Epicardial adipose excision slows the progression of porcine coronary atherosclerosis

We produced CAD in 9 castrated male Ossabaw miniature swine by feeding them 1000 g, once a day, of a high fat/cholesterol/fructose, atherogenic diet for 6 months starting at 6 months of age. The diet provided 16.3% kcal from protein, 40.8% kcal from complex carbohydrates, 19% kcal from fructose, and 42.9% kcal from fat. Fat calories derived from a mixture of lard, hydrogenated soybean oil, and hydrogenated coconut oil. It was supplemented with 2.0% cholesterol and 0.7% sodium cholate by weight (KT324, Purina Test Diet, Richmond, IN). Age- and gender-matched lean control swine were fed 750 g of regular chow (5 L80, Purina Test Diet, Richmond, IN) containing 18% kcal from protein, 71% kcal from complex carbohydrates, and 11% kcal from fat once a day. All animals had free access to drinking water. After 6 months on the diet, CT was performed to quantify EAT surrounding the heart. Surgical excision of cEAT covering the middle segment of the left anterior descending (LAD) artery, which we termed selective coronary adipectomy, was then conducted via a thoracotomy, as described below, followed immediately by intravascular ultrasound (IVUS) quantification of CAD. Following the procedure, the same baseline diet was continued during a 3-month recovery period at the end of which identical CT followed by IVUS were performed prior to sacrifice and tissue collection. Control pigs did not undergo IVUS or adipectomy because they have been shown previously to develop minimal or no CAD over the time course of this experiment [12]. In both adipectomy and control pigs at sacrifice, cEAT for mRNA analysis was resected from the adventitial surface of the LAD just proximal to the middle LAD adipectomy site, and as a comparator, paracardial fat was sampled from the mediastinal surface of the pericardium towards the apex of the heart. Samples were stored at −80°C. Biopsies of proximal, middle and distal LAD were taken and preserved for immunohistochemistry. All protocols involving animals were approved by an Institutional Animal Care and Use Committee and complied fully with recommendations in the Guide for the Care and Use of Laboratory Animals[14] and the American Veterinary Medical Association Panel on Euthanasia [15].

Following an overnight fast, swine received a 2.2 mg/kg of xylazine (Webster Veterinary, Devens, MA) and 5.5 mg/kg of telazol (Fort Dodge Animal Health, Fort Dodge, IA) intramuscular injection to induce anesthesia. Swine were intubated and anesthesia was maintained with 2-4% isoflurane in 100% O₂ as a carrier gas. The isoflurane level was adjusted to maintain anesthesia with stable hemodynamics. Heart rate, aortic blood pressure, respiratory rate, and electrocardiographic data were continuously monitored throughout the procedure. A 7 F introducer sheath was inserted into the right femoral artery and heparin (200 U/kg) was administered under sterile conditions. A 7 F guiding catheter (Amplatz L, sizes 0.75-2.0; Cordis, Bridgewater, NJ) was advanced to engage the left main coronary ostium. A 3.2 F, 30 MHz IVUS catheter (Boston Scientific, Natick, MA) was advanced over a 0.014 inch diameter percutaneous transluminal coronary angioplasty guide wire (Boston Scientific, Natick, MA) and positioned in the LAD. Automated IVUS pullbacks were performed at 0.5 mm/sec in the LAD. Throughout the procedure angiography was performed to place the guiding and IVUS catheters in their desired positions. The IVUS catheter, guide catheter, and introducer sheath were removed and the right femoral artery ligated. The incision was closed and the animal was allowed to recover. Cefazolin (1000 mg; WG Critical Care, Lake Forest, IL) was given twice a day for six days following the procedure.

All IVUS pullback images were analyzed off-line (Windows Media Player, Microsoft) at 1 mm intervals. The LAD was divided into 3 segments: proximal (0–15 mm), middle (15–30 mm) and distal (30–45 mm) with 0 mm starting at the bifurcation of the left main into LAD and circumflex. Three IVUS pullback images were captured per vessel segment. Using the Image-Pro Plus software (Media Cybernetics, Rockville, MD), lumen area (LA) was found by tracing the vessel lumen. From LA, lumen circumference (LC) was calculated for each vessel segment by the following formula: LC = 2π [√ (LA/π)]. To derive wall coverage (WC) by plaque, the circumference of each cross-sectional image was divided into 16 equal segments. Percent wall coverage was calculated as (# segments containing atheroma ÷ 16) × 100. From the product of average WC × LC, a circumferential plaque length for each vessel segment was found. As an example, the following results are presented from a proximal coronary artery segment:

LA 21.40 mm²; LC 16.40 mm; WC 29.58%; circumferential plaque length = 4.85 mm. Pullbacks were used for 7 out of 9 animals due to restrictions from vasospasms or lumen diameter. Using a 2-way analysis of variance (ANOVA), average circumferential plaque length per vessel segment was compared pre- and post-adipectomy.

