Experimental model
The study was approved by the local Ethics Committee for animal experiments. Pigs (weighing 40–50 kg) were pre-medicated with ketamine 15 mg/kg (Ketaminol, Intervet, Danderyd, Sweden) and midazolam 0.5 mg/kg intramuscularly (Dormicum, Roche AB, Stockholm, Sweden) after overnight fasting with free access to water. Anesthesia was induced with propofol 20 mg/ml (Propofol Sandoz A S, Copenhagen, Denmark) and the animals were intubated using cuffed endotracheal tubes. Anesthesia was maintained with inhalation of sevoflurane gas (Sevorane, Baxter Medical AB, Kista, Sweden) using a disposable administration system (AnaConDa, Sedana Medical AB, Uppsala, Sweden) and titrating to desired effect. Mechanical ventilation was established using a 900C ventilator (Siemens AB, Upplands Väsby, Sweden) in volume controlled mode regulated to a pCO2 of 5.0–6.0 kPa. Monitoring included arterial blood pressure, heart rate, ECG, pulse-oximetry and temperature. Arterial blood gases were drawn and analyzed directly after establishing arterial access, before occlusion, after reperfusion, and to follow up any unexpected deviation of the blood gas results or change in the clinical condition of the animal. Venous and arterial femoral access, jugular access and carotid access were established using 6–8Fr introducer sheaths. After establishing all accesses, 20,000 IU of heparin (LEO Pharma AB, Malmö Sweden) was administered intravenously and 5% glucose (Baxter Medical AB, Kista, Sweden) was slowly infused. During the experiments amiodarone (Sanofi AB, Stockholm, Sweden), fentanyl (50 mikrog/ml, B. Braun Medical AB, Danderyd, Sweden) and sodium chloride (0.9% Baxter Medical AB, Kista, Sweden) was titrated to desired effect.
Experimental protocol
Pigs were subjected to either 35 or 40-min of left anterior descending (LAD)-occlusion using a balloon-tipped catheter, followed by six hours of reperfusion. The balloon occluder was placed either after the first or the second diagonal branch of the LAD to obtain a wide range of MaR. An angiogram was acquired after inflation of the balloon and before deflation in order to confirm occlusion of the coronary vessel and correct balloon placement. Isotope (1000 MBq 99mTc tetrofosmine) for MPS imaging was administered ten minutes prior to reperfusion. After deflation of the balloon a subsequent angiogram was performed to verify restoration of blood flow. After reperfusion, animals were transported to the MR department.
CMR
In vivo
Imaging was performed on a 1.5 T MR scanner (Philips Achieva, Best, Netherlands) using a 32-channel cardiac coil. During in-vivo imaging 0.2 mmol/kg gadolinium (Gd) based contrast agent (Gd-DOTA) (Dotarem, Guerbet, Roissy, France) was administrated intravenously 15 min before LGE-CMR. Using a retrospectively gated SSFP single slice sequence, a midventricular short-axis slice of the left ventricle (LV) was repeatedly acquired during the first ten minutes after contrast injection. Then, a short-axis stack covering the entire LV was acquired, using the same sequence parameters as for the single slice (typical parameters: TE 1.40 ms, TR 2.8 ms, flip angle 60°, 25 phases, slice thickness 8 mm, no slice gap, acquisition matrix 230 x 230, reconstructed matrix 240 x 240, FOV 320 x 320 mm, pixel bandwidth 860 Hz/pixel). LGE images were acquired 15–20 min after contrast injection using an inversion recovery gradient echo sequence (typical parameters: TE 3.0, TR 6.1 ms, TI 320 ms, slice thickness 8 mm, no slice gap, acquisition matrix 200 x 158, reconstructed matrix 510 x 510, FOV 320 x 320 mm, pixel bandwidth 260 Hz/pixel).
After CE-SSFP and LGE imaging an additional single slice of SSFP in the same position as above was acquired 20–30 min after contrast injection. Both CE-SSFP and LGE slices were obtained during end-expiratory breath hold. Long axis 2-, 3 and 4-chamber views using both CE-SSFP and LGE were also acquired. Fifteen minutes before euthanization an additional 0.2 mmol/kg Gd-DOTA was administrated. Pigs were euthanized with a rapid infusion of saturated potassium chloride solution and the hearts were explanted and suspended in plastic containers with deuterated water-filled balloons in the ventricles for ex-vivo imaging.
