Cardiometabolic Risk Is Associated With Atherosclerotic Burden and Prognosis: Results From the Partners Coronary Computed Tomography Angiography Registry

Standard risk assessment guidelines regard type 2 diabetes mellitus (T2DM) as a coronary heart disease risk equivalent (1,2). Although a subgroup of patients with T2DM may have minimal coronary artery disease (CAD) (3), coronary heart disease is the most common cause of death in patients with T2DM. Moreover, the presence of T2DM is associated with a poorer prognosis after a myocardial infarction (MI) (4) As a result, the American Diabetes Association recommends primary prevention measures for CAD in all patients with T2DM, and particularly intensive secondary prevention among those with T2DM and CAD (5).

However, there is insufficient data to suggest that patients without T2DM who have cardiometabolic (CM) risk factors, such as obesity, dysglycemia, atherogenic dyslipidemia, or hypertension, should be treated as aggressively as patients with T2DM. Although such patients have been termed to be at increased CM risk (6–8), not all of them will have the same extent of anatomical CAD and risk of cardiovascular disease (5,9). Therefore, improved methods for CAD detection and risk assessment could be useful in improving patient outcome.

Nuclear myocardial perfusion imaging has been shown to be useful in risk stratification among individuals with T2DM (10), although in the contemporary era of widespread medical therapy, such testing did not lead to improvement in cardiovascular outcomes among asymptomatic patients with T2DM (11). Furthermore, in comparison with patients without T2DM, a shorter “warranty period” (i.e., favorable prognosis with normal results) has been found (12,13), possibly reflecting the fact that patients with T2DM are more likely to have diffuse coronary atherosclerosis and accelerated disease progression. More recently, the use of coronary computed tomography angiography (CCTA) has been evaluated among patients with T2DM and showed greater burden of disease in comparison with patients without T2DM as well as worse prognosis (14,15). However, these studies have been limited to all-cause mortality and did not examine patients without T2DM with CM risk factors. Therefore, the aim of this study was to evaluate the association between CM risk factors and prognosis and to determine whether differences in prognosis can be explained by CCTA findings.

This study examined data from all consecutive subjects, 18 years or older, who underwent a clinical CCTA at the Massachusetts General Hospital or the Brigham and Women’s Hospital from 2004 to 2011 with available data on baseline clinical risk factors. All scans were performed using a 64 detector (or newer generation) scanner. We excluded patients with prior known CAD (defined as prior percutaneous coronary intervention [PCI], coronary artery bypass graft surgery [CABG], or MI). The study was approved by the local institutional review board.

CCTA scans were performed according to established guidelines (16,17) and institutional protocol at the time of the scan. All CCTA exams were categorized as having no (0%), nonobstructive (<50%), or obstructive (≥50%) CAD. Vessels <2 mm were not included. The 18-segment model proposed by the American Heart Association (17) was used to categorize the presence and severity of CAD for each segment. Extent of CAD was also assessed by the number of vessels with CAD (left anterior descending, left circumflex, and right coronary artery) as one-vessel, two-vessel, and three-vessel/left main disease. Additionally, a binary CCTA finding of high-risk coronary anatomy was defined as left main or two- or three-vessel proximal obstructive coronary disease involving the left anterior descending (18). More detailed analysis of the extent and severity of the CAD was performed using previously validated scores: segment involvement score (SIS), the sum of the number of segments with CAD, which ranges from 0 to 17 (19); and segment severity score (SSS) (each segment receives a value according to the amount of disease present in that vessel [0, no CAD; 1, nonobstructive CAD; 2, 50–70% stenosis; 3, >70% stenosis]). The final score is the sum of each individual score, and ranges from 0 to 51 (20).

