Computed tomography coronary angiography was performed in 20 DM⁺CAD⁺ patients, 20 DM⁻CAD⁺ patients, 10 DM⁺CAD⁻ patients, and 20 HS referred to our institution for suspected CAD. All clinical characteristics such as laboratory parameters, presence of cardiovascular risk factors, and medical history were accurately recorded.

Diabetes was defined as treatment with drugs or fasting blood glucose ≥126 mg/dL. Dyslipidemia was defined as treatment with drugs or fasting serum total cholesterol ≥240 mg/dL, or LDL cholesterol ≥140 mg/dL, or high-density lipoprotein cholesterol <40 mg/dL, or triglyceride ≥150 mg/dL. Hypertension was defined as treatment with drugs or systolic blood pressure (SBP) ≥140 mmHg or diastolic blood pressure (DBP)≥ 90 mmHg. Anthropometrical measurements including body weight and height were recorded and body mass index (BMI) was calculated. Blood pressure and resting heart rate were measured after ≥5 min rest with a sphygmomanometer. Physical activity in HS and patients was evaluated according to the WHO guidelines for adults 18–65 aged; specifically the performance of at least 150 min of moderate-intensity aerobical physical activity per week [50–70% of maximum heart rate (MHR)] or at least 75 min of vigorous-intensity aerobic physical activity throughout the week (70–80% MHR) (37). None of the recruited subjects had physical disabilities.

Patients with known history of cancer, cardiomyopathy, active infections, chronic or immune-mediated diseases, renal failure, hepatic failure, and not suitable for cardiac imaging (atrial fibrillation, arrhythmia, or pre-scan heart rate greater than 65 bpm) were excluded from the study to avoid confounding effects due to other variables.

Peripheral venous blood samples were collected after a 12 h overnight fasting immediately before i.v. cannulation for CTCA examination. All tubes were centrifuged within 30 min of collection at 1,900 g for 10 min at 4°C to separate plasma and cellular components. Aliquots of plasma were transferred into cryostat tubes and stored at −80°C until analysis. PBMNCs were isolated by Ficoll gradient using HISTOPAQUE-1077 (Sigma Diagnostics, MO, USA) and frozen at −80°C until total RNA extraction. All biological samples were stored at the IRCCS SDN Biobank (38). The study and the protocol were approved and reviewed by the institutional ethics committee (IRCCS Fondazione SDN, protocol no. 7-13). The study was performed in accordance with the ethical standards of the institutional ethics committee and with the Helsinki Declaration. A written informed consent was obtained from all subjects enrolled.

Computed tomography coronary angiographies were performed on a CT scanner (Discovery CT750 HD, GE Healthcare), with a 64 mm × 0.625 mm collimation, 350 ms rotation time, and 228 ms temporal resolution. A prospective ECG-triggered scan without contrast medium was used for calcium score (Cascore) evaluation followed by a retrospective scan with ECG tube current modulation. Contrast enhancement was obtained by a bolus tracking technique with scan starting when a region of interest placed in the ascending aorta at the pulmonary bifurcation reached a threshold of 150 Hounsfield unit (HU). Contrast material (iomeprol 400 mg I/mL, Iomeron 400, Bracco, Milan, Italy) was injected at 5–6 mL/s through an 18-gauge intravenous antecubital catheter and was followed by saline solution at the same flow. Tube voltage and contrast agent volume were adapted to patient anatomical features such as BMI, calcifications, or stents. Images were reconstructed with a section thickness of 0.625 mm and an increment of 0.4 mm; standard and sharp reconstruction filter kernels were used; an additional sharper convolution kernel was used in patients with stents or calcification. The best data set was chosen according to the phase of the cardiac cycle with lower artifacts and coronary motions. Images were sent to a dedicated offline workstation (GE Advantage workstation 4.6, GE Healthcare) where MIP, cMPR, and 3D volume rendering were generated. Cascore was calculated by the SmartScore tool to obtain the Agatston score. Total plaque volume (TPV), non-calcified plaque volume (NCPV), calcified plaque volume (CPV), and total lumen volume were measured for the major coronaries using the HU cutoff values reported in Ref. (39). The resulting values were summed to determine a per-patient plaque volume. Total vessel volume was determined summing TPV and total lumen volume. Plaque burden was obtained dividing TPV by total vessel volume (40). The remodeling index (RI) was calculated by dividing the cross-sectional vessel area at the site of maximum luminal narrowing including plaque by the cross-sectional vessel area in the most proximal atherosclerotic free segment chosen as reference (41). The total number of coronary artery segments exhibiting plaque (NSP) was determined according to the modified American Heart Association 16-segment classification (42) for each patient (less or more than 8 segments affected). Significant coronary stenosis was defined as a decrease in the luminal diameter of >50% in one or more of the major coronary arteries; the total number of coronaries (NCS) with significant stenosis was calculated for each patient (less or more than one stenotic vessel). All scans were analyzed by two experienced, independent radiologists; therefore, a consensus interpretation was arrived to obtain a final coronary CT diagnosis according to the international SCCT guidelines (43).

