Picture a 34-year-old man in clinic with an LDL-C of 135 mg/dL, a normal blood pressure, no diabetes, and no family history he knows about. The AHA's PREVENT equations could give him a low ten-year estimate while a 30-year estimate offers a longer perspective. The older Pooled Cohort Equations start at 40 and will not score him at all. A low ten-year number does not end good care; guidelines still ask about risk enhancers, lifetime exposure, and lipid treatment where it is indicated. But none of those estimates, at any horizon, answers a different question: whether atherosclerosis is already in his arteries.

REACT put an ultrasound probe and a CT scanner on 16,808 adults spanning ages 18 to 70. One in three men in their thirties had plaque.

The mistake is treating a low ten-year risk estimate as evidence that atherosclerosis is absent. REACT shows how often those two findings can coexist.

A risk score estimates the probability that a person has a heart attack or stroke within a set horizon. An imaging study reports whether plaque is present now. Those are different questions, and a third question sits behind both: whether finding plaque earlier and treating it changes what happens to the patient. REACT measures plaque prevalence, shows how plaque findings relate to calculated risk, and leaves the treatment question unanswered.

Risk calculators were built to do one job: estimate the chance of a cardiovascular event over a defined period from age, sex, blood pressure, cholesterol, smoking, and diabetes. The Pooled Cohort Equations estimate ten-year atherosclerotic risk for adults 40 to 79. SCORE2 estimates ten-year fatal and non-fatal cardiovascular risk for Europeans 40 to 69 without prior cardiovascular disease or diabetes. PREVENT, the newest, estimates ten-year risk of total cardiovascular disease, including heart failure, for adults 30 to 79, and adds a 30-year estimate for adults 30 to 59. They do that job well in populations. None of them was designed to report whether disease is present, and in a young patient the ten-year horizon hides a great deal, because a 30-year-old with poor numbers still has a low ten-year risk when the events are decades away. The 30-year estimate narrows that gap. It still does not look at the artery.

The problem is that the disease is not decades away. It is already there.

We have suspected this since the autopsy series of the 1950s and the PESA cohort in Madrid, which in 2015 found subclinical atherosclerosis in 63% of asymptomatic 40-to-54-year-olds, including 58% of those the Framingham score called low risk. What we did not have was a large, sex-balanced cohort that started imaging at 18 and used the same three vascular territories all the way to 70. That is what REACT built.

REACT enrolled adults aged 18 to 70 with no known atherosclerotic cardiovascular disease into five prespecified age strata, with balanced numbers of women and men in each, at Rigshospitalet in Copenhagen and CNIC in Madrid. Everyone had three-dimensional vascular ultrasound of the carotid and femoral arteries plus coronary CT angiography (CCTA) with a coronary artery calcium (CAC) score. Mean age was 45, 51.4% were women, mean BMI was 26.2, and roughly one in ten was on a lipid-lowering drug at baseline: a metabolically unselected, largely untreated population, not a high-risk clinic. What the paper reports is a baseline cross-section, each participant imaged once.

Silent atherosclerosis, defined as plaque in at least one of the three territories, was present in 57.1% (95% CI 56.3 to 58.0).

The age gradient is the part worth memorizing. Among 18-to-29-year-olds, 8.7% of men and 6.7% of women had plaque. In the 30-to-39 stratum it was 34.6% of men and 21.3% of women. By 60 to 70, 98.1% of men and 91.9% of women had plaque somewhere. Prevalence follows an S-curve with age, and the odds of plaque roughly quadrupled with each decade in both sexes.

The pattern differed by sex. Prevalence among men ran ahead of women's, an offset the investigators describe as roughly five to ten years. The largest differences between adjacent strata for women fell in the forties and fifties, an age range that overlaps the menopausal transition, although REACT was not designed to test menopause as the mechanism.

Then the paper does the thing that matters for practice. It restricts to the participants SCORE2 was derived for, ages 40 to 69 with no prior cardiovascular disease and no diabetes, and asks how the score classified the ones who had plaque on imaging. The denominator is people with plaque in that eligible group, not the whole cohort. Of them, how many did SCORE2 place in the high-risk category?

The answer is 1.9%.

My 34-year-old is not in that analysis. SCORE2 does not apply below 40.