Review of all clinical data prior to the CCTA was used to verify the presence or absence of risk factors (1,21). Hypertension was defined as a systolic blood pressure ≥130 mmHg, diastolic blood pressure ≥85 mmHg, or diagnosis/treatment of hypertension. Obesity was defined as BMI ≥30 kg/m². Hypertriglyceridemia was defined as triglycerides ≥150 mg/dL. Low HDL cholesterol was defined as <40 mg/dL (male) or HDL <50 mg/dL (women). T2DM was defined by hemoglobin A_1c (HbA_1c) ≥6.5% (45 mmol/mol) (22), two fasting glucose levels ≥126 mg/dL, or diagnosis/treatment of T2DM. Dysglycemia was defined as HbA_1c ≥5.7% (39 mmol/mol) and <6.5% (45 mmol/mol) or fasting glucose level 100–125 mg/dL without a known diagnosis of T2DM (22). Insulin-requiring T2DM was defined according to the presence of exogenous insulin, excluding where used for sliding scale purposes only. All patient clinic notes were reviewed to determine the duration (years) of diagnosed T2DM. Smoking was defined as current (tobacco products used within the last month), former, or never. Family history of premature CAD was defined as any first-degree family member with a history of clinical CAD prior to age 60 years.

For each patient, we determined the total number of CM risk factors present, as clustering of these common risk factors has been recognized as the key contributor to the pathogenesis of both T2DM and cardiovascular disease (6,7). We used the following CM risk factors: 1) obesity, 2) low HDL, 3) hypertriglyceridemia, 4) hypertension, and 5) dysglycemia. All patients were verified as having T2DM or not by past medical history and laboratory testing. BMI was used as a measure of central adiposity (instead of waist circumference), although these two measures have consistently been shown to be highly correlated (23) and have a similar predictive value for future onset of diabetes (24) or clinical CV disease (25). This method is consistent with the recommendations of the American Association of Clinical Endocrinologists (26) and the World Health Organization (27), both of which include BMI as a means of defining CM risk.

All patient charts were reviewed by two cardiologists who were blinded to CCTA results for adjudication of cardiovascular events. In order to ensure that events outside of our healthcare network were captured, a standardized questionnaire was mailed to each patient in a manner similar to prior studies (28,29). In addition, patients had the option of completing a web-based version of the questionnaire via the REDCap (Research Electronic Data Capture) system (30), which is encrypted, secure, and Health Insurance Portability and Accountability Act (HIPAA) compliant. For patients who did not reply to the questionnaire upon repeated mailings, scripted phone interviews were performed based on the questionnaire. In order to avoid erroneous diagnoses from self-reported events, all self-reported events were verified via outside medical record review by two cardiologists, blinded to CTA results, with discordant events adjudicated by consensus.

The primary end point was major adverse cardiovascular events (MACE), composed of MI, late coronary revascularization, unstable angina, and cardiovascular mortality. In addition, we further evaluated the combined end point of only cardiovascular mortality and nonfatal MI to avoid inherent bias of softer outcomes (e.g., angina and coronary revascularization). As a secondary end point, we evaluated all-cause mortality. Diagnosis of MI was confirmed by two of three of the following: chest pain or equivalent symptom complex, positive cardiac biomarkers, and electrocardiogram changes typical of MI (31). Time to the first coronary revascularization procedure (PCI or CABG) was evaluated. Early revascularizations (≤90 days post CCTA) were removed from survival analysis to minimize verification bias (32–34). That is, patients with ≥50% stenosis by CCTA are likely to undergo invasive angiography and revascularization early after the CCTA, whereas late revascularizations are less likely to be associated with the CCTA and more likely to be associated with CAD progression and prognosis. Unstable angina without revascularization was defined as chest pain or chest pain equivalent with dynamic electrocardiogram changes such as ST depression or T wave inversion but without abnormal cardiac biomarkers and characterized by 1) rest symptoms, 2) new-onset angina (<2 months duration), or 3) increasing duration or severity of previously stable anginal symptoms (35).

Deaths were confirmed by the Social Security Death Index. For all patients who died, the cause of death was obtained from the National Death Index as well as the Massachusetts Department of Vital Statistics, if applicable. In addition, other pertinent clinical records (e.g., death notes, autopsy findings, and hospice notes) related to the cause of death were reviewed. Using all available data, the causes of deaths were adjudicated by two cardiologists blinded to the CTA results, with cardiovascular mortality defined as a primary cause of acute MI, atherosclerotic coronary vascular disease, congestive heart failure, valvular heart disease, arrhythmic heart disease, stroke, or other structural or primary cardiac cause of death.

