Increased epicardial adipose tissue thickness is a predictor of new-onset diabetes mellitus in patients with coronary artery disease treated with high-intensity statins

The study retrospectively enrolled patients who underwent percutaneous coronary intervention (PCI) between January 2009 and June 2013, received high-intensity statin treatment for at least 6 months, and had at least one baseline echocardiographic evaluation within 3 months after PCI at Seoul National University Bundang Hospital. We excluded patients with DM at baseline or no clinical/laboratory information regarding DM status, patients with a follow-up duration less than 6 months, and patients with a poor echocardiographic image quality for the measurement of EAT thickness (Additional file 1: Figure S1). PCI was performed using standard techniques, and follow-up was performed according to routine clinical guidelines. For each patient, the follow-up duration was calculated based on the prescriptions of high-intensity statins. High-intensity statin therapy, defined as either atorvastatin (40 or 80 mg) or rosuvastatin (20 or 40 mg), had been administered according to the 2013 American College of Cardiology/American Heart Association guidelines [8]. DM was defined as fasting blood glucose levels ≥ 126 mg/dL, glycated hemoglobin levels ≥ 6.5% (48 mmol/mol), and/or the need for oral hypoglycemic agents or insulin. If there was no clear clinical diagnosis, or if a patient had discordant results from two different tests, a second test was searched for confirmation. Prediabetes was defined as fasting blood glucose levels of 100–125 mg/dL or glycated hemoglobin levels of 5.7–6.4% (39–47 mmol/mol) [14].

The study protocol was approved by the Institutional Review Board of Seoul National University Bundang Hospital and was conducted according to the principles of the Declaration of Helsinki.

All subjects underwent echocardiographic examination performed using commercially available ultrasound machines (Vivid E9, GE Healthcare, Chicago, USA; EPIQ 7, Philips Healthcare, Amsterdam, The Netherlands), and standard examination was performed with the patient in left lateral position. Left ventricular ejection fraction was calculated with the modified biplane Simpson’s method. EAT thickness was measured at the end of systole and diastole on the free wall of the right ventricle in the parasternal long-axis view on standard transthoracic echocardiography, and was defined as an echo-free or hypoechoic area adjacent to the right ventricle (Additional file 1: Figure S2). This method was validated in previous studies and shown to be strongly correlated with various metabolic markers [1, 7]. Only the maximum EAT thickness values were measured. The measurement was performed for two beats, and the average value was retained.

Data are presented as numbers and frequencies for categorical variables and as median and interquartile ranges for continuous variables, and were compared using Student’s t-test or the Mann–Whitney U test. To compare the groups, the χ² test (or the Fisher’s exact test when any expected cell count was < 5 for a 2-by-2 table) was used for categorical variables, and the unpaired Student t-test or one-way analysis of variance was applied for continuous variables.

In the multivariate analysis performed to identify variables influencing NODM, we used the multivariable Cox proportional hazard model. Candidate variables with P < 0.10 in the univariate analyses, duration of statin treatment, and previously described risk factors of DM [i.e., age, male sex, body mass index (BMI), and hypertension] were included in the model [15, 16]. For the sensitivity analysis of predictors of progression of glucose intolerance, we used the binary logistic model based on multiple variables. Variables included in the logistic regression model were identical to those of the multivariable Cox proportional hazard model. To determine the best cutoff value of EAT thickness that would be included in the predictive model, we performed receiver operating characteristic curve analysis. To determine intraobserver variability, one of the authors (JK) measured EAT thickness at systole and diastole twice at an interval of > 30 days. Agreement was analyzed by means of the Bland–Altman plot (Additional file 1: Figure S3) and by determination of the intraclass correlation coefficient using the two-way mixed model (coefficient for EAT thickness at systole: 0.936 [0.916–0.951]; coefficient for EAT thickness at diastole: 0.943 [0.925–0.956]).

All statistical tests were two-tailed. A two-sided probability value less than 0.05 was considered to indicate statistical significance. Statistical tests were performed using SPSS version 20 (SPSS Inc., Chicago, IL, USA).

A total of 321 patients were enrolled in this study, according to the flow chart provided in Additional file 1: Figure S1. The mean age was 59.9 years, 74% of patients were male patients, and 64% presented with acute coronary syndrome. The patients received high-intensity statins for a median of 930 days; atorvastatin 40 mg in 204 patients (63.6%), atorvastatin 80 mg in 57 patients (17.8%), and rosuvastatin 20 mg in 60 patients (18.7%; Table 1).

