Increased myocardial native T1 and extracellular volume in patients with Duchenne muscular dystrophy

This single-center, prospective, observational study was approved by the Vanderbilt University Institutional Review Board and completed between January 2013 and July 2015. Consent was obtained for all participants; those under 18 years of age signed an age-appropriate assent form. Participants in the DMD group were recruited from the multidisciplinary Neuromuscular-Cardiology Clinic and were over 7 years of age. The diagnosis of DMD was confirmed by either skeletal muscle biopsy or the presence of a dystrophin mutation and skeletal muscle weakness. Exclusion criteria were: requiring sedation for CMR, renal dysfunction or other contraindication to contrast-enhanced CMR.

Healthy participants aged 18–30 with a body mass index ranging from 18 to 25 kg/m² were recruited into the control group. Exclusion criteria were: cardiovascular disease (including but not limited to congenital heart disease, cardiomyopathy, secondary heart disease such as rheumatic fever or Kawasaki’s disease), risk factors for cardiovascular disease (such as arrhythmia or hypertension), muscular dystrophy or unexplained skeletal muscle weakness, any diagnosis that could affect cardiac function or lead to myocardial fibrosis, and renal dysfunction (glomerular filtration rate <75 ml/min assessed on day of scan) or other contraindication to CMR with gadolinium.

Two DMD participants were excluded for inadequate images. Two control participants were excluded for abnormal CMRs (bicuspid aortic valve and pericardial effusion) and one was excluded for inadequate images.

Pertinent DMD clinical data were collected from the electronic medical record. For the control group, a phone screening was performed to assess suitability, including comprehensive questioning on past and family medical history. Prior to the CMR, the participant completed a health risk assessment. All participants provided a blood sample for measurement of hematocrit on day of CMR.

CMR was performed on a 1.5 Tesla Siemens Avanto (Siemens Healthcare Sector, Erlangen, Germany) with an 8 channel cardiac coil. Functional imaging was performed as previously described using balanced steady-state free precession (bSSFP) imaging [13]. Breath-held, T2-weighted (effective TE 60 ms) double inversion turbo spin-echo imaging was performed prior to contrast administration.

The control group and a subset of DMD participants underwent T2 mapping using a breath-held, electrocardiogram (ECG)-triggered, bSSFP sequence with motion correction. Typical imaging parameters were as follows: Adiabatic T2 preparation with 35° flip angle, field of view 340 × 272 mm², matrix size 192 × 144, slice thickness 8 mm, voxel size 1.8 × 1.9 × 8.0 mm³, TR/TE 2.5 ms/1.1 ms, parallel imaging factor of 2. Intravenous Gd-DTPA contrast (gadopentate dimeglumine, Magnevist®, Bayer Healthcare Pharmaceuticals, Wayne, NJ, USA) was administered through a peripheral intravenous line (PIV) at a dose of 0.2 mmol/kg. Multiple LGE techniques were used to insure detection of enhancement: 1) single shot and 2) segmented inversion recovery bSSFP with optimized inversion recovery to null the signal from the myocardium, and 3) phase sensitive inversion recovery bSSFP with an inversion time of 300 ms.

Breath-held modified Look-Locker inversion recovery (MOLLI) sequences were performed prior to and 15 min after contrast administration in the mid-ventricular level in the short axis plane at the same slice location as the T2 mapping and the T2-weighted, cine, and LGE imaging [10, 21]. MOLLI sequences were motion-corrected, ECG-triggered images obtained in diastole with typical imaging parameters: non-selective inversion with a 35° flip angle, single shot SSFP imaging, initial inversion time of 120 ms with 80 ms increments, field of view 340 × 272 mm², matrix size 256 × 144, slice thickness 8 mm, voxel size 1.3 × 1.9 × 8.0 mm³, TR/TE 2.6 ms/1.1 ms, parallel imaging factor of 2. The matrix size was decreased to 192 × 128 for heart rates >90 (approximate voxel size 1.8 × 2.1 × 8 mm³). The initial pre-contrast MOLLI acquired 5 images after the first inversion with a 3 beat pause followed by 3 images after the second inversion, or 5(3)3. This was later modified to a 3 s pause, or 5(3 s)3, to reduce bias from higher heart rates; the post-contrast protocol was acquired at a 4(1)3(1)2 [22]. Motion correction as described by Xue, et al. was performed and a T1 map was generated on the scanner [23]. A goodness of fit map was also performed in most participants at the time of the scan to evaluate data quality. One investigator (JS) was present for all scans and reviewed the MOLLI sequences with the technologist for adequacy at the time of the scan. Any image felt to be inadequate due to poor breath holds or poor motion correction was repeated at the time of the scan.

One reader (JS) traced all contours in this study. LV volume, mass, and function were calculated as previously described [13]. The presence or absence of LGE was qualitatively assessed by one reader and confirmed by a second reader. T2 maps were contoured to obtain global myocardial T2.

