Prediction of incident coronary artery disease in individuals with zero coronary artery calcium using a novel multi-ancestry, label-free polygenic risk score framework

This study demonstrates that a novel, label-free, multi-ancestry polygenic risk score framework (8 Billion) can identify individuals with a coronary artery calcium score of zero who still harbor a significantly elevated risk of future coronary artery disease, thereby refining preventive strategies beyond imaging alone.

The Gist

The Big Picture: The "Zero" Score Trap

Imagine you are driving a car, and you pull into a mechanic for a safety check. The mechanic looks under the hood and says, "Your engine is clean! No rust, no cracks, no warning lights. You're good to go for another 10 years."

In the medical world, this is what a Coronary Artery Calcium (CAC) score of 0 means. It's a scan that looks for hardened, calcified plaque (rust) in your heart arteries. If you have a score of 0, doctors usually tell you, "You are low risk. You don't need cholesterol-lowering medication yet."

But here is the problem: Some people with a "clean" engine (CAC=0) still crash (have a heart attack) later on. Why? Because the engine might be clean of rust, but the fuel system might be fundamentally flawed from the factory.

The New Tool: The "Genetic Blueprint" (PRS)

This study introduces a new way to look at that "factory flaw." It uses something called a Polygenic Risk Score (PRS).

Think of your DNA as a genetic blueprint you were born with.

• Traditional PRS: Most current tools look at your blueprint and say, "Oh, you look like a European, so we'll use the European rulebook," or "You look like an African, so we use the African rulebook." This is like trying to fit a square peg into a round hole; it ignores the fact that many people are a mix of many backgrounds (like a "mixed" blueprint).
• The "8 Billion" Framework: The researchers built a new tool called "8 Billion." Instead of forcing you into a box labeled "European" or "African," it looks at your blueprint and finds your specific genetic neighbors in a massive database. It says, "You aren't just 'European'; you are genetically similar to these specific 500 people." It treats human genetics like a smooth gradient (a spectrum) rather than distinct islands.

The Experiment: Who is actually at risk?

The researchers took a group of people who all had a "clean" heart scan (CAC=0). They then used their new "8 Billion" tool to check their genetic blueprints.

They found two groups:

1. The "Low Genetic Risk" Group: Their blueprints looked safe. They truly were low risk.
2. The "High Genetic Risk" Group: Even though their heart scans were clean, their blueprints showed they were born with a high susceptibility to heart disease.

The Results:

• The "High Genetic Risk" group was 3 times more likely to have a heart attack than the "Low Genetic Risk" group, even though they started with the exact same "clean" scan.
• Their 10-year risk of a heart attack jumped to over 8-9%, which is high enough that doctors usually would prescribe medication.
• Meanwhile, the "Low Genetic Risk" group stayed safely below that threshold.

The Twist: Why didn't the scan show it?

You might wonder, "If they are at high risk, why didn't the scan show rust?"

The study suggests that calcium (rust) is a late-stage warning sign.

• The Analogy: Imagine a house fire.
  • CAC (The Scan): Detects the charred wood (calcium). If there's no char, the house looks safe.
  • PRS (The Blueprint): Detects the faulty wiring (genetic risk).
  • The Reality: A house can have terrible, dangerous wiring (high genetic risk) that hasn't burned down the walls (calcium) yet. But if a spark happens (a heart attack), it can happen fast because the wiring is bad.

The study found that high genetic risk was linked to non-calcified plaque (soft, dangerous gunk that hasn't hardened yet). This is the "hidden danger" that the standard scan misses.

What Does This Mean for You? (The Takeaway)

1. Don't ignore the "0" score, but don't trust it blindly. A zero calcium score is great news, but it doesn't mean you are invincible.
2. Your DNA matters. If you have a "clean" scan but a "high-risk" genetic blueprint, you might still need treatment (like statins) to prevent a future crash.
3. Personalized Medicine. The new "8 Billion" tool is better because it doesn't force you into a racial box. It looks at your unique genetic mix to give you a more accurate risk score.

In short: If your heart scan says "All Clear," but your genetic blueprint says "Danger," the blueprint might be the one telling you to be careful. This new method helps doctors listen to the blueprint so they can protect you before the rust even starts to form.

Technical Summary

1. Problem Statement

• Clinical Gap: A Coronary Artery Calcium (CAC) score of 0 is currently the strongest negative predictor for short- to intermediate-term Coronary Artery Disease (CAD) and is often used to defer lipid-lowering therapy. However, a subset of individuals with CAC=0 still experience clinical events, indicating "residual risk" not captured by imaging alone.
• Methodological Limitation: Existing Polygenic Risk Scores (PRS) typically rely on predefined, discrete ancestry labels (e.g., European, African, Asian). This approach fails to account for the continuum of human genetic diversity, particularly in admixed populations, potentially obscuring genetic similarity and reducing predictive performance.
• Objective: To determine if a novel, label-free PRS framework can identify clinically meaningful residual risk in individuals with a baseline CAC score of 0, thereby refining preventive decision-making beyond imaging.

