By Michael H. Crawford, MD, Editor
SYNOPSIS: A study of the utility of a polygenetic score for coronary artery disease risk was compared to the pooled cohort equation (PCE) for determining which primary prevention patients would benefit from statin use. Only at the top 5% risk stratum did the genetic score exceed an odds ratio for coronary disease of 2. Using these data would increase statin recommendations by 4% vs. the PCE.
SOURCE: Aragam KG, Dobbyn A, Judy R, et al. Limitations of contemporary guidelines for managing patients at high genetic risk of coronary artery disease. J Am Coll Cardiol 2020;75:276-280.
Genetic risk prediction is a goal of precision medicine. Polygenomic risk scores (PRS) are available, but their performance in the clinical setting is unclear.
Aragam et al assessed more than 47,000 individuals from three northeastern U.S. medical center biobanks and their linked clinical information. The PRS were normalized for genetic ancestry. Strata, such as 5% or 20% risk for coronary artery disease (CAD), were defined. The pooled cohort equation (PCE) was used to calculate the 10-year risk of CAD based on clinical data. The entire cohort was 54% women, 62% of European ancestry, mean age 60 years, and 23% had CAD.
The PRS for CAD was associated with incident CAD (odds ratio [OR], 1.4), as was high PRS (top 20%; OR, 1.9) and the top 5% (OR, 2.3). Among those without CAD (primary prevention), high PRS did not correlate with guideline recommendations for statin use based on the PCE. Also, if a high PRS was employed as a guideline-enhancing factor, an additional 4% of primary prevention patients would be recommended for statin therapy. The authors concluded the prevention of CAD may be enhanced by considering polygenetic susceptibility to CAD in addition to clinical risk.
American Heart Association/American College of Cardiology guidelines for blood cholesterol management recommend the consideration of certain risk-enhancing factors not included in the PCE for determining statin therapy eligibility. These include a family history of premature CAD, a high sensitivity C-reactive protein > 2.0 and a lipoprotein(a) > 50. These and other such factors should meet the criteria of an OR > 2.0.
Aragam et al explored whether genetic testing should be added to the list. A propensity to CAD can be the result of a single gene or, more often, polygenetic. The PRS used in this study is based on 6 million common single nucleotide polymorphisms (SNPs) and has been shown to be predictive of CAD in population studies. In previous attempts to define the potential clinical utility of a PRS in primary prevention populations, such as ARIC, MESA, and the UK biobank, a PRS was not found to be superior to the PCE. One reason for the difference may be these studies were more heterogeneous in the subject’s ancestry. Most subjects in the Aragam et al study were of European ancestry; the same goes in the studies used to develop the PRS. Thus, it works better in a European population. Aragam et al had to adjust the PRS for patients of non-European origin. Based on these earlier studies, those authors concluded that although the risk OR for a PRS were statistically robust, they were not strong enough to be clinically useful.
Does the Aragam et al study challenge this conclusion? Of 47,000 subjects, 16,000 showed no evidence of vascular disease. Of these, about 6,000 recorded an intermediate PRS, and about 1,000 recorded a PRS risk > 20%. An additional 650 were not on statins. Using the PRS would capture 4% more candidates for statins (650 ÷ 16,000 = 0.041). Also, only the top 5% of PRS scores achieves an OR > 2.0 (2.3, to be precise). The top scorers in the Aragam et al study would be 800 individuals (0.05 × 16,000 = 800). However, the cost of a genetic profile is now < $100, and it only has to be performed once. Perhaps a PRS would be a cost-effective addition to the PCE. But what if we add other measures of risk, such as coronary calcium scores? Then, we drive up the cost.
There were limitations to the study. The data for the PCE were pulled from charts and could be inaccurate. The data were limited to three health centers with fairly homogeneous European ancestry populations. However, results from the three centers, when considered individually, were not significantly different, and one center in New York did feature a greater ethnic mix. There were limited data on other risk enhancers, so comparison studies could not be conducted. Aragam et al focused on those with a PRS strata of > 20%; other cutpoints may have changed the results. The biobank volunteers from the three centers may represent a biased sample. Finally, this was an associative study; a randomized, controlled study with cardiovascular disease outcomes would have to be conducted to establish PRS as a solid enhancer that changes outcomes. With the wide availability of genetic testing, we may be increasingly confronted by patients who have completed such assessments, along with a cardiac CT scan they underwent at the mall and perhaps other commercially available tests to assess their risk. I use these data if presented to me, but the only test I order in difficult decision cases is a coronary calcium score. I do consider family history as an enhancer. PRS may be just a more sophisticated family history, which is perhaps more accurate.