Pooled study sample
The present cross-sectional study is based on data from two independent population-based investigations, the Study of Health in Pomerania (SHIP-TREND-0) [8, 9] and the Cooperative Health Research in the Region of Augsburg (KORA FF4) [10]. Our pooled sample, from SHIP-TREND-0 and KORA FF4, comprised 1391 individuals (604 women, 43.4%) aged 21 to 81 years. Individuals with inadequate MRI image quality (n = 79), previous myocardial infarction or stroke (n = 16), left ventricular ejection fraction (determined by MRI) less than 40% (n = 9), fasting time less than 8 h (n = 217), use of hypoglycemic medication (n = 37), missing values for OGTT parameters (n = 17) or any of the covariates (n = 9) as well as individuals with extreme values (> 99.5th percentile for fasting glucose, insulin or 2-h postload glucose; n = 6) were excluded. Accordingly, our final analytical sample consisted of 1001 individuals (453 women, 45.3%), aged 21 to 80 years.
All study participants gave written informed consent. The study was approved by the ethics committees of the University of Greifswald, the Bavarian Chamber of Physicians, and the Ludwig-Maximilians-Universität München and complies with the Declaration of Helsinki.
Glucose and insulin measurements, oral glucose tolerance test and classification of prediabetes and unknown type 2 diabetes
Measurements of fasting glucose (FG) and 2-h postload glucose (2HG) were based on plasma in SHIP-TREND-0 and on serum in KORA FF4. Duplicate measurements were carried out using serum samples from all SHIP-TREND-0 participants and serum glucose from KORA FF4 and plasma glucose from SHIP-TREND-0 were considered as comparable for the current analysis (concordance correlation coefficient of r = 0.94 in a validation study comparing plasma and serum glucose measurements).
In both studies, FG was sampled and 75 g of anhydrous glucose (Dextro OGT; Boehringer Mannheim, Mannheim, Germany) was given to those participants without diagnosed type 2 diabetes or taking glucose-lowering agents. In SHIP-TREND-0, plasma FG and 2HG levels were measured using a hexokinase method (Dimension Vista 1500, Siemens Healthcare Diagnostics, Eschborn, Germany) [3] and serum fasting insulin (FI) and 2-h postload glucose insulin (2HI) values were assessed by an electrochemiluminescence immunoassay (ADVIA Centaur, Siemens Healthcare Diagnostics, Eschborn, Germany) [11]. In KORA FF4, serum FG and 2HG levels were measured using an enzymatic colorimetric method (Dimension Vista 1500, Siemens Healthcare Diagnostics, Eschborn, Germany or Cobas c702, Roche Diagnostics GmbH, Mannheim, Germany) and FI and 2HI values were measured by a solid-phase enzyme-labeled chemiluminescent immunometric assay (Immulite 2000 Xpi, Siemens Healthcare Diagnostics, Eschborn, Germany) or by an electrochemiluminescence immunoassay (Cobas e 602, Roche Diagnostics GmbH, Mannheim, Germany).
The homeostasis model assessment-insulin resistance index (HOMA-IR) was calculated as (FG [mmol/l] × FI [μU/ml])/22.5 [12].
Following the criteria of the American Diabetes Association (ADA) [13], we classified individuals as having normal glucose tolerance (NGT) when they had FG values < 5.6 mmol/l (< 100 mg/dl) and 2HG < 7.8 mmol/l (< 140 mg/dl). Unknown type 2 diabetes (UT2D) was defined as FG values ≥ 7.0 mmol/l (≥ 126 mg/dl) or 2HG ≥ 11.1 mmol/l (≥ 200 mg/dl). Participants were classified as having prediabetes if FG values were between 5.6 and 6.9 mmol/l (100–125 mg/dl, impaired fasting glucose: IFG) and/or 2HG values were between 7.8 and 11.0 mmol/l (140–199 mg/dl, impaired glucose tolerance: IGT) [3, 13]. We defined three groups of prediabetes: isolated impaired fasting glucose (i-IFG), isolated impaired glucose tolerance (i-IGT), and combined IFG and IGT (IFG + IGT) [3, 13].
