This prospective case–control study recruited a total of 81 consecutive patients with HCM who were examined in our department between January 2012 and April 2013. The diagnosis of HCM was made by echocardiography based on the criteria of the World Health Organization/International Society and Federation of Cardiology. All patients had also undergone MRI examination. Patients’ medical records were thoroughly reviewed to ensure that there was no history of hypertension, and no patient presented with hypertension during the study period. Patients were excluded if they had poor resolution of echocardiographic images; obstructive HCM (LV outflow tract gradients >30 mmHg under basal conditions or after Valsalva maneuver); left ventricular ejection fraction (LVEF) <50 % as assessed by echocardiography; cardiac muscle disease secondary to any known systemic condition; atrial fibrillation, implantation of a pacemaker or defibrillator; significant valvular heart disease; or known coronary artery disease (lesions >50 % on angiogram). Finally, after five patients were excluded because their echocardiographic images were of poor quality and not suitable for the present study, the data from 76 patients were retained for subsequent analysis.

Patients were classified as having asymmetric septal hypertrophy if wall thickness in the basal or mid anterior/inferior septum was >1.3 cm and septal/lateral wall thickness ratio was >1.3. Patients were classified as having a septal bulge if the proximal anterior/inferior septum was >1.3 cm and the mid or apical anterior/inferior septum was <1.3 cm. Apical hypertrophy was diagnosed if the ratio of maximal segmental wall thickness of the apex and the rest of the ventricle was >1.5. All patients were in sinus rhythm.

All drugs were discontinued at least 24 h before echocardiography and MRI evaluation, as previously recommended [13]. Patient history of syncope (transient and complete loss of consciousness) and family history of sudden cardiac death in a first-degree relative were recorded.

For the purpose of analysis, the HCM group (n = 76) was further divided (after imaging and echocardiographic examinations described below) into a fibrosis group and a non-fibrosis group.

In addition, a group of 46 healthy volunteers who underwent advanced LV-twist analysis by echocardiography served as controls. These subjects had no documented history of cardiovascular disease, including hypertension or HCM. These subjects did not undergo MRI evaluation.

This study was reviewed and approved by the hospital’s Institutional Review Board. All included patients and control subjects provided signed informed consent.

Magnetic resonance imaging examination was performed within 2 weeks of echocardiographic examination. MRI was performed using a 1.5-T Sonata scanner (Siemens Corporation, Erlangen, Germany). Electro cardio-gating breath-holding sequential scanning was carried out with the patient in the dorsal position. Axial and sagittal single-shot turbo spin echo imaging was carried out on the basis of conventional axial, sagittal and coronal phases. Cine-imaging was carried out for left ventricular two-chamber and four-chamber longitudinal planes as well as LV outflow tract planes and LV shot-axis view for 6–8 layers, and the scanning sequence was retrospective ECG gating true fast imaging, as described previously [14]. The left ventricular mass at end-diastole was semi-automatically calculated on the basis of Simpson’s rule, using a work station and manual revision by visual assessment. The left ventricular mass was normalized to body surface area. Images for LGE diagnostics were acquired 10 min after the injection of gadopentetate dimeglumine 0.2 mmol/kg (Magnevist; Schering Corporation, Berlin, Germany) with breath-hold segmented inversion-recovery sequences acquired in the same view. Myocardial perfusion and delayed scanning were analyzed according to the 17-segment method for the left ventricle recommended by the American Heart Association (AHA) [15]. The presence of late enhancement within each segment was assessed semiquantitatively using a 17-segment model of the LV [16]. LGE of the LV was considered present if the signal intensity of hyperenhanced myocardium was >5 standard deviations above the mean signal intensity of vital myocardium. MRI analysis was completely blinded from the echocardiographic and clinical analysis.

