often base selection of candidate genes on assumptions about biologically relevant genes. Thus, candidate gene studies are biased against identification of novel genes. For example, several groups including our own published candidate gene association studies on the relationship between SNPs in the CRPgene and CRP levels based on numerous prospective studies showing the influence of baseline CRP levels on cardiovascular risk (22,23). Three studies in particular provide comprehensive coverage of all common CRPSNPs (24–26). Figure 1shows 7 common CRPSNPs (Fig. 1A), illustrating the sequence difference at an SNP within exon 2 (Fig. 1B) and the generation of 6 common haplotypes that result when SNP alleles combine (Fig. 1C).
The SNPs associated with an intermediate phenotype are not always associated with the disease state itself. The third CRPSNP in Figure 1, rs3091244, consistently associates with higher CRP levels in multiple studies (24–27). Despite these consistent associations, the proportion of variance in CRP levels explained by CRPgenotypes is only ∼2%. These SNPs thus have only a modest influence on plasma CRP levels. Family studies indicate that there must be additional genetic factors contributing to CRP levels. To date, candidate gene SNP selection has achieved limited success in finding genetic variants of CRP level, an example of the aforementioned bias against novel gene identification. Case-control studies to date have not been adequately powered to evaluate whether or not these SNPs impact on incident disease (25,26,28).
Testing greater numbers of SNPs can increase the chances of finding a disease-causing variant. Technological advances that reduce genotyping costs now allow researchers to “supersize” their candidate gene studies. A recent study focused on 11,053 SNPs located within genes, reasoning that those SNPs more likely would affect gene function or expression (29). Using a case-control design and 3 rounds of replication, Shiffman et al. (29) identified 4 SNPs associated with risk of MI (all odds ratios [ORs] indicate carriers of 2 vs. 0 risk alleles): the cytoskeletal protein paladin (KIAA0992[OR 1.40]), a tyrosine kinase (ROS1[OR 1.75]), and 2 G-protein coupled receptors (TAS2R50[OR 1.58] and OR13G1[OR 1.40]). Although these SNPs represent risks on par with accepted risk markers such as hypertension and CRP (23), other attempts at replication for this panel of SNPs yielded mixed results with as few as 1 in 5 SNPs having a similar magnitude and direction of effect (30). The determination of the generalizability of genetic associations observed in a given study population requires replication in independent samples, as discussed in detail elsewhere (31).
Combining different genetic methodologies facilitates discovery. For example, genetic association studies help refine regions identified by linkage studies. After identifying linkage between MI and the region containing ALOX5AP, Helgadottir et al. (17) focused more closely on ALOX5AP, as a candidate gene, sequencing the entire gene in 93 affected individuals and 93 control subjects to identify a panel of 48 common SNPs in ALOX5AP. A case-control association study genotyped these 48 ALOX5APSNPs to determine differences in the inheritance of a particular pattern of SNPs, or haplotype, between subjects with MI (n = 779) and control subjects (n = 624). An inherited 4-SNP haplotype, HapA, conferred an increased risk (nearly 2-fold) for MI and stroke (relative risk 1.8, adjusted p = 0.005). The LTB4 production in stimulated neutrophils increased substantially in individuals with MI compared with control subjects, especially in male carriers of ALOX5APHapA, who showed significantly greater LTB4 production compared with control subjects (p = 0.0042). Male carriers of the at-risk haplotype had the strongest association with disease and also had significantly greater production of leukotriene-B4 (LTB4). The HapA also increased risk of stroke (relative risk 1.67, p = 0.000095). To date, an association between ALOX5APand stroke was observed in more than 1 cohort (32,33), but replication studies for ALOX5APhave provided mixed results (34).
Most recently, Helgadottir et al. (35) used additional candidate gene association studies to expand the ALOX5APstory (35). Using a pathway approach, they studied the leukotriene A4 hydrolase (LTA4H) gene, the enzyme responsible for the production of LTB4. They performed DNA sequencing of the LTA4Hgene region (42 kb) in 93 subjects with MI to identify 8 novel SNPs. Considered as single markers, none of those 8 SNPs was significantly associated with MI in further studies. Considering groups of SNPs as haplotype blocks revealed a particular haplotype, called HapK, significantly associated with MI. Each inherited copy of HapK conferred a relative risk of 1.45 (p = 0.035 after adjusting for the number of haplotypes tested).
