The Role of Carotid Arterial Intima-Media Thickness in Predicting Clinical Coronary Events

Carotid arterial intima-media thickness measured with B-mode ultrasonography is used as a noninvasive end point (that is, an outcome) in epidemiologic studies and clinical trials to gauge progression and regression of atherosclerosis [1-3]. As such, carotid arterial intima-media thickness, expressed as a single measurement (in millimeters) or a rate of change (in millimeters per year), is used as a surrogate end point for atherosclerosis of the coronary artery. However, its relation to coronary events has not been fully explored. It is well established that progression of atherosclerosis of the coronary artery determined by sequential coronary angiography is predictive of coronary events [4-6]. A close histologic relation between carotid and coronary atherosclerosis has been seen in autopsy studies [7], and the two arterial beds share many risk factors that contribute to the progression of atherosclerosis [8, 9]. Furthermore, carotid arterial intima-media thickness has been a good indicator of the presence and extent of coronary artery disease in observational studies [10, 11]. The Cholesterol Lowering Atherosclerosis Study [12] was a clinical arterial imaging trial designed to study the effects of colestipol-niacin therapy on progression of atherosclerosis in the coronary, femoral, and carotid arteries. We have reported that treatment is beneficial for all three arterial beds [3, 13-16]. In addition, long-term follow-up of the study cohort indicated that progression of coronary artery disease was predictive of coronary events [4]. The objectives of this long-term follow-up of the Cholesterol Lowering Atherosclerosis Study cohort are 1) to determine whether carotid arterial intima-media thickness [expressed as a single measurement or as a rate of change] predicts coronary events, 2) to compare the relative prognostic utility of the two carotid arterial intima-media thickness measures; and 3) to compare the relative prognostic contribution of the two carotid arterial intima-media thickness measures with an angiographic measure of coronary artery disease progression and lipid levels. Methods Study Design In the Cholesterol Lowering Atherosclerosis Study, 188 nonsmoking men 40 to 59 years of age who had previously had coronary artery bypass graft surgery were randomly assigned to receive colestipol-niacin therapy plus dietary therapy (target diet, <125 mg of cholesterol per day and 22% of energy as fat, 10% as polyunsaturated fat, and 4% as saturated fat) or placebo plus dietary therapy (target diet, <250 mg of cholesterol per day and 26% of energy as fat, 10% as polyunsaturated fat, and 5% as saturated fat) [12]. In addition to the primary end point provided by coronary angiograms at baseline and after 2 years of treatment, B-mode ultrasonography of the carotid artery was done at baseline and every 6 months during the 2-year treatment period. The cohort for this study consisted of patients who had completed the 2-year treatment period and had evaluable coronary and carotid arterial end points. Baseline and 2-year coronary artery films were processed by quantitative coronary angiography in tandem; frames were matched for orientation and degree of contrast filling [17]. For each evaluable arterial segment, three sequential frames were processed in end diastole. For each coronary lesion, percent diameter stenosis was obtained as the average over the three sequential frames. For each patient, the change in percent diameter stenosis over 2 years was averaged for all evaluable coronary artery lesions. B-mode ultrasonographic images of the carotid artery were obtained with a Diasonics CV400 system with a 7.5-MHz probe (Diasonics, Milpetas, California). Longitudinal views of the far wall of the right distal common carotid artery were recorded with the minimum gain necessary for clear visualization of structures. Common carotid arterial intima-media thickness was measured with an automated computerized edge-detection algorithm [18]. The distance between the echoes arising from the blood-intima interface and the media-adventitia interface was taken as the measure of intima-media thickness. Distal common carotid arterial intima-media thickness was the average of approximately 80 intima-media thickness measurements made over 1 cm. Measurements were made by persons blinded to treatment assignment and the occurrence of clinical coronary events. Follow-up for Coronary Events After completion of the 2-year treatment period, the occurrence of major medical events and information on all medications (including lipid-lowering agents) was determined for all patients by a clinic visit (59% of all follow-ups) or a mailed questionnaire (41%) at annual follow-up points through 30 June 1994. No ascertainment bias was associated with method of follow-up. Coronary events were nonfatal acute myocardial infarction, coronary death, and need for coronary artery revascularization [percutaneous