Echocardiography was carried out using the Phillips iE33 Color Doppler Ultrasound System with probe frequency 3.5 MHz (Philips Healthcare, Andover, MA, USA). The Qlab multiparameter analysis workstation (Phillips) was equipped with STE analysis software. The second-harmonic B-mode images of apical (4-chamber, 2-chamber) and short-axis (at the mitral valve and apical level) views were obtained. The frame rate was more than 50 frames/s. The LV endocardial border was manually traced at the end-systolic frame and the speckle tracking region of interest was automatically selected. The width of the region of interest was adjusted as necessary to accommodate the total thickness of the LV wall [17]. The computer automatically tracked stable objects in each frame using the sum of absolute differences algorithm. After these steps, the work station computed and generated strain curves, as previously described [18].

The peak values of basal rotation were defined as the maximum negative values of the curves from the short-axis view at the mitral valve level. The peak values of apical rotation were defined as the maximum positive values of the curves from the short-axis view at the apical level [19]. Every view considered was divided into six segments, giving six different values from which mean values were determined. LV twist was defined as the difference between the mean values of the peak rotation at the apical and mitral valve level (twist = mean peak apical rotation − mean peak basal rotation) [10].

Standard two-dimensional measurements were obtained using standard methods. LVEF was calculated using the two-dimensional Simpson’s biplane method. LV mass was assessed with the two-dimensional area-length method and the left ventricular mass was normalized to body surface area, as previously described [20]. Left ventricular diameter in end-diastolic (LVDd), inter-ventricular septal thickness (IVST), left ventricular posterior wall thickness (LVPWT), and relative wall thickness (RWT) were measured according to criteria of the American Society of Echocardiography. Left atrium end-systolic dimensions (LADs) were measured at end-systolic, and LA volume was calculated using the prolate ellipsoid model. The LA volume index (LAVi) was calculated as LA volume divided by body surface area, reported as milliliters per square meter. The following variables were measured at the tips of the mitral valve leaflets from the apical 4-chamber view using pulsed-wave Doppler: peak early (E) and late (A) diastolic velocities, and E/A ratio were measured. Finally, by pulsed tissue Doppler, peak early diastolic velocity on the septal part of the mitral annulus was measured (Em) and E/Em ratio was calculated as previously described [20].

Patients’ baseline characteristics in HCM and control groups are presented in Table 1. Among 122 subjects enrolled in this study, 76 were HCM patients and 46 were healthy subjects, including 82 males and 40 females with a mean age of 46.1 years. Mean heart rate was 70.5 beats/min, mean BSA was 1.7 m², mean BMI was 23. 9 kg/m², mean SBP was 122.5 mmHg andmean DBP was 80.4 mmHg. In the HCM group, 26.3 % had a history of HCM and 13.2 % had unexplained syncope.

No significant differences were found between the HCM and control group in age, gender, heart rate, BSA and BMI (p > 0.05). The means of SBP and DBP were significantly higher in the HCM group than in the control group (SBP: 124.3 vs. 119.5, p < 0.001; DBP: 81.8 vs. 78, p < 0.001). The HCM group had more patients with a history of HCM and unexplained syncope than the control group (history of HCM: 20 vs. 0, p < 0.001; unexplained syncope: 10 vs. 0, p < 0.001) (Table 1).

Comparisons of echocardiography parameters between HCM and control groups are presented in Table 2. Among 122 patients, the mean Bas-Rotation was 7.4°, mean AP-Rotation was 9.9°, mean LV-Twist was 17.3°, mean LVEF was 69.2 %, mean LADs was 37.4 mm, mean LVDd was 42.3 mm, mean IVST was 16.3 mm, mean LAVi was 43.5 ml/m², mean E/A was 1.1, mean E/Em was 12.3 and mean LVMI was 100.2 g/m². The HCM group had significantly higher values for Bas-Rotation, AP-Rotation, LV-Twist, LVEF, LADs, LAVi, IVST, LVPWT, RWT, E/Em and LVMI than the control group (all p < 0.0001). LVDd and E/A were significantly lower in the HCM group than in the control group (both p < 0.001) (Table 2). Representative examples are shown in Fig. 1.

