he two pairs of carotid arteries (internal and common, right and left) are accessible to detailed study, by sophisticated ultrasound techniques, where the vessels pass through the neck. When the carotids are "viewed" in cross-section, fine measurements can be made, not merely of the overall diameter of the vessel but of the state of its walls. Arteries are not simple cylinders like the inner tube of a bicycle tire, and one focus of cardiovascular research over the past decade has been the gap, normally around 1or 2 mm, between the inside of the vessel wall (intima) and the layer known as the media.

It has long been known that carotid intima-media thickness is an indicator of the risk of heart disease and stroke, but does this measurement tell clinicians much more than conventional risk factors do—and could non-invasive ultrasound scans be used to screen people who do not yet have clinical cardiovascular disease? These questions have been addressed by the U.S. multicenter Cardiovascular Health Study Collaborative Research Group (CHSCRG) in paper #7.

This study population of men and women aged 65 years or more was set up in 1989-90. Once those with cardiovascular disease already present had been excluded, 4,476 participants were left; high-resolution ultrasound scans were done; and the median follow-up was 6.2 years. Five quintiles of roughly 900 people were set up, depending on their intima-media thickness grouping, which, for the common carotid, ranged from below 0.87 mm up to 1.18 mm or more.

Seven-year rates for cardiovascular endpoints (heart attack or stroke) rose consistently from below 5% in the first quintile (smallest baseline thickness) to over 25% in the highest quintile. Interesting, but not that important unless the relationship holds when other factors such as blood pressure and smoking history are taken into account. It does. The unadjusted relative risk for myocardial infarction, with the risk for the lowest quintile for combined common/internal carotid intima-media thickness taken as unity, rose steadily from 1.0 to 6.30 for the highest quintile. After adjustment for some risk factors (but not cholesterol) this "dose response" held up but not quite so strongly, the highest relative risk being 3.61. With stroke as the endpoint the relationship was still present, and significant, but it was not as clear-cut as it was for heart attacks.

Carotid-artery intima-media thickness is also being used as a surrogate endpoint for the clinical ones of a fatal or non-fatal stroke or heart attack.

The CHSCRG data took over six years to accumulate, and those who seek to intervene and reduce risks, by pharmaceutical or other means, prefer not to wait that long. This is what lead author Dr. Daniel O’Leary, Tufts-New England Medical Center, Boston, and his colleagues think, too. "Selective use of non-invasive ultrasound permits us to better quantify risk," O’Leary tells Science Watch. Its future, he says, "may lie in its potential to directly quantify the impact of medical intervention and life-style changes."

A recent example of this surrogacy is one of those studies where, one is tempted to hazard, the acronym was devised before its translation. ARBITER (Arterial Biology for the Investigation of the Treatment Effects of Reducing Cholesterol) was announced in March of this year (Am. Heart J.141: 342-47, 2000). Two "statins" are being compared over just 12 months, with common carotid ultrasound measurements as the endpoints.

When O’Leary presents the data from this paper (# 7) the most common question he is asked is whether addition of this measurement would improve the predictive power of the "Framingham score," a risk equation that emerged from a well-known population study set up in Massachusetts many years ago.