A stable plaque: what the word means in a report, and what is done about it

People usually arrive at a carotid ultrasound for some other reason — headaches, neck pain, a routine check-up, "we're here anyway, let's look at the vessels" — and leave with a report that says: atherosclerotic plaque, stenosis up to 40 %, smooth contour, no signs of instability. Below that there may be a line reading "Plaque-RADS 2", or a plaque "type 2". None of those lines answers the question that actually matters: how bad is this, what happens now, and why does the doctor talk about cholesterol rather than surgery at 40 %.

Let's go through what the report actually measures, where the word "stable" comes from, what the degree of narrowing predicts in studies and what the appearance of the plaque predicts instead, and which treatments have been shown to change it in the populations studied — as opposed to those that only change a seller's confidence.

The percentage is not a measure of how much disease there is

The first thing anyone reads is the degree of narrowing. It intuitively parses as "what share of the artery is taken up by disease", and that reading is wrong in two separate ways.

First, the percentage is calculated by more than one convention. The trials of carotid surgery historically used two incompatible methods: the North American one, which compares the residual lumen with the diameter of the normal artery beyond the narrowing, and the European one, which compares it with the estimated diameter of the bulb at the same level. The same plaque gets different numbers under the two, and the difference is not cosmetic — in the moderate range it runs to tens of percentage points. A direct comparison across 1,001 angiograms (Rothwell et al., Stroke, 1994, PMID 7974586) produced a conversion: European % ≈ 0.6 × North American % + 40, so 70 % by the European method corresponds to roughly 50 % by the North American one — and in about half of the measurements the two methods placed the same artery in different severity categories. That is exactly why guidelines expect the degree of stenosis to be reported together with the method used, and why a report giving a bare percentage is ambiguous.

Second, the degree of narrowing is the geometry of the lumen in one cross-section. It says nothing about what the plaque is made of, whether its surface is intact, or how much of it there is along the vessel. As plaque accumulates the artery enlarges outward for a while, so the lumen stays nearly unchanged — meaning a substantial plaque can carry a small percentage.

Practical upshot: the percentage is not a severity score. It is one of several measurements, and it answers "is this obstructing flow", not "is this dangerous".

What the word "stable" stands for

In a report, "stable" almost never means "it isn't growing" — a single study cannot show change over time. It means "it has none of the features that studies have linked to events".

The clearest attempt to agree on what those features are is Plaque-RADS, an expert consensus document from the vascular imaging community (Saba et al., JACC: Cardiovascular Imaging, 2024, PMID 37823860). It is a 1–4 ladder, and it deliberately rests on morphology rather than on the percentage:

• 1 — normal wall, no plaque.

• 2 — plaque present, maximum wall thickness under 3 mm, no complicated features.

• 3 — thickness of 3 mm or more, or a healed ulcer; subtypes distinguish the lipid-rich core and the thickness of the fibrous cap.

• 4 — complicated plaque regardless of thickness: haemorrhage inside the plaque, a ruptured fibrous cap, or intraluminal thrombus.

One detail tends to get lost in retellings: calcification, inflammation, neovascularisation and progression are ancillary features in this scheme — on their own they do not raise the score. "Calcified" and "dangerous" are not synonyms, and that deserves its own section below.

What the ladder predicts is a separate question from how it is built. The consensus only described it; the prognosis was tested afterwards, in a retrospective cohort (Huang et al., JACC: Cardiovascular Imaging, 2025, PMID 39243231): 1,378 people, 987 of them asymptomatic, followed for about five years on average. Five-year event-free survival among the asymptomatic ran RADS 1 — 100 %, RADS 2 — 95.4 %, RADS 3 — 78.5 %, RADS 4 — 61.9 %.

And immediately, what those numbers do not mean. This is a referral imaging cohort: people are in it because someone sent them for vascular imaging, not because they represent a population. The data are retrospective, so causality is not established. And it is not "the risk for an average person with a screening finding" — transposing such a cohort onto an asymptomatic middle-aged person inflates the numbers rather than deflating them.

Morphology versus percentage: what is associated with what

This is where most of the data sit, and nearly all of it is observational — that is, it describes association, not influence.

A meta-analysis of MRI studies (Gupta et al., Stroke, 2013, PMID 23988640; nine studies, 779 people) compared the presence of individual features against their absence and reported these hazard ratios for ipsilateral stroke or transient ischaemic attack: intraplaque haemorrhage 4.59 (95 % CI 2.91–7.24), lipid-rich necrotic core 3.00 (1.51–5.95), thinning or rupture of the fibrous cap 5.93 (2.65–13.20).

