Carotid Plaque Calcification Predicts Future Cardiovascular Events in Type 2 Diabetes

Atherosclerosis is a common complication of diabetes, driving severe morbidity and mortality. Atherosclerotic plaques are complex structures with multiple tissues and cell types that differentially contribute to lesion vulnerability (1). In population-based longitudinal studies, atherosclerotic plaque characteristics have been shown to predict atherosclerosis progression at a systemic level, being associated with incident cardio- and cerebrovascular events (2–7). Based on ultrasonographic plaque characterization, echolucent (lipid-rich) plaques seem to be a better predictor of stroke and cardiovascular events than echogenic or echorich plaques containing calcified areas (8–11), although an independent association between echogenic plaques and cardiovascular events has also been demonstrated in the general population (4,12–14). In patients with diabetes, the prevalence of carotid plaques is increased (15). Diabetes traditionally is associated with vascular calcification (16), but there seem to be no gross differences in histological plaque composition compared with the nondiabetic condition (17,18). Most studies specifically in patients with diabetes have suggested that a higher cardiovascular risk is associated with the presence of echolucent plaques (19–21) rather than echogenic ones (22). Using tissue characterization of carotid plaques by the grayscale median (GSM) in a small cohort of patients with type 2 diabetes (T2D), Irie et al. (21) demonstrated that echolucent plaques can improve risk prediction of cardiovascular events. On the other hand, the Diabetes Heart Study (22) showed that multibed artery calcifications detected by computed tomography imaging significantly improves the prediction of cardiovascular outcomes, lending support to the notion that atherosclerotic calcification can be harmful.

We have previously demonstrated a significant association between microangiopathy and composition of carotid plaques in a wide cohort of patients with T2D (23). In this article, we report the follow-up (up to 9 years) examination of the same cohort to evaluate prospectively whether the ultrasonographic characteristics of carotid plaques predict cardiovascular outcomes.

The study was approved by the local institution and ethics committee of the University Hospital of Padova. Recruitment and baseline clinical evaluation data were collected between January 2004 and December 2007 and have been described previously (23). Briefly, all consecutive subjects with T2D who underwent a screening or diagnostic carotid ultrasonography examination at the diabetic outpatient clinics of the University Hospital of Padova and Montebelluna Hospital were enrolled. At baseline, the following data were collected: anthropometric parameters; smoking status; presence/absence of hypertension, dyslipidemia, and coronary heart disease; history of ischemic stroke and lower-limb amputation; pharmacological treatment; biochemical measures (lipid profile, HbA_1c, urinary albumin excretion); and retinal fundus examination. The presence of carotid plaques, respective degree of stenosis, and ultrasonographic tissue characteristics were recorded. On the basis of the GSM on ultrasound examination, plaques were classified as echolucent (low GSM, lipid rich), echogenic (high GSM, mostly occupied by calcified areas), and heterogeneous (mixed echolucent and echogenic) as previously described in the Tromsø Study (9). This classification has been validated against histopathology (24). The mean of left- and right-side carotid intima-media thickness (IMT) at 1 cm from the bifurcation was also measured in plaque-free areas and recorded as recommended by the Mannheim consensus (25).

The follow-up data collection was performed between January 2012 and August 2012 from electronic health records of the Veneto region in Italy between the baseline examination and the last follow-up visit. We recorded vital status, eventual causes of death, and occurrence of cardiovascular events. Major adverse cardiovascular events (MACE) were defined as cardiovascular death, angina, ST-segment elevation myocardial infarction (STEMI)/non-STEMI, coronary revascularization, heart failure, atrial fibrillation, pulmonary embolism, pacemaker implantation, ischemic stroke, transient ischemic attack, carotid revascularization, lower-limb amputation, peripheral ischemic wound, peripheral revascularization, and hospital admission for cardiovascular causes. Atherosclerotic MACE were defined as angina, STEMI/non-STEMI, coronary revascularization, ischemic stroke, transient ischemic attack, carotid revascularization, lower-limb amputation, peripheral ischemic wound, and peripheral revascularization. Causes of death were classified as due to cardiovascular disease, malignancy, diabetes, infectious disease, trauma, and unknown and were retrieved by the national Registry office codes.

Continuous variables are expressed as mean ± SE and categorical variables as percentages. Nonnormal variables of the Kolmogorov-Smirnov test were log-transformed before analysis. Comparisons of continuous data between two or more groups were performed with Student t test and ANOVA, respectively, and χ² test was used to compare categorical data. The least significant different post hoc test was applied. To determine the association between carotid plaque type and future MACE, univariate analyses and then multivariate Cox proportional hazards regression models were run, entering total/atherosclerotic MACE occurrence as the dependent variable and explanatory covariates chosen among those showing significant (post hoc P < 0.05) associations in the univariate group analysis. Crude incidence rate (or density) was calculated as the number of events divided by the number of person-years observed. Risk estimates were derived from multivariable regression models. Various models were built based on degrees of adjustment. Statistical significance was accepted at P < 0.05, and SPSS version 22 statistical software was used for all analyses.

