The effect of dapagliflozin treatment on epicardial adipose tissue volume

In 40 type-2 DM patients with coronary artery disease (CAD) (10 women and 30 men; mean age of all 40 patients was 67.2 ± 5.4 years), EAT volume was compared prospectively between the dapagliflozin and conventional treatment groups (both n = 20) during a 6-month period. Because there are few reports on the ability of SGLT-2 inhibitors to reduce EAT, we used a previous report on the effect of glucagon-like peptide-1 on EAT volume (approximately a 10% reduction) [14] to estimate a sample size of 17 subjects per group. Sample size estimates were based on the following: standard deviation, 10; α-level, 0.05; and power, 80%.

A total of 40 consecutive patients who received conventional medical therapy for CAD were recruited at the Tachikawa General Hospital. Patients were randomized into the dapagliflozin and conventional treatment groups. Group allocation was managed by an independent administrator. Thereafter, the addition of any medication was chosen. In the dapagliflozin group, only dapagliflozin was newly prescribed with medications before the allocation, while in the conventional treatment groups, oral hypoglycemic drugs except SGLT-2 inhibitors were newly prescribed with medications before the allocation. During the 6 months, the treatment regimen remained unchanged, except when unacceptable hyperglycemia, hypoglycemia, or adverse events occurred.

Patients who satisfied the following inclusion criteria were eligible for enrollment: [1] HbA1c level ≥ 6.5%, despite treatment with only diet and exercise therapy, or treatment with a combination of oral antidiabetic agents. Patients were excluded if they had age ≥ 75 years, insulin therapy, atrial fibrillation, iodine-based contrast agent allergy, renal insufficiency (estimated glomerular filtration rate < 45 mL/min/1.73 m²), chronic inflammatory disease, or coronary artery bypass graft surgery. Risk factors included were hypertension (oral treatment with antihypertensive drugs or documented history), hypercholesterolemia (oral treatment with lipid-lowering drugs or a total cholesterol measurement > 220 mg/dL), and cigarette smoking.

EAT volume was measured with multislice CT at baseline and 6 months after the randomization of patients. Similarly, blood samples were taken at baseline and 6 months later. The study protocol was approved by the institutional ethics committee of Tachikawa General Hospital, and written informed consent was obtained from all patients.

Documented CAD encompassed one or more of the following: stable angina pectoris that was diagnosed with stress electrocardiology, or radioisotope or coronary angiography; previous acute coronary syndrome including unstable angina, and myocardial infarction that was diagnosed with coronary angiography and treated with percutaneous coronary intervention. However, the patients who underwent percutaneous coronary intervention for acute coronary syndrome, received optimal medical therapy and were enrolled 6 months after initial percutaneous coronary intervention.

Blood samples were taken from the antecubital vein after a minimum of 12 h of fasting to determine serum levels of lipids [triglyceride (TG), high-density lipoprotein cholesterol (HDL), and low-density lipoprotein (LDL)], HbA1c [National Glycohemoglobin Standardization Program], insulin, and brain natriuretic peptide (BNP). In addition, samples of morning midstream urine were collected to evaluate the albumin/creatinine ratio. PAI-1 and TNF-α levels were measured at SRL (Tokyo, Japan), TNF-α was measured by an enzyme-linked immunosorbent assay using a commercial kit according to the manufacturer’s recommended protocol. Protein levels of total PAI-1 were determined using an LPIA-PAI test. Interactive, 24-variable model homeostatic model assessment insulin resistance-2 (iHOMA2%S) was measured as an index of insulin resistance based on a previous report [15]. In addition, HOMA-IR was calculated based on the following formula: HOMA-IR = [fasting serum insulin (μU/mL) × fasting blood sugar (mg/dL)/405].

For EAT measurements, CT angiography images were assessed using electrocardiography-gated cardiac CT scans on a 320-slice multi-detector computed tomography scanner (Aquilion One, Toshiba Medical Systems, Otawara, Japan). Subjects were examined in a supine position with their arms stretched above their heads. Nitroglycerin (0.3 mg) was administered to all subjects immediately before CT imaging. A β-blocker was also administered before CT imaging to ensure that the heart rates remained less than 65 beats/min, if required according to local institutional guidelines.

The coronary CT angiography protocol applied in the present study was as follows: slice collimation, 320 × 0.5 mm; gantry rotation time, 0.275 s; and tube voltage, 120 kV. At the time of scanning, a bolus of contrast medium (Omnipaque, 350 mg iodine/mL or iopamidol, 370 mg iodine/mL) was injected intravenously through an antecubital vein via an 18-gauge catheter, followed by 30 mL of normal saline. A site within the ascending aorta was selected as the region of interest (ROI), and the scan was initiated when the CT density reached the target level of 150 Hounsfield units more than the baseline density.