Ex vivo
ex-vivo CMR was performed on the same scanner as above, using a simulated ECG with heart rate 60 beats per minute for triggering. A full coverage LV short-axis stack was acquired using the same settings as for the in-vivo sequence above (typical parameters: TE 1.40 ms, TR 2.8 ms, flip angle 60°, 25 phases, slice thickness 8 mm, no slice gap, acquisition matrix 60 x 50, reconstructed matrix 80 x 80, FOV 100 × 100 mm, pixel bandwidth 1400 Hz/pixel). A high resolution T1-weighted short axis stack was acquired for detailed infarct visualization (TE 3.4 ms, TR 20 ms, 0.5 mm isotropic voxels, no slice gap, acquisition matrix 200 × 200, reconstructed matrix 220 × 220, FOV 100 × 100 mm, pixel bandwidth 440 Hz/pixel).
MPS
ex-vivo MPS was performed approximately 8–10 h after intravenous injection of a 1000 MBq dose of 99mTc-tetrofosmin using a dual head camera (Philips SKYlight, Best, the Netherlands) and a vertex high resolution collimator (ADAC Vertex, Milpitas, CA, USA) at 32 projections (40 s per projection) with a 64 × 64 matrix yielding a digital resolution of 4.24 mm isotropic voxels. Iterative reconstruction using maximum likelihood expectation maximization (MLEM) was performed with a low resolution Butterworth filter with a cut-off frequency set to 0.6 of Nyquist and order 5.0. No attenuation or scatter correction was applied. Finally, a short-axis image stack was reconstructed using commercially available software (AutoSPECT Plus, Pegasys software version 5.01, Philips, Best, The Netherlands).
Image analysis
All images were analyzed using the software Segment, version 1.9 R3314 (
http://segment.heiberg.se) [
13]. MaR from the in-vivo CE-SSFP images was assessed according to a previously described method [
9]. In short, LVM was defined by manual delineation of the epicardial and endocardial borders in end-diastolic and end-systolic timeframes. Hyperintense myocardium was then manually delineated and defined as MaR. The same method was used to delineate MaR in ex-vivo CE-SSFP images. The analysis of CE-SSFP images was performed both by a blinded primary observer and an independent, blinded secondary observer. When delineating CE-SSFP images over time after injection of gadolinium all images were sorted randomly both between timepoints and between experiments and observers were blinded to the randomization.
Infarcted myocardium was delineated from the in-vivo short-axis LGE images according to a previously defined method [
14]. In short, the endocardial and epicardial borders were manually traced with exclusion of the papillary muscles. The LGE myocardium was defined using a computer algorithm that takes into consideration partial volume effects within the infarcted region [
14]. Manual adjustments were made when image artefacts caused misinterpretation by the computer algorithm. Hypointense signal within the area of LGE, microvascular obstruction [
15], was included and considered as infarction.
The high resolution T1-weighted ex-vivo images were also delineated according to a previously described and validated method [
16]. In short, the endocardial and epicardial borders were traced manually including the papillary muscles. Infarct was defined as >8 standard deviations from a manually defined remote region. Manual adjustments were made if the computer algorithm was clearly wrong due to image artefacts or inclusion of intramyocardial fat as infarct.
MaR and infarct size was expressed as percentage of LVM. Myocardial salvage index (MSI) was calculated as (MaR-infarct size)/MaR = MSI.
Evaluation of MPS images was performed by using anatomical information from the high resolution T1-weighted ex-vivo images and perfusion information from MPS according to a previously described method [
17]. In short, both T1-weighted and MPS images were re-sampled to similar resolutions and were spatially matched using purpose-designed software. Delineations of endocardium and epicardium from the T1-weighted images were then used for defining the myocardium in the MPS images. MaR was then automatically defined by calculating and applying a threshold and performing manual adjustments as previously described [
17].
Statistics
Analyses were performed using GraphPad Prism (version 6.00, GraphPad Software, San Diego, USA) or IBM SPSS Statistics (version 23, IBM Corporation, New York, USA). Results for continuous variables are expressed as mean ± standard deviation. Bias according to Bland-Altman was used to compare MaR by MPS, in-vivo CE-SSFP and ex-vivo CE-SSFP. Bias according to Bland Altman was also used to compare MaR by MPS, in-vivo CE-SSFP and ex-vivo CE-SSFP to infarct size by LGE imaging [
18], and for inter-observer analysis. A paired
t-test was used to test ex-vivo CE-SSFP vs MPS and in-vivo CE-SSFP vs ex-vivo CE-SSFP. Pearson correlation coefficient was used for assessment of correlation between ex-vivo CE-SSFP, MPS, in-vivo CE-SSFP, infarct, and for inter-observer analysis. When comparing MaR over time for the single-slice images, the MaR was normalized to MaR of the first acquired slice. Subsequently a repeated measures linear mixed model using time as a nominal variable and a fixed effect was used to test for differences in MaR between different time points after contrast injection.
P < 0.05 indicated statistical significance.