A total of 581 patients of the previously studied cohort of 662 were included in the follow-up analysis (88%), whereas 81 were lost to follow-up. There was no significant difference between patients lost to follow-up and those included in the analysis. Clinical characteristics of the study subjects at baseline are summarized in Table 1.

The mean duration of follow-up was 4.3 ± 0.1 years (median 4.4 [interquartile range 2.5–6.1] years). At the end of observation, 523 patients (89.8%) were alive. Table 2 summarizes cardiovascular outcomes and causes of deaths (n = 58), the prevailing being cardiovascular (n = 27 [46.6%]). MACE, including cardiovascular death, nonfatal acute myocardial infarction, stroke, transient ischemic attack, new-onset peripheral artery disease or ischemic lower-limb amputation, coronary and peripheral artery revascularization, unstable angina, heart failure, arrhythmias, and pulmonary thromboembolism, occurred in 236 (40.6%) patients (crude annual rate 9.5%), 205 of whom had fatal or nonfatal atherosclerotic MACE. Table 1 reports clinical characteristics in patients according to the occurrence or nonoccurrence of MACE during follow-up. Patients with incident MACE were older and had longer diabetes duration; higher HbA_1c; lower HDL cholesterol; higher prevalence of hypertension, macroangiopathy, and microangiopathy; and more intense antidiabetic and cardiovascular drug therapy.

Table 1 also shows the clinical characteristics of patients according to the presence and composition of carotid atherosclerotic plaques. As determined by ultrasound, carotid plaques were categorized as echolucent (lipid rich [n = 95]), heterogeneous (partially calcific [n = 251]), and echogenic (predominantly calcific [n = 129]). Age, prevalence of hypertension, dyslipidemia, macroangiopathy, microangiopathy, and intensity of antidiabetic and other cardiovascular drug therapies were significantly different among these patient groups and in paired comparisons across types of carotid plaques.

Upon univariate Cox proportional hazards analysis, presence versus absence of any carotid plaque was associated with incident MACE (not shown). Although the hazard ratio (HR) associated with echolucent plaques was not significant (1.97 [95% CI 0.93–3.44]), there was a clear trend for progressively higher risk in heterogeneous (3.10 [2.09–4.23]) and echogenic (3.71 [2.09–5.59]) (Fig. 1A) plaques. Quite similar data were obtained for atherosclerotic MACE (Fig. 1B), although the CIs were larger because of a smaller number of events. We therefore examined whether plaque composition was predictive of incident MACE in Cox proportional hazards models adjusted for confounders derived from univariate analyses of the variables in Table 1. Considering patients without plaques as the reference group, only in patients with heterogeneous and echogenic plaques was the HR for total (Fig. 1C) and atherosclerotic (Fig. 1D) MACE significantly increased, and a trend for a linear increase in HR with plaque calcification was detected. Among patients with carotid plaques, the presence of an echogenic versus an echolucent (reference) plaque was significantly associated with incident MACE independently from confounders derived from univariate analyses (Table 3, model 1). When this was corrected for the maximal degree of stenosis, calcific plaque types were not predictive of incident events likely because stenosis severity outperformed type of plaque in terms of predictive power (Table 3, model 2). However, when the analysis was restricted to patients with a maximal degree of stenosis ≤30% (n = 279), patients with echogenic plaques had significantly higher HR of MACE compared with those with echolucent plaques after full adjustment (Table 3, model 3). Quite similar results were detected when the outcome was restricted to atherosclerotic MACE. Kaplan-Meier curves for total and atherosclerotic MACE showed similar event-free survival in patients with heterogeneous and echogenic plaques (Fig. 2A and B), allowing the two plaque types to be combined. In fact, a calcific plaque type (heterogeneous and echogenic combined vs. echolucent) was predictive of incident total or atherosclerotic MACE after fully adjusting for confounding factors (Table 3, model 4). When models 3 and 4 were further adjusted for maximal carotid IMT, results did not significantly change, and a calcific plaque type was still associated with incident total or atherosclerotic MACE (Table 3, model 5). Sex was not associated with the outcome (Table 1), and forcing sex into the models did not significantly change results (data not shown).

In the current study, we demonstrated that carotid plaque calcification predicts future cardiovascular events and death in patients with T2D. The risk of incident MACE progressively increased with increasing plaque calcification from noncalcified echolucent, to partially calcified heterogeneous, to heavily calcified echogenic plaques. This consistent association was independent of age, diabetes duration and control, hypertension, and micro- and macroangiopathy.