Volumetric measurements were performed on axial views with a 0.5-mm slice thickness. The superior border for measuring the EAT volume was set at the lower surface of the left pulmonary artery origin. The inferior border for the measurement was set at the left ventricular apex [16]. EAT area was calculated by tracing an ROI that included the heart and EAT. The ROI was manually placed outside the line of the visceral pericardium on the cross-sectional axial image. The area outside the traced pericardium was excluded (Fig. 1). EAT volumes were quantified by calculating the total volume of tissue having a CT density between − 150 and − 30 Hounsfield units (Fig. 1). EAT volume measurement by CT were performed by an experienced cardiologist who was blinded to the quantitative analysis data as well as baseline clinical characteristics.

All statistical analyses were performed using SPSS version 22 (IBM Japan, Tokyo, Japan). Continuous data with a non-parametric statistical distribution are presented as mean ± standard deviation, and categorical data as count percentages. Comparisons between the dapagliflozin and conventional treatment groups were performed by Mann–Whitney’s U test for continuous data and Fisher’s exact test for categorical data. Furthermore, the correlations between the change in EAT volume and those in body weight or TNF-α level were evaluated using Spearman rank-order correlation coefficient analysis. A two-sided p-value of < 0.05 was considered statistically significant for all analyses.

The overall mean age of the 40 patients was 67.2 ± 5.4 years. The average HbA1c value was 7.3 ± 0.3%. There were no significant differences in age, renal function, body weight, and medication use between the two groups at baseline (Table 1). Tables 2 and 3 shows the therapies for the patients in both groups.

At baseline, no significant differences were observed in HbA1c, LDL, TG, and HDL levels between the dapagliflozin and conventional treatment groups (7.2 ± 0.6 vs. 7.4 ± 1.1%, p = 0.51; 91 ± 26 vs. 84 ± 20 mg/dL, p = 0.44; 152 ± 92 vs. 151 ± 81 mg/dL, p = 0.83; 46 ± 14 vs. 40 ± 10 mg/dL, p = 0.17, respectively).

Compared with the baseline, at the 6-month follow-up, the decreases in body weight in the dapagliflozin group were significantly greater than in the conventional treatment group (− 2.9 ± 3.4 vs. 0.2 ± 2.4 kg, p = 0.01). However, the change in HbA1c level in the dapagliflozin group was comparable with that in the conventional treatment group (− 0.41 ± 0.21 vs. − 0.19 ± 0.25%, p = 0.22). In addition, the change in the TG/HDL ratio in the dapagliflozin group tended to be greater than that in the conventional treatment group (− 0.8 ± 1.7 vs. 0.2 ± 0.8, p = 0.06).

Regarding insulin resistance, the changes in HOMA-IR and iHOMA2 values in both groups were similar, and both groups showed improved insulin resistance compared with both the baselines.

Overall EAT volume was 112 ± 24 cm³. At baseline, the EAT volumes in the dapagliflozin and conventional treatment groups were 115 ± 22 and 108 ± 25 cm³, respectively, and were comparable with each other. At the 6-month follow-up, the EAT volume in the dapagliflozin group decreased significantly at follow-up compared with the baseline. Furthermore, the change in EAT volume in the dapagliflozin group was significantly greater than that in the conventional treatment group (− 16.4 ± 8.3 vs. 4.7 ± 8.8 cm³, p = 0.01).

In addition, the serum PAI-1 level tended (non-significantly) to decrease at the 6-month follow-up, from 42.2 ± 16.1 to 32.9 ± 14.4 ng/mL (p = 0.07), in the dapagliflozin group. However, the change in the serum PAI-1 level in the dapagliflozin group was similar to that in the conventional treatment group. On the other hand, the TNF-α level decreased significantly from 2.4 ± 0.7 to 1.9 ± 0.5 pg/mL (p = 0.04) in the dapagliflozin group. In addition, the change in the TNF-α level in the dapagliflozin group was significantly greater than that in the conventional treatment group (− 0.5 ± 0.7 vs. 0.03 ± 0.3 pg/ml, p = 0.03).

Furthermore, the changes in the EAT volume and body weight showed significantly positive correlation (r = 0.71, p = 0.01) as did the changes in the EAT volume and TNF-α level (r = 0.51, p=0.04) (Fig. 2).