Under general anesthesia, cardiovascular monitoring (both as described above), and using aseptic surgical techniques, all adipectomies were performed on the LAD with access via left thoracotomy by the cardiac surgeon. Figure 1A shows the anatomical structures and distribution of EAT over the epicardium of a representative pig heart ex vivo. A Thoratrak Retractor (Medtronic; Minneapolis, MN) was utilized for minimal rib spreading and as a base for the Octopus cardiac stabilizer (Medtronic; Minneapolis, MN) (Figure 1B). The pericardium was opened and cradled to the drapes exposing the LAD. The Octopus was then positioned to stabilize the middle LAD segment. Using the bifurcation of the left main into LAD and circumflex as the reference point, cEAT was excised sharply starting 15 mm distal to this landmark for a distance of 15 mm along the length of the LAD and 3–5 mm on both the right and left lateral aspects of this part of the vessel (Figure 1C). Non-absorbable Prolene sutures were placed to control epicardial venous bleeding from resected fat typically to the left side of the LAD. The ligatures demarcated this site for later tissue collection at sacrifice. When possible, the pericardium was loosely approximated with interrupted sutures. The chest was then closed in layers in the standard fashion. Pleural air was evacuated by means of suction on an intrapleural red rubber catheter being withdrawn during a valsalva breath and closure of the chest wall musculature. One of the 9 animals succumbed intra-operatively from an acute myocardial infarction complicated by refractory ventricular fibrillation documented by ECG and by subsequent autopsy that showed acute LAD thrombosis. There were no noticed post-operative complications in the remaining swine.

Sections of the LAD were prepared using techniques previously described by us [16]. Samples of LAD were dissected and immersed in neutral-buffered 10% formalin for a minimum of 24 h and then processed to paraffin embedment. The proximal, middle, and distal samples of LAD were obtained consistently from the same location in all pigs. Five-micrometer sections were cut with an automated microtome (Microm, Thermo Fischer Scientific, Bellefonte, PA), floated onto positively charged slides (Thermo Fischer Scientific), and deparaffinized. The slides were then steamed in citrate buffer at pH 6.0 (Dako target retrieval solution S1699; Dako, Carpenteria, CA) for 30 min to achieve antigen retrieval and subsequently cooled for 30 min. Next, the slides were stained manually with sequential Tris buffer or water wash steps performed after each protocol step. Sections were incubated with avidin biotin two step blocking solution (Vector SP-2001; Vector Laboratories, Burlingame, CA) to inhibit background staining and in 3% hydrogen peroxide to inhibit endogenous peroxidase. Nonserum protein block (Dako X909; Dako) was applied to inhibit nonspecific protein binding. Immunostaining was performed using scavenger receptor A (SRA-ES; 1:100 dilution, Transgenic Inc. #KT-022), tryptase (1:600 dilution, Dako #M7052), CD-3 (1:600 dilution, Dako #A0452), monocyte chemoattractant protein-1 (MCP-1; 1:100 dilution, PeproTech #500-P76), resistin (1:800 dilution, Phoenix Pharmaceuticals #H-02840), T-cadherin (1:1000 dilution, Abgent #CDH13, AP14346), adiponectin (1:200, Chemicon #MAB 3604) or tumor necrosis factor-alpha (TNF-alpha; 1:100 dilution, R&D #AF690). All the primary antibodies were diluted using antibody diluent (S0809; Dako) and were incubated with the tissue sections overnight at 4-C. After the appropriate washing was completed, sections were incubated with biotinylated antimouse or rabbit link secondary antibody and then peroxidase-labeled streptavidin (LSAB + kit, peroxidase, K0690; Dako). Diaminobenzidine (K3468; Dako) was applied for 5 min and allowed visualization of primary antibody staining. Sections were counterstained with Mayer hematoxylin for 1 min, dehydrated, and coverslipped. For negative controls, histological sections were prepared as described, but incubation in primary antibody was replaced with Tris buffer.

