Clinical use of CV genetic risk information may become common in near future

Since President Barack Obama announced the Precision Medicine Initiative in his January 2015 State of the Union address, the idea of personalized medicine for each individual is becoming more of a reality. In this new era of precision medicine, the role genetic risk scores can play in prevention of CVD is receiving a lot of attention.

Researchers have experimented with different algorithms using a variety of combinations of single nucleotide polymorphisms (SNPs) to determine whether there is any value of adding genetic information to traditional risk scores like Framingham.

“Genetic risk factors are independent of age and can be predicted at any time of life, so it gives you the chance to prevent the disease from happening in the first place, instead of after a first cardiac event,” Cardiology Today Editorial Board member Robert Roberts, MD, FRSC, FRCPC, MACC, LLD (Hon.), chair of the International Society for Cardiovascular Translational Research at University of Arizona, said in an interview.

This issue was highlighted in July, when the NIH awarded $55 million to the Precision Medicine Initiative’s cohort program to help generate information about the prevention and treatment of cancer and other diseases based on individual lifestyle, environmental and genetic differences (Read more about the Precision Medicine Initiative, click here).

While members of the cardiology community find the data gathered to date promising, some are more optimistic than others about how quickly genetic information will be implemented into clinical practice.

For Roberts, the time has arrived to look at a prospective study in which genetic risk is used to determine who should receive statin therapy. “I suspect that in the next year or two that indeed is going to happen,” he said.

Julie A. Johnson, PharmD, dean, College of Pharmacy and distinguished professor of pharmacy and medicine (cardiology) at University of Florida, Gainesville, whose work focuses on CV drug pharmacogenomics, including the genetic basis for different responses to CV medications such as warfarin and antihypertensive drugs, said she has a more cautious view of the future of CV genetic risk scores.

Photo credit: UF Health; reprinted with permission.

“A lot of the studies show that numerically, there is a benefit, but it is not clear that that numerical benefit is clinically relevant,” Johnson told Cardiology Today.

Reclassification of risk

In the last 2 years, several studies analyzed whether the addition of genetic information increased reclassification rates for CV risk. One such study, MI-GENES, was presented at the American Heart Association Scientific Sessions in November 2015 and subsequently published in the March 2016 issue of Circulation. The MI-GENES researchers investigated whether disclosing genetic risk for CHD based on 28 variants not related to BP or lipids would lower LDL.

Iftikhar J. Kullo, MD, professor of medicine, College of Medicine, Consultant Division of Cardiovascular Diseases, Mayo Clinic, and colleagues reported that patients whose risk for CVD was disclosed on the basis of a genetic risk score plus conventional risk factors were more likely to have a lower LDL 6 months after disclosure than patients whose risk was estimated based on conventional risk factors alone. The LDL lowering was driven by increased statin prescriptions, as opposed to changes in fat intake or physical activity, and was most pronounced in people who learned they have high genetic risk for CVD.

“Individuals who got the genetic risk score had lower LDL at the 6-month time point, and because LDL is a pretty good surrogate one can expect such an approach would reduce CV events; however clinical trials would be needed to establish an improvement in outcomes,” Kullo said in an interview.

Jessica L. Mega, MD, MPH, chief medical officer of Verily, on leave from the TIMI Study Group of the cardiovascular division at Brigham and Women’s Hospital and Harvard Medical School, Nathan O. Stitziel, MD, PhD, assistant professor from the cardiovascular division of the department of medicine and the McDonnell Genome Institute at Washington University School of Medicine, St. Louis, and colleagues analyzed 48,421 adults from five studies: the Malmo Diet and Cancer Study, a community-based cohort study; ASCOT and JUPITER, randomized controlled trials of statin therapy for primary prevention; and CARE and PROVE IT–TIMI 22, randomized controlled trials of statin therapy for secondary prevention. The analysis was published in The Lancet in 2015.

Adults with the highest genetic risk for CHD derived the greatest benefits from statin therapy, regardless of traditional risk factors. In those cohorts, 3,477 CHD events were reported. Using a genetic risk score based on 27 variants and the endpoints of incident or recurrent CHD, compared with the lowest-genetic-risk quintile, individuals in the three quintiles of intermediate genetic risk (HR = 1.34; 95% CI, 1.22-1.47) and those in the highest-genetic-risk quintile (HR = 1.72; 95% CI, 1.55-1.92) had elevated risk for CHD.

“We showed that the genetic risk score was indeed able to stratify risk for CHD within placebo arms even after adjusting for traditional risk factors. Then we also showed that individuals at higher genetic risk appeared to derive greater benefit from statin therapy compared with those at low genetic risk, and they appeared to have both greater absolute and greater relative benefit from statin therapy,” Stitziel told Cardiology Today.

According to Roberts, this analysis demonstrated that approximately two-thirds of genetic risk is unrelated to traditional risk factors like BP and cholesterol, and that the 27 genetic risk variants used were better at stratifying for risk than known risk factors, including family history.