2. Methodology

• Study Population:
  • Cohort: Multi-Ethnic Study of Atherosclerosis (MESA).
  • Inclusion Criteria: Participants with a baseline CAC score of 0 and at least one follow-up CAC measurement.
  • Exclusion Criteria: East Asian and Admixed American groups were excluded from the primary analysis due to insufficient event counts after filtering.
  • Final Sample: 1,700 participants of African, European, or Admixed ancestry.
• The "8 Billion" Framework:
  • Concept: A label-free approach that models genetic similarity as a continuum rather than assigning individuals to fixed ancestry groups.
  • Mechanism: For each MESA participant, a "genetically similar reference neighborhood" is identified within the UK Biobank dataset using a kinship-based distance matrix.
  • Optimization: Neighborhood size is optimized via logistic regression and Brier score minimization (using 1,000 bootstrap resamples) to calibrate individualized PRS estimates without pre-labeling ancestry.
• Risk Stratification: Two PRS-based groups were evaluated:
  1. Top 5%: Individuals with the highest 5% of PRS-derived odds ratios in the cohort.
  2. Individualized High-Risk: Defined by a personalized threshold (≥2-fold risk) derived from the 8 Billion framework.
• Statistical Analysis:
  • Primary Outcome: Incident CAD events.
  • Secondary Outcome: Progression to CAC ≥100.
  • Models: Multivariable Cox proportional hazards models adjusted for Principal Components (PC1–PC4), age, sex, smoking, systolic blood pressure, lipids (total/HDL), diabetes, and antihypertensive medication.
  • Risk Estimation: 10-year absolute risk estimates were derived from adjusted models.

3. Key Contributions

• Novel Framework: Introduction of the "8 Billion" framework, which eliminates the need for discrete ancestry labels, allowing for individualized PRS calibration across diverse and admixed populations.
• Resolving Residual Risk: Demonstrated that genetic risk stratification can identify high-risk subgroups within the "low-risk" CAC=0 population, a group traditionally considered safe from immediate intervention.
• Decoupling Calcification from Events: Provided evidence that high polygenic risk predicts clinical CAD events even when it does not predict the progression of calcified plaque (CAC ≥100), suggesting PRS captures susceptibility to non-calcified, vulnerable plaque.

4. Key Results

• Incident CAD Risk:
  • Top 5% PRS Group: Showed a significantly higher risk of incident CAD compared to the remainder (Hazard Ratio [HR] = 3.12; 95% CI, 1.05–9.31; P=0.041).
  • Individualized High-Risk Group: Also showed significantly increased risk (HR = 2.52; 95% CI, 1.12–5.66; P=0.025).
• 10-Year Risk Estimates:
  • High-PRS individuals exceeded the 7.5% threshold (commonly used to initiate statin therapy) despite having CAC=0.
  • Top 5% Group: ~9.0% 10-year risk.
  • High-Risk Group: ~8.1% 10-year risk.
  • Remainder: ~4.2–4.6% 10-year risk.
• Conventional Risk Factors: In the fully adjusted models for the CAC=0 subset, conventional risk factors (smoking, BP, lipids, diabetes) were not statistically significant, whereas PRS remained a strong independent predictor.
• CAC Progression: PRS was not significantly associated with progression to CAC ≥100 (Top 5% HR = 1.50, P=0.386). Conversely, smoking, diabetes, and antihypertensive use were significant predictors of calcification progression.
• Survival Analysis: Kaplan-Meier curves showed significantly reduced event-free survival for high-PRS groups over the mean 10-year follow-up (Log-rank P < 0.05).

5. Significance and Clinical Implications

• Re-evaluating CAC=0: A CAC score of 0 should not be interpreted as uniformly protective. High polygenic burden identifies a subgroup with residual risk comparable to or exceeding standard treatment thresholds.
• Precision Prevention: Integrating PRS with CAC imaging allows for better risk stratification. Patients with CAC=0 but high PRS may benefit from earlier initiation of lipid-lowering therapy or closer monitoring.
• Biological Insight: The dissociation between PRS and CAC progression suggests that high genetic risk may drive the formation of non-calcified, lipid-rich plaques (which are more vulnerable to rupture) rather than calcified plaque. This supports the potential use of Coronary CT Angiography (CCTA) to detect non-calcified plaque in high-PRS, CAC=0 patients.
• Equity: The label-free "8 Billion" framework offers a pathway to accurate genetic risk assessment in multi-ethnic and admixed populations, addressing a major limitation of current PRS methods that often underperform in non-European groups.

Conclusion: The study validates that a label-free, multi-ancestry PRS framework can identify individuals with CAC=0 who are at high risk for future coronary events, challenging the current paradigm of deferring treatment based solely on a zero calcium score.