Author + information
- †Intermountain Heart Institute, Intermountain Medical Center, Salt Lake City, Utah
- ‡Department of Internal Medicine, University of Utah, Salt Lake City, Utah
- ↵∗Reprint requests and correspondence:
Dr. Benjamin D. Horne, Intermountain Heart Institute, Intermountain Medical Center, 5121 South Cottonwood Street, Salt Lake City, Utah 84107.
Discoveries in human genetics have occurred rapidly over the last 15 years. The Human Genome Project launched a new era of biotechnology that empowered previously impossible research, including genome-wide association studies and whole-genome sequencing (1). Despite early failures and limited advances in medical applications of genetics over this same period, recent rapid advances have kept alive the hope of genetically driven personalized medicine (2). Although personalized care has not developed as dramatically as was hoped in cardiovascular medicine, advances have been made in understanding the inheritance of cardiac risk factors (e.g., cholesterol, blood pressure, and metabolic factors), identifying previously unrecognized risk pathways (e.g., chromosome 9p21.3), evaluating medication safety via pharmacogenetics (e.g., warfarin, clopidogrel), and developing medications for reducing cardiovascular risk (e.g., PCSK9 inhibitors) (1). These innovations would not have occurred without the ability to replicate an initial genetic association, the development of hypothesis-free designs such as genome-wide association studies, and the collaborative work in international consortia that sometimes evaluate dozens of populations (1).
Historically, genetic discoveries in adult cardiovascular medicine were rejected because of the inability to validate an initial finding, in part due to gene selection and limited sample size (3). Often those studies began with a positional candidate gene identified via linkage analysis or a biologically functional candidate gene, and then were carried out in 1 or a few populations (3). Although useful for monogenic diseases, this approach represents a poor design for evaluating genetic influences on common, complex diseases such as coronary heart disease (CHD).
Haptoglobin (Hp) is a plasma protein produced in the acute response to infection or inflammation. Hp binds oxygenated free hemoglobin, protecting the body from oxidative damage. An Hp genetic variant (Hp2-2) produces a dysfunctional protein that is less able to protect against oxidation than the products of Hp1-1 and Hp2-1 variants. Evidence suggests that Hp2-2 is associated with intermediate phenotypes that lead to worse CHD outcomes. Studies of the association of Hp2-2 with clinical CHD endpoints unfortunately have produced mixed results (4). Some evidence suggests that Hp2-2 is associated with a greater risk of myocardial infarction (MI) and coronary artery disease overall, whereas other data suggest that the Hp2-2 effect may be limited to patients with elevated glycosylated hemoglobin (HbA1c), and still other studies failed to find an association (4–7).
Specifically, the proposed interaction between Hp2-2 and HbA1c has produced mixed results. In 2013, Cahill et al. (5) reported that Hp2-2 was associated with CHD in 3 populations but only among subjects with HbA1c ≥6.5%. The number of CHD events (nonfatal MI or CHD death) was relatively low across the 3 studies (93 CHD events total among Hp2-2 subjects) (5). In 2014, Pechlaner et al. (6) did not replicate the finding in a fourth population, with results indicating a weak trend toward a protective effect of Hp2-2 in the context of HbA1c ≥6.5%; however, only 8 total events were found in the Hp2-2 group, and CHD events accounted for fewer than one-half of the composite endpoint (stroke was the other event) (6). We also were unable to validate the Cahill et al. (5) finding of an interaction of Hp2-2 and HbA1c for CHD events in Intermountain Healthcare data, but due to the challenges of reporting negative genetic results (3), these data remain unpublished.
In this issue of the Journal, an evaluation of a new study population (29 CHD events in Hp2-2/HbA1c ≥6.5% subjects) and a reanalysis of the first population (24 CHD events) are presented (7). Although on the surface, this paper is another validation of the Hp/HbA1c interaction, it should be noted that Cahill and colleagues have altered the original hypothesis (Figure 1): they examine whether HbA1c ≥6.5% is associated with CHD outcomes in subjects with the Hp2-2 genotype (instead of whether Hp2-2 predicts events exclusively in people with HbA1c ≥6.5%) (7). In both study populations, elevated HbA1c (a known risk factor) was verified to be associated with poor CHD outcomes overall and was strongly associated with poor outcomes in Hp2-2 patients. However, whereas HbA1c ≥6.5% tended to show a greater ability to stratify risk in Hp2-2 than Hp1-1/2-1 subjects, a significant statistical interaction was not shown (7). In secondary analyses, HbA1c ≥6.5% was especially strong for CHD risk prediction among Hp2-2 subjects in the first 8 years of follow-up compared with the last one-half of the 16-year study period (7).
