Study Population
This observational study consisted of participants selected from the National Heart Centre Singapore Biobank, who had at least one cardiometabolic risk factors defined as: hypertension, type 2 diabetes mellitus (T2DM), hyperlipidemia, fatty liver, increased BMI and abdominal obesity. Individuals with inherited cardiomyopathies (hypertrophic, dilated and infiltrative cardiomyopathies) were excluded from the study. Selected participants had available fat and cardiac MRI data.
Variables evaluated included anthropometric measurements, body fat mass by bioimpedance analysis, MRI-quantified adipose tissue (abdominal VAT and SAT areas, liver fat fraction and EAT volume) and cardiac metrics (mass, volumes, myocardial fibrosis markers, wall stress) by cardiovascular magnetic resonance (CMR).
Ethics approval was granted by the SingHealth Biobank Research Scientific Advisory Committee (SBRSA 2019/001). The study was performed in accordance with ethical principles that have their origin in the Declaration of Helsinki. All study participants provided written informed consent.
Anthropometric Indices and Bioimpedance Body Fat Analysis
Anthropometric measurements acquired with standard methods included BMI, waist circumference (WC) and waist-hip ratio (WHR). BMI was calculated as weight (kg)/height (m)2. WC was measured at just above the navel. Hip circumference was taken at the widest portion of the hip area. The ratio between waist and hip circumferences was calculated as WHR. All measurements were taken in standing position. Bioimpedance analysis (InBody, Cerritos, California, USA) was used to measure body fat mass (BFM), calculated as the difference between total body mass and fat free mass that was made up of water, protein and mineral.
Local BMI thresholds guided by the recommendations made by the WHO Expert Consultation Panel were used to define normal (<23 kg/m2), overweight (23-30 kg/m2) and obese (>30 kg/m2) (5). Abdominal obesity was defined as WHR >0.90 and >0.85 or WC >90cm and >80cm for males and females, respectively (6).
Abdominal and Cardiac MRI Acquisition
MRI was performed in all participants using the 1.5T Siemens Aera (Siemens Healthineers, Erlangen, Germany). Abdominal MRI examination was performed according to the LiverMultiScan protocol (Perspectum Ltd, Oxford, London) (7,8). A series of contiguous cross-sectional abdominal water- and fat-separated images were obtained from the two-point Dixon method.
For CMR, balanced steady-state free precession cine images were acquired in the long-axis 2-, 3-, and 4-chamber views (acquired voxel size 1.6×1.3×8.0 mm; slice gap 2mm; 30 phases per cardiac cycle).
Short-axis cines were acquired extending from the mitral valve annulus to the apex (acquired voxel size 1.6×1.3×8.0 mm3; slice gap 2mm; 30 phases per cardiac cycle). Late gadolinium enhancement imaging was performed 8 minutes after 0.1 mmol/kg of gadobutrol administration (Gadovist; Bayer Pharma AG, Germany). A breath-held inversion-recovery fast gradient echo sequence was used, and the inversion time was optimized to achieve appropriate nulling of the myocardium. The native and 15-minute postcontrast myocardial T1 maps were acquired with modified Look-Locker inversion-recovery sequence, applying a heartbeat acquisition scheme of 5(3)3 and 4(1)3(1)2, respectively.
MRI-based Fat and Cardiac Analysis
De-identified abdominal and cardiac images were analyzed at Perspectum and the National Heart Research Institute (NHRIS) Core Laboratory, respectively, by trained individuals who were blinded to clinical data.
Cross-sectional areas of VAT and SAT were segmented from the abdominal Dixon MRI image at the L3 vertebral level using ITK-SNAP software version 3.8 (PICLS, University of Pennsylvania, USA) (Figure 1A). This single-slice approach of quantifying VAT and SAT has been shown to correlate strongly with total SAT and VAT volumes (9, 10). The proportion of VAT relative to SAT was calculated as the VAT/SAT ratio.
Liver fat was quantified as proton density fat fraction (PDFF), expressed as a percentage, computed as fat/(fat+water) based on MRI-visible fat and water signals in the regions of interest placed on the PDFF parametric map, avoiding image artifacts and vessels (11, 12) (Figure 1B). Fatty liver was defined as PDFF >5.6% (13, 14).
EAT volume on the left and right ventricles was quantified at end systole on short axis cines extending from the mitral valve annulus to the apex using CVI42 (Circle Cardiovascular Imaging, Calgary, Canada). The bright layer between the myo-epicardial border and the pericardium constituted the EAT (Figure 1C).
LV mass and cardiac volumes were analyzed according to standardized protocols (15). LV concentricity was defined as the ratio of LV mass over end-diastolic volume (EDV) (16). The Remodeling Index (RI) is a surrogate marker of global myocardial wall stress, calculated as: RI= , where EDV is the LV end-diastolic volume (mL) and t is the maximal wall thickness (cm) across the 16 myocardial segments (17). A lower RI denotes increased global myocardial wall stress and predicts worse cardiovascular outcomes in individuals with hypertension (18).
Extracellular volume (ECV) was quantified from native and post-contrast T1 maps. Interstitial volume (mL) was calculated as extracellular volume (ECV) fraction × myocardial volume (mL), where myocardial volume was calculated by dividing the myocardial mass by the specific gravity of the myocardium (1.05 g/mL). Myocyte volume (mL) = myocardial volume – interstitial volume (19).
Statistical Analysis
Normality of data distribution was assessed with Shapiro-Wilk test. Continuous variables were presented as means ± SD if normally distributed or median (interquartile range) if otherwise. Categorical variables were expressed as frequency (percentage) and were analyzed using the χ2 test.
Depending on the continuous or categorical nature of the dependent variable, multivariable linear or logistic regression analyses with adjustment for potential confounders were performed to evaluate the associations between: (i) anthropometric indices and adipose tissues; (ii) abdominal, epicardial and liver adipose tissues; and (iii) adipose tissues and cardiac remodeling markers. Clinically important potential confounders included age, sex, BMI, systolic blood pressure (SBP), hyperlipidemia and T2DM status that were adjusted for where applicable.
Mean differences in adipose tissues between categorical groups (WHR categories, BMI categories, T2DM status) were adjusted for confounders and compared by one-way analysis of covariance (ANCOVA). Post hoc Bonferroni was performed for pairwise comparison between BMI categories (normal, overweight and obese). Adjusted mean difference with 95% confidence interval (CI) were reported. We compared the ability of VAT, SAT and VAT/SAT ratio to differentiate individuals with and without T2DM using the area under the receiver operating characteristic curve (AUC). Depending on the variables of interest in each analysis, only cases with complete data were included in analysis.
Statistical significance was defined as P<0.05. Statistical analyses were performed using IBM SPSS Statistics Version 26 (IBM Corp, Armonk, NY, USA) and GraphPad Prism Version 7.05 (GraphPad Software, Inc, La Jolla, CA, USA).