Haemorrhage has its own individual-participant-data analysis across seven cohorts (Schindler et al., JACC: Cardiovascular Imaging, 2020, PMID 31202755): 560 symptomatic people and only 136 asymptomatic ones. Among the asymptomatic with haemorrhage the annual rate of ipsilateral events was 5.4 % against 0.8 % without it, hazard ratio 7.9 — but with a confidence interval of 1.3–47.6, meaning the asymptomatic subset is small and the estimate imprecise. That caveat is worth carrying every time the "eightfold" figure is quoted without it.

For "high-risk plaque" as a bundle — meaning any feature the source studies treated as such, a list broader than the morphological five: haemorrhage, lipid core, thin cap, ulceration, thrombus, but also echolucency, neovascularisation and microembolic signals — in asymptomatic stenosis, there is a meta-analysis of 64 studies and 20,751 participants (Kamtchum-Tatuene et al., JAMA Neurology, 2020, PMID 32744595): 4.3 events per 100 person-years when such features are present against 1.2 when they are not.

Now calcification, where intuition misleads most. A meta-analysis of nine studies covering 2,239 arteries (Homssi et al., Stroke, 2023, PMID 37638399) found no overall association between calcification and ischaemic events (RR 0.75, CI 0.44–1.28). In stenosis above 50 % calcification was associated with ischaemia less strongly (RR 0.56, CI 0.38–0.85), while in the subgroup of largely non-stenosing plaques the direction reversed (RR 1.72, CI 1.01–2.91). The authors' caveat matters: most of the included work consists of case-control and cross-sectional comparisons, not prospective follow-up. The conclusion that follows is narrow: "densely calcified" is not the same thing as "dangerous", and in most comparisons the association runs the other way.

Why "plaque type 2" means nothing without the name of the scale

There is a separate trap here that makes the same report readable in opposite ways.

The historical echogenicity classifications of carotid plaque number their types from soft to dense. Gray-Weale (1988, PMID 3062007) had four: type 1 predominantly echolucent with a thin cap, type 4 uniformly dense and echogenic. Geroulakos (1993, PMID 8242296) made the scale five steps by adding a type 5 for plaques that cannot be classified because calcium casts an acoustic shadow. The direction is the same in both: the lower the number, the softer and more lipid-rich the plaque.

But there is also a grading by plaque height, from the consensus statement of the American Society of Echocardiography (Johri et al., Journal of the American Society of Echocardiography, 2020, PMID 32600741), where the roman grades describe millimetres and say nothing about composition.

The result: a line reading "plaque type 2" without the name of the classification permits two incompatible readings — "soft, lipid-rich" and "1.5–2.4 mm tall, composition not stated". The difference is substantive: a soft plaque is the kind in which studies show therapy has something to change, a dense one is the kind in which there is almost nothing. The only sound move is not to guess but to ask the physician who performed the study which scale they used and what the echogenicity of this particular plaque was. The question is specific and gets a specific answer.

When surgery is discussed and when it is not

The threshold at which intervention — endarterectomy or stenting — even enters the conversation was set historically by the surgical trials, and this matters: the degree of stenosis was an entry criterion there.

• The large randomised trials in asymptomatic stenosis enrolled people from roughly 60 % narrowing upward: ACAS (JAMA, 1995, PMID 7723155; 1,662 people, stenosis ≥60 %) and ACST-1 (Lancet, 2004, PMID 15135594; 3,120 people).

• Moderate asymptomatic stenosis — 40–49 % — was not enrolled in those trials at all. Their numbers therefore do not apply to it: this is not "sparse data", it is a different population — the same transfer error that shows up when treatment evidence from one group is read onto another.

It is also worth knowing that the absolute risk in asymptomatic narrowing has fallen over time, and markedly so. A systematic review of 73 studies covering 28,625 people with asymptomatic stenosis of 50 % or more (Poorthuis et al., Stroke, 2023, PMID 37309688) reported an ipsilateral ischaemic stroke rate of 0.98 per 100 person-years, declining by roughly 24 % with every five years of later enrolment. A population-based cohort under contemporary therapy (Howard et al., Lancet Neurology, 2021, PMID 33609477) recorded zero ipsilateral strokes over five years among 154 people with 50–69 % stenosis, against 14.6 % at 70–99 %. Who is in that cohort matters especially here: this is OxVasc, meaning people who presented after a recent suspected stroke or transient ischaemic attack and in whom an asymptomatic stenosis — usually on the opposite side — was found along the way, at a mean age of about 77. It is not a person whose plaque turned up on an elective scan. The data are observational, and the fall is not attributable to any single factor but to a whole set of changes in treatment over thirty years. The direction, though, is consistent across samples.