Continuous variables are reported as mean ± SD. Categorical variables are reported as counts and proportions. Continuous variables were compared between groups using ANOVA techniques. Median segment scores were compared between groups using the Kruskal-Wallis test. Categorical variables were compared using χ² or Fisher exact test, where appropriate. Univariable and stepwise multivariable logistic regression was used to assess the association between covariates and the presence of obstructive CAD (defined as above). For purposes of multivariable logistic regression for variables associated with obstructive CAD only, missing HbA_1c (HbA_1c was available in 78% of patients with T2DM) was imputed based on linear regression of HbA_1c with available fasting glucose levels for patients without HbA_1c, as has been previously validated (36,37). The Kaplan-Meier method was used to assess event-free survival for the various outcomes of interest (as defined above). To account for baseline differences in patient characteristics for patients with T2DM who are treated or not treated with exogenous insulin, a propensity-weighted analysis was conducted. In order to adjust for these baseline differences, we used a two-step process (38,39) of first evaluating a priori all variables associated with insulin therapy in a logistic model and including those baseline characteristics with a P value <0.10 in a final propensity score, which represented a summary statistic for prediction of treatment with insulin therapy. Second, we included the propensity score in a propensity-weighted Cox proportional hazards model for prediction of MACE. Such an approach controls for confounding due to baseline differences in each cohort in addition to the effect of nonrandomized treatment allocation to statin therapy. Univariable and multivariable propensity-weighted Cox proportional hazards models were constructed to compare risk between strata. Predictors with a P < 0.10 in univariable analysis were included in the propensity-weighted multivariable model. All statistics were performed using Stata version 12.1 (Statacorp, College Station, TX).

A total of 1,148 patients met inclusion criteria, of which 10 patients (0.9%) had no available outcomes follow-up and 20 patients (1.7%) had uninterpretable CTA, leaving 1,118 patients available for analysis (Table 1). The most common indication for referral was chest pain symptoms in 641 (57%) followed by abnormal stress testing in 278 (25%) and preoperative assessments in 66 (6%) patients.

There were 483 (43%) patients without T2DM with fewer than three CM risk factors and 187 (17%) patients without T2DM and three or more CM risk factors. Among the remaining 451 (40%) patients with T2DM, 367 did not require insulin and 84 required insulin.

Over a mean follow-up of 3.1 years, there were 46 (4.1%) all-cause deaths, 21 (1.9%) cardiovascular deaths, 13 (1.2%) MIs, 13 (1.2%) unstable angina without revascularization, and 34 (3.1%) late coronary revascularizations.

As expected, significant differences in the prevalence of baseline risk factors occurred across CM risk groups (Table 1). HbA_1c was progressively higher across groups of increased CM risk (P < 0.001).

Across a spectrum of increasing CM risk, there was increasing prevalence of any CAD from 51% among those with no T2DM and fewer than three CM risks to 60% among those with no T2DM and three or more CM risks, to 71% among those with T2DM not requiring insulin, to 80% among those with T2DM requiring insulin (P < 0.001). Although statistically significant differences occurred (P < 0.001), the prevalence of nonobstructive CAD was numerically similar across all patient groups, ranging from 36, 42, 41, and 33%. Finally, obstructive CAD was markedly increased according to CM risk, ranging from 15 to 46% (P < 0.001). When considering the extent of disease, SIS and SSS both increased significantly across groups of increasing CM risk (Fig. 1).

In logistic regression models to identify predictors of obstructive CAD across the entire population (Fig. 2), CM risk factors and T2DM exhibited a strong and significant association with obstructive CAD. Typical angina was also associated with obstructive disease (odds ratio [OR] 1.6, P = 0.03). A stepwise increase in association with obstructive CAD was seen with low HDL (OR 0.98 for each 1 mg/dL increase, P < 0.001), dysglycemia (1.34 for each 1% [11 mmol/mol] increase in HbA_1c, P < 0.001), hypertriglyceridemia (1.55, P = 0.004), and hypertension (3.50, P < 0.001). The OR for obesity was not significant (1.14, P = 0.3). Diabetes was a significant univariable predictor of obstructive CAD (2.34 for patients with T2DM not requiring insulin and 4.79 for T2DM requiring insulin, P < 0.001 for both). In the absence of diabetes, the presence of three or more CM risk factors was not associated with a numeric increase in the odds of obstructive CAD (univariable OR 1.21, P = 0.4) but was associated with increased SIS and SSS (Fig. 1).

