Genomics, Lipids, Inflammation at Midlife Predict Future CAD

A four-pronged approach to CV risk assessment that integrates genomics with lipid and inflammatory biomarkers tested in midlife is predictive of future coronary artery disease, new data from the UK Biobank confirm.

Notably, Raysha Farah, MD (Massachusetts General Hospital, Boston, and the Broad Institute of MIT and Harvard, Cambridge, MA), and colleagues say, the associations are both age- and sex-specific, allowing for tailored prevention strategies. The findings were published online recently in JACC.

For their four-biomarker model, the researchers considered the coronary artery disease polygenic risk score (CAD PRS), “a measure of genetic predisposition to disease [that] enables earlier risk prediction for CAD across the life course,” plus three other factors: LDL cholesterol, lipoprotein(a), and high-sensitivity C-reactive protein (hs-CRP) levels.

Senior investigator Akl C. Fahed, MD (Massachusetts General Hospital and the Broad Institute of MIT and Harvard), pointed out that all the biomarkers they looked at aren’t obscure and are mentioned or even recommended by guidelines, most recently those for dyslipidemia management.

“It gets very confusing. If you are a patient, you have a clinical risk calculator and then there’s three or four or five other things you can measure [as risk enhancers], but each one gives you a different value. One is high, one is low: how do you put it all together? . . . It’s actually very hard to [do],” Fahed told TCTMD. “What we really wanted to do in this study is say, what if you just measure these four biomarkers together once? How much prediction does that give you if you’re a middle-aged person?”

As it turns out, their model outperforms the more-traditional pooled cohort equations (PCEs) at gauging risk in this context, he said. “What we found is that not only is each one of them predictive, which is something we’ve known, but also when you [combine] them together, they provide additive information.”

Sumeet A. Khetarpal, MD, PhD (UVA Health, Charlottesville, VA), led a similar study published late last year exploring the same four-biomarker concept and using the same CAD PRS, but with a different dataset from the UK Biobank.

He told TCTMD that that study, in turn, had been sparked by an earlier paper by Paul M. Ridker, MD (Brigham and Women’s Hospital, Boston), and colleagues, published in the New England Journal of Medicine, showing that LDL cholesterol, hs-CRP, and Lp(a) levels are strongly linked to MI risk in middle-aged women.

This latest report  is “a very strong paper, a very comprehensive analysis” that usefully confirms the performance of the four-biomarker strategy, said Khetarpal. The consistency among studies is “good for the field,” he added. “Hopefully it provides some strength for the concept of a four-biomarker score in clinical practice.”

UK Biobank Data

The model is based on 215,695 individuals ages 40 to 69 years (mean age 55.9 years; 56.0% women; 88.1% white) at the time of their enrollment in the UK Biobank study. All had their CAD PRS, LDL cholesterol, Lp(a), and hs-CRP levels assessed at baseline.

Over a median follow-up of 12 years, 3% developed CAD. Adjusted for age, sex, and other biomarkers, likelihood of incident CAD was increased when any one of the biomarkers was elevated:

• CAD PRS (HR 1.79; 95% CI 1.70-1.89)
• LDL cholesterol (HR 1.60; 95% CI 1.48-1.66)
• Lp(a) (HR 1.20; 95% CI 1.12-1.29)
• hs-CRP (HR 1.64; 95% CI 1.57-1.72)

Patients were most vulnerable to developing CAD when all four biomarkers were elevated (HR 4.65 versus none; 95% CI 3.90-5.54).

With CAD PRS in particular, the association per standard deviation was stronger for men (HR 1.49; 95% CI 1.45-1.54) than for women (HR 1.37; 95% CI 1.31-1.44; P for interaction ≤ 0.001). With each of the biomarkers, associations were stronger among younger versus older patients.

Researchers then compared their four-biomarker model against the PCE, finding it had a higher C-statistic (0.753 vs 0.740). However, the highest C-statistic (0.756) was reached when integrating both PCE and biomarkers (excluding LDL, because it is included in PCE).

For both men and women, the predictive ability of the four-biomarker model was greater in younger patients. Overall. it led to a 32% continuous net reclassification index in comparison to the PCE.

Using a single measure sampled in midlife, this “biomarker-based prediction model achieves comparable performance with clinical risk calculators and may identify populations who are not detected using a traditional clinical risk calculator, creating an opportunity for improved primary prevention, particularly in younger age groups, where the performance of biomarker-based screening is highest,” the researchers conclude.

“Interpreting risk as a cumulative burden of genetic, lipid, and inflammatory pathways,” they say, “provides a straightforward approach to prioritizing earlier, targeted primary prevention.”

They note that several developments have made this framework possible: the clinical accessibility of CAD PRS testing, progress to phase III trials of Lp(a)-lowering therapies, and the validation of inflammation as a therapeutic target.

Real-world Adoption

Although it’s not yet commonly used, the CAD PRS has begun to trickle into clinical practice. In theory, a physician could use a DIY approach to the four-biomarker model—especially since for each one there’s an understanding of the mechanisms by which it drives CAD—but Fahed said ongoing research will help fine-tune the specifics.

On the plus side, “polygenic risk testing is certainly becoming cheaper,” said Khetarpal. “As adoption grows, I think hopefully . . . it will be easier and easier for practitioners,” eventually reaching beyond specialty use to primary care. Currently, clinicians can order the CAD PRS used in this study and others by contacting Mass General Brigham. The price is $255.

“At the moment, it seems to be that academicians within preventative cardiology are probably the ones that are mostly applying [the CAD PRS], but I think that is changing,” he noted, highlighting Endeavor Health, located in the Chicago, IL, area as well as his own institution, UVA Health, as two systems that have begun to leverage the genetic information.

Khetarpal, to explain how this might guide management, gave the example of a patient who he recently saw. The person was in their early 50s and had experienced three separate CV events despite trying to optimize their risk factors, such that their LDL cholesterol wasn’t especially high. Something like the four-biomarker model might enable better prediction of that risk before such events occur, when the patient is in their 20s or 30s and hasn’t yet raised any red flags, he said.

“You need a test like this to be able to offer,” even if the specific biomarkers chosen for testing evolve over time with better understanding of what to target, said Khetarpal, who mentioned apolipoprotein B and lipid size/composition as possibilities.

You can imagine a world where you can use biomarkers to just decide who in the population should get imaging. Akl C. Fahed

Fahed suggested that any prediction tool in cardiovascular medicine—whether the four-biomarker model or another approach—ideally should have two characteristics. “One, it should be integrated,” meaning that it should combine all the information derived from the included risk factors into a single estimate, he explained. Two, it should be “dynamic,” in that “it changes with you over your lifetime.”

In the more immediate future, the next step will be implementation research, both Fahed and Khetarpal said.

For Fahed, the key will be conducting “clinical trials that start demonstrating that you can use these measurements, particularly, I would say, genetics, to prospectively [show] that there’s benefit in identifying people a better way and [offering] better treatment.” He pointed to the PROACT trials using CAD PRS to identify patients at high risk who are then invited to undergo coronary CT angiography for plaque assessment.

“You can imagine a world where you can use biomarkers to just decide who in the population should get imaging,” in order to spot silent disease before it manifests, said Fahed.