Hence what "the threshold has not been reached" actually means: with a moderate asymptomatic plaque, the subject of the conversation is not surgery but the factors that drive the disease onward — chiefly low-density lipoprotein and blood pressure.

What treatment has been shown to change — and what exactly changes

Two questions get fused into one in popular accounts and have to be separated: does the volume of the plaque change, and does its composition change.

The carotid studies are the most directly relevant by vessel and the most modest in result.

• METEOR (JAMA, 2007, PMID 17384434): randomised, 984 people, mean age 57, low calculated risk, two years of high-intensity statin therapy; LDL cholesterol fell by roughly half. Carotid intima-media thickness stopped increasing (−0.0014 mm/year against +0.0131 in the control group), and on the primary endpoint — the average across twelve sites — the authors state plainly that the drug did not induce regression. On one secondary endpoint, the common carotid artery, regression relative to "no change" was in fact shown (p = 0.004). And the endpoint there was wall thickness, not plaque volume.

• ORION (American Heart Journal, 2008, PMID 18294500): randomised, 43 people, paired MRI in 33, two years, two statin doses. Carotid plaque volume did not change significantly in either arm. Among those who had a lipid-rich necrotic core at baseline, the share of the wall it occupied fell by 41.4 % (p = 0.005).

• A compositional analysis within a trial of intensive regimens (Zhao et al., JACC: Cardiovascular Imaging, 2011, PMID 21920335): lipid core volume fell from 60.4 to 37.4 mm³ over three years. The authors' caveat is important — the published analysis is a pooled set of 33 people with a measurable core, not a comparison between arms.

The coronary studies are more numerous and larger, and they cannot be transferred to the carotid artery directly. The effect size there is consistent and small: percent atheroma volume shifts by roughly one absolute point over one to two years. REVERSAL (JAMA, 2004, PMID 14996776) — −0.4 % against +2.7 % in the comparator arm, that is, arrest of growth rather than regression. SATURN (NEJM, 2011, PMID 22085316) — about −1 % in both intensive arms. GLAGOV (JAMA, 2016, PMID 27846344) — −0.95 % against +0.05 % when a PCSK9 inhibitor was added to a statin in people with established coronary disease. PACMAN-AMI (JAMA, 2022, PMID 35368058) and HUYGENS (JACC: Cardiovascular Imaging, 2022, PMID 35431172) — around −2 % in people after a myocardial infarction, that is, the population with the highest baseline risk.

Note the arithmetic: these are single percentage points, not "dissolving the plaque". None of these studies shows atheroma disappearing, and none of them was conducted in asymptomatic middle-aged people with a moderate carotid finding.

What shifts more reliably than volume is composition. Fibrous cap thickness increased in several randomised studies using intravascular optical coherence tomography: EASY-FIT (JACC, 2014, PMID 25456755) — an increase of 69 % against 17 % in the lower-dose arm; PACMAN-AMI — +62.7 against +33.2 µm; HUYGENS — +42.7 against +21.5 µm. The lipid arc shrank. All of this is coronary arteries and post-acute-event populations.

Calcium rises on treatment — and that is not deterioration

Worth knowing in advance, before the next scan. A post-hoc analysis of eight intravascular ultrasound studies covering 3,495 people (Puri et al., JACC, 2015, PMID 25835438): on high-intensity statin therapy atheroma volume fell while the calcification index rose more than it did without therapy; the authors' own phrasing is that statins promote coronary atheroma calcification independently of their plaque-regressive effects. The same direction appears in IBIS-4 (European Heart Journal, 2015, PMID 25182248): atheroma volume fell, and the calcified fraction grew.

So a higher calcium score on a follow-up study after starting therapy does not automatically read as progression. It is expected behaviour of that measurement — and for someone who was not warned about it, the number ruins months for nothing.

What has not been shown — the most important part

The temptation here is obvious: if the cap is thicker and the core smaller, the plaque is "stabilised", so there will be fewer events. The logic looks flawless, which is precisely why it deserves checking.

• The link between change in atheroma volume and events has been shown, but as an association within studies (Nicholls et al., JACC, 2010, PMID 20488313; six intravascular studies, 4,137 people), and largely through repeat revascularisation rather than deaths and infarctions. A meta-regression of 23 studies (Iatan et al., JAMA Cardiology, 2023, PMID 37647074) gives a similar picture at the level of a surrogate.

• There is no randomised evidence that stabilisation in itself, separately from the level of low-density lipoprotein achieved, lowers event rates. The clinical benefit of lowering LDL has been demonstrated separately — on outcomes, not on pictures.