Replication of genetic associations among subjects of various ethnicities addresses variability in SNP allele inheritance patterns among ethnic groups that could influence the way a particular variant affects disease risk. As one example, Helgadottir et al. (35) studied LTA4Hin 3 independent MI cohorts of European-Americans and African Americans from the U.S. The HapK was significantly associated with MI among the European-American subset of this cohort (relative risk 1.19; p = 0.006), and also conferred higher relative risk among the African-American subset (relative risk 3.57; p = 0.000022). Presumably, underlying genetic and environmental differences between the European-American and African-American subsets explain the difference in LTA4H-associated MI risk, underscoring the need to include different ethnic populations in genetic association studies (36). Rare variants in one population may be relatively common in another, for example, proprotein convertase subtilisin/kexin type 9 serine protease (PCSK9), where 2 nonsense mutations not present in Caucasians confer lower levels of LDL cholesterol and protection from CAD to African Americans (37).
Other efforts to determine genetic associations with atherosclerosis have achieved limited success in explaining a significant proportion of the population-attributable risk (38,39), including several associations summarized elsewhere (3,40). This inherent problem, not limited to the study of cardiovascular disease, occurs in all genetic association studies of complex disease in which many genetic markers individually explain only a fraction of the genetic contribution to disease. Moreover, results from a genetic association study require replication through independent studies, as discussed in detail elsewhere (41).
Genome-wide association studies
As the per-SNP costs of high-throughput genotyping decline, cardiovascular researchers are turning to genetic association studies on an even larger scale. In large prospective cohorts, so-called genome-wide association studies examine hundreds of thousands of SNPs throughout the genome. Such studies, not hypothesis-driven, are suited ideally to discovering previously unimagined pathways for particular diseases.
The most significant disadvantage of genome-wide association studies involves the statistical conundrum of multiple comparisons inherent in simultaneously performing association tests on thousands of markers. A study of 500,000 SNPs with a false-positive rate of 0.1% would generate 500 false-positive results, a very large number that necessitates multiple rounds of replication to confirm an association. Thus, thoroughly replicated results from genome-wide association studies applied to cardiovascular disease and related phenotypes have just now begun to appear (42).
Gene Expression Profiling in Cardiovascular Disease
Validating new discoveries related to atherosclerosis will require multiple complementary approaches (Fig. 2),including gene expression profiling. The SNP association studies cannot determine which SNPs may affect gene expression or function. Gene expression microarray technology, reviewed elsewhere (43), provides a way to scan rapidly specific tissues for patterns of expression among thousands of genes across the genome. Several studies in both human and animal models have applied expression profiling technology to atherosclerosis. The general approach is exemplified by studies such as that by Ma and Liew (44), which used transcriptional profiling of peripheral blood leukocytes among subjects with coronary heart disease (80% to 90% stenosis) versus healthy control subjects to identify 108 differentially expressed genes.
Although SNPs represent static information (the genome), expression profiles are dynamic (the transcriptome) and may show physiological fluctuations (Fig. 2). Expression profiling of atheromata in human aortic tissue collected from 63 heart donors indicated significant differences in gene expression patterns for 212 genes, based on disease severity and lesion location within the thoracic aorta (45). Importantly, the investigators verified transcriptome reliability and accuracy by comparing results obtained by more traditional histological methods. Indeed, expression profiling correctly classified disease severity and location in 93% of samples tested.
Extending those findings, Karra et al. (46) recently reported studies of gene expression during atherosclerotic lesion progression. These studies compared a histologically determined disease state with patterns of gene expression in mouse aortic tissue from wild-type and apolipoprotein E−/− C57BL6/J mice exposed to different diets and at different ages. Gene expression patterns varied according to disease stage: no disease to early disease (197 genes, including many involved in lipid metabolism), early to intermediate disease (146 genes, including many involved in inflammation), intermediate to moderate disease (110 genes), and moderate to severe disease (650 genes).