Total RNA was isolated from PBMCs using TRIzol Reagent (Thermo Fischer Scientific, USA) as previously described (44). The quantity and quality of RNA were measured using the NanoDrop 1000 (Thermo Fischer Scientific, USA). Total RNA (0.5 µg) was reversed transcribed (RT) using the SuperScript^® III First-Strand Synthesis SuperMix for qRT-PCR (Thermo Fischer Scientific, USA) according to the protocol of the manufacturer. The RT was performed using the Bio-Rad iCycler Thermal Cycler with the following protocol: incubation at 25°C for 10 min (primer annealing), 42°C for 30 min (cDNA synthesis), and 85°C for 5 min (termination of cDNA synthesis). Immediately after, the samples were cooled down and stored at −20°C.

The optimal reference genes for the study were selected as previously reported (45). Gene expression was quantified on the MyiQ™ Single-Color Real-Time PCR Detection System (Bio-Rad Laboratories, USA). Primers pairs were designed through OLIGO 6.7 program, and their specificity was verified with the BLAST program for test of sequence homology, a test for secondary structures and optimization of multiplex setup. All primers were purchased from Life Technologies. All samples were run in triplicate for genes of interest and reference genes using 1 µL of cDNA and iQ™ SYBR^® Green Supermix (Bio-Rad Laboratories, USA) in a 25 µL final volume reaction. The thermal profile employed was 3 min of initial step of denaturation at 95°C followed with denaturation for 15 s at 95°C, annealing at 60°C for 30 s, and elongation at 72°C for 30 s for 40 cycles. Melt curve analysis was performed to verify a single product species. Relative expression (fold change) was calculated by the 2^−ΔΔCT method (46). Mean and SE were determined by averaging relative expression levels across three independent experiments, each determined in triplicate.

Plasmatic levels of HA were determined by enzyme-linked immunosorbent assay (ELISA) using Quantikine Hyaluronan Immunoassay kit (DHYAL0) (R&D Systems, Abingdon, UK), in accordance with the protocol supplied by the manufacturer. Briefly, samples were incubated with HA binding protein coated on microplates for 2 h at room temperature. After incubation, the microplates were washed five times with wash buffer, and further incubated with 100 µL of peroxidase labeled HA binding protein for 2 h at room temperature. After incubation, the microplates were again washed five times, and further incubated with 100 µL of peroxidase substrate for 30 min at room temperature in a dark room. The reaction was stopped by the addition of 100 µL of stop solution. The optical density of each well was determined using a microplate reader set to 450 nm within 30 min. HA concentration in each sample was calculated using the standard curve obtained with the purified HA solutions, included in the kit as references.

