Associations of Extracranial Carotid Atherosclerosis Progression With Coronary Status and Risk Factors in Patients With and Without Coronary Artery Disease

Patients, Clinical Evaluation, and Study Design

Equal numbers of male and female patients were enrolled in this study, as described previously.^20–23⇓⇓⇓ They consisted of 139 controls free of any evidence of CAD at coronary angiography and 141 cases with obstructive CAD manifest by at least 1 coronary artery with 50% or more stenosis. Approximately 70% of controls underwent cardiac catheterization because of chest pain and 20% because of valvular heart disease.²³

Trained interviewers collected pertinent medical history and risk factor profiles from all participants within 6 to 8 weeks after angiography. Clinic coordinators also measured height, weight, and blood pressure. Blood was drawn for laboratory analyses. Hypertension was defined by history of the disease, a systolic blood pressure >150 mm Hg, or a diastolic blood pressure >90 mm Hg. Diabetes was defined by history of the disease or by a fasting plasma glucose level of >140 mg/dL. Smoking status was recorded as the number of pack years smoked. Plasma concentrations of lipids and lipoproteins were measured using Center for Disease Control–standardized methods.^20–23⇓⇓⇓ Apolipoprotein E polymorphism and lipoprotein (a) were determined, as described previously.²² B-mode ultrasound was used to quantify IMT of the extracranial carotid arteries, as described below. Subsequently, patients completed a brief health, risk factor, and medication use questionnaire and underwent repeat B-mode ultrasound evaluation on a yearly basis for 3 years (total of 4 examinations).

Of the 280 individuals who were originally enrolled, 256 (91%) returned for at least one follow-up visit. Use of lipid-altering therapy at baseline was an exclusion from participation in the study. This likely resulted in exclusion of a few markedly hyperlipidemic participants from the baseline examination (before 1993). However, at study end, 19% of cases but only 4% of controls who returned for follow-up were treated with lipid-altering therapy. Because lipid-altering therapy has been shown in several clinical trials to retard the progression of extracranial carotid IMT,¹ analysis for this study excluded all visits after participants began taking lipid-altering therapy. We similarly excluded visits when participants were taking antioxidants because of suggestions that this therapy might similarly influence IMT progression.²⁴ These decisions excluded 72 visits in a total of 32 participants from the database. We also censored follow-up visits for 1 individual after carotid endarterectomy. All these decisions pared the analysis database to 241 individuals (86% of the original cohort) with an average of 2.80 of 3 planned annual follow-up visits. Risk factors and baseline IMT of the extracranial carotid arteries of the subset of the population in this cohort study were not statistically different from those of the overall population evaluated at baseline.

B-Mode Ultrasound

The ultrasound machine used was a Biosound 2000 II, s.a. with an 8-MHZ annular array transducer (Biosound, Indianapolis). The patient’s head position, sonographer position, scanning angles, and scanning angle sequence were all standardized. The extracranial carotid artery was evaluated at 3 1-cm segments, the common carotid, the carotid bifurcation, and the internal carotid artery. In longitudinal arterial images, the distances between the interfaces generated from the boundaries between adventitia/media, intima/lumen on the near wall, and the interfaces from lumen/intima, media/adventitia on the far wall were identified as the IMT. The right and left carotid arteries were scanned as circumferentially as possible to search for the thickest IMT on both near and far walls for each segment. The scan was recorded on a super VHS videotape and reviewed offline. While reviewing the image tape, the reader selected the most appropriate frame showing the thickest IMT for each site of the carotid artery (which was used for analysis^20–23⇓⇓⇓).

Statistical Methods

Baseline characteristics of CAD cases and controls were compared using t tests and χ² tests. IMT progression was described by fitting mixed models to the longitudinal data using maximum likelihood.²⁵ In these models, IMT measurements from each wall segment were included and blocked within subjects. Progression of IMT over time was modeled with a fixed effect; random intercepts terms were included to describe subject-specific differences. The use of fixed, rather than random, effects for IMT progression may inflate type I error rates if IMT progressions are heterogeneous²⁶; however, we found little difference in the results in our analyses from the two approaches and chose to avoid models in which explicit distributions must be assumed for any heterogeneity. For simplicity, correlations among arterial sites and longitudinal measurements of IMT from individual segments were modeled with compound symmetry covariance structure. This approach, rather than using unstructured covariance matrices, slightly decreased the efficiency of comparisons but seemed to provide more stable estimates. This general approach of using maximum likelihood and mixed models protects against biases related to missing measurements.²⁷ Differences in the progression rates between CAD cases and controls were contrasted using Wald tests.²⁵