• And a counterpoint from observation: in ELSA (Zanchetti et al., Circulation, 2009, PMID 19738145) baseline wall thickness and the presence of plaques predicted events, whereas the change in thickness under treatment did not.

The conclusion is dull and true: "plaque stabilisation" is a plausible interpretation of what the images show, not a separately demonstrated mechanism of benefit. Decisions are made on demonstrated outcomes; imaging helps understand what is happening, not the other way around.

What has been tested outside of medication

Briefly, because this area carries many promises and little data on plaque specifically (as opposed to wall thickness, which is often substituted for it).

• Diet. The ultrasound sub-study of PREDIMED (Sala-Vila et al., ATVB, 2014, PMID 24285581): 164 people with complete data, 2.4 years on average, in a population at high cardiovascular risk rather than an average asymptomatic one. In the Mediterranean-diet-with-nuts arm maximum plaque height changed by −0.091 mm against +0.106 mm in the control arm (p = 0.047). The olive oil arm showed no effect on plaque or wall thickness in that sub-study.

• Alcohol. The randomised CASCADE trial (European Journal of Clinical Nutrition, 2018, PMID 29379143): two years, participants with type 2 diabetes, wine against water. Total carotid plaque volume did not differ between arms (p = 0.9).

• Smoking. No randomised trial of quitting versus continuing with carotid plaque volume as an endpoint could be found in the available literature. There is extended follow-up of participants in a cessation trial (Stein et al., Atherosclerosis, 2020, PMID 33227549): the degree of abstinence predicted less growth in the plaque score over seven years. That is observation, not randomisation — people who quit may differ in other ways too.

• Physical activity. Meta-analyses show an effect on intima-media thickness, a surrogate for the wall rather than the plaque. Randomised studies with carotid plaque volume as a primary endpoint are nearly absent, and those that exist are small.

How it is followed, so that the change is about the plaque and not the machine

Carotid ultrasound depends on the machine, its settings and the operator — much as echocardiography does, where the same dependence complicates reading change over time. A difference of about a millimetre in wall thickness between two conscientious studies is ordinary. So a repeat study intended to show change is worth doing in the same laboratory and, where possible, with the same physician — and comparing against the original report rather than someone's summary of it.

Second: professional documents answer differently the question of what an asymptomatic carotid plaque means for treatment targets. The European dyslipidaemia guidelines (Mach et al., European Heart Journal, 2020, PMID 31504418) count significant plaque on carotid ultrasound as atherosclerosis documented on imaging, with the risk category that follows. The word "significant" carries weight there: unlike the coronary case, where specific percentages are named, the document sets no numeric threshold for carotid ultrasound. Elsewhere the same European documents describe plaque burden as a risk modifier in people at low or moderate risk, which is a weaker statement. The American cholesterol guideline (Grundy et al., Circulation, 2019, PMID 30586774) does not include asymptomatic carotid plaque in its definition of clinical atherosclerotic disease, nor in its list of risk-enhancing factors. The divergence is real, and the target number depends on which framework the treating physician uses. That is a good subject for a direct question at the appointment — citing the document, not an article on the internet.

What to take away

• The percentage answers a question about flow, not about risk, and without the name of the method it is ambiguous.

• "Stable" describes the absence of specific features — haemorrhage, cap rupture, thrombus, ulceration — not the absence of disease and not the absence of change.

• Calcification is an ancillary feature in these scales; the meta-analysis found no overall association with ischaemia, a weaker one in marked stenosis, and a reversed direction among non-stenosing plaques.

• A line reading "plaque type N" is unreadable without the name of the classification: ask which scale, and what echogenicity.

• In the randomised study that measured it directly, carotid plaque volume did not fall; its composition changed. In a pooled set of treated participants in another study, wall volume did fall in segments containing a lipid core — but without a between-group comparison. Coronary studies show on the order of one percentage point of atheroma volume over one to two years, and they cannot be transferred to the carotid artery directly.

• A rise in calcium after starting therapy is expected behaviour of that measurement, not a sign of deterioration.

• "Stabilisation" as an independently demonstrated mechanism of benefit is not confirmed in randomised trials; what is demonstrated are outcomes from lowering LDL.

Your own numbers — lipids, blood pressure, inflammatory markers, weight, previous vascular reports — are easier to hold in one place and read side by side: change shows up in a series, not in the latest single report. That is how the personal record in Lonevi is built.

This material is informational and does not replace a consultation with a physician. Decisions about treatment, investigation and target values are made by the doctor who follows you — on the whole of your data, not on one line in a report.