Cardiac MR imaging
In SHIP-TREND-0, cardiac MR imaging was performed on a 1.5 Tesla MR system (Magnetom Avanto; Siemens Medical Systems, Erlangen, Germany) [11, 14] and in KORA FF4, on a 3 Tesla MR system (Magnetom Skyra; Siemens Medical Systems, Erlangen, Germany) [15, 16]. In both studies imaging of cardiac function and morphology was performed using cine steady-state free precession (cine-SSFP) sequences.
Image analysis
LV end-diastolic volume (LVEDV) was determined during the first image of the acquisition. LV end-systolic volume (LVESV) was measured by determining the phase in which the LV intra-cavity blood pool was at its smallest by visual assessment at the midventricular level. LV myocardial mass (LVM) was calculated at the end-diastole using the specific density of the myocardium (1.05 g/cm3) [14]. Papillary muscles were included in the LVM and excluded of the LV end-diastolic and systolic volumes. Basal slices were included if at least half of the LV circumference blood pool was confined by myocardium [17]. Inclusion or exclusion of apical slices depended on the visibility of myocardium. LV wall-thickness (LVWT) was determined in the 16-segment model (according to the AHA-segment model) [18]. LV concentricity (LVC) was calculated as LVM/LVEDV. LV stroke volume (LVSV), LV cardiac output (LVCO) and LV ejection fraction (LVEF) were calculated following the formulas below:
$${\text{LVSV }}\left( {\text{ml}} \right) \, = {\text{ LVEDV }}{-}{\text{ LVESV}}$$
$${\text{LVCO }}\left( {{\text{l}}/{ \hbox{min} }} \right) \, = {\text{ LVSV }} \times {\text{ heart rate}}$$
$${\text{LVEF }}\left( \% \right) \, = \, \left( {{\text{LVEDV }} - {\text{ LVESV}}} \right) \, /{\text{ LVEDV}}$$
LVM, LVEDV, LVESV, LVWT, LVSV and LVCO were indexed for body height in meters, normalized to the allometric power of 2.7, which linearizes the relations between the cardiac anatomic and functional parameters with height and identifies the impact of obesity [19]. This resulted in LVM index (LVMI), LVEDV index (LVEDVI), LVESV index (LVESVI), LVWT index (LVWTI), LVSV index (LVSI) and LVCO index (LVCI).
Arterial stiffness index (ASI) was calculated as (systolic blood pressure − diastolic blood pressure)/LVSI [20].
Interview, medical and laboratory examinations
In both studies, information on socio-economic variables (including years of school education [< 10, 10, or > 10 years]), smoking status (never, former or current smoker) [21], alcohol consumption (in grams per day) and medical history was collected by trained and certificated medical staff during a standardized interview. Sedentary lifestyle was defined as individuals who did not participate in leisure time exercise, for at least 1 h/week, during summer or winter [22]. Participants were asked to bring the original packaging of their medications that were taken during the last 7 days before the examination date. Unique identifiers and drug names were recorded according to the ATC classification system.
All participants underwent an extensive standardized medical examination. Anthropometric measurements included height and weight based on recommendations of the World Health Organization (WHO) [23]. Weight was measured to the nearest 0.1 kg in light clothing and without shoes using standard digital scales. Body mass index (BMI) was calculated as weight (kg)/height2 (m2). Waist circumference (WC) was measured to the nearest 0.1 cm using an inelastic tape midway between the lower rib margin and the iliac crest in the horizontal plane, with the participant standing comfortably with weight distributed evenly on both feet [24]. While in SHIP-TREND-0 body fat-free mass (FFM) and fat mass (FM) were measured by bioelectrical impedance analysis (BIA) using a multifrequency Nutriguard-M device (Data Input, Pöcking, Germany) and the NUTRI4 software (Data Input, Pöcking, Germany) [25,26,27], in KORA FF4, BIA scans were obtained by BIA 2000-S device (Data Input, Pöcking, Germany) with an operating frequency of 50 kHz at 0.8 mA. Ohmic resistance was measured at the dominant hand (between wrist and dorsum) and the dominant foot (between angle and dorsum).