Importantly, Karra et al. (46) and Tabibiazar et al. (47) provide additional validation to these gene expression patterns by showing consistent cross-species pattern comparisons. Referring the results of their mouse experiments to earlier results in humans (45), Karra et al. (46) identified 40 genes among 650 (p < 0.0001) that significantly correlated with disease progression in both mice and humans. Analyses based on the Gene Ontology database, a bioinformatics tool designed to facilitate clustering of genes based on functional pathways (48), bolstered the concordance between mice and humans in both studies.
Overlapping genes identified in the 2 studies may provide a reliable molecular signature for atherosclerotic disease states (Table 2).Many overlapping genes—including chemokines, chemokine receptors, and cytokine-related genes, which consistently increased in both studies and at various stages of atherosclerotic lesion progression, and major histocompatibility complex molecules such as H2-Eb1and H2-Ab1—associate with inflammation. Both studies identified increased levels of oncostatin M receptor (Osmr) in atherosclerosis. Oncostatin M (Osm) belongs to a cytokine family that regulates endothelial cell production of other cytokines, including interleukin-6, granulocyte colony-stimulating factor, and granulocyte macrophage colony-stimulating factor. Also Osminduces Abca1in HepG2 cells (49) and Mmp3and Timp3gene expression via Janus kinase/signal transducers and activators of transcription signaling (50). Both studies observed that increased levels of Osmr, Jak3, and Abca1associate with atherosclerosis progression. Both studies also detected increased levels of osteopontin (Spp1), a component of cell-mediated immunity (51), and a gene intensively studied in relation to atherosclerosis.
Such unbiased transcriptional profiling approaches discovered genes such as Runx2, an Spp1-related transcript, previously not implicated in atherosclerosis. In addition, Tabibiazar et al. (47) identified several other Spp1-related transcripts not previously associated with atherosclerosis. Both studies identified genes previously associated with atherosclerosis such as Vcam1and Timp1. Additionally, each study identified several other novel genes, and some appeared in both studies. For example, Tabibiazar et al. (47) and Karra et al. (46) both determined that the differential regulation of Xin, a gene involved in cardiac and skeletal muscle development, associates with atherosclerosis and arterial repair, respectively. Both studies also determined that differential regulation of Sgcg, a gene strongly expressed in skeletal and heart muscle as well as proliferating myoblasts, associates with atherosclerosis.
Transcriptional profiling is especially advantageous in genomic studies designed to detect an environmental stimulus. This extension of the concept to cell culture has permitted investigation of the influence of vascular dynamics on transcripts related to atherosclerosis. Studying cultured human vascular endothelial cells exposed to varying shear stress, oxygen gradient, and oxidized LDL, Warabi et al. (52) showed differential expression of several Nrf2-related genes in response to laminar flow, underscoring the advantage of precise environmental regulation in cell culture for transcriptional profiling.
Detection of transcripts in circulating cells offers convenient clinical application of transcriptional profiling, but may not reflect gene expression in atherosclerotic lesions. Yet one such strategy may nonetheless offer novel insight into the mechanisms of coronary thrombosis. Transcriptional profiling of platelets from patients with acute coronary syndromes offers the ability to examine gene transcription that occurred in megakaryocytes some weeks before the onset of symptoms. This approach has identified myeloid-related protein-14 as a novel marker of coronary risk (53).
How do transcriptional profiling experiments impact cardiovascular therapeutics? Validated markers that emerge from gene expression profiling might improve risk stratification and personalization of treatment decisions. Gene expression profiling may aid identification of drug targets and validation of candidate drugs for the management of atherosclerosis. As proof of principle, Tuomisto et al. (54) showed that HMG-CoA reductase (the target of statin drugs) increased in atherosclerotic plaque. Using laser capture microdissection to isolate macrophage-rich tissue from human atherosclerotic lesions, they compared expression profiles to normal intimal tissue and THP-1 macrophage-like cells and identified augmented expression of 72 genes including HMG-CoA reductase. Studying the transcriptional effect of statin exposure on peripheral monocytes, Waehre et al. (55) provided evidence showing that statins inhibit expression of inflammatory cytokine interleukin-1β, normally present at high levels among subjects with CAD (55). Such findings show the potential of genomic techniques to identify additional drug targets and provide greater understanding of their effects.
