The present study did not find significant sex differences in the ability of a range of imaging measures to predict CVD overall and CHD, stroke, and HF separately despite the clinically relevant differences between men and women with respect to CVD. In addition, CAC was most strongly associated with CHD and CVD; LV mass and LV concentric remodeling were most strongly associated with stroke; and LV mass showed the strongest association with HF after adjustment for traditional risk factors in the overall MESA sample and in men and women separately. Imaging measures showed these associations with incident CVD independently of one another. According to the AUC analysis, CAC provided the highest incremental CHD prediction in both men and women, and LV mass added the most to HF prediction beyond traditional risk factors in men compared with the other imaging measures.
A number of prior studies have reported a strong positive association between CAC score and incident CHD after inclusion of conventional coronary risk factors.3
A prospective analysis in MESA has addressed the potential utility of CAC versus IMT for CVD risk prediction and found that CAC better predicted CHD and total CVD than did IMT.8
Similarly, CAC has been shown in detail to significantly improve CHD risk classification in MESA using the NRI statistic.18
Previous ECG- and echocardiography-based studies have demonstrated LV hypertrophy and abnormal LV geometry to confer an independent, increased risk of stroke.19,20
Although the highest quartile of IMT was linked to an increased risk of CHD independent of traditional risk factors as well as of other imaging tests, IMT did not show an independent association with stroke, in contrast to prior reports.4,8
This finding could be variously attributable to methodological differences among these studies with respect to population demographics, carotid segment definition, ultrasound protocols, adjustment for confounding covariates, and definition and number of clinical end points.
Sex and Ethnic Considerations
Imaging measures of subclinical CVD are associated similarly with incident CVD in men and women as indicated by the lack of any effect modification by sex. This finding is despite the multifactorial influence of female reproductive hormones on the cardiovascular system and lower sensitivity and specificity of imaging modalities for disease detection in women due to smaller vessels and heart size,15,21
and the resulting potential for larger errors in measurements. Our study offered an important advantage over previous studies by calculating and comparing hazards in relation to standardized increments in CAC score used as a continuous measurement, ensuring a greater degree of comparability between the 2 sexes. Furthermore, other studies were based on specialized study samples consisting of largely white patients selected by primary care physicians for screening, thereby limiting the applicability of their findings to population-based settings.
Past studies have demonstrated racial differences in the prevalence and severity of many of these imaging risk measures, favoring the due consideration of ethnicity in any cardiac risk prediction tool. Whites have been reported to have a higher prevalence and extent of CAC than blacks and other racial/ethnic groups22
but lower LV mass and lower prevalence of concentric LV hypertrophy than blacks.23
In our multiethnic cohort, absence of ethnicity interactions in the prognostic value of imaging measures implies that imaging tests may have a similar ability of risk stratification across different ethnicities to sex. Further research is needed to determine whether the same levels of these imaging measures (eg, CAC) translate to differences in prognosis depending on race/ethnicity, which may result in improved characterization of CVD risk across different ethnic groups.
Strengths and Limitations
Strengths of this study include its large, ethnically diverse cohort free of symptomatic CVD at baseline, standardized imaging protocols and risk factor assessments, and reliance on systematically defined symptomatic end points. We assessed imaging measures as both continuous and categorical variables, which enabled more complete assessment of their relationship to CVD outcomes.
Several limitations exist. First, our study had limited power to detect stroke and HF associations, particularly incremental risk prediction, and to assess sex and ethnicity interactions. Women showed relatively fewer cardiovascular events. Coronary artery disease is known to lag by approximately a decade in women compared to men, with an observed increase in coronary calcification in women after the sixth decade.3
The cardiovascular risk and event rates should predictably increase in women with the aging of this population on longer-term follow-up. For the same reason, the relationship between imaging measures and cardiovascular risk could not be evaluated by specific sex and ethnic subgroups. Second, carotid plaque area (CPA) is known to be an independent predictor of coronary artery disease and future cardiovascular risk and is believed to be a more direct measure of the global atherosclerotic process and stronger predictor of cardiovascular risk than carotid IMT.24–26
CPA >0 emerged to be moderately more sensitive in detecting underlying coronary artery plaque stenosis than CAC >0. This finding is potentially attributable to the reliance of CAC testing on the presence of calcium in the plaque, which may not characterize early lesions because of the correlation of calcification with disease evolution over time.25
Although this prior evidence could underlie the scope of a direct comparative assessment of CAC and CPA for their predictive ability of CHD and CVD, it could not be accomplished in the present study because of the nonavailability of measures of plaque area. Third, we could not compare imaging measures for their NRI for prediction of events other than CHD. Reclassification measures depend on particular risk categories used. However, unlike CHD, there are no clearly defined clinical risk categories for stroke, HF, and all CVD, rendering any choice of categorization as arbitrary and movement among categories not clearly meaningful. This issue was compounded by the limited number of strokes and HF events in MESA. Finally, the cardiovascular events considered were an admixture of both hard (myocardial infarction, resuscitated cardiac arrest, stroke, CHD death, stroke death) and soft end points (angina, HF, other atherosclerotic death, other CVD death) to maximize statistical power.
Bioimaging tests differ in their association with incident CVD. Their ability to predict cardiovascular events does not seem to differ by sex and ethnicity.
Cardiovascular disease (CVD) is the leading cause of morbidity and mortality in both men and women. We found that (1) coronary artery calcium (CAC) was most strongly associated with coronary heart disease and total CVD, (2) left ventricular (LV) mass and concentric remodeling best predicted stroke, and (3) LV mass best predicted heart failure. Refining the role of potentially competing bioimaging tests in CVD risk assessment can guide the selection of the optimal screening test for a given population. Although CAC best predicted overall incident CVD events, it may be more relevant in populations where coronary heart disease is common, and LV mass and concentric remodeling may be more suitable if demographics and risk factor profile favor an increased likelihood of subsequent heart failure or stroke. Practical considerations such as cost, availability, patient preference, local expertise, and technological limitations are important factors that also may dictate the potential choice among these imaging measures. Although the number of events in women was limited, our results provide population-based evidence that the independent prognostic utility of CAC, LV mass, and LV concentric remodeling in women is similar to men and suggest that CAC offers incremental risk prediction beyond traditional risk factors in women as in men. Like sex, imaging tests also may have a similar ability of risk stratification across different ethnicities.