This observational cross-sectional autopsy study was approved by the local ethics committee. Written informed consent is obtained from the next-of-kin (NOK) of all participants before starting any study procedure.

The Sao Paulo Autopsy Service (SPAS) is a general autopsy service based at the University of Sao Paulo that performs approximately 13,000 autopsies per year. In Sao Paulo city, autopsy is mandatory when an individual dies from a natural cause, and where the cause of death is unclear or the individual did not have medical assistance. Since 2004, the Brazilian Aging Brain Study Group has sourced autopsy material from SPAS to investigate the normal and pathological aging brain [18]. The present study employed the same approach to collect adipose tissue samples and hearts.

While the NOK waits for the autopsy of their deceased family member, they are invited to participate in this study. Nurses explain the aim of the study to the NOK and ask for their signature on an informed consent form. For all subjects to be included in this study, the NOK needs to have at least weekly interactions with the deceased and can therefore provide reliable information. A semistructured interview relating to sociodemographic and clinical information is conducted with each NOK prior to the collection of tissue samples (Figure 1).

The inclusion criteria for this study are: individuals of age ≥30 years at the time of death; a post mortem interval of ≤24 hours; a NOK with at least weekly contact with the deceased in the 6 months prior to death; and the availability of heart, SAT, and PrAT tissues from the medical pathologist. We included participants aged 30 years or older because advanced atherosclerotic plaques could be found in individuals with at least 30-years old at death [19].

The exclusion criteria employed in this study include: inconsistent clinical data provided by the NOK; adipose tissue fixation >72 hours; the use of corticosteroids or immunosuppressants, the presence of autoimmune disease, pericarditis, pericardial effusion, myocarditis, endocarditis, Chagas’ disease, hemopericardium, cardiac tamponade, coronary artery stent, or previous cardiac surgery. Patients with these conditions are excluded because they could induce or suppress inflammation in an atheroma plaque-independent manner. During the autopsy, samples of the major organs (heart, lung, spleen, kidney, and liver) are collected, processed, and stained with hematoxylin and eosin. Individuals that have an infection and show 2 or more criteria associated with sepsis in these samples, as defined previously [20,21], are also excluded from this study.

The cause of death is determined by certified pathologists, based on the autopsy. In addition to the semistructured interview relating to the deceased’s sociodemographic information (including age, race, and education in years completed) [19,22]. The NOK provide clinical information relating to diagnosis (hypertension, diabetes, CAD, heart failure, dyslipidemia, or stroke) and lifestyle (physical activity, alcohol use, and smoking). Body mass index is calculated using the weight and height measured in a supine position, while the deceased have no clothes or shoes before the autopsy exam.

The heart is collected to obtain PAT and coronary arteries. SAT is removed from the region of the umbilical scar and PrAT from the kidney. The heart, SAT, and PrAT is washed in running water to remove blood and clots.

Two processes were compared for heart fixation. The first involved fixation of the heart (with EAT) by immersion in 70% alcohol for at least 24 hours. After this period, we dissected the right coronary artery, left coronary artery trunk, anterior descending coronary artery, and circumflex coronary artery, together with 1 cm of the adjacent PAT and all EAT was removed from myocardium. PAT was removed systematically and cut into 1.5 cm sections, starting from the ostium (Figure 2). After these dissections, PAT and EAT were weighted using an electronic scale. Subsequently, agar was injected into the lumen of each coronary artery, which was then stored in 10% formaldehyde solution for 5 days. However, we observed flattening of the coronary arteries (Figures 3 D, 3 E), even with agar injection.