T1 maps, obtained prior to and after contrast administration as described by Messroghli et al., [21] were used along with the subject’s hematocrit to calculate an ECV map using software programmed in MATLAB 2014a (The MathWorks, Natick, MA, USA). Affine two-dimensional image registration was performed using a normalized mutual information cost and search function. Native and post-contrast T1 maps were registered to the native T1 map. The field of view was cropped to the region of interest to prevent spurious maximization of mutual information that would result in poor alignment. In cases of suboptimal T1 map image registration on the scanner, the sequence was either repeated without motion correction or image registration was performed using MATLAB software. T1 maps were computed from magnitude reconstructed MR data using iterative sign inversion Levenberg-Marquardt optimization, as complex data were not available. This process results in multiple optimizations and T1 estimates per voxel. The best fit, and therefore best estimate of T1, was determined by the optimization that had the smallest root mean squared error. Affine two-dimensional image registration was again used and all image frames, both pre and post-contrast, were registered to the first pre-contrast frame for consistency. The T1 and ECV calculations were validated using an a priori defined dataset (Fig. 1). The ECV was calculated as:

Regions of interest (ROIs) were manually drawn on T1 and ECV maps within the LV mesocardium in 6 segments using the standard AHA/ACC model of segmentation (Fig. 2) [24]. Global myocardial T1 and ECV values were calculated from segmental values. ROIs were carefully traced to avoid partial volume averaging with blood-pool or epicardial fat. Based on the T1 mapping consensus statement, areas of LGE were included in the ROIs as these areas were felt to be the most focal areas in a continuum of diffuse ECM expansion [10]. For cases where the ROIs were difficult to trace, LGE images were used to help trace the ROIs. Imaging artifact was not contoured. Segments were not included in the analysis if the bounds of the myocardium could not be distinguished from surrounding tissue and blood pool or if image registration was inadequate in those segments. In order to assess the ECV of large areas of LGE, targeted ROIs were traced in areas of clear LGE on ECV maps. These ROIs were confirmed by a second reader (DP).

A random sample of 10 ECV maps was re-contoured by a separate reader (DP) to assess inter-observer variability. The same sample was re-contoured by the initial reader one month later to assess intra-observer variability.

The quality of each ECV map was graded using methods reported by Kellman et al. [14] whereby ECV maps were assigned a score ranging from 1 to 5, with 1 representing a non-diagnostic map and 5 representing an excellent map. None of the maps with a score less than 3 were included.

Demographic variables were compared using either a Wilcoxon rank-sum (continuous variables) or a Chi-square or Fisher’s exact test (categorical variables). Inter- and intra-observer variability were assessed using an intraclass correlation coefficient. Correlations between continuous variables were evaluated using Spearman’s rho (2 variables) or linear regression (controlling for multiple variables). The segmental and global native T1 and ECV were compared between groups using Wilcoxon rank-sum tests. Subset analyses included comparisons of DMD participants with normal LVEF (≥55 %) and without LGE to controls. The variance of ECV measurements within each segment was calculated by squaring the standard deviation. We then averaged the variances from each segment and took the square root, which estimates the average within-subject standard deviation.

Analyses were performed with IBM SPSS statistics, version 22.0 (Armonk, NY: IBM Corp). Study data were collected and managed using REDCap (Research Electronic Data Capture) electronic data capture tools hosted at Vanderbilt [25].

The study enrolled 31 DMD and 11 healthy participants. The DMD participants were younger than the controls (Table 1). DMD participants had faster heart rates and were shorter and weighed less, resulting in smaller body surface areas. There were no other significant demographic differences.

The DMD group had a lower LVEF than controls (54.8 % ± 8.1 vs 60.3 % ± 2.7, p = 0.033) (Table 2). All control participants had normal LVEF (≥55 %). Twelve of the 31 DMD participants (38.7 %) had abnormal LVEF. All DMD participants with abnormal LVEF had regional wall motion abnormalities; 21 DMD participants had regional wall motion abnormalities. The DMD group had a smaller indexed left ventricular end diastolic volume (63.8 ml/m² ± 14.8 vs 79.3 ml/m² ± 11.0, p = 0.001). This is consistent with the minimal early LV dilatation previously reported in DMD [26]. There were no other significant differences in chamber dimensions.

No control patients had LGE. Nineteen of the 31 DMD participants had LGE in at least one segment (Table 2); 16 of those participants had LGE in the mid-ventricular slice used for T1 mapping. Eighteen of the 19 DMD participants with LGE had regional wall motion abnormalities. DMD participants with LGE had significantly lower LVEF than DMD without LGE (51.5 % ± 8.6 vs 60.2 % ± 2.7, p = 0.001). All 12 DMD participants with abnormal LVEF had at least one segment with LGE. LGE was present in 7 DMD participants with normal LVEF.

Figure 3 shows example native T1 maps and ECV maps, with corresponding LGE images. The mean native T1 value was significantly higher in DMD than in controls (1045 ms ± 57 vs 988 ms ± 14, p = 0.001) (Table 3, Fig. 4a). These differences remained significant when comparing DMD participants with normal LVEF (1039 ms ± 49 vs 988 ms ± 14, p = 0.002) and DMD participants without LGE (1038 ms ± 55 vs 988 ms ± 14, p = 0.011) (Fig. 4b, c).