Importantly, the Cahill et al. (7) paper examines Hp genotype data from a new perspective that may be useful for precision medicine when using genetic markers from across the genome. The study utilized a genetic marker (Hp2-2) to identify a subpopulation within which to evaluate a standard clinical risk marker (HbA1c). This approach is a paradigm shift from the traditional concept wherein a clinical subpopulation (e.g., those with HbA1c ≥6.5%) is studied to determine whether a genetic risk marker (Hp2-2) predicts events (5,6). If the gene or genes utilized for this unusual approach are carefully tuned to the clinical therapeutic area and the cost–benefit analysis is favorable, genetic testing could represent a powerful tool for risk stratifying patients based on the likelihood of responding to treatment (e.g., identifying groups in which lowering of HbA1c is more successful using current medications) or based on the risk of CHD events (e.g., identifying high-risk groups in which a more intensive patient care plan is justified by a greater potential benefit).
Although this risk stratification approach is intriguing, it creates a plethora of new questions: Can the trends reported be replicated and strengthened (or refuted)? Is Hp the best gene to use for this clinical application when many other genes could be considered? Should a genetic risk score of HbA1c-related genes be used instead? Also, would it not be better to use a genome-wide or genome-sequencing approach to find all important variants? The use of genetics as a first-line stratification tool before or simultaneously with evaluating standard risk markers could also be applied elsewhere (e.g., cholesterol- or blood pressure–lowering). By contrast, is it possible to obtain equally excellent, but lower-cost, risk stratification using tools (8) that do not require genetic information? The hurdles for this gene-based stratification method are high.
Finally, this study of Hp genotypes highlights the difference between scientific validity and clinical usefulness (7). The drive to validate genetic variants for prediction of CHD events (e.g., Hp or the chromosome 9p21.3 CHD locus) continues to overshadow the limited demonstrated practical value of clinical genetic testing (i.e., proven ability to guide preventive therapy leading to improved outcomes). In the case of warfarin pharmacogenetics, even the consistent replication of genetic associations with health outcomes has not yet translated into clinical usefulness, in part due to the costs and timeliness of genetic testing in the presence of marginal incremental clinical benefit. Considerable influx of capital investments into cardiovascular genetics has not yet overcome the clinical usefulness issues. At present, human genetics remains primarily a source of knowledge regarding human biology and potential drug targets (2).
In conclusion, the application of genetics for patient risk stratification is a relevant ongoing goal in adult cardiovascular medicine. The potential use of genetic information to better target diagnosis and treatment, as evaluated by Cahill et al. (7), is timely in this era characterized by an emphasis on efficient medical practice, but it will require substantial additional methodological development. Although we agree with Cahill et al. on the need for further replication of their specific findings, we propose that hypothesis-free genetic methods should be applied to identify all genetic variants that may be useful for risk stratification, including methods that evaluate copy number variants. The clinical usefulness of genetic testing should be particularly considered in research studies to determine whether adding the complexity and financial costs of genetics to medical care is warranted.
↵∗ Editorials published in the Journal of the American College of Cardiology reflect the views of the authors and do not necessarily represent the views of JACC or the American College of Cardiology.
Both authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- American College of Cardiology Foundation
- Ginsburg G.S.,
- Shah S.H.,
- McCarthy J.J.
- Cahill L.E.,
- Levy A.P.,
- Chiuve S.E.,
- et al.
- Pechlaner R.,
- Kiechl S.,
- Willeit P.,
- et al.
- Cahill L.E.,
- Jensen M.K.,
- Chiuve S.E.,
- et al.
- Horne B.D.,
- Anderson J.L.,
- Muhlestein J.B.,
- Ridker P.M.,
- Paynter N.P.