Significant baseline characteristics associated with insulin use among patients with diabetes included age, male sex, low HDL cholesterol, hypertriglyceridemia, hypertension, higher BMI, and years diagnosed with T2DM. In the final propensity model for insulin use, the C statistic was 0.91 (P < 0.001).

As shown in Supplementary Fig. 1, patients with T2DM requiring insulin exhibited the highest annualized risk of cardiovascular death or MI (3.1%), whereas patients without T2DM and fewer than three CM risk factors exhibited the lowest risk (0.3%, P < 0.001). Patients without T2DM and three or more CM risk factors had an equivalent annualized rate of cardiovascular death or MI as compared with patients with T2DM not requiring insulin (0.7 vs. 1.1%, P = 0.2). Similar results were found when we assessed annualized risk of all-cause mortality (Supplementary Fig. 2).

When each CM risk group was stratified by severity of CAD, we found that increased stenosis by CCTA was associated with increased MACE (Fig. 3, top), although differences in cardiovascular death and MI (Fig. 3, bottom) and all-cause mortality (Supplementary Fig. 2) were less pronounced.

To evaluate whether early revascularizations (occurring <90 days post-CCTA) influenced prognosis, we compared revascularization rates by metabolic group. Those with T2DM requiring insulin had a significantly increased rate of early revascularization (16%, P < 0.001), but other groups did not differ significantly from one another (P = 0.3): 5.0% for those without T2DM and fewer than three CM risk factors, 5.6% for those with three or more CM risks, and 7.6% for those with T2DM not requiring insulin.

When analysis was restricted to patients with T2DM, those not requiring insulin and without stenosis by CCTA had the greatest long-term event-free survival, whereas patients requiring insulin had an unfavorable prognosis that was further worsened in the presence of stenosis by CCTA (Fig. 4, top).

In patients without T2DM, the poorest survival occurred in patients with both three or more CM risk factors and significant stenosis (Fig. 4, bottom). Notably, patients with significant stenosis and lower CM risk (fewer than three CM risk factors) had similar outcomes as patients with three or more CM risk factors and no stenosis (P = 0.3).

To further examine the association between CM risk, CAD severity, and prognosis, we constructed several univariable and multivariable Cox regression models for combined MACE (Supplementary Fig. 3). In univariable Cox models, diabetes, extent of CAD, presence of high-risk anatomy, and smoking were associated with risk of combined MACE and cardiac death or MI. In addition, the presence of diabetes was also strongly associated with combined MACE (hazard ratio [HR] 2.55, P < 0.001) and cardiac death or MI (2.89, P = 0.004). A significant interaction (P interaction = 0.008) occurred for diabetes and insulin use.

The final propensity-weighted multivariable Cox regression model adjusted for age, smoking, propensity score for insulin use, and high-risk anatomy. In this model, age, current smoking, increasing CM risk category, and high-risk CCTA were independently associated with adverse events (Supplementary Fig. 3). Analysis according to number of CM risks demonstrated that among those with T2DM, the presence of three additional CM risks was associated with increased risk for MACE (HR 1.98, P = 0.035). Among patients without T2DM, the presence of four or more CM risks was borderline significantly associated with MACE (2.47, P = 0.052) independent of age (1.03, P = 0.049) and high-risk CAD (7.20, P < 0.001).

A finding of high-risk CAD by CCTA significantly increased the propensity-weighted multivariable HRs for MACE in each group with increased CM risk (Supplementary Fig. 4). Similar trends were obtained from the univariate and multivariable models for CV death and MI in addition to all-cause mortality (data not shown).

Our study demonstrates that across categories of increasing CM risk, there is a graded, dose-dependent association with the presence, extent, and severity of CAD as well as adverse cardiovascular prognosis. A particularly important finding is that among patients without T2DM, the presence of multiple CM risk factors is increasingly associated with risk of future cardiovascular events. Furthermore, among patients without T2DM, CM risk appears to modify the association between obstructive CAD and adverse prognosis, such that the presence of obstructive CAD most strongly impacts patients with higher CM risk (Fig. 3). Finally, among patients with T2DM (already at “high CM risk”), insulin use identifies a subgroup with a worse prognosis. Collectively, these results suggest that CM risk is strongly associated with both CAD and cardiac events, and may be an important risk factor in CAD pathogenesis and prognosis before the onset of frank T2DM.

