Intra-thoracic fat volume is associated with myocardial infarction in patients with metabolic syndrome

Ninety-four patients with metabolic syndrome (MetS) and/or myocardial infarction (MI) undergoing Late Gadolinium Enhancement (LGE) were identified from a clinical CMR registry at the Cardiovascular MR Clinical Research (CMCR) program between March 2009 and March 2011. MetS was defined in accordance with published guidelines [9] as any three of the following five criteria: body mass index (BMI) ≥30 kg/m², serum triglycerides ≥1.7 mmol/L, high density lipoprotein C (HDL-C) ≤1.0 mmol/L for men or ≤1.3 mmol/L for women, Hemoglobin A₁C ≥6.5%, systolic blood pressure ≥130 mm Hg or diastolic blood pressure ≥85 mm Hg. Current pharmacologic therapy for hyperlipidemia, diabetes, or hypertension was considered equivalent to their respective criteria. The presence of MI was defined as the unequivocal presence of subendocardial-based (ie: ischemic) enhancement on LGE CMR, confirmed in two imaging planes. In addition, 16 normal subjects were identified to serve as a control population for ITFV measurement, each required to have no history of MetS or overt cardiovascular disease.

Based upon the established inclusion criteria 4 groups were available for analysis: MetS without MI (MetS+/MI-) [N = 32], MetS with MI (MetS+/MI+) [N = 32], No MetS with MI (MetS-/MI+) [N = 30], and No MetS and No MI (MetS-/MI-) [N = 16].

Patients with a Glomerular Filtration Rate (GFR) of ≤30 ml/min/1.73 m² were excluded due to FDA recommendation against the administration of gadolinium-based contrast agents in this population [10]. Patients with standard contraindications to CMR were similarly excluded. Ethics approval was obtained for this study from the University of Western Ontario Research Ethics Board.

CMR was performed using 3 T MRI scanners (TIM Trio or Verio, Siemens, Erlangen, Germany). Images were obtained using a 32-channel phased-array radiofrequency coil with ECG-gating. All patients had thoracic imaging performed in the sagittal orientation inclusive of both humeral heads using a Half-Fourier Acquisition Single-shot Turbo-spin Echo (HASTE) pulse sequence during shallow breathing (slice thickness 8 mm, gap 2 mm, matrix 256 x 155, FOV 400–450 x 350-400 mm, TE 50 ms, TR 600 ms, iPAT 2). The typical imaging time required for this survey was 25–30 seconds (1 image slice per cardiac cycle), depending upon heart rate (See Figure 1). Cardiac function was assessed in the short-axis orientation at 10-mm intervals from the atrioventricular annulus to apex and in the 2, 3 and 4-chamber views using a standard SSFP-based pulse sequence (slice thickness 6-mm, gap 4-mm, TE 1.3 ms, flip angle 10 degrees, matrix 256x205, iPAT 2, temporal resolution 28-38 ms). Ten to fifteen minutes following the intravenous administration of Gadolinium contrast (0.1 to 0.2 mmol/kg, Gadovist®, Bayer, Inc), LGE imaging was performed using a standard segmented inversion-recovery gradient echo pulse sequence in imaging planes identical to cine images. Typical imaging parameters were: slice thickness = 6 mm; gap = 4 mm; TR = 800 ms; TE = 3.9 ms; flip angle = 20 degrees; matrix 256 x 205; trigger pulse = 2; segments 13–21; iPAT = 2. The inversion time was optimized to null normal myocardium, as previously described [11].

All cine and LGE-CMR datasets were blindly analyzed by an experienced CMR interpreter as part of a standardized, core-laboratory protocol. All quantitative analysis was performed using the same commercially available analysis software (CMR42, Circle Cardiovascular Inc, Calgary). The LV end-systolic volume (ESV), LV end-diastolic volumes (EDV) and LV mass were obtained using semi-automated contour tracing of sequential short axis cine image datasets. LGE imaging was visually scored for the presence of any abnormal enhancement and its transmural distribution. Only patients with unequivocal subendocardial contrast enhancement in two or more contiguous / orthogonal image slices not associated with artifact were labeled as having MI. Signal quantification of LGE images was incrementally performed using a Signal Threshold Versus Reference Myocardium (STRM) technique to measure infarct mass, as previously described [12]. This technique was performed by manual tracing of the endocardial and epicardial contours with the tracing of the largest contiguous region of normal reference myocardium. Total scar volume was determined using a signal threshold of ≥5SD above the mean signal of normal reference myocardium. Total infarct mass was expressed as a percentage of the total LV mass.