Statistical analysis was performed using R Core Team (version 3.03 Austria, Vienna). Continuous variables were expressed as mean ± SD or as median (1 quartile and 3 quartile). Data were tested for normality through the Shapiro–Wilk test and for homoscedasticity through the Levene test. For comparison between two groups, t-test was used if gaussianity was met; otherwise the Mann–Whitney U test was chosen. For comparison among four groups, the one-way analysis of variance was used if both gaussianity and homoscedasticity were met; otherwise the Kruskal–Wallis test was chosen. In case of statistical significance, the Tukey–Kramer test and the Conover test were used for multiple comparisons as parametric and non-parametric test, respectively. Categorical variables were expressed as percentage and were compared using the Fisher’s exact test. The Spearman correlation test was performed to assess linear relationship between variables; in case of binary variables, the association was tested by the Wilcoxon rank sum test. A P < 0.05 was considered for statistical significance (rounded to the third decimal place).

The baseline demographic and clinical characteristics of the study population are summarized in Table 1. Heart rate was significantly different between HS and DM⁺CAD⁺ (P < 0.01) and HS and DM⁻CAD⁺ (P < 0.01) since only 10% of HS was in treatment with beta blocker agents, while no statistical significance was found between both CAD groups and DM⁺CAD⁻. Considering the metabolic markers, glycemia was significantly higher in DM⁺CAD⁺ and DM⁺CAD⁻ patients compared to HS (P < 0.01 and P < 0.001, respectively) and DM⁻CAD⁺ subjects (P < 0.01 and P < 0.001, respectively). Of diabetic patients, in DM⁺CAD⁺ group, 16% were insulin users, 64% were in treatment with antihyperglycemic agents, and 20% were not in treatment; in DM⁺CAD⁻ group, the percentage of treatments were, respectively, 10% for insulin, 80% for antihyperglycemic drugs, and 10% were not treated. Total cholesterol, LDL- and HDL-cholesterol plasmatic concentrations did not significantly differ among the four groups, reflecting the effects of statin therapy to which 73.68% of DM⁺CAD⁺, 50% of DM⁻CAD⁺, 40% of DM⁺CAD⁻, and 5% of HS were subjected. Furthermore, SBP and DBP were not statistical different among the groups (P = 0.50 and P = 0.52, respectively). In this regard, hypertensive patients were in treatment with beta blocker agents (P = 0.008), calcium channel blockers (P = 0.68), and ACE inhibitors (P = 0.06).

There was significant difference between DM⁻CAD⁺ and DM⁺CAD⁺ according to NCS and NSP (P = 0.026, P = 0.04, respectively). Cascore was significantly higher in DM⁺CAD⁺ compared to DM⁻CAD⁺ (Figures 1 and 2): 1,068.7 (517.2–2,086.85) vs 214.05 (72.98–970.15) P = 0.019. As regards plaque characterization, CPV was significantly higher in DM⁺CAD⁺ [105.85 (51.2–341.73) mm³] compared to DM⁻CAD⁺ [42 (7.2–105.9) mm³] P = 0.014, but there was no significant difference according to NCPV and TPV: 519.85 (411.93–1,064.85) mm³ for DM⁺CAD⁺ and 421.85 (240.10–689.58) mm³ for DM⁻CAD⁺ P = 0.37 and 688.95 (470.05–1,436) mm³ for DM⁺CAD⁺ vs 454.45 (257.78–820.83) mm³ for DM⁻CAD⁺ P = 0.16, respectively. RI was 1.40 ± 0.24 for DM⁺CAD⁺ and 1 ± 0.19 for DM⁻CAD⁺ P < 0.001, and plaque burden was 0.45 ± 0.14 for DM⁺CAD⁺ and 0.27 ± 0.15 for DM⁻CAD⁺ P < 0.001. Results are summarized in Table 2.

In our population, RI highly correlated with plaque burden (ρ = 0.65, P < 0.001). Cascore showed a positive correlation with NCPV (ρ = 0.83, P < 0.001), CPV (ρ = 0.96, P < 0.001), TPV (ρ = 0.88, P < 0.001), and plaque burden (ρ = 0.60, P < 0.001). Moreover, a significant correlation was found between plaque burden and NCPV (ρ = 0.57, P < 0.001), CPV (ρ = 0.64, P < 0.001), and TPV (ρ = 0.60, P = 0.001).