Therefore, we changed our fixation protocol. The second fixation process involved the coronary artery dissection with the adjacent PAT immediately after heart collection to test whether the dissection of fresh coronary arteries and PAT would be more effective. The heart was incubated at −20°C for 2 minutes prior to injecting agar inside each coronary artery, and then the heart was placed at −20°C for another 2 minutes to solidify the agar. The coronary arteries were then dissected as described above and PAT was removed. The second fixation protocol has two advantages. First, the injection of agar before artery dissection allows agar infiltration until minor branches. Second, when we dissect the coronary arteries, there are already solidified agar inside the arterial lumen, facilitating arterial opening after repeated injections of agar (Figures 2 B, 2 C). The problem of injecting agar only after heart fixation is that coronary arteries are already hardened by alcohol. Moreover, when we injected agar after coronary dissection, most of the agar did not stay inside the lumen, preventing arterial opening (Figures 3 D, 3 E).

We also investigated whether the 10% formaldehyde solution and 4% buffered paraformaldehyde solution (pH 7.2‑7.4) influenced coronary artery fixation. Coronary arteries samples were identified and then fixed for 5 days. We used the same fragment of coronary artery divided in 3 to compare agar injection procedures and fixatives. Agar prevented artery flattening (Figures 3 B, 3 C), which could interfere with the measurement of arterial stenosis. Moreover, the presence of the agar protected the intima layer from 10% formaldehyde solution damage (Figure 3 A, 3 B). In addition, the 4% buffered paraformaldehyde solution was less aggressive to the tissue than 10% unbuffered formaldehyde solution (Figure 3 C). Therefore, we chose to dissect fresh coronary arteries and PAT and used 4% buffered paraformaldehyde solution as the fixative.

We compared adipose tissue fixation times of 24, 48, and 72 hours in 4% buffered paraformaldehyde solution. Fixation times >72 hours were not tested because they can lead to a loss of antigenicity or the production of formic acid, which can impair identification of immune cells [23,24]. All primary antibodies were tested in positive control tissue sections (lung, tonsil, and lymph node) to facilitate standardization (Table 1). No difference in inflammatory cell staining was found using the 3 different fixation times (Figure 4). Thus, although 24 hours was chosen as the fixation time for subsequent analyses, this could be extended for up to 72 hours if necessary.

After we complete the tissue processing, participants will be classified into 3 groups: chronic CAD, with ≥50% obstruction in at least one artery [7], and an AHA classification different than VI [25]; acute CAD, with at least one atheroma plaque with an AHA classification of VI (unstable plaque) [19]; with no meaningful CAD [7], and an AHA classification different than VI [19]. If the sample size allows, we will perform a subanalysis with participants without arterial obstruction as the control group.

PAT, SAT, and PrAT are cut into 4 µm sections, applied to silanized slides (3-aminopropyltriethoxysilane), and immersed in paraffin. Prior to immunochemistry, the sections of PAT that correspond with sampled coronary artery fragments, SAT, and PrAT are deparaffinized by placing the slides in hot xylene in an oven at 60°C to 65°C for 5 minutes and then dipping in 3 baths of cold xylene. The sections are then hydrated in 95% alcohol, followed by 70% alcohol, washed in tap water and deionized water, and placed in phosphate buffer, pH 7.4. Antigen recovery is performed in 10 mM citric acid, pH 6, at a high temperature in a pressure cooker. Endogenous peroxidase is blocked using 3% hydrogen peroxide prior to incubating the slides with the indicated primary antibodies in the presence of 1% bovine serum albumin for 24 hours at 4°C. The slides are then incubated with the appropriate horseradish peroxidase-conjugated secondary antibody and EasyLink One and the signal is generated using the chromogen, diaminobenzidine. The sections are counterstained using Harris hematoxylin.

The slides are scanned and analyzed at 40× magnification using the Pannoramic Viewer software. This program hides the slide identification so that the operator is blinded to the subject’s diagnosis, the source and type of adipose tissue, and hotspots (ie, accumulation of inflammatory cells). Twenty random fields with 600 µm of diameter were also analyzed with no magnification (cells can only be observed with precision at 20× magnification). This process is systematic and large cell agglomerates (identified in >20× magnification) are avoided. Inflammatory cells are counted per the primary antibody staining and these results were expressed as the number of cells per micrometers squared.