New-onset diabetes mellitus occurred in 40 patients (12.5%), with the incidence of NODM increasing gradually over the course of a mean follow-up of 3.9 years. Regarding baseline characteristics, patients with NODM had a higher frequency of baseline prediabetes, and higher levels of fasting blood glucose and HbA1c. Furthermore, among echocardiographic variables, EAT thickness at diastole and systole were significantly larger in the NODM group than in the non-NODM group (Additional file 1: Figure S4), whereas other variables did not show significant difference between the two groups. Regarding the relationship between variables, we found a moderate positive correlation of EAT thickness with HbA1c (Pearson correlation coefficient 0.307, P < 0.001, Additional file 1: Figure S5).

Regarding factors associated with NODM, univariate Cox regression analysis showed that EAT thickness and prediabetes at baseline were significant predictors of NODM (Table 2), which remained significant after multivariate adjustment for significant covariates (for each 1 mm of EAT thickness at systole: hazard ratio (HR) of 1.580, 95% confidential interval (CI) of 1.346–1.854, P < 0.001; for prediabetes at baseline: HR 4.321, 95% CI 1.998–9.349, P < 0.001; Table 2). Using binary cutoff values derived from the receiver operating characteristic curve analysis, EAT thickness at systole equal to or larger than 5.0 mm showed a sensitivity of 52.5%, and specificity of 80.8% for predicting NODM (Fig. 1). Moreover, the HR for NODM in patients with EAT thickness ≥ 5 mm was 3.402 (95% CI 1.751–6.611, P < 0.001), showing that EAT thickness remained as an independent predictor of NODM after adjusting clinical variables, including prediabetes at baseline.

When we stratified the patients into four groups according to EAT thickness at systole and presence of prediabetes, the incidence of NODM was highest in patients with EAT thickness ≥ 5 mm and prediabetes at baseline (17 out of 40 patients; 42.5%), which was 12.0-fold higher than that in patients with EAT thickness < 5 mm and without prediabetes at baseline (5 out of 153 patients; 3.3%, Fig. 2a). On Kaplan–Meier curve analysis and Cox regression analysis, patients with both risk factors had a significantly higher risk for NODM (Fig. 2b, Table 3). Meanwhile, patients with either one of the risk factors (i.e., those with EAT thickness ≥ 5 mm without prediabetes and those with EAT thickness < 5 mm with prediabetes) had a similar risk for NODM (P = 0.509).

To evaluate the association between EAT thickness and gradual impairment in glucose tolerance, we stratified the patients into two groups according to the progression of glucose intolerance (progression vs. no progression; Additional file 2: Table S1). Progression of glucose intolerance was noted in about 40% of the total study population (progression group), among whom 81 patients (25.2%) showed normal glucose tolerance at baseline and developed new-onset prediabetes during the follow-up period. The progression group had more prediabetic patients, higher levels of fasting blood glucose, and marginally longer statin treatment duration. Additionally, analysis of echocardiographic findings revealed larger EAT thickness at systole in the progression group (4.2 [3.2, 5.4] vs. 3.8 [2.9, 4.8] mm, P = 0.014; Additional file 2: Table S2). A logistic regression model including age, sex, BMI, hypertension, statin duration, prediabetes, and EAT thickness at systole showed that EAT thickness was an independent predictor of progression of impaired glucose tolerance (for each 1 mm of EAT thickness at systole: odds ratio of 1.309, 95% CI 1.117–1.534, P = 0.001), and so was prediabetes (odds ratio of 3.265, 95% CI 1.919–5.555, P < 0.001; Additional file 2: Table S3).

Statins are effective therapeutic agents for prevention of cardiovascular events, and can reduce mortality in patients with coronary heart disease [17]. However, recent studies reported that statin treatment may be associated with an increased risk of NODM. A meta-analysis of 13 trials involving 91,140 individuals showed that statin treatment was associated with a 9% increase in the 4-year risk of NODM [10]. Regarding the dose–effect relationship in statin-associated NODM, some studies have shown a higher risk of incident diabetes in patients on higher-intensity statin therapy [9, 18], while some suggest that there might be difference in the incident diabetes by statin class [19]. Large scale studies focusing on the occurrence of NODM by statin intensity or by statin class should be conducted to give us clear answers on this issue. Our present study found a similar rate of NODM in patients using high-intensity statins during 4 years of follow-up. Regarding previous reports showing that NODM is associated with a substantial risk for mortality [20], further efforts should be allocated to the early prediction and prevention of NODM [21].