ECV values demonstrated excellent intra- and inter-observer agreement (ICC = 0.963 for inter-observer, ICC = 0.957 for intra-observer, p < 0.001 for both). When compared with controls, DMD participants had significantly increased ECV (0.31 ± 0.05 vs 0.24 ± 0.01, p < 0.001) (Fig. 5a, Table 3). When comparing LVEF-normal DMD participants with controls, this difference remained significant (0.28 ± 0.03 vs 0.24 ± 0.01, p < 0.001) (Fig. 5b). LGE-negative DMD participants had a significantly higher ECV than controls (0.29 ± 0.03 vs 0.24 ± 0.01, p = 0.001) (Fig. 5c). DMD participants with systolic dysfunction had a greater mean ECV than DMD participants with normal function (0.35 ± 0.05 vs 0.28 ± 0.03, p < 0.001). There was no significant difference in ECV between DMD participants with and without LGE (0.32 ± 0.05 vs 0.29 ± 0.03, p = 0.306). However, 3 of those participants had no LGE in the mid-ventricular slice position used for T1 mapping. When DMD subjects were divided into those with visible LGE at the mid-ventricular slice and those without, those with visible LGE had significantly elevated ECV (0.33 ± 0.05 vs 0.29 ± 0.03, p = 0.03). The mean ECV of the targeted ROIs of LGE was significantly higher than the ECV of DMD participants without LGE (0.51 ± 0.08 vs 0.29 ± 0.03, p < 0.001). The range of ECV in these targeted ROIs was 0.39 to 0.63. DMD participants had a higher within subject standard deviation of ECV compared with controls (0.039 vs 0.024, p < 0.001). DMD participants without LGE also had a higher within subject standard deviation of ECV compared with controls (0.033 vs 0.024, p = 0.001).

There was no correlation between duration of angiotensin converting enzyme inhibitor, angiotensin receptor blocker, or beta-blocker therapy and either native T1 or ECV in DMD participants. There was a trend towards a negative correlation between duration of corticosteroids and ECV (rho = −0.351, p = 0.053), but this did not reach statistical significance. There was no difference in ECV between DMD participants with more than 6 months of steroid therapy and those with less than 6 months of therapy. There was a negative correlation between ECV and heart rates (r = −0.463, p = 0.009) suggesting that higher heart rates are associated with lower ECV. However, when adjusting for use of beta-blockers or LVEF, this association was no longer significant. There was no significant association between native T1 and heart rate within DMD (p = 0.345). There was a significant negative correlation between heart rate and T2 (r = −0.64, p = 0.025), though DMD participants with heart rates less than 100 bpm also had T2 values below the mean of controls (47 ms vs 49 ms, p < 0001). Global ECV had a significant negative correlation with LVEF (rho = −0.618, p < 0.001).

When analyzing the entire cohort, linear regression demonstrated no change in the association between global ECV and diagnosis when correcting for age, height, weight, or heart rate (Table S1, Additional file 1). There was also no change in the association between native T1 and diagnosis when correcting for age, height, or heart rate (Table S2, Additional file 1). There was a correlation between weight and native T1 using multivariable regression with no association between native T1 and diagnosis. The significance of this finding is unclear.

This is a single center study and in principle, this could make the results less generalizable. The Neuromuscular-Cardiology Clinic at our institution, however, draws DMD patients from across the country, and a large percentage of DMD patients are followed in a small number of centers similar to ours. The primary objective of this study was the detection of ECM expansion and not the identification of modifiers of this expansion; therefore, adjustment for medications was not performed. Moreover, because the study included control participants not taking any medications, the analysis could not be adjusted for current medications. We evaluated the effects of medications within the DMD population, though this analysis was likely underpowered, and this adjustment will be important for any analysis performed exclusively in DMD participants. Studies with a larger numbers of patients, likely multi-center, will be necessary to determine the effects of medications on ECM expansion, but this is beyond the scope of the current manuscript. The heart rates were significantly faster in DMD subjects. However, data suggest that the heart rate effects on T1 mapping are minimized with the sequence parameters used in this study [22]. Interestingly, there was a significant negative correlation between heart rate and ECV, suggesting that higher heart rates were associated with lower ECV. While this may have made it more difficult to achieve statistical significance, we suspect that this association is related to confounding from severity of disease; the association was no longer significant after correction for either beta-blocker usage or LVEF. There was no significant correlation between heart rate and native T1. The T2 values of DMD participants did correlate with heart rate and this may have led to some bias. The negative association between weight and native T1 may be related to myocardial fat deposition and should be explored in future studies.

Gadolinium contrast agents were not administered to healthy children in order to minimize the potential for adverse events. Therefore, the ages of the control group did not match perfectly with that of the DMD group. The progressive nature of DMD cardiomyopathy, however, suggests that older DMD participants are at higher risk of ECM expansion. Moreover, ECV does not vary significantly with age in healthy controls [36]. Because of the difficulty in obtaining CMR without anesthesia or sedation in young children, the youngest age of DMD participants was 8 years. The reported ECVs were only in the 6 myocardial segments at the mid portion of the LV; we anticipate that increasing the myocardial coverage will improve sensitivity.