Polygenic Risk Score IDs People at Risk for Early MI, Death

Genetic susceptibility to CAD, in the form of a polygenic risk score (PRS) incorporating more than 1.14 million variants, is predictive of incident MI and all-cause death, although its performance varies by age and sex, a UK Biobank analysis shows.

The PRS had the strongest relationship with outcomes in younger individuals and in men, with the tightest link found in men ages 41 to 50, researchers led by Hasanga Manikpurage, MSc (Université Laval, Québec, Canada), report. There was also significant net reclassification improvement when the PRS was added to models that included only clinical risk factors, particularly in younger men.

The findings, published online last week in Circulation: Genomic and Precision Medicine, open the possibility that the score could be used to identify at-risk individuals early on, giving clinicians enough time to implement more-aggressive preventive measures to stave off adverse outcomes.

“Of course, lifestyle modifications remain the basis of the interventions,” senior author Sébastien Thériault, MD (Université Laval), told TCTMD, noting that prior research has shown that positive behavioral changes—like eating healthier and getting more exercise—are associated with better outcomes even in people with high genetic risk.

Studies also have shown that people with elevated risks according to a PRS derive more benefit from medications. Thus, a PRS like the one evaluated here might be used to select patients who would most benefit from costly therapies like PCSK9 inhibitors, Thériault suggested.

He acknowledged, however, that even though there is potential, PRS-guided prevention is not yet ready for clinical use. Genotyping arrays are not currently used in clinical settings, and there are several polygenic risk scores that have been developed, he pointed out. So, he explained, the methods for measuring the genetic variants underlying the scores need to be made available to clinicians; there needs to be a “standardization and harmonization” step to determine the best way to deploy a PRS-guided approach; and work needs to go into the optimal way to communicate the information to patients and their doctors.

“It’s in an early stage, but I think it’s showing very promising results that could be eventually applied,” Thériault said.

UK Biobank

To develop a PRS, the investigators turned to the UK Biobank, a prospective study with more than 408,000 participants ages 40 to 73 (mean age 56.9 years; 54.1% women) that has provided genotype data to the scientific community. Using a Bayesian approach called LDpred, they created a PRS incorporating more than 1.14 million single nucleotide polymorphisms previously associated with CAD.

In a validation analysis, the PRS was associated with the likelihood of CAD in the overall cohort—OR 1.56 (95% CI 1.54-1.58) per standard deviation (SD) increase in score. The relationship, however, was stronger in men than in women (OR 1.62 vs 1.45). Findings were similar when looking at MI, MI/revascularization, and atherosclerotic CVD as outcomes; the association was weaker but still significant for ischemic stroke (with no difference between men and women).

Manikpurage et al then assessed how well the PRS performed when it came to predicting incident events over a median follow-up of 11 years. They found that each SD increase in PRS was associated with a greater incident MI risk (HR 1.53; 95% CI 1.49-1.56), with a stronger relationship in the youngest age group, 41 to 50 years (HR 1.89; 95% CI 1.77-2.02).

Sex made a difference, too, with the tightest correlation seen in the youngest men (HR 2.00; 95% CI 1.86-2.16). It was in this group that the consideration of the PRS on top of the pooled cohort equations led to the greatest net reclassification improvement when it came to predicting incident MI.

The optimal cutoff for picking out high-risk individuals was determined to be the 90th percentile of PRS values. At this level, the HR for incident MI was over 2 for both men and women, although the relationship remained stronger in men. “All in all, the 90th percentile . . . could be used as a threshold to identify high-risk individuals for selection in clinical trials and the use of more-aggressive preventative measures,” the authors write.

Significant predictive ability of the PRS also was seen for all-cause mortality, with associations of lesser magnitude than those seen for incident MI.

Why the Sex, Age Differences?

Thériault said his team expected to see a correlation between the PRS and incident MI, but it was more surprising to observe a sex-based difference in the strength of the association. He speculated that the disparity stems from the makeup of the genome-wide association studies on which the PRS was based; those studies included more men than women.

“The use of sex-specific association data could potentially improve the predictive accuracy of PRS, notably for women,” he and his colleagues note in their paper.

The superior performance of the PRS in younger age groups is consistent with prior research showing that these scores are more accurate for early-onset versus late-onset events, Thériault noted. He explained that at younger ages, the genetic component of risk may take on a disproportionately larger role compared with environmental factors, which could be more important later in life.

One potential use for PRSs could be to identify patients with very high risks of heart disease comparable to those seen in people with monogenic CVDs like familial hypercholesterolemia, Thériault suggested. Putting the PRS cutoff at a high level might identify patients who would benefit from early treatment but who would otherwise not be identified by other risk-prediction tools. “There aren’t that many other ways to identify these people other than using these polygenic risk scores,” he said.