The characterization of carotid artery lesions is feasible by conventional B-mode ultrasound imaging, allowing a description of plaque composition. Echolucent or hypoechogenic plaques mainly comprise high lipid content, inflammatory cells, and neovessels, whereas echogenic plaques comprise calcific tissue (26,27). An accelerated arterial calcification has been described in diabetes wherein metabolic and inflammatory factors contribute to intimal and medial calcifications (28–30). Several large population studies have demonstrated that the identification of vulnerable carotid plaque characteristics is a significant predictor of cardiovascular events (4,14,31). In the Multi-Ethnic Study of Atherosclerosis (MESA), ultrasound-derived plaque metrics independently predicted cardiovascular events and improved risk prediction for coronary heart disease events when added to the Framingham risk score (7). Similarly, the identification of vulnerable plaque characteristics by magnetic resonance imaging has been demonstrated to improve cardiovascular disease prediction (31). In the Atherosclerosis Risk in Communities (ARIC) study (4), where ∼10% of >12,000 subjects had T2D, and in the Northern Manhattan Study (NOMAS) (14), where ∼20% of 1,118 subjects had T2D, arterial calcific lesions predicted cardiac and cerebrovascular events. In T2D, atherosclerotic plaque composition has been evaluated in several arterial districts: In the coronary bed, mixed or noncalcified plaques detected by computed tomography seem predominant with respect to calcified lesions (32,33) and significantly associated with acute cardiac events. At the carotid level, predominance of both echolucent plaques (9) and calcified plaques (15) has been reported. A histological analysis performed on carotid endoarterectomy specimens from 295 patients with T2D has demonstrated the presence of marked calcifications in 53.9% and of a large lipid core in 69.8% of subjects, findings similar to the nondiabetic counterpart (17). In T2D, although most studies suggest that echolucent plaques of the extracranial arteries are better predictors of vascular events than calcific ones (19–21,34), Cox et al. (22) recently demonstrated that calcified plaques predicted cardiovascular and all-cause mortality in a cohort of 699 subjects with T2D.

The current study shows for the first time in our knowledge a progressive rise in the risk of cardiovascular events with increasing plaque calcification as determined by ultrasound in patients with T2D. Such data contrast with several other observations that highly calcified plaques are more stable, less prone to rupture, and weakly associated with future vascular events. It should be noted that the debate about whether calcifications stabilize plaque is still open. Although histological assessment indicates that heavily calcified plaques do not predict the clinical outcome (35), intravascular ultrasound studies suggest that spotty calcifications are typical of culprit lesions in acute myocardial infarction (36), which are, by definition, unstable. In addition, as shown by frequency domain optical coherence tomography, the presence of spotty calcification is associated with features of plaque vulnerability (37). The size, number, and location of calcified areas likely determine plaque stability versus instability. Plaques occupied by a large medial calcific nodule are likely to be stiffer and more stable.

On the other hand, plaques with small, dispersed, multiple intimal subendothelial calcifications may be severely unstable and prone to thrombosis when the endothelial layer is damaged, owing to exposure of procoagulant calcium. In fact, mathematical modeling predicts that plaques with the embedded calcified spots display higher wall stress concentration in the fibrous cap a bit upstream to the calcified spot, even in the presence of mild stenosis (38). We emphasize that when patients with a degree of stenosis ≤30% are considered, the presence of echogenic plaques is still significantly associated with future MACE, even after adjusting for mean IMT. This finding has an important clinical value, implying that a worse cardiovascular risk may be attributed to patients even in the presence of mild stenosis when calcium is predominant in the plaques. In the present cohort, patients with heavily calcified or heterogeneous plaques showed similar event-free survival curves, which were significantly worse compared with that of patients with echolucent plaques. Macro- and microcalcifications coexist in the atherosclerotic process, and the relative importance of each in modulating plaque vulnerability may be different in the T2D versus the nondiabetic population. Indeed, several specific pathways triggered by diabetes can lead to calcification beyond inflammation (16), and the biological relationship between calcification and plaque instability should be explored further.

We have previously reported in this cohort that the presence of calcified plaques was associated with diabetic microangiopathy (23), the latter being itself a determinant of cardiovascular disease (39). Plaque microangiopathy, with neovascularization and hemorrhage (35), may be particularly important in diabetes and may modify the association between plaque calcification and vulnerability. It should be noted that only 48% of patients were on lipid-lowering medications, suggesting that the relationship between calcification and risk of MACE may be modified by intensification of statin use.

A limitation of the current study is that data on socioeconomic status and alcohol intake, which can modify the outcome, were not available. In addition, although we adjusted for several factors in the multivariable analysis, we cannot definitively rule out confounding due to other unmeasured variables. This may partly explain differences between the current study and previous ones on this topic.

In conclusion, in this prospective study conducted in patients with T2D, plaque calcification predicted incident MACE, even in patients with mild stenosis. Because this finding contrasts the common notion derived from studies in the general population that calcified plaques are stable, the biology of atherosclerotic calcification in patients with diabetes may be different from that in subjects without diabetes.

Funding. The study was supported by a grant from the Italian Ministry of Education, University and Research (MIUR) Projects of National Interest (PRIN 2012) to A.A.

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. S.V.d.K. contributed to the study concept and design; data collection, analysis, and interpretation; and manuscript writing. G.P.F. contributed to the study concept and design, data analysis and interpretation, and manuscript writing. S.G. contributed to the data analysis and interpretation and manuscript revision. M.M., A.V., and A.C. contributed to the data collection, analysis, and interpretation and manuscript revision. A.A. contributed to the study concept, data interpretation, and manuscript revision. S.V.d.K. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.