transluminal coronary angioplasty or coronary artery bypass graft surgery] because of recurrence or worsening of angina pectoris. For all patient-reported events, hospital records were obtained for confirmation, and all causes of death were confirmed by hospital records and death certificates. Myocardial infarction was diagnosed by a cardiologist who was blinded to treatment assignment and ultrasonographic and angiographic end point measures. Myocardial infarction was confirmed if two of the following three criteria were substantiated: typical chest pain, positive creatine phosphokinase-MB, and a new Q wave on electrocardiogram. In order to include only events that were clearly related to clinical symptoms, we did not count 1) coronary artery revascularizations that were related to the reading of the 2-year coronary angiogram or 2) silent myocardial infarctions noted on electrocardiograms at annual follow-up examinations as clinical coronary events. Statistical Analysis The two dependent variables were time from completion of the trial to nonfatal myocardial infarction or coronary death and time from completion of the trial to the first coronary event (nonfatal myocardial infarction, coronary death, or coronary artery revascularization). The two independent variables were the absolute carotid arterial intima-media thickness (in millimeters), measured at the end of the 2-year trial, and the annual rate of change in carotid arterial intima-media thickness (in millimeters per year), evaluated over the 2-year trial. The intima-media thickness change rate was computed for each patient by fitting a least-squares regression line relating intima-media thickness measurements to time in the study. The average number of ultrasonographic examinations per patient was 2.8 0.4. The absolute carotid arterial intima-media thickness is a cumulative measure of carotid atherosclerosis, whereas the intima-media thickness change rate represents the speed with which atherosclerosis of the carotid artery is changing. Univariate and multivariate proportional hazards models were used to test for relations (overall and within each treatment group) between the intima-media thickness variables and coronary event rates. Covariates included the baseline value for intima-media thickness (for analyses of intima-media thickness change rates) and treatment group (for analyses of the total sample). Because patients had the option to continue their randomized, blinded treatment in a 2-year extension of the Cholesterol Lowering Atherosclerosis Study, an additional covariate indexed whether a given patient was treated in the 2-year extension period. Likelihood ratio tests for trend in coronary event rates used each intima-media thickness variable as continuous data. Hazard ratios (as estimators of relative risks) and 95% CIs were expressed per SD (0.03 mm/year for the carotid arterial intima-media thickness change rate and 0.13 mm for the absolute carotid arterial intima-media thickness). Absolute carotid arterial intima-media thickness values were categorized by quartiles based on the distribution of the baseline intima-media thickness for all patients; carotid arterial intima-media thickness change rates were categorized by quartiles based on the distribution of changes in the placebo group. Hazard ratios were then computed for each of the upper quartiles relative to the first. Because earlier analyses of the study cohort showed a significant relation between progression of coronary artery disease (assessed by the change in percent diameter stenosis using quantitative coronary angiography) and coronary events [4], we also evaluated the relative prognostic contributions of this angiographic measure of coronary artery disease progression and the ultrasonographic measure of carotid arterial intima-media thickness progression. For the change in percent diameter stenosis, hazard ratios and 95% CIs were expressed per 10% change in percent diameter stenosis. The cutoff of 10% change is double the measurement error for percent diameter stenosis on short-term repeated angiography. The joint prognostic contribution of the carotid artery and coronary artery measures of atherosclerosis with lipid levels that were found to be significantly different between patients with and without coronary events was also evaluated. Values given are the mean SD unless otherwise indicated. Role of the Funding Source The authors were responsible for data collection, data management, statistical analyses, and data interpretation. Research was supported by the National Heart, Lung, and Blood Institute through investigator-initiated grants to Dr. Hodis (RO1-HL-49885) and Dr. Mack (RO3-HL-54532). The funding source had no role in deciding whether the study would be submitted for publication. Results Characteristics of the Cohort at Baseline and after Treatment Of the 188 patients randomly assigned to treatment, 42 (22%) were excluded from the study: Eleven did not have a baseline ultrasonogram, 13 had no ultrasonographic fo