The EAT is known as the true visceral fat deposit of the heart, lying directly on the epicardial surface of the myocardium within the pericardial sac [22]. Due to its close proximity to the coronary vessels, the EAT exerts profound effect on the local physiology of the myocardium and the coronary vasculature by expressing various cytokines [23]. Additionally, EAT was associated with cardiovascular risks and further development of cardiovascular complications [24]. EAT thickness has also been shown to be related to the metabolic status of the individual. Specifically, Yorgun et al. reported that EAT thickness was significantly increased in patients with metabolic syndrome, and that age and BMI, which are factors related to metabolic syndrome, were the strongest independent predictors of EAT thickness [25]. A recent meta-analysis also showed that EAT thickness was significantly higher in patients with metabolic syndrome [26]. Other studies have explained that the association between metabolic syndrome and EAT thickness may be attributed to the endocrine action of the EAT, which also affects insulin sensitivity [27], designating EAT as a biologically active organ. Conversely, some studies suggested the beneficial effect of EAT by protecting the heart against myocardial stress, hypertension, and local inflammation. EAT may even function as a brown adipose tissue store which can protect adjacent tissues from hypothermia, while showing high degrees of white adipose tissue lipolysis allowing the buffering of high toxic levels of free fatty acids [6]. Additionally, a genetic study explored the EAT transcriptome, unveiling a majority of genes involved in coagulation, endothelial function, phospholipase activity, apoptosis, and immune signaling [28]. Despite these beneficial effects, EAT may shift from being protective to detrimental for obesity and cardiovascular homeostasis [6]. Although the mechanisms that regulate the balance between protective and harmful effects of EAT are not clearly understood, epicardial fat can serve as target for pharmaceutical agents targeting the adipose tissue [29]. Furthermore, the association of EAT and diabetes has been studied in a few studies. Increased EAT thickness was independently associated with the prevalence of diabetes, insulin resistance and cardiac contractile dysfunction in diabetes [30, 31]. In the present study, we found that EAT thickness was closely associated with NODM. Aside from prediabetes, which is a well-known risk factor for DM, EAT thickness was the only other significant predictor of statin-associated NODM.

Previous studies have reported conflicting results regarding the potential predictors of NODM. Specifically, the IDEAL study suggested that only patients who already have elevated risk for DM are at increased risk to develop statin-associated DM [32]. Furthermore, the Justification for Use of Statins in Prevention: an Intervention Trial Evaluating Rosuvastatin (JUPITER) study reported that the risk of statin-associated NODM was independent of baseline glucose levels, whereas Waters et al. reported that the development of NODM can be predicted based on baseline fasting glucose levels and other components of the metabolic syndrome (i.e., triglyceride levels, BMI, and hypertension) [33]. On the other hand, a cohort-based study by Woestijne et al. found that the increase in the risk of type 2 DM with statin therapy was independent of metabolic syndrome or insulin resistance [13]. The discrepancies in these previous observations may be attributed to the differences in the study population, as well as to variable statin dosage and duration.

In the present study, we limited our study population to patients with CAD who underwent PCI and required strict lipid-lowering therapy and we were able to draw a conclusion based on a relatively homogeneous small sample of patients. Moreover, considering that echocardiography is performed in nearly all CAD patients receiving PCI, our finding that EAT thickness is a predictor of statin-associated NODM may have considerable clinical implication.

Despite the risk of statin-associated NODM, the general consensus is that the positive effects of statins outweigh the negative effects on metabolic control [34]. The TNT study showed that patients at risk of statin-associated NODM also obtained substantial benefit from high-intensity statins [11]. This finding may be related to the traditional effect of statins, which lower blood cholesterol levels, and have favorable pleiotropic effects on endothelial function, oxidative stress and inflammation [35]. Nevertheless, in clinical practice, it remains important to identify factors that can estimate the risk of statin-associated NODM.

Several limitations should be noted. The study population was relatively small compared to cohorts investigated in previous studies or randomized controlled trials. Furthermore, we may have introduced selection bias by excluding patients prescribed with high-intensity statins for less than 6 months and those only prescribed with low- or moderate-intensity statin. Moreover, the patients in our study had various patterns of statin usage; specifically, some were not statin-naïve, and had been prescribed low- or moderate-intensity statins before receiving high-intensity statins. We also used a single method to measure EAT thickness. A previous study proposed a method to measure EAT thickness at the anterior interventricular groove, which may be more accurate than our method [36]. However, this measurement was not applicable in our retrospective analysis, because this echocardiographic view was not routinely used in our institute. Furthermore, our study did not include a control arm, which made it impossible to investigate the association between high-dose statin and NODM. However, this was not the purpose of our study, but rather its background. Therefore, the findings of our investigation may be considered as hypothesis-generating, and further large-scale studies are warranted.