A total of 8 domestic female pigs weighing between 20 and 25 kg were used for this study. Type I DM was induced in four pigs by intraperitoneal injection of streptozotocin at a dose of 150 mg/kg.1112 Consecutive measurements of serum glucose were performed at pre-determined time points. Sustaining blood glucose levels > 250 mg/dL were considered as having diabetes. If blood glucose exceeds 400 mg/dL, 4 to 8 units of long-acting insulin (Lantus^®; Sanofi-Aventis, Paris, France) were injected to avoid acute complications such as diabetic ketoacidosis or hyperosmolar coma. The pigs underwent coronary angiography (CAG) with optical coherence tomography (OCT) at 1 and 2 months and CMR imaging and coronary computed tomography angiography (CCTA) at 2 months. After invasive and non-invasive imaging the pigs were sacrificed for histologic analyses (Fig. 1A).

The animals were given aspirin (300 mg) and clopidogrel (300 mg) 1 day before the procedure. The pigs received premedication with intramuscular injection of atropine sulfate (0.5 mg/kg), and then Zoletil (4–6 mg/kg) and Xylazine (4.4 mg/kg). After endotracheal intubation, the anesthesia was maintained with 70% N₂O and isoflurane (1%–2%) in oxygen (2 L/min) by an experienced veterinarian.

A 6-French vascular sheath was introduced into the femoral artery using the Seldinger technique under sonographic guidance. Under fluoroscopic guidance, 6-French Judkin right 3.5 guiding catheters were positioned in the left coronary ostium. After the intracoronary administration of nitroglycerin (200 μg), CAG was performed. Intracoronary administration of 5,000 units of unfractionated heparin was given to prevent thrombus formation.

OCT images were acquired using the OCT system consisting of an optical fiber, a proximal low pressure occlusion balloon catheter (Helios™ Goodman; Advantec Vascular Corp™, Sunnyvale, CA, USA), and an OCT system mobile cart containing the optical imaging engine and computer for signal acquisition, analysis, and image reconstruction (M2CV OCT Imaging System; LightLab, Westford, MA, USA).

The left anterior descending artery (LAD) and left circumflex artery were wired with a 0.014-inch Runthrough guide wire (Terumo, Tokyo, Japan) and then replaced by an OCT ImageWire™ and the occlusion balloon. The occlusion balloon was pulled-back and positioned in a proximal-segment of the arteries. Inflation was performed gently up to 0.4 to 0.7 atmosphere with a dedicated inflation device. The vessel segments were imaged using an automated pull-back (1.0 mm/sec) from the distal to proximal segments. Images were acquired at 15 frames per second and displayed in a real-time 2D array at different transverse positions.

All pigs underwent CCTA with a 64-slice multi-detector computed tomography (CT) scanner (Brilliance 64; Philips Medical Systems, Best, The Netherlands) in the prone-position. Prior to CCTA, all animals with a baseline heart rate > 70 beats/min received 10 to 30 mg of intravenous esmolol (Jeil Pharm, Seoul, Korea). CCTA was performed with a 64 × 0.625 mm section collimation, 420-ms rotation time, 120-kV tube voltage, and 800-mA tube current. During CCTA acquisition, a bolus of 80 mL iomeprol (Iomeron 400; Bracco, Milan, Italy) was injected intravenously (4 mL/sec). The animals' electrocardiogram (ECG) was simultaneously recorded to allow for retrospective segmental data reconstruction. Images were initially reconstructed at the mid-diastolic phase (75% of R-R interval) of the cardiac cycle. Compartment analysis was performed by using the cardiac CT viewer tool of the workstation (IntelliSpace Portal ISP, version 8.0; Philips Medical Systems). Center lines were rotated to the sagittal plane to acquire suitable short-axis pig heart image. Hounsfield units (HU) of region of interest were measured using the dedicated software in the 16 segments (American Heart Association [AHA] model, true apex not imaged) of the heart, ascending and descending aorta, and left ventricular (LV) cavity in 2 or 3 phases of the heart cycle. We calculated the ratio of HU of the myocardial segments to the averaged HU of the aorta and LV cavity.

All CMR studies were performed using a 1.5 T magnetic resonance (MR) imaging scanner (Siemens Avanto, Erlangen, Germany or Intera CV release 10; Philips Medical Systems) with a phased array cardiac coil. Cine images were acquired retrospectively with ECG-gated steady-state free precession (SSFP) sequences (image matrix 128 × 128, read field of view [FOV] 340 mm, phase FOV 75%–100% of the FOV read, echo time 1.12 ms, flip angle 70°) with a slice thickness of 8 mm and intersection gaps of 12–15 mm.

