When this factor was added to standard risk factors, the area under the ROC curve for predicting CVD events was unchanged, at 0.74 (Circulation 2009;120:502-9).
But flow-mediated dilation independently predicted CVD events. Also, it yielded a net 29% improvement in the correct classification of participants according to whether they experienced an event (P less than .001), mainly because of a 52% improvement in correctly classifying which participants would not experience events.
“Here is a marker with no improvement at all in the ROC curve, and yet you are substantially able to better predict who will get an event,” Dr. Criqui observed.
However, this substudy did not adjust for the other subclinical CVD markers. “If it had adjusted for those, I think it would have been less impressive,” he commented.
Ankle-Brachial Index
A recent analysis, to be published in full later this year, has shown ABI to be an independent predictor of CVD events in MESA participants, according to Dr. Criqui.
“We sort of gave the ankle-brachial index the third degree, if you will,” he elaborated. The investigators adjusted the analyses for standard risk factors, novel risk factors, coronary artery calcium, carotid IMT, and major electrocardiographic abnormalities at baseline.
After these adjustments, participants with a low ABI (lower than 1.0) and those with a high ABI (1.4 or greater) both had a roughly doubled risk of CVD events relative to their counterparts with a normal index.
In addition, after the exclusion of participants with a high ABI, each 0.1-unit increase in this index was associated with a reduced risk of CVD events across the four racial/ethnic groups studied (hazard ratios, 0.74–0.87). This index also increased the area under the ROC curve for predicting such events (P = .02).
“Better classification of cardiovascular disease risk is indeed an important goal clinically,” concluded Dr. Criqui. And to that end, MESA has helped by identifying a set of subclinical markers that improve risk prediction.
However, “I think the parsimonious assessment of CVD risk is still elusive, since – much like the cardiovascular disease risk factors themselves – different subclinical cardiovascular measures appear to contribute independently to risk,” he commented. “Everything we try seems to add a little bit, to a greater or lesser degree, to risk prediction.”