To calculate the sample size, we used a previous study that found an effect size of 0.93, with a mean standard deviation of 44 (SD 21) inflammatory cells/µm² in a CAD group and 24 (SD 22) inflammatory cells/µm² in a control group [7]. Assuming an alpha of 5% and a power of 90%, we estimated that 26 subjects would be needed in each group, giving a total sample size of 78. The independent variable of our study is CAD and the dependent variables are the numbers of macrophages, polarized macrophages, and lymphocytes (B and T) in PAT, PrAT, and SAT. The groups will be compared regarding demographic and clinical variables, using chi-square test for categorical variables and one-way analysis of the variance (ANOVA) for continuous ones. The weight of EAT will be compared among groups using one-way ANOVA. A multivariate linear regression model will be used to compare the dependent and independent variables, adjusted for age, hypertension, diabetes mellitus, body mass index, alcohol use, and smoking. The significance level for all tests will be set at 5% in two-tailed tests. We will use STATA 13.0 to perform these analyses.

Although EAT thickness and volume can be evaluated using imaging methods [26], the number and type of inflammatory cells in PAT can only be determined by pathological examination. In addition, autopsy studies can employ morphometric methods to calculate the degree of plaque stenosis and plaque composition. Finally, they can exclude other inflammatory diseases, which could bias the results. Here, we describe a protocol for sample processing, immunochemical analyses, and morphometric measurements of coronary artery stenosis and inflammatory cells in adipose tissues.

Although the association of CAD with macrophages and T lymphocytes has been investigated previously [6,8,9,15], a study of this association using a range of controls, including the analysis of SAT, PrAT, and EAT distal from the atheroma plaque will help to determine the extent of the inflammatory process. Moreover, the contribution of B lymphocytes in PAT to the atherosclerotic process has not yet been investigated, and the association between EAT inflammation and acute CAD has only been investigated using imaging methods, thus precluding the direct quantification of inflammatory cells [6].

It is possible that inflammation in PAT could contribute locally to the development of the atherosclerotic plaque, as suggested by previous imaging [7,6] and autopsy studies [15]. The mechanism underlying these findings is not yet established, but it is biologically plausible. Infiltration of adipocytes in the PAT was found in the adventitial layer, which could have direct influence on the inflammation in coronary arteries [12,27]. In addition, the vasa vasorum, which is in close contact with the PAT [10,26], grows in the direction of the intimal layer [28] when intima media thickness is present. The infiltration of inflammatory cells in adventitial layer may contribute to this angiogenesis.

PAT inflammation may be a measurable and modifiable risk factor that could be used in clinical practice. Some studies have investigated interventions that aim to reduce the inflammatory burden, for example by reducing the EAT volume via a reduction of the total body weight or by promoting the conversion of white adipose tissue into brown adipose tissue, which is associated with a decreased risk of obesity-related disorders. Other studies have investigated drugs that modulate immune receptors to reduce inflammation [29,30].

However, while inflammation and PAT thickness have been suggested to show positive associations with the degree of arterial stenosis [31], a paradox has been observed in clinical practice, whereby a low EAT volume was associated with a reduced myocardial salvage area and a larger infarct size in patients with a first ST-segment elevation myocardial infarction [32]. Therefore, studies of the roles of different inflammatory cells in PAT are important to elucidate CAD pathophysiology and identify new therapeutic targets.

Nevertheless, one limitation of the present study is that the sociodemographic and clinical information are collected after death from the NOK. To improve the accuracy of this information, we only include NOK with daily or weekly interactions with the deceased. Moreover, the reliability of the post mortem interview had been demonstrated by a previous study from our group, which showed a high sensitivity (87%) and specificity (94%) using this approach [33]. In addition, the main variables analyzed in the present study (atherosclerotic burden and inflammatory cell numbers) are measured objectively.

Here, we described the protocol that we are using to investigate the association between CAD and inflammation in adipose tissues. Particularly, we described in details the pilot studies that we performed to fixate and process the arteries and adipose tissues. These measures are important to allow for unbiased morphometric measures of atherosclerosis and inflammatory cell counting.