NEW YORK (GenomeWeb News) – Traditional risk factors still trump genetic markers for predicting heart disease risk, according to a study of more than 19,300 women appearing online today in the Journal of the American Medical Association.
As part of the Women's Genome Health Study, a prospective study following thousands of female health professionals from around the US, researchers from Brigham and Women's Hospital, McMaster University, and Amgen tested two sets of SNPs — one involving a dozen variants directly linked to cardiovascular disease and another based on more than 100 SNPs linked to related phenotypes such as blood pressure or cholesterol.
After a follow-up time of roughly 12 years, they found that traditional risk information based on factors such as family history and plasma biomarker levels were better for predicting heart disease than either set of genetic variants.
"It was surprising and a little disappointing," lead author Nina Paynter, an epidemiologist and preventative medicine researcher at the Brigham and Women's Hospital, told GenomeWeb Daily News.
By identifying genetic risk factors associated with heart disease and related conditions, Paynter and her colleagues noted, researchers hope to find ways to better predict and prevent such disease. Even so, they added, relatively little research has been done to test the predictive value of multiple risk loci in the general population.
"[T]he predictive abilities of recently discovered genetic markers have not been tested," the researchers noted. "In particular, there has been no evaluation of literature-based genetic risk score for cardiovascular disease."
For the current study, researchers came up with two genetic risk scores: one based on a dozen SNPs directly tied to cardiovascular disease risk and a second involving 101 SNPs linked to either cardiovascular disease itself or related risk factors, including blood pressure, cholesterol and triglyceride levels, diabetes, and so on.
All of the SNPs used came from the National Human Genome Research Institute's GWAS catalog and represented SNPs published between 2005 and June 2009.
To test the predictive power of such markers, the team followed 19,313 Caucasian women for between 11.6 and 12.8 years.
Using blood samples collected from the individuals at the beginning of the study, the team measured traditional heart disease plasma biomarkers — such as cholesterol levels, C-reactive protein levels, and triglyceride levels — and isolated DNA for risk SNP genotyping.
Collaborators at Amgen genotyped samples with the Illumina HumanHap300 Duo+ platform. The team also imputed SNPs not included on the array using the MACH 1.0.16 program and used a questionnaire to glean additional non-genetic information for each individual, including age, smoking behavior, family history, and blood pressure.
During a median follow-up time of 12.3 years, 777of the study participants experienced a cardiovascular event. Both the 101 SNP and 12 SNP sets were roughly associated with cardiovascular events.
But this crude association didn't hold up after researchers adjusted for traditional risk factors, Paynter explained.
In addition, incorporating the genetic information into risk models based on non-genetic risk factors didn't significantly improve the team's ability to classify women based on their heart disease risk.
"Our study finds no clinical utility in a multilocus panel of SNPs for cardiovascular disease based on the best available literature," they wrote.
Meanwhile, in a paper appearing in the same issue of JAMA, a trio of researchers from Brown University and the US Centers for Disease Control and Prevention focused specifically on a chromosome 9 region called 9p21, which houses some of the best known and replicated heart disease variants.
Their meta-analysis of 47 datasets from 22 published papers suggests the 9p21 region shows a small but significant association with heart disease. Still, they noted, studies so far don't support the clinical utility of testing SNPs in this region.
"[S]howing that a genetic test has clinical validity does not necessarily lead to improved health," the team wrote. "Clinical trials need to demonstrate that use of the test is associated with changes in physician management, patient motivation, and long-term behavioral changes, improved health outcomes, and/or reduced costs to the health care system."
The studies come on the heels of a British Medical Journal paper last month, in which researchers from University College London reported that non-genetic factors were more useful for predicting type 2 diabetes than a set of 20 SNPs.
The similar findings for cardiovascular disease and type 2 diabetes likely reflect the fact that both are chronic diseases with complicated risk factors and complex genetics, Paynter explained.
Still, she emphasized, the findings do not diminish the value of finding genetic risk variants. For instance, Paynter noted, such studies can reveal new biological pathways underlying heart disease — some of which may eventually make attractive targets for new treatments or preventative measures. And, she explained, such variants may still prove useful for helping to target existing therapies.
Finally, Paynter didn't rule out the possibility that SNPs with greater predictive power could be identified down the road. "There are definitely more SNPs being discovered," she said. "It'll be a little bit of a 'wait and see' to see what comes out in the next couple years."