In another study based on expression profiling, CRP augmented the expression of matrix metalloproteinase (MMP)-1 and MMP-10 in human endothelial cell transcriptional profiling experiments (56). The MMPs are participants in plaque disruption and thrombosis. In another study, purified CRP caused increased expression of genes related to programmed cell death such as GADD153in cultured vascular smooth muscle cells (57). Apoptosis of vascular smooth muscle cells occurs in atheromata (58). Thus, expression profiling described a molecular link between inflammation and atherosclerotic lesions and provided a means for testing the effect of statins or other compounds on apoptosis-associated genes, pointing to novel therapeutic approaches to atherosclerosis.
Proteomic and Metabolomic Profiling in Atherosclerosis
Even if replicated, association or transcriptional profiling studies cannot assess potentially important post-transcriptional variables including alternative splicing of mRNA, control subjects on protein translation, and post-translational processing of proteins (Fig. 2). Protein markers may provide more accurate real-time information about pathophysiology than stable germ-line markers such as SNPs. As with genomics, potential benefits to the clinical community include better tools for diagnosis (cardiac biomarkers) and identification of therapeutic targets. The National Heart, Lung, and Blood Institute has supported this effort by establishing several Proteomics Centers (59), and several recent reports highlight progress and further challenges in this area of cardiovascular research.
Challenges in the application of proteomics to cardiovascular disease begin with the selection of tissue samples. For example, one study that sampled endarterectomy sections containing atherosclerotic plaque determined decreased expression of heat shock protein-27 (HSP27) in plaque compared with healthy tissue, and confirmed these results by showing a similar trend for the amount of soluble HSP27 in plasma of subjects with atherosclerotic cardiovascular disease (60). Additionally, recent evidence suggests a possible connection between HSP27 and atherosclerosis at the level of estrogen signaling (61). Others have pointed to confounding in studies of atherosclerotic plaque because of the heterogeneity of cell types (62). Yet breaking down the components of a plaque without inducing experimental artifact also presents challenges. Although circulating plasma may not reflect all aspects of an atherosclerotic lesion, it offers accessibility and reproducibility of sample collection. Given the overwhelming abundance of certain proteins such as albumin in blood, ferreting out changes in levels of much less abundant proteins such as cytokines and growth factors presents a formidable technical challenge.
As in the case of genomic markers, proteomic studies require rigorous validation of technology platforms and experimental results. Providing a foundation for study interpretation, the Human Proteome Organization initiated a Plasma Proteome Project. The pilot phase identified 345 cardiovascular disease-related proteins in human plasma (63,64). These catalogs determine additional novel proteins that might associate with cardiovascular disease in future proteomic discovery experiments. Such databases accelerate the identification of unknown markers present in atherosclerotic cardiovascular disease.
Proteomic studies on isolated plasma lipid fractions have yielded new insight into the composition of LDL and high-density lipoprotein particles, identifying 3 proteins previously not associated with LDL and 2 proteins not previously associated with high-density lipoprotein (65,66). Davidsson et al. (67) described unique patterns of LDL-associated apolipoproteins in subjects with type 2 diabetes and subclinical peripheral atherosclerosis compared with healthy control subjects, suggesting that particular distributions of LDL-associated apolipoproteins in subjects with type 2 diabetes could contribute to the increased incidence of atherosclerotic cardiovascular disease in that group.
Metabolomics seeks to quantify small molecules that serve as physiological indicators within circulating plasma or particular cells and tissues. The potential list of small molecules runs into the thousands and includes carbohydrates, peptides, lipids, and metabolic intermediates such as amino acids, organic acids, and drug metabolites. Quantification is generally based on various methods of spectroscopy or chromatography, techniques that could provide rapid high-throughput results at relatively low cost in clinical practice. Noninvasive sampling of plasma to identify real-time markers of CAD has been reported (68), but clinical applications of this technology must first overcome technological and statistical limitations in simultaneously detecting myriad compounds across a broad concentration range.
