Study design and population
T2DM patients were recruited from the Department of Endocrinology at Næstved, Ringsted, and Slagelse Hospitals in Denmark. After written informed consent, patients were included between January 2016 and August 2019. The study complies with the Declaration of Helsinki and was approved by The Zealand Ethics Committee (SJ-490). T2DM patients aged 18–80 years and diagnosed with T2DM for at least three months prior to inclusion were eligible for inclusion. Exclusion criteria were persistent or permanent atrial fibrillation at the time of echo or CMR, eGFR < 30 mL/min/1.73 m2, claustrophobia or other contraindications to CMR. Sex-matched normal subjects served as a control group, all without T2DM (HbA1c < 40 mmol/mol). Antihypertensive and statin therapy were allowed in the control group.
Patients underwent full medical history, assessment of blood pressure, electrocardiogram (ECG), and echocardiography. Blood and urinary samples were collected prior to the CMR scan, which in general was performed within 14 days of other assessments. History of retinopathy (latest fundoscopy), peripheral neuropathy (latest chiropodist report), ischemic heart disease, hypertension (BP > 140/90 mmHg), and medication were obtained from the patient journal. An ECG obtained during five cycles of in- and expiration was performed to allow for beat-to-beat variation being analysed. Orthostatic blood pressure measurements were obtained ½, 1½, 3, 5, and 7 minutes after acutely standing up(9). Patients with an orthostatic decrease in systolic blood pressure > 25 mmHg in any of the measurements or a beat-to-beat variation ≤ 6 bpm, or both, were diagnosed with impaired autonomic neural function(9, 10). Albuminuria was defined as a urinary albumin/creatinine-ratio of > 30 mg/g.
Echocardiography protocol and data analysis
Standard 2D echocardiography was performed using a Vivid S5 probe on a GE healthcare Vivid E9 cardiovascular ultrasound system. The GE EchoPAC workstation was used for analysis. Peak early mitral inflow (E) was measured in the 4-chamber view using pulsed-wave Doppler with the sample volume placed between the tips of the mitral valve leaflets. Also, in the 4-chamber view, the early diastolic myocardial velocity (e’) was measured using pulsed-wave tissue velocity Doppler with the sample volume placed in the septal and lateral mitral annulus.
Cardiac MRI protocol
CMR studies were performed on a 1.5T scanner (Siemens Avanto, n=247 patients and all healthy controls or GE healthcare Optima 450w, n=19 patients) with patients in a supine position on the back using a surface- and spine cardiac coil with ECG gating. Cardiac short-axis steady-state free precession cine sequences were acquired during end-expiratory breath-holds in 25 phases (covering the whole heart; slice thickness 8mm, no gap) and in 2-, 3- and 4-chamber views. Sequence settings were TE 1.16-1.25ms, TR 46.24-49.98ms, Matrix 210x208, FoV 258x320-485x481. Myocardial perfusion images were obtained at the basal, mid-ventricular and apical cardiac short-axis level during rest and pharmacological stress (adenosine 140 µg/kg/min) using an iv. dose of 0.075 mmol/kg of gadobutrol (Gadovist®, Bayer AG, Germany) for each sequence as previously described(11). LGE imaging was performed using a phase-sensitive segmentation gradient echo inversion recovery sequence following selection of the appropriate TI time based on a Look-Locker sequence. Approximately 5 min after the last contrast dose, a short axis LGE stack of the LV was performed (slice thickness 8mm, no gap, TE 3.16-3.20ms, TR 537-748ms, Matrix 224x256-256x256, FoV 318x340-443x458). This was followed by LGE images in the 2-, 3- and 4-chamber views. A trained physician reviewed LGE images during the scan and any LGE lesions seen in the short axis images were confirmed from perpendicular slices through the lesion in question. Additionally, native and post-contrast T1-mapping were performed as previously described(11) (not performed in the 19 patients scanned with the GE healthcare Optima 450w).
Cardiac MRI analysis
CMR images were analysed using cvi42© (Circle Cardiovascular Imaging Inc., Calgary Canada, v. 5.5.1). LV volumes and mass were determined by semi-automatic tracing of the endo- and epicardial contours in end-diastolic and end-systolic phases. Papillary muscles were excluded. LA maximum volume was measured by manual tracing in the LV end-systolic frame. For all CMR parameters we chose to report both the absolute values and indexed to body surface area because the majority of our patients were severe obese, which could bias the indexed results. A LGE lesion was considered to be present if seen in at least two perpendicular slices. Blinded to each other, two observers analysed all LGE images (AB and MS). Blinded to clinical data, the analyses were reviewed and finalized by a third observer with 10 years of CMR experience (PLvM). A LGE lesion was classified as ischemic if subendocardial or transmural and respecting the perfusion territory of a coronary artery(12), otherwise, the lesion was considered to be non-ischemic. Hinge-point fibrosis was not included, and these patients were categorised as patients without LGE lesion. A semi-automatic toll marking LGE lesions with increased myocardial signal intensity 5SD above remote myocardium was used to quantify lesion mass. Myocardial mass and mass of myocardial LGE lesions were determined from end-diastolic volumes applying a density of 1.05 g mL-1. On native and post-contrast T1-maps endocardial and epicardial borders were traced semi-automatically, and the average ECV was calculated from areas outside LGE lesions(13). For determination of the ECV within a LGE lesion, care was taken to exclude any myocardium without LGE in the segment. Myocardial perfusion scans before and after adenosine infusion were visually inspected for perfusion defects. Using a five-point linear fit model of signal intensity vs. time myocardial perfusion up-slopes were calculated and normalized to the arterial input function. Regions with infarcts, sub-endocardial perfusion defects and dark-rim artefacts were excluded. Myocardial perfusion reserve index (MPRI) was defined as the ratio between normalized stress and rest perfusion(12).
Statistical analysis
For association analysis, patients with both ischemic and non-ischemic LGE lesions were included in the group with ischemic lesions. Quantitative variables were analysed for normal distribution and are presented as mean and standard deviation (SD) or median and interquartile range (IQR) as appropriate. Categorical variables are presented as absolute values and percentages. A one-way analysis of variance (ANOVA) was applied for the comparison of continuous variables with a normal distribution. If an unequal variance was seen either visually or by a significant Levene’s test, Welch’s ANOVA was applied. Non-normal distributed continuous variables were compared using a Kruskal-Wallis test. Categorical variables were compared using a Chi-squared test. If an ANOVA was significant additional analyses were performed with the comparison of DM2 patients without LGE lesions to patients with non-ischemic LGE lesions using either the Student’s T-test, the Mann-Whitney U test, Fisher’s exact test or a Chi-squared test as appropriate. For all tests, a two-tailed P-value <0.05 was considered statistically significant. All calculations were made in SAS Enterprise Guide v. 7.15 (SAS Institute inc., Cary, NC, USA).