“I mention family history because all along people have felt that if you took a family history you would probably capture most of the genetic risk,” Roberts said. “That is not true.”

A Stanford University paper published in Genetic Epidemiology in 2015 also has potential to impact clinical practice. The researchers evaluated how to choose the optimal set of SNPs for a genetic risk score for the outcome of CHD and discovered that overall, P-value threshold had the largest impact, with the optimum threshold being around .001. Evaluation metric and how the genetic risk score is to be used also played a role.

Other research published in July in Heart focused on the effectiveness of a genetic score based on 53 SNPs associated with CHD or stroke, either alone or in conjunction with the QRISK-2 tool consisting of several phenotypic measures (See related article, click here).

The net reclassification improvement index was 0.25% (95% CI, –1.33 to 1.83) when the genetic risk score was combined with QRISK-2 for participants with a QRISK-2 of at least 10%. The researchers estimated that applying the genetic risk score to those with a QRISK-2 score of 10% to less than 20% and prescribing statins to those with a QRISK-2 score of greater than 20% would prevent one additional event for every 462 people screened.

“It is important to continue to push on to make the predictors better and to better understand how these can add information on top of existing clinical risk scores,” Stitziel said.

Patient perspective

With any genetic testing, the primary concern is the benefits and drawbacks for the patient, and the main goal is the prevention of disease, experts said.

Roberts used the example of a 48-year-old premenopausal woman with an LDL level of 160 mg/dL, but no other risk factors.

“She would not qualify for treatment according to our guidelines, but if you were to assess genetic risk and she was at increased genetic risk, then you would treat that LDL,” he said.

Roberts said most women prior to menopause have very minimum disease in their coronary arteries, which makes that a good time to treat them. While not everyone should be given a statin, the group at highest risk should be treated, he said.

“Primary prevention is what genetics is made for,” he said.

Stitziel agreed, noting that there now is the potential to look at young adults before clinical risk would be apparent and allocate preventive therapy for those at high genetic risk before the onset of disease.

With an increase in the number of people being prescribed preventive therapy, however, one concern is patient adherence. Roberts referenced asymptomatic individuals in their 30s, 40s or 50s who may not want to take medical therapy.

“People will say, ‘I feel fine, there are side effects to these drugs and I won’t take them.’ And this will be unfortunate,” he said. “Statin therapy is very safe, but what will happen is that these people won’t take a statin until they have first MI or coronary event. At that point, it is certainly effective, but it going to be much more effective if taken before it occurs.”

Other concerns are patient anxiety over the diagnosis and issues of insurability and cost. Kullo said, however, that MI-GENES investigators did not observe any heightened anxiety in participants who received genetic risk information.

A factor, he said, is how the genetic information is presented. Whether presented by a genetic counselor or the physician, the designated health care professional needs to emphasize to the patient that it is a probabilistic estimate, and that they can make other changes to reduce their overall risk.

Johnson said genetic testing for CHD is different from genetic testing for other diseases like Alzheimer’s disease, where there might be undue stress over something that can not be prevented, and discrimination that might affect insurability.

“People usually know if they have a strong family history of CV risk, so I am not sure genetic testing is really going to significantly change someone’s understanding of their risks. And, I think it is not really going to change their insurance risk because the insurance company is already looking at your family history to some degree,” she said.

Building the evidence

The gold standard for clinical implementation in cardiology is the randomized clinical trial, but an undertaking of that scope for a CV genetic risk score would be complex.

“It is hard to imagine such a clinical trial where you would genotype a large number of young people and follow them for decades,” Stitziel said.

Johnson is a principal investigator of Genomic Medicine Implementation, the Personalized Medicine Program under the National Human Genome Research Institute (IGNITE) consortium. “For any genomic medicine implementation we have to first build the evidence, but it doesn’t necessarily have to be a [randomized] controlled design,” she said.

According to Johnson, a randomized controlled trial is costly and not completely real-world, and the study design would dictate what the physician did with the genetic information. In a pragmatic design, she said, the researchers could still randomly assign participants at a system level or a clinic level, but the physician would only be provided the genetic information and the tools to understand it, not given explicit instructions to follow.

“It does have limitations, though,” Johnson said. “It is not as clean and pure as a [randomized] controlled trial, but I think many people feel that for filling that information gap for genomic medicine including pharmacogenomics, the design probably makes more sense just from an efficiency and cost perspective. If we had to do a [randomized] controlled trial to test for every genomic medicine implementation, we will never really get to where we need to be, because it will be too hard and too expensive.”

She said the NIH appears to agree with the sentiment that “relatively large but more pragmatic approaches will probably be the way to go for figuring out the clinical and economic benefits of doing a genomic medicine implementation clinically.”