As proof of principle, Sabatine et al. (69) used mass spectrometry-based technology to identify differences in plasma metabolites among 18 subjects with ischemia induced by exercise stress testing compared with nonischemic individuals who also exercised; specifically, changes in 6 metabolites, including citric acid, accurately differentiated cases from control subjects (p < 0.0001). Although requiring further study, combining small molecule profiling with the arsenal of genomic and proteomic technologies should provide additional diagnostic information and improve our ability to identify new therapeutic targets. Identifying relevant proteomic and metabolomic markers will benefit from comparison with genetic and genomic data. Future experiments should compare fluctuations of cardiovascular biomarkers with other epidemiologic factors such as age, gender, ethnicity, and a variety of environmental exposures already recognized to influence disease risk and outcome.
Scientific Community Genomic Resources
Genetic and genomic studies of atherosclerosis are moving forward rapidly thanks to sharing of resources such as the Human Genome Project and The International HapMap (i.e., haplotype map) Project. HapMap recently completed systematic genotyping of 3 million SNPs in each of 269 subjects representing 4 different ethnic groups, offering high-resolution information about inheritance patterns of common SNP alleles throughout the genome (70). Such information facilitates development of high-throughput genotyping technologies allowing researchers to perform genome-wide association studies at ever lower costs per sample. Detailed genotype information in HapMap facilitates fine mapping of areas identified in linkage studies and provides a valuable community resource for patterns of genetic variation in diverse ethnic groups, thus simplifying the statistical analysis of genetic association studies. Community efforts such as the MicroArray Quality Control Project aim to improve reliability and validity of gene expression data (71), and the same will be necessary for proteomics and metabolomics.
Genomic Data Integration
Integrating various forms of genomic data, so-called genomic convergence, would yield a whole greater than the sum of its parts. Overlapping signals from linkage analysis, association studies, and expression profiling would facilitate identification of relevant markers. Identifying markers through multiple experimental approaches reduces bias and strengthens validity. Such nonbiased approaches are of particular interest in the discovery of genomic markers that impact complex traits, such as atherosclerosis. More importantly to clinical translation, cross-referencing with proteomic and metabolomic profiles could identify markers more easily measured in a clinical setting.
Comparisons of such large data sets are biased toward false-positive results. Statistical methods for meeting this challenge are manifold and reviewed elsewhere (72,73). In general terms, approaches to large genomic data sets include data normalization, filtering, and correction algorithms for the numerous significant p values generated by thousands of comparisons. At best, these measures will help refine the list of prime candidates for further replication. Identifying clinically applicable genomic, proteomic, and metabolomic markers requires consistent replication across populations, a challenge best addressed through collaborative research efforts.
Clinical Applications of Genomics to Cardiovascular Medicine
Genomic-based cardiovascular risk prediction models
Cardiovascular genomic, proteomic, and metabolomic research will continue to identify molecular markers—genotypes, RNA expression levels, proteomic and metabolomic markers—associated with cardiovascular disease, thus revealing new therapeutic targets. In atherosclerosis, many factors influence disease predisposition and therapeutic response, providing an appropriate testing ground for a personalized medicine approach based on convergence of information. Future risk assessment models will likely incorporate a patient’s genomic, proteomic, metabolomic, and environmental information, using statistical models to identify marker-disease associations and correct for confounders such as gene-environment interactions (i.e., patients with a particular marker will have high risk only if exposed to a particular stimulus). Avoidance of that stimulus offers a more tractable intervention for both patient and physician.
Data networks that combine various forms of information will greatly assist this process by identifying, for example, metabolomic markers that vary in response to a patient’s genetic makeup and environmental exposures. Decision algorithms will harness information generated by medical and family histories, clinical criteria such as the Framingham Score, and multiple genetic, genomic, and biomarker tests (Fig. 3).Clinicians might wonder why we are not yet testing patients for genetic markers such as ALOX5APor TNFSF4SNPs. Quite simply, our current knowledge of the risk attributable to these variants is applicable only to a population, not to individuals. Providing patient-specific risk information based on genomic markers awaits a more comprehensive understanding of how these variants interact with each patient’s genetic background and other risk factors.