After a resting period of at least 5 min, systolic and diastolic blood pressures as well as heart rate were measured three times on the right arm of seated subjects using an oscillometric digital blood pressure monitor (HEM-705CP, Omron Corporation, Tokyo, Japan) with an interval of 3 min between readings. The mean of the second and third measurements was used for the present analyses. Antihypertensive medication was defined as use of agents with the ATC-code C02, C03, C07, C08 and C09 [28]. Hypertension was defined as systolic blood pressure ≥ 140 mmHg and/or diastolic blood pressure ≥ 90 mmHg and/or current self-reported use of any anti-hypertensive medications.
Fasting blood samples were obtained from all study participants while sitting [29]. In SHIP-TREND-0, glycated hemoglobin was determined by high-performance liquid chromatography (Diamat, Bio-Rad Laboratories, Munich, Germany). Total serum cholesterol, low-density lipoprotein cholesterol (LDL-C) and high-density lipoprotein cholesterol (HDL-C) were measured photometrically (Dimension RxL or Dimension VISTA 1500, Siemens Healthcare Diagnostics, Eschborn, Germany). Serum creatinine concentration was assessed using a modified kinetic Jaffé method (Dimension RxL or Dimension Vista 1500, Siemens Healthcare Diagnostics, Eschborn, Germany). In KORA FF4, glycated hemoglobin was measured in hemolyzed whole blood using the cation-exchange high performance liquid chromatographic, photometric VARIANT II TURBO HbA1c Kit-2.0 assay on a VARIANT II TURBO Hemoglobin Testing System (Bio-Rad Laboratories Inc., Hercules, USA). Total serum cholesterol, low-density lipoprotein cholesterol (LDL-C), high-density lipoprotein cholesterol (HDL-C) and serum creatinine concentrations were measured using an enzymatic colorimetric method (Dimension Vista 1500, Siemens Healthcare Diagnostics, Eschborn, Germany or Cobas c702, Roche Diagnostics GmbH, Mannheim, Germany). Because of the changes from Siemens to Roche, the Siemens measurement results were calibrated to the Roche measurements using the following formulas (in mg/dl): Total_Cholesterol_Roche = 3.00 + (Total_Cholesterol_Siemens * 1.00); HDL_Cholesterol_Roche = 2.40 + (HDL_Cholesterol_Siemens * 1.12); LDL_Cholesterol_Roche = antilog (− 0.13328 + [log LDL_Cholesterol_Siemens * 1.03051]); Creatine_Roche = − 0.037568 + (Creatinine_Siemens * 1.02703) [16].
Hypercholesterolemia was defined as use of lipid-lowering medication defined by the ATC-code C10 and/or total serum cholesterol ≥ 6.2 mmol/l and/or LDL-C ≥ 4.1 mmol/l and/or total cholesterol/HDL-C ratio ≥ 5.0. The estimated glomerular filtration rate was estimated according to the CKD-EPI formula [30] and expressed in ml/min/1.73 m2.
Statistical analysis
To characterize the study population, data was reported as median (with 25th and 75th percentiles) for continuous variables and as percentages for categorical variables stratified by OGTT classification.
We used linear regression models to associate FG, FI, HOMA-IR, 2HG and 2HI levels and OGTT groups with LVMI, LVEDVI, LVESVI, LVWTI, LVC, ASI, LVSI, HR, LVCI and LVEF. The basic multivariable models were adjusted for age, sex, body fat-free mass and body fat mass (both assessed by BIA), systolic blood pressure, use of antihypertensive medication, smoking status, alcohol consumption, sedentarism (defined as individuals who did not participate in leisure time exercise for at least 1 h/week during summer or winter [22]), estimated glomerular filtration rate, fasting time and study sample (SHIP-TREND-0, KORA FF4). We used fractional polynomials to test potential non-linear relationships between exposure and outcomes [31].
A two-sided p-value p < 0.05 was considered as statistically significant. Statistical analyses were performed using Stata 14.2 (Stata Corporation, College Station, TX, USA).
Please see Additional file 1 for a more detailed description.