Our findings that patients with T2DM requiring insulin have a greater risk of events have also been shown by Berman et al. (40), who showed that individuals with T2DM requiring insulin have higher cardiac mortality, particularly if they also have an abnormal adenosine myocardial perfusion imaging study, when the annual cardiac mortality was 9%. However, our study had a lower event rate in this group (3% annual rate of cardiovascular death or MI), reflecting the lower risk associated with patients that undergo CCTA versus MPI. Moreover, whereas our study excluded all patients with prior CAD, nearly half of the patients with diabetes in the study by Berman et al. (40) underwent prior coronary revascularization.

The mechanism underlying the higher event rate of patients with T2DM requiring insulin is not known. Some have hypothesized that direct harmful effects of insulin therapy on endothelial dysfunction (41) could account for poorer long-term prognosis, independent of epicardial disease severity. Alternatively, the use of insulin may serve as a marker for patients with more prolonged duration of diabetes and worse CM risk status and who are more likely to have hyperglycemia. In turn, hyperglycemia leads to increased inflammation and oxidative stress as well as trapping of proatherogenic particles in the vascular endothelium (42). These mechanisms also account for the increased instability of atherosclerotic plaque among patients with T2DM (43), potentially explaining the higher event rate observed in patients with T2DM on insulin. Although recently it has been shown that patients with T2DM with stenosis in at least two coronary vessels have a significantly lower risk of death or MI when treated with CABG versus PCI (44), the optimal treatment for individuals with nonobstructive disease has not been defined in prospective studies and warrants further investigation.

Prior studies have demonstrated that patients with increased CM risk have a higher prevalence of CAD and worse prognosis. Investigators from the Multi-Ethnic Study of Atherosclerosis (MESA) demonstrated that T2DM and metabolic syndrome are associated with higher prevalence of nonzero coronary arterial calcium (CAC) score and higher CAC (9). Recently, Wong et al. (45) also demonstrated that T2DM and the metabolic syndrome are associated with a higher rate of CAC progression and adverse cardiovascular outcomes. Extending these observations, it has also been shown that the absence of CAC identifies patients at low risk. Malik et al. (3) showed that even among patients with T2DM or the metabolic syndrome, the absence of CAC was associated with a favorable prognosis; similar findings have also been found by Raggi et al. (46)

Our findings extend the results of these prior studies utilizing CAC. Whereas these prior studies have focused on coronary calcifications in predominantly asymptomatic populations, our study is among the first to investigate the association of CM risk across patients with T2DM and patients without T2DM with CCTA, a more sensitive test for coronary plaque, which can also assess for coronary stenosis. Recent important studies by Rana et al. (14) and Hadamitzky et al. (15) demonstrated that CCTA has prognostic utility among patients with T2DM, although these studies used an end point of all-cause mortality and were unable to evaluate glycemic status or distinguish differences according to insulin use. Our analysis specifically evaluated the end point of cardiovascular mortality and nonfatal MI. Importantly, these prior studies and our analysis demonstrate that CCTA identifies a subgroup of patients (those with no CAD) with an excellent prognosis in spite of guideline recommendations for consideration as being “high risk” by virtue of a diagnosis of T2DM. However, since the proportion of patients with T2DM requiring insulin who had normal studies (i.e., no plaque and no stenosis) was only 20%, the yield of testing in this group may be low (i.e., the number needed to scan to identify one normal study is five).

