Myocardial Effective Transverse Relaxation Time T2* is Elevated in Hypertrophic Cardiomyopathy: A 7.0 T Magnetic Resonance Imaging Study

Hypertrophic cardiomyopathy (HCM) is the most common genetic disease of the myocardium and bares the risk of progression to heart failure or sudden cardiac death. Identifying patients at risk remains an unmet need. Recognizing the dependence of microscopic susceptibility on tissue microstructure and on cardiac macromorphology we hypothesized that myocardial T2* might be altered in HCM patients compared to healthy controls. To test this hypothesis, myocardial T2*-mapping was conducted at 7.0 Tesla to enhance T2*-contrast. 2D CINE T2*-mapping was performed in healthy controls and HCM patients. To ensure that T2* is not dominated by macroscopic magnetic field inhomogeneities, volume selective B0 shimming was applied. T2* changes in the interventricular septum across the cardiac cycle were analyzed together with left ventricular radius and ventricular septal wall thickness. The results show that myocardial T2* is elevated throughout the cardiac cycle in HCM patients compared to healthy controls. A mean septal T2* = 13.7 ± 1.1 ms (end-systole: T2*,systole = 15.0 ± 2.1, end-diastole: T2*,diastole = 13.4 ± 1.3 ms, T2*,systole/T2*,diastole ratio = 1.12) was observed in healthy controls. For HCM patients a mean septal T2* = 17.4 ± 1.4 ms (end-systole: T2*,systole = 17.7 ± 1.2 ms, end-diastole: T2*,diastole = 16.2 ± 2.5 ms, T2*,systole/T2*,diastole ratio = 1.09) was found. Our preliminary results provide encouragement that assessment of T2* and its changes across the cardiac cycle may benefit myocardial tissue characterization in HCM.

SCIENTIFIC RepoRts | (2018) 8:3974 | DOI: 10.1038/s41598-018-22439-x Cardiovascular magnetic resonance (CMR) imaging has emerged as an indispensable tool in the diagnosis and risk stratification of HCM 3,[5][6][7] . Vigorous research has resulted in an enormous body of literature that documents the merits of CMR and showed that the degree of hypertrophy or presence of myocardial fibrosis are associated with a poor outcome in HCM [8][9][10][11] . Notwithstanding this success, understanding the pathophysiologic mechanisms of the imaging findings and identifying patients at risk of SCD or progression to heart failure remains an unmet clinical need 4 . So far it has been extremely challenging to connect the molecular and cellular defects that characterize HCM to the level of major organ systems at which they play themselves out.
Quantitative mapping of the effective transversal relaxation time T 2 * provides valuable means for myocardial tissue characterization without the need for exogenous contrast 12,13 . A growing number of reports refers to mapping T 2 * in basic research and emerging clinical CMR applications 14,15 . Myocardial T 2 * is commonly assumed to provide a surrogate for myocardial tissue oxygenation 14 . Yet, the factors influencing T 2 * are of multiple nature 16 . Further to blood oxygenation, blood volume fraction per tissue volume, hematocrit, the oxyhemoglobin dissociation curve, main magnetic field inhomogeneities, tissue pH, tissue susceptibility, tissue iron content and tissue microstructure or micromorphology were reported to govern T 2 * 13,17,18 . Cardiac macromorphology including ventricular radius and ventricular wall thickness constitutes another category of physiological parameters that orchestrate T 2 * 19 .
The linear relationship between magnetic field strength and microscopic susceptibility effects renders ultrahigh field (B 0 ≥ 7.0 T) CMR conceptually appealing for myocardial T 2 * mapping 20,21 . The enhanced susceptibility effects at 7.0 Tesla (T) may be useful to lower the detection level and to extend the dynamic range of the sensitivity for monitoring T 2 * changes 19 . Moving to ultrahigh magnetic fields also enables cinematic T 2 * mapping in scan times feasible for breath held acquisitions 22 . Taking advantage of this gain, temporally resolved T 2 * mapping at 7.0 T showed cyclic changes of myocardial T 2 * , demonstrated a close correlation with myocardial wall thickness and myocardial wall stress and suggested an association with alterations in myocardial blood volume fraction across the cardiac cycle 23 . These findings hold the potential to exploit T 2 * mapping for non-invasive probing of myocardial (patho)physiology in vivo 24 .