To determine a standardized signal threshold for fat quantification on HASTE imaging we performed a visual calibration study, referencing CT-based fat segmentation performed using previously validated techniques [7, 13, 14]. This was performed in 10 study patients who had also received cardiac CT imaging within one month of their CMR study. In these patients their CMR and CT datasets were simultaneously displayed, the latter re-formatted using a 3D multi-planar reconstruction (MPR) to provide anatomically matched slice orientations of three randomly selected MR imaging planes (total of 30 slices evaluated) (OsirX, Version 4.0). Manual contour tracing of the intra-thoracic borders was performed for all selected images, the anatomic limits defined as the level of the clavicle superiorly, the diaphragm inferiorly, the sternum anteriorly, and the vertebral bodies / ribs, posteriorly. An example of this is shown in Figure 2. A validated signal threshold fat signal range (−190 to −30 Hounsfield units [7, 13, 14]) was applied to highlight adipose tissue on the CT image. Five randomly selected regions of paravertebral muscle were selected on the corresponding CMR dataset to serve as a signal reference, the mean signal and SD of these regions determined using a 5–10 mm region of interest. This was followed by manual adjustment of the signal threshold on the CMR image, described by SD above this mean value, until a visual match of intra-thoracic fat segmentation was obtained between the two images. Using this process a signal threshold of ≥10 SD above the mean signal of paravertebral muscle was consistently found to be most effective in identifying intra-thoracic adipose tissue on HASTE images compared to CT-based segmentation.

CMR-based quantification of the ITFV was blindly performed on sequential sagittal HASTE images using manual contour tracing of the intra-thoracic border, as described above, followed by applying the signal threshold of ≥10 SD above the mean signal of paravertebral muscle. The total ITFV was derived as the sum of fat signal area from all slices multiplied by the total slice thickness. ITFV was indexed to the patients’ body mass index (BMI). Total analysis time for measurement of ITFV from HASTE imaging ranged from 8 to 14 minutes.

Intra-observer and inter-observer variability of ITFV quantification was evaluated by the repeated measurement of 10 randomly selected study patients by two interpreters. This was performed in random order on two separate occasions.

Multivariable linear regression analysis was used to determine significant predictors of total scar volume. Reproducibility of ITFV analysis was assessed by linear regression analysis, and the Pearson correlation coefficient (r) is reported for inter- and intra- observer variability of both measures. Bland and Altman analysis was performed to evaluate for systematic bias.

All statistical tests were two-tailed and p <0.05 was regarded as significant. S-Plus (version 8.0, Insightful Software, Seattle, WA) was used to perform the statistical analyses.

Baseline clinical characteristics are shown in Table 1. The mean age of the study population was 59.8 ± 12.5 years with 78% being male and 71% being Caucasian. Patients with MetS were more likely to be receiving oral hypoglycemic therapy, lipid lowering therapy and anti-hypertensive therapy. Higher triglyceride levels were seen in those with MetS while LDL values were modest due to concurrent pharmacologic therapy.

Of the 62 patients demonstrating LGE-CMR evidence of prior MI 10 (16%) were clinically silent with no history of a clinical event. The mean time interval between the most recent clinically recognized MI and CMR was 0.9 ± 2.3 years.

As a retrospective cohort study, performed within a single-center CMR clinical registry, these findings require confirmation within a larger multi-center setting.

The CMR protocol incorporated the use of gadolinium-based contrast for the performance of LGE imaging. Accordingly, patients with significant renal insufficiency (GFR <30 ml/min/1.73 m²) were not included due to FDA warnings of its use in such patients. Extrapolation of study findings to such patients may not be appropriate.

The available healthy control population (MetS-/MI-) from the employed registry was modest in size due to the registry’s mandate to recruit patients with known or suspected cardiovascular disease. Accordingly, case matching of age and sex variables was not feasible and potentially introduces bias for comparisons among this group. However, significant differences in ITFV were identified between disease cohorts (ie: MetS+/MI- versus MetS+/MI+), supporting that ITFV is indeed associated with the occurrence of MI among those with MetS.

Finally, the small number of female patients, most notably in the MetS+/MI- and MetS-/MI- groups, presents a limitation for generalizability of our current findings. Therefore, we recommend validation of our findings within a larger population with more balanced gender representation.

Written informed consent was obtained from the patient for the publication of this report and any accompanying images.