We evaluated, by quantitative real-time PCR, SOD2 and LXRα gene expression in PBMCs from our population (Table 3). For both genes, ΔCT was computed and compared between the four groups. Molecular analysis showed that SOD2 mRNA was downregulated in DM⁺CAD⁻ (ΔCT = 5.70 ± 3.28; fold change = 0.10 ± 0.03; P = 0.001), DM⁺CAD⁺ (ΔCT = 4.57 ± 3.56; fold change = 0.22 ± 0.08; P = 0.03), and DM⁻CAD⁺ (ΔCT = 5.75 ± 3.04; fold change = 0.10 ± 0.05; P = 0.001) compared to HS (ΔCT = 2.36 ± 2.61), with no statistically significant difference between the two CAD groups (with and without DM) (Figure 3A). LXRα gene expression was significantly upregulated in DM⁺CAD⁻ (ΔCT = 2.77 ± 1.36; fold change = 4.51 ± 0.20 P = 0.006), DM⁺CAD⁺ (ΔCT = 3.27 ± 1.79; fold change = 3.19 ± 0.42; P = 0.008), and DM⁻CAD⁺ (ΔCT = 2.15 ± 1.16; fold change = 6.93 ± 0.70; P < 0.001) compared to the HS (ΔCT = 4.94 ± 2.14), with a significant difference between the two CAD groups (P = 0.03) (Figure 3B). No statistically significant correlation was found between SOD2 and Cascore (ρ = −0.04, P = 0.81), NCPV (ρ = −0.03, P = 0.85), and TPV (ρ = −0.02, P = 0.89) as well as between LXRα and Cascore (ρ = 0.13, P = 0.44), NCPV (ρ = 0.10, P = 0.55), and TPV (ρ = 0.10, P = 0.57).

In the HS group, mean concentration of plasma HA was 46.90 ± 23.79 ng/mL. Compared with HS, HA concentrations were higher in DM⁻CAD⁺ (90.05 ± 35.11 ng/mL; P = 0.001), DM⁺CAD⁻ (105.56 ± 18.13 ng/mL; P = 0.001), and DM⁺CAD⁺ (120.74 ± 21.17 ng/mL; P = 0.001) (Figure 3C). When compared to DM⁻CAD⁺, HA concentration was significantly higher in DM⁺CAD⁻ (P = 0.008) and DM⁺CAD⁺ (P < 0.001) with a significant difference between two diabetic groups (P = 0.003). Correlation of HA levels with Cascore, NCPV, and TPV revealed ρ = 0.29, P = 0.076, ρ = 0.30, P = 0.073, and ρ = 0.31, P = 0.063, respectively.

Analysis on risk factors and molecular data showed no significant association between sex and SOD2, LXRα, and HA (P = 0.85; P = 0.21; P = 0.75, respectively) as well as regarding familiarity (P = 0.83; P = 0.64; P = 0.55, respectively) and smoke (P = 0.73; P = 0.17; P = 0.49, respectively). Furthermore, no significant correlations were found between BMI and the three molecular parameters (SOD2 ρ = 0.12, P = 0.35; LXRα ρ = −0.11, P = 0.35; HA ρ = 0.19, P = 0.15). HA was significantly correlated with age (ρ = 0.46, P < 0.001), unlike SOD2 (ρ = 0.03, P = 0.97) and LXRα (ρ = 0.03, P = 0.95). HA showed also a significant association with dyslipidemia (P = 0.01) and diabetes (P = 0.01) in the study population, while statistical analysis on hypertension revealed a P = 0.08. On the other hand, SOD2 and LXRα were not significantly associated with the previously mentioned risk factors (SOD2 vs dyslipidemia P = 0.45, SOD2 vs hypertension P = 0.63, SOD2 vs diabetes P = 0.47, LXRα vs dyslipidemia P = 0.85, LXRα vs hypertension P = 0.53, and LXRα vs diabetes P = 0.61). Correlation between HA levels and glycemia was statistically significant (ρ = 0.73, P < 0.001) (Figure 3D), while no significant correlation was found between SOD2 (P = 0.62) and LXRα (P = 0.55) gene expression and glycemia.