It is established in the literature that HCM can cause alterations in the microstructure of myocardial tissue and can induce changes in cardiac macromorphology including myocardial wall thickening and reduction in left ventricular inner radius. Based on the dependence of microscopic magnetic field pertubations on such changes 25,26 we hypothesize, that myocardial T 2 * and its time course across the cardiac cycle might be altered in HCM patients compared to healthy controls and hence might provide an imaging based marker for HCM. To test this hypothesis myocardial T 2 * of the intraventricular septum was examined at 7.0 Tesla using high spatio-temporally resolved, susceptibility weighted 2D CINE techniques in healthy controls and in HCM patients. This approach was paralleled by an assessment of the patterns and degree of myocardial hypertrophy.

Methods
Study population. Six healthy volunteers without any known history of cardiac disease (4 male, age = 50.0 ± 12.4 years (mean ± sd), BMI = 23.9 ± 2.9 kg/m 2 ) and six patients with confirmed HCM (4 male, age = 52.7 ± 17.5 years (mean ± sd), BMI = 25.2 ± 1.9 kg/m 2 ) were included in the study (Table 1) after due approval by the local ethical committee (ethics committee name: Ethikkommission der Charité, Ethikkommission 1 am Campus Charité-Mitte, Berlin, Germany; registration number DE/CA73/5550/09, Landesamt für Arbeitsschutz, Gesundheitsschutz und technische Sicherheit, Berlin, Germany). In vivo studies were approved by the local ethical committee. The diagnosis of HCM was based on clinical parameters including echocardiography 3 . Informed written consent was obtained from each volunteer prior to the study in compliance with the local institutional review board guidelines. All experiments were performed in accordance with the Declaration of Helsinki and the local institutional review board guidelines.
Late Gadolinium enhancement imaging. All HCM patients included in the study had previously undergone a clinical MRI exam including LGE imaging for detection of myocardial fibrosis 27 . For this purpose a 3.0 T MR system (Magnetom Verio, Siemens, Erlangen, Germany) was employed using a 32-channel RF receive array.
LGE images were acquired 10-15 minutes after application of gadobutrol (0.2 mmol/kg body weight) using a FLASH inversion recovery gradient echo technique to detect fibrosis. Imaging parameters were: TE = 5.4 ms, TR = 10.5 ms, flip angle 30°, spatial resolution (1.5 × 1.5 × 6.0) mm 3 . LGE imaging at 3.0 T was performed for five of the six patients. For the sixth patient LGE imaging data were available from a 1.5 T system (Magnetom Avanto, Siemens, Erlangen, Germany). The slice planning during the current study was based on the previous clinical exams and carefully adjusted to achieve the same slice positioning. B 0 shimming. To minimize the influence of macroscopic magnetic field inhomogeneities on T 2 * and to ensure that T 2 * is not dominated by macroscopic magnetic field inhomogeneities but rather governed by microscopic susceptibility effects, volume selective B 0 shimming was carefully carried out prior to T 2 * mapping 23 . A B 0 field map was acquired in a single breath hold in end-diastole using an axial stack of slices covering the entire heart. A cardiac triggered multi-echo gradient-echo technique (TEs = 2.04 ms and 4.08 ms, TR = 5.4 ms, spatial resolution (4.2 × 4.2 × 8.0) mm 3 , 18 slices) was employed. Based on this field map second order shimming was applied for a shim volume accommodating a four chamber view and a mid-ventricular short axis view of the heart 23 . This shimming approach yielded magnetic field homogeneity in the heart comparable to what has been reported at 3.0 T 22 . Data processing. Data processing ( Fig. 1) was performed offline using MATLAB (The Mathworks, Natick, MA, USA) routines. Prior to T 2 * fitting, all T 2 * sensitized images were de-noised using a spatially adaptive non-local means (SANLM) filter 30,31 (VBM8 toolbox (http://dbm.neuro.uni-jena.de/vbm8/)). SANLM filtering decreased the estimated fit standard deviation of T 2 * in the myocardium by approximately 32% in healthy volunteers and 46% in patients in comparison to fits from unfiltered images. No artifacts due to filtering were observed in the fitted maps 23,31 . After de-noising, the T 2 * weighted CINE images from the three breath holds were co-registered employing a non-rigid registration provided by the MIRT MATLAB toolbox (https://sites.google. com/site/myronenko/research/mirt). Images of the three registered scans were combined to form multi-echo series covering nine echoes with increasing T 2 * weighting (Fig. 1II). Subsequently, non-linear T 2 * fitting was performed using the MATLAB trust region algorithm in combination with a mono-exponential signal decay model (Fig. 1III). Goodness-of-fit was evaluated by R 2 . Additionally the fit standard deviation (T 2 * -STD) was estimated 32 . Voxels with decreased fit quality (R 2 < 0.7 or T 2 * -STD > 3 ms) or unnaturally high/low T 2 * (T 2 * ≤ 1 or T 2 * ≥ 50 ms) were considered unreliable and excluded from further analysis.