Finally, we show that HbA_1c has utility not only to define glucose intolerance and diabetes as CM risk factors but also as a predictor of increasing severity of CAD. Specifically, we found a 34% increase in the odds of obstructive CAD per 1% (11 mmol/mol) increase in HbA_1c (Fig. 2). Interestingly, however, increasing HbA_1c was not associated with worse prognosis. This finding is in keeping with prior randomized trials that demonstrated that aggressive control of hyperglycemia among patients with T2DM reduced microvascular complications such as retinopathy but was not associated with a benefit in cardiovascular outcomes, perhaps due to off-target effects of insulin therapies (47). Specifically, in the Action to Control Cardiovascular Risk in Diabetes (ACCORD) trial (48), the cardiovascular mortality rate of the intensive glucose-lowering group (0.8% per year) was not significantly different from the standard therapy control subjects (0.6% per year). Notably, these event rates were comparable to the overall CV mortality rate in our study (0.8% per year among all those with T2DM). However, our finding is exploratory, as HbA_1c assessment was cross-sectional and its association with risk may be influenced by disease-specific therapy.

The results of our study must be viewed in context with its design. Our study was retrospective, thus limiting full ascertainment of some risk factors for all patients who underwent CCTA. Thus, our sample size was reduced in order to include only patients with completely available data for all risk factors. Additionally, we did not use measures of central adiposity, such as waist-to-hip ratio, but rather used BMI. BMI is highly correlated (23) with and has similar predictive value for future onset of diabetes (24) or clinical CV disease (25) as waist-to-hip ratio. Also, we did not obtain follow-up creatinine in order to assess for contrast nephropathy, as the rate of this complication is exceedingly low in clinical practice. For instance, among patients with T2DM referred for CT angiography for pulmonary embolus (a study that requires a higher dose of contrast than coronary CT; 120 vs. 70 mL), there was no clinical case of renal failure, although 4% of patients experienced an increase in creatinine of unknown clinical significance (49). Because contrast nephropathy after CCTA is a rare event even among patients with T2DM (50), we followed Society of Cardiovascular Computed Tomography guidelines for performance of CCTA, which do not recommend any active surveillance for this rare event (16). Finally, the current study only includes patients who underwent CTA, and thus these results can only be generalized to individuals who do not have contraindications for intravenous contrast and those with no history of prior CAD. In addition, this study involved patients clinically referred for CTA at tertiary care referral centers, who may therefore represent an inherently higher-risk group of patients. Although not the focus of our study, it is well known that anatomic measures of coronary stenosis, whether by CCTA or invasive angiography, only have a modest correlation with physiologic tests that are designed to detect ischemia (51). Furthermore, although anatomic stenosis (52,53) and extent of CAD (54) have been shown to predict outcomes, many adverse CV events occur in the setting of plaque rupture at sites of previously <50% luminal stenosis (55), demonstrating that anatomy alone does not exclusively determine prognosis. Nevertheless, as has been demonstrated in prior research (52,53), anatomical markers of plaque and stenosis do provide important prognostic data. Notwithstanding these limitations, we demonstrate a strong and independent association of CM risk, CAD extent, and prognosis, suggesting that across a diverse spectrum of baseline CM risk, CCTA may stratify clinically important risk.

In conclusion, our study shows that CCTA identifies patients across a spectrum of baseline CM risk factors with differential CAD prevalence and prognosis at 3.1 years follow-up. Patients without T2DM with increased CM risk have a similar prognosis as patients with T2DM but not requiring insulin. On the other hand, among patients with T2DM, the need for insulin therapy identifies individuals with greater burden of CAD and worse prognosis. Although using CCTA to more effectively stratify patients without T2DM but with increased CM risk and those with overt diabetes warrants additional investigation, our findings suggest that integrating data on CM risk factors together with data on CAD presence and severity may have an important role in efficient allocation of aggressive preventive therapies.

Duality of Interest. U.H. reports research support from Siemens Medical Systems. F.R. reports research support from Toshiba Medical Systems. No other potential conflicts of interest relevant to this article were reported.

Author Contributions. E.H. was involved in all steps of this study. M.S.B. and R.B. were involved in data collection, study design, and reviewing and editing the manuscript. D.O., M.P.C., P.M., and J.H. assisted in data collection. M.S., R.S., Q.A.T., K.N., F.R., and T.J.B. reviewed and edited the manuscript. M.F.D.C., B.G., U.H., and S.A. assisted in study design and reviewed and edited the manuscript. E.H. 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.

Prior Presentation. This study was presented at the 8th Annual Scientific Meeting of the Society of Cardiovascular Computed Tomography, Montréal, Québec, Canada, 11–14 July 2013.