Protein Risk Score Alongside Genetic, Clinical Factors Modestly Improves Cardiovascular Disease Prediction

NEW YORK – Using data from an Icelandic population, researchers have developed a protein risk score to predict atherosclerotic cardiovascular disease (ASCVD).

While the score by itself was significantly associated with ASCVD events, when it was added to a model that included clinical risk factors and polygenic risk scores for common heart conditions, it only led to a modest improvement in ASCVD prediction. The findings were published by a team led by researchers from Decode Genetics, a subsidiary of Amgen, in the Journal of the American Medical Association on Tuesday.

“Further research is needed to determine whether the protein risk score and polygenic risk scores are clinically useful for screening purposes,” the authors wrote in their paper. 

In their study, the researchers included two cohorts. One was a group of 13,540 Icelandic individuals with no history of major ASCVD events at recruitment but 1,507 of whom subsequently had a heart, ischemic or hemorrhagic stroke, or coronary heart disease. The other group of participants was from the placebo arm of the FOURIER clinical trial and all had known heart conditions and some of whom were on statin therapy.

For both cohorts, the researchers analyzed the levels of 4,963 proteins in participants’ blood plasma using SomaLogic’s SomaScan platform. They developed a protein risk score using a subset of the Icelandic cohort as a training set and the other subset as the test set, identifying 70 proteins for the risk score. When they applied a model of this protein risk score with age and sex to the test set, the researchers found it was statistically associated with ASCVD events.

But when they added the protein risk score into a model including clinical risk factors — such as statin use, hypertension treatment, type 2 diabetes, and body mass index — and with or without the addition of polygenic risk scores, the researchers only noticed a small improvement in risk prediction.

Meanwhile, the researchers noted that the protein risk score could reclassify individuals between the low- and intermediate-risk groups but they did not find a statistically significant improvement in predicting ASCVD events among those reclassified. Because of this, the researchers said they “cannot conclude that the protein risk score, with or without including polygenic risk scores, has clinical utility for screening programs.”

However, Decode Genetics noted in a statement that protein levels rise and fall dynamically, which might make protein risk scores well-suited to predict the timing of disease-related events. According to the company, a protein risk score could become an important tool in clinical trials to get an early assessment of the efficacy of therapeutic intervention or for monitoring risk.