For Dan Roden, MD, Cardiology Today Editorial Board member and assistant vice chancellor for personalized medicine at Vanderbilt University Medical Center, other barriers include a clear picture of the usefulness of the risk scores, the cost and how cumbersome it will be to implement into clinical practice.

“It will take a lot of medical-record and technical support to interpret those results for health care practitioners. I can order a creatinine and get the result back and know what to do with it, but ordering a genetic risk score is a little more complicated,” he said.

Choosing one particular algorithm to standardize into practice and getting more data on individuals of non-European ancestry are other issues that need to be addressed, Stitziel said.

“There are some nuts-and-bolts practical issues to be worked out before we can think about expanding this clinically,” he said.

Kullo said other logistic barriers exist, like recalculation of the algorithm if new SNPs are identified, and lag time between blood sample and test results, which can take 1 to 2 weeks.

According to Johnson, the cost and whether the value added justifies that cost is the biggest barrier to clinical implementation.

“In our current health system, if there is little value [but cost added], even if it is just only a few hundred dollars, if it is done on thousands or millions of people, that adds a lot to the health system. And if it doesn’t bring clinical value then I think it is still a tough sell,” Johnson said.

Goal to prove clinical value

To prove clinical value, there needs to be a better understanding of reclassification rates, according to Roden.

“The patients who should be targeted are those at medium risk, not high or low risk,” he said. “If you take someone at very high risk because of clinical risk factors — hypertension, current smoking — and the genetic risk score is low, they still have high risk. For people with low risk, the addition of genetic information might not change their classification at all.”

Even if the effect of genetic risk scores is not large in magnitude, it could still be clinically useful, Kullo said.

“These scores still remain modest in terms of relative risk,” he said. “However, genetic risk score is typically independent of conventional risk factors and therefore has promise in clinical application. In contrast, several of the circulating biomarkers such as C-reactive protein are correlated with conventional risk factors.” Even though modest, a 20% to 30% change in relative risk based on a genetic risk score can still be meaningful, given that it is independent of conventional risk factors, he said.

Another important step before implementation can occur, according to Roberts, is the further education of clinicians.

“Genetics is still a fringe area for most physicians, who did not grow up with an understanding or any significant knowledge about genetics. So that is something too that physicians will have to learn, especially as we move farther along with precision medicine,” he said.

Visions for the future

While most experts agree that CV genetic risk information will at some point be implemented into clinical practice, how long that will take is still under debate.

While the national Precision Medicine Initiative will provide a lot of genomic marker information on a lot of patients, Roden said routine implementation “will be a journey, not a leap.”

Roberts predicts that within 5 years, genetic testing will become a routine clinical procedure — not necessarily in all fields, but for preventable diseases like CHD and diabetes.

Kullo, however, sees a shorter journey ahead. “Some would say there is enough evidence at this point to consider pilot studies of its use in clinical practice, particularly in patients at intermediate risk. We are going to study such an approach at Mayo Clinic and have genetic testing available for individuals at intermediate risk based on conventional risk factors.”

The biggest advantage to genetic risk scores, Roberts said, may be that they offer different information from what is currently available, and could help prolong the lives of people who otherwise would not be considered high risk for CHD.

“What all of this really comes down to is genetic risk is now capable of predicting CHD and since it is independent of known risk factors, it offers a whole new ball game,” he said. – by Tracey Romero

• References:
• Goldstein BA, et, al. Genet Epidemiol. 2015;doi:10.1002/gepi.21912.
• Kullo IJ, et al. Circulation. 2016;doi:10.1161/CIRCULATIONAHA.115.020109.
• Mega JL, et al. Lancet. 2015;doi:10.1016/S0140-6736(14)61730-X.
• Morris RW, et al. Heart. 2016;doi:10.1136/heartjnl-2016-309298.

• For more information:
• Julie A. Johnson, PharmD, can be reached at University of Florida, PO Box 100484, Gainesville, FL 32610; email: johnson@cop.ufl.edu.
• Iftikhar J. Kullo, MD, can be reached at Mayo Clinic, 200 First St. SW, Rochester, MN 55905; email: kullo.iftikhar@mayo.edu.
• Robert Roberts, MD, FRSC, FRCPC, MACC, LLD (Hon.), can be reached at University of Arizona, College of Medicine-Phoenix, 550 E. Van Buren St., Phoenix, AZ 85004; email: bobrobertsx2@gmail.com.
• Dan Roden, MD, can be reached at Vanderbilt University, 1285-B MRB4 37232-0575, 2215B Garland Ave., Nashville, TN 37232-0575; email: dan.roden@vanderbilt.edu.
• Nathan O. Stitziel, MD, PhD, can be reached at Washington University School of Medicine, 660 S Euclid Ave., Campus Box 8086, St. Louis, MO 63110; email: nstitziel@wustl.edu.

Disclosure: Johnson, Kullo, Roberts, Roden and Stitziel report no relevant financial disclosures.