Progression of IMT in Individuals With and Without CAD

Intimal-medial thickness of the extracranial carotid arteries of CAD cases has been previously shown to exceed that of controls in cross-sectional analysis.^1–3⇓⇓ We observed faster IMT progression rates in CAD cases (33.7 μm/year) than in controls (8.9 μm/year). As a qualitative comparator, we also estimated progression from cross-sectional analyses of the associations of IMT with age in the same population at baseline (not presented) to determine how well the estimated progression rates compared with those actually observed. We inferred progression rates of 20.6 and 10.4 μm/year for cases and controls, respectively. Thus, the progression rates inferred from the cross-sectional analyses are qualitatively comparable to those directly measured for controls, but they underestimate progression for cases, as would be expected if older fast progressors were underrepresented in the CAD group at baseline. The progression rate in controls observed and inferred from cross-sectional baseline data in this study (10 μm/year) is similar to that inferred in a previous review that included other healthy populations.³ One prior publication demonstrated more rapid carotid IMT progression in patients with CAD compared with those without significant CAD.⁴ However, in that study IMT was quantified as the sum of carotid IMT from various sites and cannot be compared with the now more widely used mean maximum IMT (as in the present communication).

Associations of Risk Factors With Progression of IMT

Several epidemiologic studies have explored the associations of risk factors with progression of extracranial carotid IMT in healthy populations. Cigarette smoking,^5–8⇓⇓⇓ dyslipidemia,^5,6,8–10⇓⇓⇓⇓ blood pressure,^6,8,11,12⇓⇓⇓ factors related to thrombosis and thrombolysis,^5,6,8,13⇓⇓⇓ alcohol,¹⁴ diabetes control,¹⁵ Chlamydia,¹⁶ lutein,¹⁷ cardiorespiratory fitness,¹⁸ and psychosocial factors¹⁹ have all been related to progression. Clinical trials indicate that reduction of LDL cholesterol,¹ antioxidant vitamins,²⁴ angiotensin-converting enzyme inhibition, ²⁸ and β-blocker treatment²⁹ retard progression of IMT.

In the present study population, CAD status was the most significant correlate of IMT progression. In addition, cigarette smoking and reduced HDL cholesterol were the only risk factors that were associated with greater IMT progression. After control for CAD status, only reduced HDL cholesterol remained associated with faster progression.

The lack of an association of many risk factors with IMT progression in the overall cohort obscured the fact that sex, waist to hip ratio, smoking, HDL cholesterol, and triglycerides were related to IMT progression in CAD cases. Low HDL cholesterol was associated with more rapid progression of carotid IMT in male but not female CAD patients. Because control status was associated with overall attenuated progression, the power to detect the impact of risk factors on progression in the control group was similarly reduced, and we cannot precisely discuss the relative impact of risk factors on progression in them compared with cases. Nonetheless, the data are consistent with a hypothesis that the impact of risk factors on progression is quantitatively greater in patients with CAD than in those free of CAD. It is possible that the discrepancy between the apparent impact of risk factors on progression in asymptomatic population-based samples^5–14⇓⇓⇓⇓⇓⇓⇓⇓⇓ and the lack of effect in our controls partly reflects the presence of individuals with subclinical disease and more rapid progression in apparently healthy populations (see below).

Strengths and Limitations of This Investigation

Reed and Yano³⁰ have previously described the limitations of studies where case and control status were identified at angiography. Angiographically defined controls are a unique population referred for catheterization to evaluate chest pain, valvular heart disease, and heart failure. They also differ from clinically asymptomatic individuals because of objective exclusion of obstructive coronary disease.³¹ For these reasons, risk factor profiles in them may not be precisely representative of those in clinically asymptomatic populations.

The primary prespecified outcome for this study was progression of the aggregate measure of IMT. An alternate approach would have been to focus on the common carotid alone. Separate analysis of influence of risk factors on progression of the common carotid IMT (data not presented) showed only an association of smoking with progression in the entire cohort and in those with CAD, and it is possible that risk factors may have a differential impact on progression of different segments.

Strengths of the study, however, were its single-center nature, the accurate assessment of IMT, and precise definition of case/control status. We excluded patients with mildly obstructive CAD (<50% stenosis) by design because of the likelihood that they had underlying CAD.³¹

Conclusions

A single measurement of vascular structure better reflects past risk factor exposure than present risk burden (and thus future risk). By way of example, the Framingham investigators observed stronger associations of carotid disease with time-integrated risk factor levels than with concurrently measured risk factors.³² On the other hand, progression of a stable marker of atherosclerosis may be a better index of future CAD risk.³³ Our data support the likelihood of large differences in progression in patients with CAD compared with CAD-free controls. It is unlikely that this technique could provide a specific or sensitive clinical predictor of incident CAD for an individual patient; however, additional studies of the ability of change in IMT to predict clinical events in subsets of symptomatic as well as asymptomatic individuals are needed.