Data analysis.
For each subject the left ventricular (LV) myocardium was manually segmented for all cardiac phases. LV wall thickness and inner radius were calculated for 2.5° wide radial sections covering the whole myocardium. This procedure was executed for all cardiac phases resulting in a total sample size of 18720 sections for the healthy controls and in 17712 sections for the HCM patients. Median T 2 * and mean wall thickness were calculated for each cardiac phase to allow assessment of temporal changes. For this purpose only the anteroseptal and inferoseptal segments 33 (6326 septal sections in controls and 5904 sections in HCM patients) were considered, because T 2 * measurements have been shown to be most reliable in the ventricular septum 20 (Fig. 1IV). Mean inner LV radius was calculated per phase by averaging over all sections. The averaged values of T 2 * , wall thickness and LV radius per cardiac phase were determined for all subjects. For calculation of group averages cardiac cycle duration was normalized and the number of phases was unified using linear interpolation. The overall distribution of septal T 2 * was analyzed using the histogram of the relative frequencies of T 2 * for all septal voxels in all subjects within the respective group. Evaluation of LV-morphology was described recently 27 .
Statistical analysis. Statistical analysis was performed using R 34 and MATLAB. Continuous data are expressed as mean ± SD. Group differences were analyzed for significance using a student's t-test for normally Availability of data. The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.

Results
Assessment of cardiac morphology. All scans could be performed without any complication. Mean examination time for the assessment of cardiac morphology was 22 ± 7 minutes in healthy volunteers and 21 ± 2 minutes in patients. Ample blood myocardium contrast with a contrast-to-noise-ratio of about 55, image sharpness and signal uniformity across the heart were achieved for all subjects. Mean septal wall thickness (SWT) averaged over all subjects for all cardiac phases was found to be 7.3 ± 1.2 mm in healthy controls and 14.1 ± 2.5 mm in HCM patients. For end-systole mean SWT was 9.8 ± 1.4 mm in healthy controls compared to 16.6 ± 1.8 mm in HCM patients (Table 1). LV analysis confirmed the substantial difference in LV-mass between healthy volunteers (93 ± 17)g and HCM patients (169 ± 68) g (P = 0.04). No significant differences in left ventricular end-diastolic volume (LVEDV) (P = 0.64), left ventricular end-systolic volume (LVESV) (P = 0.85) and left ventricular ejection fraction (LVEF) (P = 0.47) were found. CINE T 2 * mapping. All volunteers and patients tolerated the breath-hold 2D CINE T 2 * mapping acquisitions (mean examination: 1:28 ± 0:19 minutes in healthy volunteers and 1:26 ± 0:13 minutes in patients). B 0 shimming resulted in macroscopic field dispersions of ΔB 0 < 3 Hz per mm (in-plane) and ΔB 0 < 1 Hz per mm (through-plane) in the ventricular septum which is similar to what has been previously reported for cardiac MRI at 7.0 T and 3.0 T indicating, that T 2 * was not dominated by macroscopic magnetic field variations 22,23 . Averaging T 2 * over all healthy controls for all cardiac phases (6323 septal sections) revealed a mean T 2 * = 13.7 ± 1.1 ms (Table 1). Figure 2 surveys LV T 2 * maps obtained for all healthy volunteers during systole and diastole. For this cohort a mean septal T 2 * = 15.0 ± 2.1 ms was observed for end-systole. For end-diastole mean septal T 2 * = 13.4 ± 1.3 ms was determined. The T 2 * ,systole /T 2 * ,diastole ratio was 1.12 for healthy subjects. The mean range (max -min) of T 2 * over the cardiac cycle was 4.0 ± 1.2 ms which is significantly higher (P < 0.01) than the 0.7 ± 0.4 ms T 2 * change attributed to the periodic macroscopic B 0 variation 23 Fig. 3) which was significantly higher (P < 0.001) than in healthy controls. For patients mean septal T 2 * was 17.7 ± 1.2 ms at end-systole and 16.2 ± 2.5 ms at end-diastole (Fig. 2). The T 2 * ,systole /T 2 * ,diastole ratio was 1.09. No significant difference was found for the mean range (max -min) of T 2 * over the cardiac cycle compared to controls (T 2 * (max-min) = 3.8 ± 1.2 ms).  