Genomics for Identifying Therapeutic Targets
Newer genome-wide approaches may uncover additional novel markers of disease risk that represent new drug targets. The human genome likely contains more than 22,000 genes (see the preceding text), of which current medications target only ∼500 (74). Genomic approaches can identify novel drug targets. Unfortunately, the path from potential target discovery to development of useful therapeutics presents considerable challenges. Microarray experiments might identify multiple potential targets, but prioritization of that list presents a challenge. Given a list of 100 or 1,000 candidate genes, researchers likely will focus on the ones they recognize; however, this bias could overlook viable targets.
However, progress in CAD gene identification drives development of potential therapeutics. For example, treatment of 191 subjects who carried at-risk variants in the ALOX5APor LTA4Hgenes with a FLAP inhibitor reduced levels of LTB4, a biomarker associated with MI risk (75), PCSK9 inhibitors seem a ripe target for adjunct therapy in patients in whom LDL targets are not achieved on statins either as monotherapy or combined with cholesterol absorption inhibitors, or who cannot tolerate high-dose statins. Moreover, studies of LTA4Hand PCSK9among African Americans may increase our understanding of ethnicity’s contribution to CAD.
Summary for Clinicians
Rapid advances in genomic, proteomic, and metabolomic technology provide researchers with valuable tools for understanding the genetic predisposition to atherosclerotic cardiovascular disease. Genetic, proteomic, and metabolomic markers will add diagnostic accuracy and perhaps predict the onset and severity of atherosclerotic cardiovascular disease. Early results from genetic association studies and genomic profiling of gene expression provide proof of concept for consistent and reproducibly associated predisposition of genomic markers to cardiovascular disease.
Clinical translation will require validation in prospective studies that encompass large, ethnically diverse cohorts. Combining classical epidemiology with modern genomics will yield unprecedented insight into mechanisms of disease, and also will generate ever more risk markers. Performing tests on patients is relatively easy. Despite test sophistication, high-throughput automation will allow clinical laboratories to offer testing at costs similar to most diagnostic imaging studies. Interpreting such test results will truly provide a challenge. Indeed, prospective validation of new risk markers likely will limit the rate of progress more than any technical or experimental factors.
Genetic, genomic, proteomic, and metabolomic tests will add to rather than replace the clinical utility of traditional cardiovascular risk markers. Elements of the Framingham Score will remain relevant. However, newer information will temper and refine the relative importance of traditional markers. Framingham Scores do not include family history data, leaving room to incorporate genetic and genomic information into the screening process. Because DNA-based genetic markers remain static throughout life, predictive genetic tests could be performed at very young ages to facilitate early intervention. On the other hand, gene expression profiling and serum biomarkers provide real-time information that integrates current environmental influences, and likely aid diagnostic categorization and the early detection of disease. Again, a coordinated screening approach that maximizes the benefits of each screening modality will provide the most clinical utility.
Finally, genomic approaches have already identified potential drug targets. Inhibitors of FLAP, OX40L-OX40 binding, MMP, and PCSK9 are all attractive candidates for CAD risk modifiers. Although inherent challenges of bringing any new drug to the clinic will influence the pace of introduction of therapeutic changes, genomic approaches to cardiology promise to bring lasting improvements to patient care. Clinical translation of genomic, proteomic, and metabolomic information will require collaborative efforts within the cardiovascular disease community at both bench and bedside.
↵1 Dr. Miller received fellowship support from the National Heart, Lung, and Blood Institute (NHLBI 1 F32 HL78274-01).
↵2 Dr. Ridker is also supported by grants HL43851, HL63293, and HL58755 from the NHLBI, with additional support from the Leducq Foundation (Paris, France) and a Distinguished Clinical Scientist Award from the Doris Duke Charitable Foundation (New York, New York). Dr. Ridker is listed as a co-inventor on patents held by the Brigham and Women’s Hospital that relate to the use of inflammatory biomarkers in cardiovascular disease and diabetes. Dr. Ridker also has received investigator-initiated research support from AstraZeneca, Roche, Dade-Behring, Novartis, and Sanofi-Aventis, and has been a consultant to AstraZeneca, Novartis, and ISIS. Cardiovascular Genomic Medicine series edited by Geoffrey S. Ginsburg, MD, PhD.
