CardioExchange Archive

CardioExchange welcomes Jeffrey Anderson to discuss his recent editorial in the Journal of the American College of Cardiology on  the 9p21 locus and CHD. Dr. Anderson and co-author Benjamin Horne carefully evaluated the relationship of 9p21 to CHD  and conclude that 9p21 appears to be an initiator of and may be a promoter of CHD, but is not a precipitator of disease. (Scroll to the bottom of this post for more background information: Dr. Anderson has helpfully placed 9p21 within the broad perspective of recent efforts to use genetics to better understand risk.)

As you point out in your editorial, very few genetic variants have been consistently observed in association with CHD aside from 9p21, despite multiple investigative efforts.  What do you think are the main reasons for this?  Do you think that additional variants may be identified in the future with the advent of newer technologies?

Frankly, the main reasons for this failure to discover the major proportion of genetic risk are unknown. However, a growing number of investigators see the alternative “common disease—uncommon variant” hypothesis as increasingly appealing and with growing experimental support. Other explanations may involve undiscovered insertions/deletions, copy number variations, epigenetic factors, and gene-gene and gene-environment interactions. Advanced genome-wide association studies (GWAS) chip technology is now becoming available to identify insertions/deletions and copy number variations as well as to more densely cover regions of special interest (“hot spots”) of disease association. An even more exciting and rapidly emerging technology, which holds promise to unravel these possibilities, is whole-genome sequencing. Its cost has fallen dramatically within the last decade, so that now it can be done for as little as $5,000 per subject, with an anticipated further fall to <$1000 in the next 5 years, a figure comparable to the current cost of GWAS. Further, GWAS is limited to finding associations with common variants (minor allele frequency >3-5%), and then only identifies a genomic region of interest (the associated single nucleotide polymorphisms [SNPs] are rarely pathogenic but simply disease markers). In contrast, whole-genome sequencing allows for direct discovery of even rare variants/mutations, including the pathogenic one(s).

Given all that we have learned about 9p21, these newer data supporting its association with atherogenesis versus unstable plaque are interesting.  Notwithstanding the need to validate these findings prospectively, how do you envision these findings might eventually be applied in clinical practice at some point in the future?

The discovery of the 9p21.3 locus already has begun to increase scientific insights into the stage-wise development of CHD, with studies such as that of Dandona et al. and others showing that it impacts atherogenesis, but not plaque instability, with precipitation of acute coronary syndromes. A second application might involve its use in risk assessment as a novel risk factor, even without knowledge of its mechanism of action. Several groups, including our own, already have explored this potential application, but with mixed results: Some, but not others, have found it to provide useful predictive information incremental to standard risk factor (e.g., Framingham) assessment. In general, when predictive, it has been more useful in improving the categorization of risk at the level of the individual subject (i.e., from intermediate to high or low), than global risk at the population level (i.e., increasing the area under the receiver operating curve). This pattern of behavior also has been observed with many other novel biomarkers proposed in recent years to augment risk assessment. Additional assessment and validation will be required before professional societies such as the American Heart Association (AHA) and the American College of Cardiology (ACC) will be willing to endorse 9p21 for clinical application. However, with additional scrutiny, and if application is restricted to appropriate patient populations (i.e., appropriate intermediate-risk patient groups), it may become a clinical prognostic tool in the future. Finally, with future discovery of the underlying biological basis for its associated risk, the 9p21 discovery may allow for the development and application of novel, targeted therapies.

Given the current state of cardiovascular genetic research, how far away do you think we may be from applying genetic data like these in the clinic or at the bedside?

