Patients and Protocol
The protocol was reviewed and approved by the Johns Hopkins Institutional Review Board. The patient population included 43 patients with MMD (35 with type I and 8 with type II) diagnosed by genetic testing (82%) or by clinical examination plus genetically proven MMD in a first-degree family member (18%). Patients with examination findings of MMD but negative genotype were not included. All consecutive MMD patients who were referred to the electrophysiology service for arrhythmia risk stratification, without history of atrioventricular block, resuscitated sudden death, or contraindications to CMR, were enrolled in the study. All patients underwent CMR and standard 12-lead ECG. Additionally, five control patients without evidence of arrhythmia, structural cardiac disease, or neuromuscular disease underwent CMR and ECG. Twenty-three of 43 total MMD patients underwent signal-averaged ECG with frank orthogonal leads at a sampling rate of 1 kHz/channel and enough QRS complexes to reduce the noise level to <1 mcV (PC ECG 1200; Norav Medical Ltd., Thornhill, Ontario, Canada).
The median time between ECG recording and MRI was 15.5 days (interquartile range 9–50 days). Images were acquired with a 1.5-T (Magnetum Avanto, Syngo MR B13; Siemens Medical Systems, Erlangen, Germany) MR scanner and a six-channel body phased-array surface coil. After localization of the heart, base-to-apex, short-axis, cine, steady-state free precession, gradient echo images (temporal resolution, 46.80–49.14 msec; echo time, 1.1–1.2 msec; image resolution, 256 × 192 pixels; field of view, 360 × 360mm; slice thickness, 8mm; spacing, 2mm; flip angle, 69–80° ; repetition time, 48.2–55 msec; views per segment, 18; breath hold, 6 sec per slice) were obtained with retrospective ECG gating. Patients then received 0.2 mmol/kg intravenous gadodiamide (Omniscan; Amersham Health/General Electric Healthcare, Waukesha, Wisconsin, USA). Fifteen minutes after the contrast bolus, short-axis delayed images were acquired with an inversion recovery fast gradient echo pulse sequence (repetition time, 5.4–7 msec; echo time, 1.3–3 msec; image resolution, 256 × 256 pixels; field of view, 400 × 400mm; slice thickness, 8mm; spacing, 2mm; flip angle, 20–40° ; inversion time, 175 to 250 msec; repetition time, 500–940 msec; views per segment, 24; breath hold, 12–18 sec per slice). Inversion times were optimized for each patient to null normal myocardium. Given the spatially heterogeneous nature of myocardial fibrosis in MMD, the selection of normal areas required particular attention. In our experience, basal lateral areas of left ventricular myocardium were often spared of fibrosis and were used as the standard for null myocardium in selection of inversion times. The number of image planes acquired depended on the length of the ventricular long axis in each patient (six to 10 planes per patient).
The QMass MR Version 6.2.2 (Medis Medical Imaging Systems, Leiden, The Netherlands) software package was used for image analysis. Steady-state free precession images obtained in the short-axis plane were used for functional analysis. The left ventricular endocardium and epicardium were manually contoured at end diastole and end systole at each short-axis level to calculate the left ventricular ejection fraction, end-diastolic volume, and mass. Postcontrast delayed enhancement images in the short-axis plane were analyzed by manual contouring of the left ventricular endocardium and epicardium. The myocardium was then divided into 20 sectors per slice, starting from the posterior right ventricular insertion point (). The signal-to-noise ratio was used to standardize intensity measurements between patients. The signal-to-noise ratio of each of 20 sectors per inversion-recovery-prepared gradient echo image plane (120–200 sectors per patient, ) was calculated by the dividing the mean sector intensity by the standard deviation of the background noise. Background noise was measured from a large region of interest (3 × 4 cm) placed anterior to the chest wall. The SNRV of the entire left ventricle was calculated by obtaining the variance of the signal-to-noise ratio for all sectors for each patient (). Contours were reviewed and confirmed by two independent observers who were blinded to patient identities and clinical characteristics. Discrepancies were resolved by the senior observer (D.A.B.) with more than 10 years’ experience in CMR.
FIG. 2 The figure illustrates methodology for dividing the myocardium into 20 sectors per plane, starting from the right ventricular insertion point (a). This procedure was performed for all myocardial planes (i) for each patient (b) and the mean intensity of (more ...)
Continuous variables are summarized as median and interquartile range. Categorical and dichotomous variables are expressed as percentages. The unpaired Student's t test was used to compare SNRV among MMD and control patients. Univariate and multivariate linear regression were used to assess the association of SNRV with standard and signal-averaged ECG findings while adjusting for potential confounders. Potential confounders (age, left ventricular ejection fraction, and body mass index) were selected a priori due to known associations with both the dependent and independent variables under study (). Ten-fold cross-validation was used to calculate the area under the receiver operator characteristic curve to assess the value of SNRV in predicting effective bundle-branch block (QRS duration >120 msec). To perform 10-fold cross-validation, the sample was partitioned into 10 subsamples. One subsample was retained as the validation data for testing the model, and the remaining nine subsamples were used as training data. The cross-validation process was then repeated 10 times, with each of the 10 subsamples used exactly once as the validation data. The 10 results for the area under the curve were then averaged to produce a single area under the curve. Statistical analyses were performed using R software (version 2.7.1; R Foundation for Statistical Computing, Vienna, Austria).