We performed pharmacological stress and rest perfusion studies. After the acquisition of cine scans at rest in the standard views, adenosine was intravenously infused at a dose of 140 μg/kg/min for up to 6 minutes via a 20-gauge cannula using a syringe pump (Graseby® 3500; Smiths Medical, Minneapolis, MN, USA). Within the last minute of infusion, a bolus of gadodiamide (Omniscan; GE healthcare, Chicago, IL, USA) was injected (dose, 0.1 mmol/kg; injection rate, 4 mL/sec) and stress perfusion MR images were obtained with a gradient-echo sequence by using saturation-recovery SSFP. After 15 to 25 minutes, an identical MR perfusion scan was performed at rest.

Data were analyzed using a dedicated software (QMass; Medis, Leiden, the Netherlands). First-pass perfusion images at stress and rest states were analyzed semiquantitatively. The AHA 16-segment model was used (true apex not imaged), resulting in a total of 32 segments per pig (16 for the stress examination and 16 for the resting examination). Signal intensity (SI)-time curves were generated for all segments. The relative upslope at rest or at stress (RRU or SRU) was calculated by the maximum upslope of the LV myocardium divided by the maximum upslope of the LV cavity at stress and rest states. The myocardial perfusion reserve index (MPRI) was calculated as the ratio of the segmental upslope values during adenosine stress and rest (MPRI = SRU/RRU).

After euthanasia of the pigs, the coronary arteries were perfused with phosphate-buffered saline (PBS) and 2% paraformaldehyde (PFA) in PBS at room temperature. The isolated tissues were embedded in an optimal cutting temperature compound (Sakura Finetek, Torrance, CA, USA) and stored in a −80°C deep freezer. Frozen sagittal sections were cut into 25 to 30 μm slices using a cryostat (HM 550 MP; Microm, Walldorf, Germany). The sections were fixed with 4% PFA in PBS for 10 minutes and permeabilized with 10% fetal bovine serum and 0.5% Triton-X100 in PBS for 4 to 6 hours. We incubated the slices with tetramethyl rhodamine isothiocyanate-conjugated Bandeiraea simplicifolia 1-lectin (Sigma Aldrich, St. Louis, MO, USA) and fluorescein isothiocyanate-conjugated alpha smooth muscle actin (Sigma Aldrich) overnight at 4°C. We also stained the slices overnight at 4°C with specific primary antibody, anti-VE-cadherin (Santa Cruz Biotechnology, Santa Cruz, CA, USA), followed by incubation with fluorescent-tagged secondary antibodies (Invitrogen, Carlsbad, CA, USA) overnight at 4°C. The images were acquired by a confocal laser scanning microscope system (LSM 710; Carl Zeiss AG, Oberkochen, Germany) and processed with Zen 2008 (Carl Zeiss AG) and ImageJ (NIH, Bethesda, MD, USA) software. A water- or oil-immersion objective (40× or 63×, 1.4 numerical aperture) with the pinhole set for a section thickness of 0.8 to 1.2 μm (1 airy unit in each channel) was used. To visualize the three-dimensional reconstructed image, a Z-stack of 20-μm thickness was obtained. Diode 405 nm, Multi-Argon 488 nm, HeNe 543 nm, and HeNe 633 nm laser lines were selected and images were sequentially acquired using separate laser excitations to avoid cross-talk between different fluorophores.

For comparative analysis we created images under same conditions of light, contrast, and magnification. Total capillary length and average diameter of the capillaries in images of tissues from mid-ventricular six segments of each pig were measured with ImageJ software (NIH). Therefore, 24 fields for each group were selected for image analysis. For evaluation of the capillary diameter at various points of interest in z-stack images, we measured the width of the lectin-positive area that was perpendicular to the capillary length. As consequence, the capillary diameter in this study corresponds to the abluminal diameter.

We retrospectively enrolled patients with ST-segment elevation myocardial infarction (STEMI) who underwent CMR 5 to 7 days and 6 months after the onset of myocardial infarction (MI) from January 2005 to November 2012. We then compared the myocardial perfusion in non-infarct segments between patients with and without DM.