As genetic discoveries accumulate, polygenic genetic risk scores (GRS) may be developed that are sufficiently well-validated and incremental to standard risk assessment tools in their predictive ability to be approved and recommended for clinical application. We recently assessed the ability of a GRS to predict angiographic coronary artery disease (CAD). Promising variants from the literature and internal discovery efforts from lipid, inflammatory, thrombotic, and vascular development or unknown pathways (i.e., 9p21) were combined into a polygenic GRS. Preliminary results have suggested highly significant predictive ability, with a quartile 4 (vs. quartile 1) GRS increasing CAD risk 2-fold. In multivariable risk models, the GRS contributed independently and similarly in strength to hypertension as a risk factor. Furthermore, the GRS improved traditional Framingham risk classification, yielding a net reclassification improvement (NRI) of 18% overall, and importantly, of 30% in those in the key intermediate risk categories. These NRI improvements compare favorably with literature reports for NRIs for blood pressure and HDL cholesterol. Identifying a high GRS might lead to more aggressive risk-factor management. This would initially entail more aggressive treatment of known risk factors (e.g., lipids, blood pressure), but as the pathophysiological basis of the GRS factors become better known, “personalized” and specifically targeted and highly effective approaches to prevention and treatment should emerge.

Pharmacogenetics represents an even more immediate opportunity for genetic application to cardiovascular disease. The impact of drug-gene interactions is being increasingly recognized as a determinant of individual variability in therapeutic response to many drugs. Indeed, the FDA recently has amended the drug labels for warfarin and clopidogrel to include pharmacogenetic information. Based on the individual genetics of VKORC1 and CYP2C9, modified dosing regimens are proposed for warfarin initiation. For clopidogrel, a “boxed warning” indicates that patients with decreased CYP2C19 function because of genetic polymorphisms metabolize clopidogrel poorly and have higher rates of cardiovascular events after acute coronary syndrome (ACS) and percutaneous coronary interventions (PCIs). Prospective randomized trials have not yet established the efficacy or clinical and cost effectiveness of pharmacogenetic testing for these 2 applications (and AHA and ACC guidelines have not yet formally endorsed testing), however, it seems reasonable, in view of accumulating data and FDA labeling, to consider CYP 2C9 and VKORC1 genotyping as an option when initiating warfarin to improve dosing efficiency, and to consider CYP2C19 genotyping, or alternatively, platelet function testing directly, in patients with ACS and those undergoing PCI if results of testing will alter management. The future undoubtedly holds many other opportunities for pharmacogenetic applications to “personalize” management of cardiovascular disease.

Background: It is believed that patients are predisposed to CHD equally by contributions from environment and genetics. In early studies, the genetics of relatively uncommon, early-onset familial CHD conditions, such as familial hyperlipidemia but not common CHD, were found to follow Mendelian patterns and discovered to be due to high-effect size but rare mutations (e.g., in the LDL-receptor gene). Thinking about genetics of the more common forms of CHD then evolved to a “common disease—common variant” hypothesis, the basis for discovery efforts over the past several years. This assumed that for common, polygenic diseases such as CHD, modest contributions from many common variants in multiple genes explained disease heritability. Under this assumption, a candidate gene approach first was undertaken, exploring common variants, for example, in lipid structural and metabolic genes. For CHD and other common diseases, this approach yielded many initial reports of associations but was frequently fraught with failures of replication or with validation but with diminished effect size.

More recently, GWAS have been applied to this search. Based on high-density arrays (“chips”) that can simultaneously test for 500,000 to 1 million SNPs randomly distributed throughout the genome, GWAS makes no a priori assumptions about the location of genetic risk markers, hence it is not limited by our incomplete knowledge of CHD pathophysiology. Despite great enthusiasm heralding this approach, the yield has been modest. The 9p21.3 locus, with moderate effect size and unknown function, was the first GWAS discovery for CHD to be widely replicated, and it remains the foremost and one of only a few validated GWAS discoveries, despite much subsequent effort. Further, subsequent GWAS discoveries generally have involved loci whose effect-sizes have grown progressively smaller. Indeed, it now appears that current plus future efforts with traditional GWAS and candidate gene research will provide only modest further improvement in, and, overall an incomplete accounting of, the genetic underpinnings of CHD.