The distribution of characteristics in HCM patients with and without fibrosis is presented in Table 3. In the 76 HCM patients, means for Bas-Rotation, AP-Rotation, LV-Twist, LADs, LAVi, IVST, LVPWT, RWT, E/Em, and LVMI were all significantly higher in the fibrosis group than in the non-fibrosis group (p < 0.001). Means of age and E/A were significantly lower in the fibrosis group compared with non-fibrosis group (age: 44.6 vs. 51.7 years, p = 0.007; E/A: 1.0 vs. 1.2, p = 0.001). No significant differences were found between non-fibrosis vs. fibrosis groups in heart rate, BSA, BMI, SBP, DBP and LVEF (p > 0.05). Representative examples are shown in Fig. 2.

In addition, LV-twist showed a significant, positive and linear correlation with IVST, LVPWT, RWT, LAVi and E/Em in HCM patients with fibrosis. The greater the increase in LV-twist, the greater were IVST, LVPWT, RWT, LAVi and E/Em. No significant linear correlations were found between LV-twist and E/A, untwisting velocity (Supplemental Table 2).

Univariate logistic regression revealed that the significant effect factors for fibrosis in HCM patients were age, Bas-Rotation, AP-Rotation, LV-Twist, LADs, IVST, LAVi, E/A, E/Em, LVMI, LVPWT and RWT. When these variables were recruited into multivariate logistic regression, stepwise model selection revealed that LV-twist would be the best factor to detect myocardial fibrosis (Supplemental Table 1). When ROC curve analysis was used to examine the LV-Twist performance in HCM patients with and without myocardial fibrosis, the AUC showed high discriminatory power for LV-twist to distinguish between myocardial fibrosis or no fibrosis (AUC 0.99, 95 % CI 0.99–1.0, p < 0.001, Fig. 3); with an optimal cut-off value of 18.5 for LV-Twist as the lower limit at which to detect patients with fibrosis. Youden index determined sensitivity and specificity to be 98 and 100 %, respectively. The AUC also showed high discriminatory power for GLS and E/Em to distinguish between myocardial fibrosis or no fibrosis (GLS: AUC 1.0, 95 % CI 1.0–1.0, p < 0.001; E/Em: AUC 0.99, 95 % CI 0.99–1.0, p < 0.001) (Fig. 3); with an optimal cut-off value of -14.7 for GLS and 13.4 for E/Em as the lower limit at which to detect patients with fibrosis. Youden index determined sensitivity and specificity to be 100 and 100 % for GLS, and 98 and 96.3 % for E/Em (Table 4).

In the present study, left ventricular myocardial fibrosis was detected in 49 HCM patients (65.5 %), and LV twist and apical rotation were significantly higher in the fibrosis subgroup than in the non-fibrosis subgroup. This may be the result of subendocardial ischaemia with dysfunction of subendocardial fibres causing rotation of the apex in a clockwise direction [15]. Microvascular myocardial ischemia, abnormal calcium handling and mechanical dyssynchrony are other possible mechanisms that could explain abnormalities of active relaxation, while fibrosis and myocardial hypertrophy may increase passive LV stiffness [27]. In addition, results of that study showed that the amount of segmental LGE correlated strongly with segmental wall thickening measured by MRI. Similarly, the present study showed that peak LV twist in fibrosis subgroups was significantly higher than in non-fibrosis subgroups, which suggests that LV twist, as well as LV rotation, is also affected by the extent of myocardial fibrosis.