Continuous variables are expressed as mean ± standard deviation or mean ± standard error and categorical variables are presented as absolute numbers and proportions (%). Overall comparisons between groups were performed with the Student's t-test or the Wilcox rank sum test for continuous variables and the χ² or Fisher exact test for categorical variables. In the MR perfusion experiments, we measured RRU, SRU, and MPRI values for 16 segments according to the AHA 16-segments model in all animals and tested whether the measured values differed between the groups. Because the measured values may also be influenced by the segments, the segment was treated as a factor as well. As consequence, we included both segments and groups as factors in the analyses and applied the two-way analysis of variance (ANOVA) to evaluate ‘the effect of the groups’ after adjusting ‘the effect of the segments.’ To check the normality of the data, we analyzed the residuals of each models using the following two methods. First, we confirmed that SRU and MPRI satisfy the normal distribution with the Shapiro tests. Second, we also visually confirmed the normality using histogram with normal density and QQ-plot. Both methods confirmed the normal distribution of the data. The appropriateness of two-way ANOVA models was tested with the Levene test and the Bartlett test.

Analyses were performed with SPSS version 22.0 (IBM Corp., Armonk, NY, USA) and R programming version 3.1.0 (The R Foundation for Statistical Computing, Vienna, Austria). A two-sided P value < 0.05 was considered statistically significant. The statistical analyses were performed by a professional statistician (SHK).

The present study protocol was reviewed and approved by the Institutional Review Board (IRB) of Seoul National University Bundang Hospital (IRB No. B-1703-388-102). Informed consent was waived because of the retrospective design and inability to obtain by the IRB.

The current study protocol was approved by the Institutional Animal Care and Use Committee (IACUC) of Seoul National University Bundang Hospital (IACUC No. BA1111-093/071-01). All animals were treated according to the study protocol and the principles of care of laboratory animals.

There was no difference in pigs' weights between the groups on days 0 and 56 (Fig. 1B). Both groups showed similar weight gain during the experiment. Blood glucose level in the diabetic pigs was significantly higher than that in the control pigs during the experiment (Fig. 1C). For the evaluation of microvasculature, we first evaluated the capillary anatomy. The two components of interest were the capillary lengths and diameter (Fig. 1D and E). There was no difference in the capillary lengths (129.1 ± 0.44 μm vs. 129.4 ± 0.48 μm; P = 0.365) between diabetic and control pigs; however, diabetic pigs had a significantly shorter capillary diameter (11.7 ± 0.33 μm vs. 13.5 ± 0.53 μm; P < 0.001) as we previously reported (Fig. 1F and G). We also found that VE-cadherin expression in the diabetic capillary was decreased and discontinued compared to that in the control (Fig. 1H and I, Supplementary Videos 1 and 2). Overall, the capillaries in the diabetic pigs were more irregular and disconnected; the capillary diameter was highly variable with frequent acellular capillaries, so called string vessels.

We retrospectively enrolled 79 patients with STEMI who underwent CMR. Baseline characteristics, MI-culprit lesions, baseline LV function, and infarct size did not differ between the patients with and without diabetes (Table 1). Patients with diabetes reported the duration of DM as median 3 and interquartile range 1–10 years. They had a higher serum glucose 250 ± 113 mg/dL vs. 146 ± 34 mg/dL and hemoglobin A1c (HbA1c) level (7.61% ± 0.28% vs. 5.95% ± 0.11%; P < 0.001) at the time of MI. Patients with diabetes had lower RRU and SRU than those without diabetes, whereas MPRI showed no difference (Fig. 5), as we had observed in the pig model.

During follow-up, despite a significant decrease in HbA1c, patients with diabetes had higher HbA1c than those without diabetes (6.78% ± 0.18% vs. 5.92% ± 0.09%; P < 0.001), indicating adequate diabetes control. The follow-up CMR revealed that LV end-systolic volume in patients with DM was significantly larger than that in non-DM patients (55.8 ± 14.9 mL vs. 61.5 ± 14 mL; P = 0.047). LV end-diastolic volume (142 ± 48 mL vs. 130 ± 29 mL, P = 0.267) was numerically larger and ejection fraction (55% ± 14% vs. 61% ± 14%, P = 0.180) was numerically lower in DM patients than in non-DM. Kaplan-Meier analysis showed that the composite of all-cause death, re-MI, and revascularization did not differ between the groups during a median follow-up of 3 years (Fig. 6).