Identification of Myocardial Disarray in Patients With Hypertrophic Cardiomyopathy and Ventricular Arrhythmias

Background Myocardial disarray is a likely focus for fatal arrhythmia in hypertrophic cardiomyopathy (HCM). This microstructural abnormality can be inferred by mapping the preferential diffusion of water along cardiac muscle fibers using diffusion tensor cardiac magnetic resonance (DT-CMR) imaging. Fractional anisotropy (FA) quantifies directionality of diffusion in 3 dimensions. The authors hypothesized that FA would be reduced in HCM due to disarray and fibrosis that may represent the anatomic substrate for ventricular arrhythmia. Objectives This study sought to assess FA as a noninvasive in vivo biomarker of HCM myoarchitecture and its association with ventricular arrhythmia. Methods A total of 50 HCM patients (47 ± 15 years of age, 77% male) and 30 healthy control subjects (46 ± 16 years of age, 70% male) underwent DT-CMR in diastole, cine, late gadolinium enhancement (LGE), and extracellular volume (ECV) imaging at 3-T. Results Diastolic FA was reduced in HCM compared with control subjects (0.49 ± 0.05 vs. 0.52 ± 0.03; p = 0.0005). Control subjects had a mid-wall ring of high FA. In HCM, this ring was disrupted by reduced FA, consistent with published histology demonstrating that disarray and fibrosis invade circumferentially aligned mid-wall myocytes. LGE and ECV were significant predictors of FA, in line with fibrosis contributing to low FA. Yet FA adjusted for LGE and ECV remained reduced in HCM (p = 0.028). FA in the hypertrophied segment was reduced in HCM patients with ventricular arrhythmia compared to patients without (n = 15; 0.41 ± 0.03 vs. 0.46 ± 0.06; p = 0.007). A decrease in FA of 0.05 increased odds of ventricular arrhythmia by 2.5 (95% confidence interval: 1.2 to 5.3; p = 0.015) in HCM and remained significant even after correcting for LGE, ECV, and wall thickness (p = 0.036). Conclusions DT-CMR assessment of left ventricular myoarchitecture matched patterns reported previously on histology. Low diastolic FA in HCM was associated with ventricular arrhythmia and is likely to represent disarray after accounting for fibrosis. The authors propose that diastolic FA could be the first in vivo marker of disarray in HCM and a potential independent risk factor.

P redicting sudden cardiac death (SCD) in hypertrophic cardiomyopathy (HCM) remains a challenge because most patients have a normal life expectancy (1). Current clinical risk stratification guidelines lack the high sensitivity and specificity needed to accurately predict this devastating complication, because they are based on clinical features alone (2,3). The hallmark feature specific to patients who die suddenly with HCM is extensive myocardial disarray (4,5), and this is a likely focus for re-entrant ventricular arrhythmias leading to SCD (6). Post-mortem histology has shown that the left ventricular (LV) mid-wall contains a band of circumferentially orientated myocytes (13)(14)(15), which in HCM is infiltrated by disarray and fibrosis, typically at the interventricular junctions and the hypertrophied segments (Central Illustration panel E) (14). Deranged myocyte orientations within disarray are not randomly distributed, and the overall average midwall orientation remains circumferential (16). Hence, HA in HCM are likely to be normal despite disarray ( Figure 2 Streeter's classic micrograph of transmural variation of helix angles from endocardium (0% wall thickness) to epicardium in a canine left ventricle (15) (A). Approximately 3 DT-CMR imaging voxels (2.8 Â 2.8 Â 8 mm 3 ) span the myocardium transmurally in diastole (B). In the endo-and epicardial voxels, myocytes are progressively changing from a longitudinal to circumferential orientation and vice versa, respectively. There are no marked transmural differences in myocyte diameter or fibrous tissue in healthy myocardium (40). Therefore, the wide distribution of myocyte orientations in the endo-and epicardium will reduce FA. Conversely, the narrow distribution of orientations in the mid-wall where myocytes are consistently circumferentially orientated will elevate FA (C). Myocyte disarray and fibrosis in the mid-wall will reduce FA, due to the wide distribution of disorganized myocyte orientations and expanded extracellular space (39), but the overall mean voxel helix angle remains in the circumferential orientation (D). Thus, helix angle is expected to be normal despite abnormal FA. LV ¼ left ventricular; other abbreviations as in Figure 1.
Ariga et al.     Figure 3B), suggestive that low FA may also be sensitive to disarray ( Figure 3C).  (Figure 4) To examine the relationships between FA, fibrosis, and maximum wall thickness in patients with and without ventricular arrhythmia, we used logistic