This work was supported by a grant from the Donald W. Reynolds Foundation (Las Vegas, Nevada).
- Abbreviations and Acronyms
- 5-lipoxygenase activating protein gene
- coronary artery disease
- C-reactive protein
- 5-lipoxygenase activating protein
- 5-hydroxy-3-methylglutaryl-coenzyme A
- heat shock protein-27
- low-density lipoprotein
- leukotriene A4 hydrolase gene
- leukotriene B4
- myocardial infarction
- matrix metalloproteinase
- odds ratio
- oncostatin M
- oncostatin M receptor
- proprotein convertase subtilisin/kexin type 9 serine protease
- single nucleotide polymorphism
- tumor necrosis factor (ligand) superfamily, member 4
- Received June 6, 2006.
- Revision received October 30, 2006.
- Accepted December 4, 2006.
- 2007 American College of Cardiology Foundation
- Topol E.J.
- Bijnens A.P.,
- Lutgens E.,
- Ayoubi T.,
- Kuiper J.,
- Horrevoets A.J.,
- Daemen M.J.
- Breslow J.L.
- Harrap S.B.,
- Zammit K.S.,
- Wong Z.Y.,
- et al.
- Mehrabian M.,
- Allayee H.,
- Wong J.,
- et al.
- Kathiresan S.,
- Larson M.G.,
- Vasan R.S.,
- et al.
- Zee R.Y.L.,
- Michaud S.,
- Hegener H.H.,
- Diehl K.A.,
- Ridker P.M.
- Lohmussaar E.,
- Gschwendtner A.,
- Mueller J.C.,
- et al.
- Zee R.Y.,
- Cheng S.,
- Hegener H.H.,
- Erlich H.A.,
- Ridker P.M.
- McCarthy J.J.,
- Parker A.,
- Salem R.,
- et al.
- Herbert A.,
- Gerry N.P.,
- McQueen M.B.,
- et al.
- Cook S.A.,
- Rosenzweig A.
- Seo D.,
- Wang T.,
- Dressman H.,
- et al.
- Karra R.,
- Vemullapalli S.,
- Dong C.,
- et al.
- Langmann T.,
- Porsch-Ozcurumez M.,
- Heimerl S.,
- et al.
- Li W.Q.,
- Dehnade F.,
- Zafarullah M.
- Ashkar S.,
- Weber G.F.,
- Panoutsakopoulou V.,
- et al.
- Healy A.M.,
- Pickard M.D.,
- Pradhan A.D.,
- et al.
- Tuomisto T.T.,
- Korkeela A.,
- Rutanen J.,
- et al.
- Waehre T.,
- Yndestad A.,
- Smith C.,
- et al.
- Montero I.,
- Orbe J.,
- Varo N.,
- et al.
- Blaschke F.,
- Bruemmer D.,
- Yin F.,
- et al.
- Geng Y.J.,
- Libby P.
- Granger C.B.,
- Van Eyk J.E.,
- Mockrin S.C.,
- Anderson N.L.,
- Working Group Members
- Martin-Ventura J.L.,
- Duran M.C.,
- Blanco-Colio L.M.,
- et al.
- Miller H.,
- Poon S.,
- Hibbert B.,
- Rayner K.,
- Chen Y.-X.,
- O’Brien E.R.
- Davidsson P.,
- Hulthe J.,
- Fagerberg B.,
- et al.
- Sabatine M.S.,
- Liu E.,
- Morrow D.A.,
- et al.
- Drews J.
- Genetic Studies of Atherosclerotic Cardiovascular Disease
- Gene Expression Profiling in Cardiovascular Disease
- Proteomic and Metabolomic Profiling in Atherosclerosis
- Scientific Community Genomic Resources
- Genomic Data Integration
- Clinical Applications of Genomics to Cardiovascular Medicine
- Genomics for Identifying Therapeutic Targets
- Summary for Clinicians