Figure 3 highlights the time course of mean septal wall thickness, median septal T 2 * and mean inner ventricular radius over the cardiac cycle for HCM patients and healthy controls. Cyclic T 2 * changes across the cardiac cycle were found. A T 2 * increase during systole which is paralleled by an increase in the SWT and a decrease in the left-ventricular radius was observed for healthy controls and HCM patients. Also a decrease of T 2 * during diastole was noted for healty controls and for HCM patients. The overall shape of the curves was similar for patients and controls, with the the diastolic T 2 * decrease being less pronounced in patients. Plotting mean septal T 2 * against mean septal wall thickness revealed two clearly separable clusters for HCM patients and normal controls (Fig. 4a). Analysis of the overall distribution of T 2 * across all septal sections and all cardiac phases showed a significant increase of T 2 * in patients compared to healthy controls (Fig. 4b).

Discussion
This study sought to test the hypothesis, that myocardial T 2 * and its time course across the cardiac cycle are altered in HCM patients compared to healthy controls. The main finding of this study is that septal T 2 * is significantly elevated in HCM patients versus healthy controls. Cyclic variations of T 2 * across the cardiac with T 2 * increasing in systole and decreasing in diastole were observed in both, healthy controls and HCM patients with similar ratios of T 2 * ,systole /T 2 * ,diastole of about 1.1. This is in line with previous reports on healthy volunteers 23 . These novel insights advance the capabilities of MR imaging based myocardial tissue characterization at a level of non-invasive interrogation not previously available in humans.
Following previous reports suggesting myocardial blood oxygenation level dependent contrast or describing T 2 * as a surrogate for tissue oxygenation would lead to the conclusion that T 2 * prolongation in HCM can be attributed to an increase in oxygen supply in hypertrophic versus healthy myocardium. This interpretation does not match the present knowledge though. Previous studies demonstrated that T 2 * strongly depends not only on blood oxygenation but also on the tissue blood volume fraction 16,23,35 . Cyclic changes in myocardial tissue blood volume fraction which are a known physiologic phenomenon related to the contraction and relaxation of the cardiac muscle 36 have been suggested as the main cause behind perodic variations of myocardial T 2 * across the cardiac cycle in healthy volunteers 23 . It is fair to assume, that the periodic changes of T 2 * observed in the HCM patient group are caused by the same mechanism. But how can the overall T 2 * increase in the patient cohort be explained? Microvascular dysfunction and subsequent ischemia are a known feature in HCM 37 and have been reported to scale with greater degrees of hypertrophy and the presence of fibrosis 38,39 . Along this line Johansson et al. described a 33% decrease in myocardial capillary density in HCM 40 . A reduction of vascularity in presence of scar tissue, which is also found in HCM patients, has been shown by histologic studies 41,42 . These conditions result in a reduced myocardial blood volume fraction in HCM and consequently decrease the impact of the deoxygenated hemoglobin on T 2 * , which could explain the observed T 2 * prolongation in hypertrophic regions versus normal or remote myocardium.