Immunostaining analysis was performed in the entire vessel wall (i.e. intima to external elastic lamina boundaries) in order to quantify the presence of SRA, tryptase, CD3, MCP-1, resistin, T-cadherin, adiponectin or TNF-alpha. Sections were examined using an Olympus BX61 photomicroscope (Olympus, Melville, NY), and pictures were captured at 20x magnification. Image-Pro Plus (version 6.2.0.424; Media Cybernetics, Inc., Silver Spring, MD) was used to identify and quantify the positive area of staining. Percent staining of the vessel wall was calculated from the data for positive staining and total area. An experienced investigator that was blinded to treatment group’s identities performed the selection of target areas, photography, and image analysis.

CT scans to quantify changes in EAT were analyzed by one observer. EAT was defined as fat between the inner surface of the visceral pericardium and epicardial surface of the myocardium [18, 19]. Paracardial fat was defined as fat lying within the mediastinum superficial to the parietal pericardium [18, 19]. Representative images are shown in Figures 2A, B and C. The pigs were scanned using a Philips Brilliance 64 detector scanner (Cleveland, OH) while sedated using 4% Isoflurane (Webster Veterinary, Devens, MA) supplied by mask. Imaging was performed with retrospective electrocardiogram (ECG) gating and without the administration of intravenous contrast from above the aortic arch to below the heart. Additional scan parameters included a 64 × 0.625 collimation, pitch of 0.2, rotation time of 0.4 seconds, 120 kVp, and 800–1000 mAs. Images were reconstructed with a slice thickness of 0.67 mm and increment of 0.67 mm using a CB filter. A batch was first created to ensure equal amounts of data were analyzed and included for each image set. The baseline and pre-sacrifice images were then aligned side-by-side to ensure similar magnification and orientation. Transverse images were analyzed by starting at the superior aspect of the atrial appendage. Sternal, vertebral body and cardinal landmarks were utilized to insure similar alignment. The slice thickness of each image produced was 3 mm without overlap, yielding about 40 images per heart in each batch. Regions of interest (ROI) were drawn by hand using quantitative computer tomography angiography (Q-CTA) software on a Philips Extended Brilliance Workspace (Version 3.5) to include the entire heart. Vessels leaving and entering the heart were included until they were noticeably detached from the heart. The Hounsfield unit (HU) value is similar for blood, muscle, and soft tissue, thus making the vessels appear as part of the heart until a visible distance is evident between the heart and the vessel. In all cases vessels were split in the ROI through their center as the pericardium encapsulates the most proximal regions of the vessels leaving and entering the heart. The software highlighted all tissue that matched the HU intensity of a reference ROI drawn in known subcutaneous fat (at the level of the carina). The fat area measurements of the images were then multiplied by their slice thickness to determine fat volume. Volumes of EAT were compared before adipectomy and at the end of the recovery period. Inter-operator coefficient of variation was 6.2%, while intra-operator variation was 1.0%.

Statistical analysis was performed via Microsoft Excel 2010, utilizing ANOVA, regression analysis and post hoc 1- and 2-tailed paired student t-tests to examine significance. All values are given as mean ± SE. A p value <0.05 was considered statistically significant.

At adipectomy, pigs on the atherogenic diet had developed CAD and its risk factors including hypertension, obesity, increased LDL/HDL ratio, insulin resistance, glucose intolerance and increased triglyceride levels compared to age-matched lean controls (Table 2). Other than the increase in body weight, these characteristics remained constant during the 3 month recovery following adipectomy (data not shown).