Widen to moderate-or-high risk and 34.1% of plaque-positive participants are included, which still leaves roughly two in three below that line. Among those with plaque in both a coronary and a peripheral territory, the group carrying the most disease, 3.8% were classified high risk; among those with single-territory plaque, 0.5%. None of this is a failure of SCORE2. The score classifies ten-year event risk, and it may be doing exactly that in people whose arteries already contain plaque. What the analysis shows is where people with imaging-detected plaque land on a scale that was never built to see it.

Two more findings change how we should think about screening.

First, isolated coronary disease was uncommon at every age: at most 9.3% of men and 5.0% of women in any stratum had plaque in the coronaries and nowhere else. In the younger strata, plaque was more often peripheral and confined to a single territory, and among participants with coronary plaque, 82% also had carotid or femoral plaque. As a single baseline snapshot, REACT cannot say where any individual's disease started. It can say that in this cohort the neck and legs usually carried plaque when the coronaries did.

Second, the calcium score did not capture much of the coronary disease seen in younger participants. Among 30-to-39-year-olds with coronary plaque on CCTA, 41.8% of men and 48.4% of women had a CAC score of zero. That is the lesson of the natural-history study I wrote about in June, now visible at baseline in a cohort of nearly 17,000: in a young adult, a CAC of zero rules out calcified plaque, not plaque.

Plaque volume, where present, rose steeply with age in every territory; the paper describes the association as exponential. In the younger strata, plaque was mostly small and confined to one territory. In the older strata it was larger and more often present in two or three territories. That shift spans the age range in which a ten-year score still classifies most people as low risk.

Free cardiovascular reference

LDL-C, ApoB, Lp(a), and key outcomes trials in four pages, with absolute event rates, clinical context, and linked sources.

Get the free Cardiovascular Evidence Cheat Sheet →

None of that makes the finding weak. It makes it early, and early is the whole point.

I have been treating cumulative LDL exposure rather than ten-year risk for a while, and REACT is a particularly clear way to show a patient why. Here is how I apply this in practice. These are clinical judgments, not strategies tested by this analysis.

In adults under 50, I no longer let a low calculator output end the conversation. I use it as what it is, a population estimate, and I put the REACT numbers on the table: roughly a third of men and a fifth of women in their thirties had plaque on imaging, and prevalence was higher in the forties. For my 34-year-old, that is the conversation the calculator alone would have skipped.

I practice virtually, so ultrasound is not practical for me, and imaging means what a patient can get locally: a coronary artery calcium score or, when the question warrants it, a CCTA. REACT shapes how I choose between them. Among its 30-to-39-year-olds with coronary plaque on CCTA, nearly half had a CAC score of zero, so in a patient that age I say before ordering a calcium score that a zero rules out calcified plaque, not plaque, and I have a lower threshold to go to CCTA when the concern is real. When a patient arrives with a carotid or femoral ultrasound from another clinician, I read it with REACT in mind: in this cohort peripheral plaque accompanied coronary plaque in 82% of the people who had it, so a positive peripheral study carries information, even though that coexistence does not make ultrasound the proven screening strategy. One caution applies to those outside studies: REACT used three-dimensional vascular ultrasound under a research protocol, and a routine community carotid study has not been shown to match that performance.

A single positive territory changes the plan. Imaging-detected plaque is not the same as guideline-defined clinical ASCVD, and it does not make someone a secondary-prevention patient on paper. In my practice it changes how I weigh cumulative LDL exposure, and it generally pushes me toward more intensive LDL lowering than the calculator alone would justify. The right threshold and target for small-volume subclinical plaque remain unsettled.

I check lipoprotein(a) once in everyone. It is a causal risk factor for atherosclerotic disease that the standard SCORE2 and PCE-style ten-year equations do not include, and it matters most when the disease has arrived earlier than expected.

REACT does not tell us that everyone with a speck of femoral plaque needs a statin tomorrow, and it cannot, because this analysis reports no clinical event outcomes. What it tells us is more basic. More than half of adults with no known cardiovascular disease had plaque on imaging, and among those with plaque in the age range SCORE2 covers, the score classified two in a hundred as high risk. The estimate and the image answered different questions. Whether acting on the image earlier slows the disease is what the randomized phase of REACT is built to measure; whether it prevents events is a question that will outlast it.

The calculator answered the question it was asked. Nobody asked it the other one.

REFERENCES

If these articles help you understand the evidence, you can support the research and writing for $5 a month or $50 a year. The articles remain free for everyone, twice a week. Support this publication, or subscribe free if a colleague forwarded this.