An association between microvascular dysfunction and late gadolinium enhancement (LGE) has been described in HCM in the literature 38,43 . Chiribiri et al. reported a coincidence of mid-myocardial LGE with reduced resting state perfusion and greater degrees of hypertrophy 44 .
LGE was also present in all HCM patients in the current study. This finding supports the hypothesis that the observed T 2 * increase might be related to microvascular dysfunction. Reduced myocardial blood flow as observed in microvascular dysfunction and subsequent ischemia has been associated with an unfavorable outcome in HCM and suggested as a strong predictor of clinical deterioration and death 37,39 . This hints that myocardial T 2 * mapping might provide an element of risk stratification in HCM. Arguably, increased myocardial T 2 often coinciding with LGE has been described in HCM and was associated with inflammation and edema as well as ischemia and fibrosis 45,46 . A T 2 increase would also result in elevated T 2 * , providing a further explanation for the observed results. Admittedly these interpretations cannot be ultimately proven at this point. Further investigations including animal models of HCM are required to better understand the relationship of MR parameter changes and underlying microstructural and pathophysiologic mechanisms which cannot be definitely verified in vivo. Yet, this study provides new insights and directions for further investigations which will help to link pathological processes to MR findings also in humans.
It is no secret that the increase of magnetic susceptibility effects at higher fields not only affects microscopic susceptibility changes of (patho)physiological origin but can also result in stronger macroscopic field distortions e.g. at air-tissue interfaces. This is often a concern about T 2 * mapping at high and ultrahigh fields. While careful B 0 shimming is certainly of high importance for this, in fact, dedicated shimming techniques have been shown to provide B 0 field uniformities in the heart at 7.0 T similar to what has been described at 3.0 T 22,23 . Further to this, an often overlooked benefit of performing T 2 * mapping at higher fields is the possiblity to invest the signal gain due to the higher net magnetization to realize smaller voxel sizes. This effectively reduces intravoxel dephasing due to macroscopic field gradients and their influence on T 2 * and also decreases partial volume effects. The voxel size used for T 2 * mapping in this study of (1.1 × 1.1 × 4.0) mm 3 is multiple times smaller than voxel sizes commonly used for T 2 * mapping in at 1.5 T ot 3.0 T which further helps to reduce the impact of macroscopic field inhomogeneities 22 .
While we recognize a limitation due to the limited number of healthy subjects and patients studied, we believe this feasibility study to be an essential precursor to a larger 7.0 T study involving healthy and HCM patient cohorts. Such a study is needed to further elucidate the underlying causes for the reported T 2 * increase and further analyze the temporal behavior of T 2 * over the cardiac cycle in health and in the presence of pathology. We are only beginning to understand the meaning of the dynamics of T 2 * across the cardiac cycle, but the technique holds the promise to enable further insights into myocardial (patho)physiology in vivo. Admittedly, the continuous motion of the myocardium currently limits temporal T 2 * assessment to larger regions of interest like the septum which have to be defined for each cardiac phase. Application of motion compensation techniques to CINE T 2 * mapping data would enable us to derive information on T 2 * dynamics at single voxel level. Parameters like T 2 * temporal standard deviation, slope, time to peak, relative T 2 * change, etc. could be derived from such times series and converted to quantitative maps that may help to highlight areas of change and pathology. This approach provides an entirely new set of potential (patho)physiological imaging paramters but requires further efforts to be implemented and evaluated.

Conclusions
Myocardial T 2 * is elevated throughout the cardiac cycle in HCM patients compared to healthy controls at 7.0 T. A reduction in tissue blood volume fraction in the hypertrophied myocardium and a T 2 increase related to inflammatory processes were suggested as potential causes for this finding. These factors have been associated with a higher risk for a poor outcome of HCM patients. Our preliminary results provide encouragement, that assessment of T 2 * and its changes across the cardiac cycle may benefit myocardial tissue characterization in hypertrophic cardiomyopathy.