Predicting Heart Trouble

Ruzanna Harutyunyan's picture
Advertisement

As a long-time Joslin Diabetes Center investigator, Alessandro Doria is well aware that nearly one in three patients with type 2 diabetes will suffer from clinically significant coronary artery disease. He would like to see the day when physicians like himself can accurately predict which patients are at the greatest risk before their vasculature starts to clog with dangerous plaque. Using familiar measures like blood pressure, cholesterol level, and a detailed medical history, physicians can roughly estimate a patient’s risk of disease. But soon, they may be able to refine their assessments by routinely incorporating other variables into the metric of risk—genotyped risk loci.

As the fruits of genomewide association studies continue to accumulate, Doria, who is an HMS associate professor of medicine at Joslin and an HSPH associate professor in the Department of Epidemiology, and other researchers are scrambling to put the information to use—both in the lab and in the clinic. While some researchers explore how genetic associations illuminate the molecular pathogenesis of complex diseases like coronary artery disease, others attempt to determine when and how genetic information can improve patient care directly through early identification and intervention.

“We’ve gone from a relatively small number of genetic associations to a large number in a very short amount of time,” said Christopher O’Donnell, an HMS associate clinical professor of medicine at Massachusetts General Hospital, who is the associate director of the Framingham Heart Study and scientific director of the SNP Health Association Resource (SHARe) Program of the National Heart, Lung, and Blood Institute. “We are in a phase now of taking stock of what these associations mean and how to use them in clinical practice.”

Dangerous Intersections

Ironically, although the identification of more genetic risk factors for complex diseases like coronary artery disease is expected to improve the accuracy of risk assessment, it makes the computation itself quite daunting. “We really need to figure out how to most appropriately put these single nucleotide polymorphisms together into risk scores,” said O’Donnell, explaining that it is not just a matter of adding up the risk factors because they can interact in unexpected ways.

A recent study by Doria and his colleagues at HMS, HSPH, and Caritas St. Elizabeth’s Medical Center in Boston illustrates how risk factors intersect. In this case, they pinpoint a diabetic subgroup at particularly high risk for coronary artery disease. Reporting in the Nov. 26 Journal of the American Medical Association, the researchers describe how diabetes interacts with a genetic variant on a region of chromosome 9, known as 9p21.

“The 9p21 risk variant is the strongest and most reproducible genetic risk factor [for coronary artery disease] identified so far,” said co-author and HSPH professor of nutrition and epidemiology Frank Hu by e-mail. The risk variant, first reported in 2007, has been associated with a 25 to 50 percent increase in coronary artery disease in the general population, depending on the number of copies carried by an individual. “The present study is the first to show that this [risk variant] also predicts coronary artery disease in [patients with] diabetes.”

“We weren’t sure what to expect,” said Doria. “Since diabetes is such a strong risk factor, we thought perhaps it didn’t matter whether you have the risk variant or not.”

He and his colleagues genotyped the 9p21 locus in 734 diabetic patients, roughly half with angiographically diagnosed coronary artery disease and half with no evidence of the disease. As previous studies have reported for the general population, the researchers found that diabetic patients homozygous for the 9p21 risk variant were more likely to have coronary artery disease. But, in diabetics, the increased risk conferred by harboring the variant was even greater: 50 percent for carriers of one copy and 140 percent for carriers of two copies.

Advertisement

“We were intrigued when we saw that the effect of the two risk factors, diabetes and the 9p21 risk variant, was more than additive,” said Doria. “Why would these two factors synergize?”

Delving deeper into their data, they discovered a clue: only those patients in the highest bracket of glycated hemoglobin, a measure of average blood glucose levels over the preceding months, were at an elevated risk of developing coronary artery disease.

“At this point, we do not know how poor glycemic control synergizes with the 9p21 risk variant,” said Doria. Uncovering the mechanism will probably first require the identification of the causal variant behind the association with the 9p21 polymorphism, which is noncoding.

“Finding the mechanism behind the interaction is important,” said Doria. “That could guide the development of new drugs designed specifically to prevent or delay atherosclerosis in diabetic patients.”

In a related study, Doria and his colleagues found that patients with the 9p21 risk variant and poorly controlled blood sugar also had increased mortality. Following a group of diabetic patients prospectively for 10 years, the researchers found that simply having the risk variant did not increase mortality. Only those individuals who were homozygous for the variant and who had high levels of glycated hemoglobin were more likely to have died. The increased mortality was due specifically to cardiovascular causes, like heart attack and stroke.

“This study illustrates that genetic variants, like 9p21, might have special predictive value in certain subgroups of patients,” said O’Donnell, who was not a co-author on the paper. “And it is likely that these common, relatively low-risk variants are going to be most informative when we know more about these meaningful interactions that can be exploited in prediction and prevention.”

A New Era

The ability to predict an individual’s risk of developing a complex multifactorial disease is one of the ultimate goals of genomic medicine. This predictive power could guide healthcare decisions, enabling physicians and public health professionals to focus resources on those who are most at risk, especially if treatment is costly, has side effects, or is difficult to implement. But developing accurate models of risk assessment is not easy.

“We are entering a new era of modeling risk prediction in which multiple risk factors—genetic and otherwise—must be considered simultaneously,” O’Donnell said. “In addition, clinical trials may be needed to determine if drug treatments personalized to the results of the genetic tests are justified.” In some cases, polymorphisms may not provide additional predictive power or change treatment decisions.

Doria agrees that many of the genetic risk factors identified to date make relatively small contributions to overall coronary artery disease risk compared to traditional risk factors like high blood pressure. “However,” he said, “the magnitude of the risk conferred by the interaction of the 9p21 variant with poor glycemic control is similar to that of major risk factors such as smoking and warrants further investigation into the clinical utility of this interaction.”

Advertisement