Representative IVUS images in Figure 3 show a normal coronary lumen (A) as well as partial (B) and complete circumferential (C) neointimal plaque formation. The results of IVUS pullbacks revealing atherosclerosis in each 15 mm length proximal, middle and distal portions of the LAD from its origin are shown in Figure 3D. Over 3 months, baseline mean circumferential plaque length in the proximal LAD increased from 4.45 ± 0.54 to 8.24 ± 1.03 mm (p < 0.01, Two-way ANOVA and Bonferroni post-hoc) and in the distal LAD from 1.98 ± 0.35 to 5.74 ± 0.75 mm (p < 0.01, Two-way ANOVA and Bonferroni post-hoc). By contrast, circumferential plaque length in the middle LAD increased from 5.60 ± 0.90 to 7.64 ± 1.03 mm which was not statistically different (p > 0.05, Two-way ANOVA and Bonferroni post-hoc). There was no difference between baseline mean proximal and middle LAD plaque lengths (p = 0.26), whereas middle LAD was higher than distal LAD (p = 0.01) and proximal was also greater than distal (p = 0.04).

On light microscopy of the LAD middle segment sections, a thinner layer of cEAT covered the adventitia than the adjacent two segments, suggesting no regrowth of remnant fat 3 months after adipectomy. Figure 4A shows representative photomicrographs of the presence, distribution and intensity of immunostaining of T-cadherin, SRA and adiponectin in the three segments of the LAD. T-Cadherin was strongly positive in the endothelial layer, but also present in the sub-endothelial neointima in proximal and distal LAD, but to a lesser extent in the middle vessel wall. SRA was present in the endothelial layer and neointima. Adiponectin was present in the endothelial layer and very dispersed over the media. As shown in Figure 4B (panel A), when these markers in all samples were quantitated, there was a significant reduction in mean percentage area of T-cadherin staining in the intima-media from 4.77 ± 0.92% in the proximal and 4.71 ± 0.66% in the distal LAD segments to 2.67 ± 0.66% in the middle LAD (proximal to middle LAD p = 0.047 and distal to middle LAD p = 0.01 using 2-tailed t tests). SRA staining (Figure 4B, panel B) was 0.73 ± 0.25% in the proximal LAD and 0.26 ± 0.08% in the middle LAD which was significantly different (p = 0.035) by a 1-tail t test with a trend towards significance (p = 0.07 in a 2-tailed t test), and 0.74 ± 0.45% in the distal LAD which although higher than the middle LAD was not significantly different. Adiponectin staining (Figure 4B, panel C) was significantly lower in the middle LAD 3.28 ± 1.04% versus the proximal 8.94 ± 2.32% (p = 0.04 in a 1-tailed t test versus a trend for significance p = 0.08 in a 2-tailed t test). In the distal LAD, adiponectin measured 5.12 ± 1.57% which was not different to either the proximal or middle LAD. There were no differences in coronary wall per cent staining for tryptase, CD3, TNFα, MCP-1 and resistin between the proximal, middle and distal LAD.

In Figure 5, CT scans done on 6 of the 8 surviving swine showed an increase in total EAT volume from 9.7 ± 1.1 cm³ at adipectomy to 13.5 ± 0.7 cm³ at time of sacrifice, representing a 39% increase (p = 0.006) over 3 months. Due to inadequate numbers of data points, CT data could not be correlated with the IVUS data to determine if a relationship existed between changes in EAT volume and total plaque burden.

Table 1 shows the expression in cEAT and paracardial fat of genes linked to atherogenesis, namely those involved in inflammation, reactive oxygen species and redox reactions, angiogenesis and endothelial cell function. The data in each fat depot are expressed as the ratio of experimental to control pig fat depots. In EAT, there were significant and substantial fold increases in mRNA expression of pro-inflammatory 11β-hydroxysteroid dehydrogenase (11βHSD-1), CCL19, CCL21 and PGD2S; anti-inflammatory adiponectin; pro-oxidant gp91phox [NADPH oxidase], angiogenic VEGF and VEGFGR1 and vascoconstrictive angiotensinogen. There were no changes in pro-inflammatory toll-like receptor 4 [TLR4], CCR7, TNFα, IL-1β, IL-6, IL-8, leptin, Unc5b, netrin-1 and NGF; redox genes eNOS, superoxide dismutase [SOD], cyclooxygenase 2 and cytochrome oxidase; and vasoconstrictor endothelin-1 and vasodilator adenosine receptor 1. By contrast, in paracardial fat relative to cEAT, 11βHSD-1 and CCL21 were normal; CCL19, PGD2S, adiponectin pro-oxidant gp91phox [NADPH oxidase] VEGF, VEGFGR1 and angiotensinogen were up-regulated to much lesser degrees; and the remaining genes were similar except for TLR4 and endothelin-1 which were modestly (~2 fold) increased.