Obesity interacts with DM closely and is associated with cardiovascular complications [2]. Adipose tissue has a high capacity to secrete many pro-inflammatory substances including PAI-1 and TNF-α [3]. Systemic micro-inflammation causes insulin resistance and is located in the central of the initiation and the propagation of atherosclerosis [17]. Furthermore, excessive adipose tissue increases the infiltration of pro-inflammatory immune cells into metabolic tissues and causes phenotypic shifts in macrophages [18, 19]. In fact, overweight people with enhanced accumulation of visceral fat have increased serum levels of pro-inflammatory cytokines such as TNF-α and PAI-1 [4‐6]. Several reports described that weight loss through diet and exercise therapy in obese patients significantly increased adiponectin and decreased PAI-1 levels [20, 21]. Glucosuria induced by SGLT-2 inhibitors is also typically associated with a net loss of ~ 200–300 kcal, leading to weight reductions of ~ 2–3 kg over 24–52 weeks. Furthermore, previous studies also reported that treatment with an SGLT-2 inhibitor even improved the inflammatory state [10, 22]. In addition, significant weight loss in obese patients has been associated with noteworthy reduction in the EAT volume [23, 24]. Taken together, the SGLT-2 inhibitor treatment in the present study might support an improved inflammatory state via a decrease in fat accumulation presumably due to loss of body weight, because changes in the EAT volume and body weight or inflammatory marker levels (TNF-α) were significantly correlated. In fact, recent studies have reported the effect of dapagliflozin, ipragliflozin, and luseogliflozin treatment on changes in body weight, EAT volume, and inflammatory marker levels [25‐27], the results of which are consistent with those of the present study.

Although adiponectin was not measured in the present study, the decreased PAI-1 and TNF-α levels seen might be associated with the improvement in adipocyte function, finally leading to changes in the HbA1c level and iHOMA2%S.

However, a previous report has also described that an SGLT-2 inhibitor reduced fat accumulation (i.e., fatty liver) without changing body weight [28]. In addition, a previous report described that brown adipose tissue weights were lower in a high-fat-diet plus high-dose empagliflozin group than in a high-fat-diet plus low-dose empagliflozin group [22]. Therefore, further studies are warranted to clarify the mechanism of SGLT-2 inhibitor-mediated improvement in the inflammatory state and the ectopic fat.

Several major trials have reported that treatment with an SGLT-2 inhibitor reduced the occurrence of cardiovascular events [13, 29]. The potential mechanisms through which these events were reduced by an SGLT-2 inhibitor include: improvement in glucose perturbations and insulin sensitivity; reduction in blood pressure and arterial stiffness; reduction in body fat and fat mass; reduction in the levels of uric acid; positive effects on proteinuria and kidney function; and positive effects on lipid parameters [30]. The present study did not evaluate blood pressure, arterial stiffness, or uric acid levels. However, glycemic parameters such as the HbA1c level and iHOMA2%S were improved by treatment with dapagliflozin. Interestingly, the LDL level increased slightly compared with the baseline in patients treated with dapagliflozin. However, a previous report indicated that, although an SGLT-2 inhibitor slightly increased the overall LDL level, it also significantly reduced small dense LDLs [31]. An increase in small dense LDLs was associated with the occurrence of myocardial infarction [32]. Though the present study did not evaluate the small dense LDL level, this level correlates with the TG/HDL ratio [33]. Thus, the decrease in the TG/HDL ratio seen in the present study probably indicates a reduction in the small dense LDL level.

As above-mentioned, the treatment by SGLT-2 inhibitors lead to weight reductions of ~ 2–3 kg over 24–52 weeks. Recently, specific genetic manipulations in adipose tissue and administration of pioglitazone, sitagliptin and metformin, or glucagon-like peptide 1 receptor agonist have also improved EAT volume or the inflammatory state [24, 34‐36]. However, a previous report described that diet (calorie restriction) was the most effective therapy to reduce EAT volume, compared with surgery and medications [37]. Therefore, calorie loss through glucosuria by SGLT-2 inhibitor treatment might contribute to the reduction in EAT volume in the present study. The increase in glucagon by SGLT-2 inhibitors was also associated with a decrease in fat accumulation [38].

Visceral adiposity is associated with an increased risk of type 2 DM, cardiovascular complications, and overall mortality, primarily related to abnormal adipocyte biology. There is also altered production of adipocytokines, leading to modulation of cardiovascular pathways that could promote atherosclerosis [39, 40]. EAT is an atherogenetic property and reflects visceral adiposity. Interestingly, EAT thickness is significantly higher in patients with type 2 DM [41]. In general, EAT-derived bioactive molecules, including inflammatory and oxidative stress mediators, are pathophysiological candidates for atherogenesis [42]. In addition, epicardial and pericardial fat volumes are both strongly correlated with the number of atherosclerotic plaques [8]. In fact, the inflammatory status of EAT in obese patients who were candidates for coronary artery bypass graft was more severe than that of subcutaneous fat located in the legs [43]. Furthermore, EAT volume itself was associated with an increased risk of a cardiovascular event [44]. Taken together, though various mechanisms to decrease cardiovascular events have been reported, the present study indicates that the decrease in EAT volume itself might also directly contribute to plaque stability and finally a reduction in the occurrence of myocardial infarction.