Noninvasive Imaging Tests
The imaging tests most commonly used to assess prognosis are 99m
Tc-sestamibi or thallium-201 stress single-photon emission computed tomography perfusion scans and stress echocardiography. Numerous studies have shown that stress imaging confers added predictive value over Duke treadmill scores for the diagnosis of coronary artery disease,44
and coronary artery calcium scores have been shown to more accurately predict coronary event rates than FRS alone in intermediate-risk individuals.45
However, to date, none of these tests have been shown to reliably predict events over a short interval of 1 year. Stress imaging identifies chronic abnormalities of coronary perfusion resulting from hemodynamically significant lesions and possibly the effect of impaired endothelial function but not the vulnerability of plaques most likely to trigger acute coronary events.
Coronary computed tomography angiography with the latest-technology scanners appears promising in diagnosing the presence of significant stenoses. However, approximately two thirds of myocardial infarctions arise from plaque rupture in coronary artery segments that previously had <50% stenosis.46
Some measure of total plaque volume may be more helpful in predicting such events. It has been demonstrated that computed tomography angiography correlates well with intravascular ultrasonography for measuring coronary artery plaque area.47
However, systematic longitudinal studies linking the extent of plaque and plaque morphology, as assessed by computed tomography angiography, with acute events are not yet available.
Plaques that ultimately rupture or erode, with superimposed thrombosis leading to ACS and other ischemic events, often do not limit flow, and current techniques cannot yet distinguish plaques that will remain silent from those that will trigger thrombosis and clinical events. Advances in anatomic imaging technologies may eventually allow an assessment of plaque architecture and composition. Future functional imaging approaches that assess, for example, the activities of plaque macrophages may also help identify plaques likely to trigger clinical events. More detailed studies of the mechanisms of plaque disruption and the cellular and biochemical events associated with this process are needed.
There has been progress in the development of new magnetic resonance imaging contrast agents that target various aspects of atherosclerosis and thrombosis.48
They have potential to identify the vulnerable plaque, but their diagnostic accuracy and prognostic accuracy have not yet been established. New imaging approaches such as in vivo 2-photon microscopy49–51
are being developed to provide new cell-level information on disease processes that should provide a whole new perspective on the remodeling of the vascular wall in vivo rather than inference from pathology, genomics, proteomics, or cell culture approaches.
The commonly assessed cardiovascular risk factors—lipid profile, hypertension, diabetes mellitus, homocysteine, lipoprotein (a), body mass index, small dense low-density lipoprotein particles, and fibrinogen level—are all inherited to some degree. Human genetics studies are now focusing on identifying sites in the genome, or loci, with variations associated with each of these quantitative factors. The list of loci implicated in these phenotypes and, by extension, for cardiovascular end points is growing.
Until recently, most studies of genetic predictors of vascular disease were largely unsuccessful because they considered too few candidate loci and/or too few functional variants in each locus. This has changed dramatically with the completion of the Human Genome Project, the continuing efforts of the International HapMap Project, and the availability of resources for deep sequencing of candidate genomic loci in large numbers of individuals.
The latest studies attempt to consider all common variations in the loci of interest through dense genotyping of single-nucleotide polymorphisms, consideration of all common haplotypes (sets of physically linked polymorphisms), or selection of representative single-nucleotide polymorphisms that act as proxies for all polymorphisms in a locus. Because the population impact of a disease-associated single-nucleotide polymorphism is a function of both the magnitude of the effect size and the frequency of the single-nucleotide polymorphisms, attention has initially focused on common variants. Whole-genome association studies, with ≥500 000 single-nucleotide polymorphisms tested across the genome, are underway; however, these studies represent a starting point, with identification of a list of candidate loci, that then requires more conventional functional studies of the genes in or near the implicated loci.
Although it is intuitive that a genetic “chip” summarizing genotype data at many risk alleles may improve prediction of cardiovascular risk, it remains to be demonstrated that genetic studies will be useful for this purpose. Prior investigations have succeeded in finding highly penetrant, mendelian rare genetic variants that result in familial dyslipidemia disorders that cause premature CVD (eg, LDLR
) or result in electrophysiological syndromes that predispose to sudden death (eg, KCNQ1
, and KCNJ2
for long-QT syndrome; KCNH2
for short-QT syndrome; SCN5A
for Brugada syndrome; PRKAG2
for Wolff-Parkinson-White syndrome).52,53
The more recent whole-genome association studies have identified common genetic variants that are associated with modestly increased cardiovascular risk (eg, chromosome 9p21 locus),54,55
although the responsible genes remain to be identified. These common variants may explain much of the inherited basis of CVD and sudden death.
Although knowledge of these DNA variants may eventually be useful in improving risk prediction algorithms, they will most likely be relevant to predicting lifetime cardiovascular risk because the variants do not change over time and represent genetic “exposures” to which a given individual has been subjected while in utero; through infancy, childhood, and adolescence; and into adulthood. Thus, the variants themselves are unlikely to meaningfully predict risk in the time frame of months to years. However, an individual’s set of genetic variants may provide the milieu on which other risk factors may confer increased near-term cardiovascular risk. For example, an individual with a particular variant of a QT syndrome gene may have normal risk of ventricular arrhythmia at baseline but may be at severe risk of arrhythmia if given a QT-prolonging drug, whereas the same drug would promote little risk in a normal individual. As so-called “pharmacogenomic” information becomes available, there may be utility to its inclusion into near-term risk algorithms.
Proteomics is the study of the proteome or the protein complement of a sample comprising all or part (subproteome) of cells, tissue, or a body fluid such as serum or plasma. Although proteomic analysis can provide insights into the molecular mechanisms of disease at the protein level, it also has the potential to identify specific disease biomarkers.
The 2 proteomic strategies for biomarker discovery include a broad-based “direct” approach, in which proteomic techniques are used to screen large numbers of proteins directly in serum or plasma to identify those that correlate to a disease phenotype, and the candidate “indirect” biomarker approach, in which proteins are preselected on the basis of known biological assumptions or from prior discovery. Either way, all biomarkers must be validated, most often with immunobased assays on a series of large independent cohorts. This validation phase is critical in the cardiovascular system, in which biomarker identification is complicated by the fact that heart function is influenced by and influences many other organ systems, making identification of robust markers difficult without an understanding of this interplay. Hence, it is important to identify and eliminate biomarkers that are generic “illness markers” or that overlap with other potentially confounding disease origins (eg, diabetes mellitus).
In contrast to DNA variants, protein expression and activity in cells, tissue, and body fluids can be quite mutable over time, with fluctuations over time intervals as brief as minutes. Thus, it is more plausible for variations with proteins to be causative and predictive of near-term cardiovascular risk than variations in DNA. As such, proteomics approaches are much more likely than genomics approaches to identify novel factors that will improve near-term risk prediction algorithms.
Gene Expression Studies
Although the genetic information encoded in the genome is stable and, for the most part, does not change over an individual’s lifetime, expression of the roughly 25 000 genes at the RNA level is highly variable and, like proteins, can readily reflect short-term physiological changes. Although it is not practical to obtain samples of most tissues to measure gene expression profiles, easily accessed cells may permit large clinical studies. For example, data from other fields of medicine suggest that gene expression data from whole blood or isolated mononuclear cells may have significant predictive power.56
Blood gene expression profiling can classify individuals with atherosclerosis, heart failure, and early allograft rejection after cardiac transplantation.57–59
Thus, gene expression analyses may offer a whole new class of biomarkers for use in near-term risk prediction and is an important area for future investigation.