DISCUSSION
In this paper, we show that DT-CMR provided in vivo visualization of normal and HCM myocardial architecture, which demonstrated high concordance with previously described post-mortem findings (14).
Despite the relatively coarse spatial resolution of in vivo DT-CMR, which was on the scale of millimeters, diastolic FA was sensitive to microscopic structure and was able to image the well-documented circumferential alignment of myocytes in the midwall (13). FA was reduced in HCM compared with control subjects, but there was heterogeneity between HCM patients. Only 34% of patients demonstrated FA values below control subjects, some of whom had LGE. Using multimodal cardiac magnetic resonance (CMR), we were able to account for the contribution of fibrosis to FA. FA, adjusted for fibrosis, remained reduced in HCM, indicating that low FA may also demonstrate sensitivity to disarray. We showed that focal reduction of FA in the maximally hypertrophied segment of HCM patients was associated with ventricular arrhythmia. This association of low FA with ventricular arrhythmia persisted even after adjusting for fibrosis and hypertrophy, suggesting that adjusted FA could be measuring disarray and hence may provide a novel, independent risk factor.

CONCORDANCE WITH OTHER IN VIVO DT-CMR
STUDIES. Our quantitative DT-CMR parameters, ADC, FA, and HA in control subjects were comparable with previous intercenter reproducibility results using the same imaging protocol (11). Furthermore, we confirm abnormal diastolic SA in HCM as previously described (22). However, no significant SA differences were detected in HCM patients with and without ventricular arrhythmia, using our protocol. Our FA results in HCM are consistent with a previous, small study that also reported reduced FA in the septum compared with the free wall in 5 HCM patients using DT-CMR acquired at mid-systole (23). However, more recently, McGill et al. (12) found no difference in septal FA in 10 HCM patients when DT-CMR was acquired from end-systole to end-systole. This may in part be due to the small sample size. More than 10 patients within our cohort had no detectable difference in FA, which is likely to represent the significant proportion of HCM patients who have a benign clinical course with no adverse effect to life expectancy (24). Another factor is that FA measured using a stimulated echo diffusion sequence is affected by cardiac strain, which is most marked at peak systole (20). Although this Values are slice mean AE SD, and for SA, circular mean AE SD. Bold p values are significant.
Ariga et al. diffusion encoding. Furthermore, ectopy, which is often seen in HCM, prolonged scan times because optimal diffusion encoding is dependent on a regular RR interval. Shorter scan times are required before DT-CMR can be adopted for routine clinical use in symptomatic patients. This may be achievable with stronger CMR gradient systems and advanced DT-CMR sequences such as second-order motioncompensated spin echo (34,35) and whole-heart methods (36). This also provides the potential to assess novel DT-CMR measures of disarray such as tractographic propagation angle (37) in HCM.
Our patients were relatively low risk because we could not include those with ICDs; thus, in this hypothesis-generating study, we could only test for association with predictors of risk rather than events.
To document clinical utility of low FA for deciding whom to treat with ICD, we will need to measure its association with hard endpoints such as SCD, aborted SCD, and appropriate therapies from ICDs implanted after initial FA measurement. This will need to be tested prospectively in large-scale multicenter studies and registries such as the Hypertrophic Cardiomyopathy Registry (38).

CONCLUSIONS
DT-CMR provided in vivo assessment of normal and HCM myocardial microstructure, which demonstrated concordance with histology. FA was reduced in HCM compared with control subjects. Low FA in HCM was associated with ventricular arrhythmia and is likely to represent disarray after accounting for fibrosis. We propose that diastolic FA could be the first in vivo marker of disarray in HCM and thus a potential independent risk factor for SCD.
ACKNOWLEDGMENTS. The authors thank Joanne Sellwood for nursing support, Rachel Given for data management, Dr. Elizabeth Ormondroyd and the Oxford Medical Genetics Laboratories for genetic data, Dr. Joanne Bates for the simulated diffusion images in Figure 1, and Dr. Lei Clifton for advice on statistics.  DT-CMR in HCM Disarray and Ventricular Arrhythmia