We also found differences between groups in the left ventricular torsion angle, which was significantly larger in the fibrosis group than in the non-fibrosis group. This may be ascribed to subendocardial myocardial hypoperfusion due to myocardial fibrosis, affecting endomyocardial function and reducing negative torsional deformation. In addition, the apical horizontal rotation angle was markedly larger than that in the non-fibrosis group, but the base rotation angle was comparable between the two groups. Thus, the overall left ventricular rotation angle in the fibrosis group increased significantly compared to that in the non-fibrosis patients. Popovic et al. [28] applied STE to analyze the relationship between the left ventricular longitudinal strain and the number of myocardial fibrotic segments. Their results showed that the left ventricular longitudinal strain was negatively associated with the number of myocardial fibrotic segments. This suggests that the larger the area of myocardial fibrosis, the smaller the left ventricular longitudinal strain. In addition, our findings also revealed that the ventricular septum in the fibrosis group was thicker than that in non-fibrosis group. It appears that the thicker the ventricular wall, the more severe is the ischemia of myocardial microcirculation; in turn, myocardial microcirculation ischemia may induce myocyte apoptosis and deposition of collagens, which facilitates myocardial fibrosis.

In the present study, the E/Em and left atrial volume index in the fibrosis group were significantly higher than those in the non-fibrosis group. This may be ascribed to the myocardial interstitial fibrosis and the subsequent increase in ventricular wall stiffness and ventricular wall tension. In addition, the reduction in myocardial relaxation capability and compliance may have resulted in restricted ventricular filling in the early diastolic phase, with an increase in left atrial pressure and subsequent elevation in left atrial volume and E/Em. Additionally, the anteroposterior diameter in the left atrial systolic phase was comparable between the fibrosis group and non-fibrosis group, but marked difference was observed in left atrial volume index. This implies that the frequently used left atrial anteroposterior diameter may not accurately reflect the actual size of the left atrium.

Results of the present study revealed no differences in LVEF between the two subgroups. However, we observed that patients with fibrosis demonstrated larger left atrial volumes, LAVi and diastolic dysfunction as measured by E/A-ratio and tissue Doppler-derived methods (i.e., E/Em-ratio) as early signs for LV pathology and LV diastolic dysfunction. These data agree with results of previous studies showing early onset of markers of diastolic dysfunction in HCM patients [21, 29]. As stated above, several other mechanisms are potential causes of this functional impairment, including microvascular myocardial ischemia, abnormal calcium handling and mechanical dyssynchrony [27]. Recently, Nagakura et al. [30] evaluated LV dyssynchrony in HCM patients, finding that although LVH is not always associated with LV dyssynchrony, greater reductions in regional strain and severe dyssynchrony may be responsible in part for the adverse outcomes frequently occurring in HCM. Findings from that study showed that even when global LV systolic function is preserved, LV dyssynchrony is characteristic in HCM; genesis is typically at the apex and is believed to be due to heterogeneous myocardial wall thickening and relaxation. In an earlier study [31], application of STE revealed that deterioration of LV strain and twist mechanics may be due to a high LV outflow gradient (LVOT), since LVOT peak velocity was associated with LV mid-rotation and LV untwist in HCM patients compared to normal controls. Such evidence suggests that myocardial fibrosis appears to contribute significantly to the progressive deterioration of LV diastolic function.

In the present study, the left ventricular torsion angle was closely related to the number of myocardial fibrotic segments, suggesting that LV twist analysis has good sensitivity and specificity in reflecting the severity of myocardial fibrosis in HCM patients. The area under the ROC curve was 0.814 using LV twist to detect myocardial fibrosis (95 % CI 0.625–0.890), which indicates that LV twist can sensitively and specifically reflect the degree of myocardial fibrosis in HCM patients. Our results are consistent with those of Saito et al. [32], who demonstrated that 2D global strain could identify subclinical global systolic dysfunction even though LV chamber function was normal in HCM subjects. Those authors concluded that global 2D strain was an important parameter of myocardial fibrosis in HCM patients. The present study also showed that the cutoffpoint for predicting myocardial fibrosis using LV twist was 18.22°, showing it to be a convenient and rapid quantitative parameter for clinical treatment and prognosis. Essentially, the larger the left ventricular torsion angle, the more severe is the myocardial fibrosis in HCM patients.

The present study has several limitations. First, the number of patients in the study was small, which results in relatively weak statistical power. Secondly, LV twist measured by 2D speckle tracking was not actually global twist because LV twist was measured from only three short-axis views (at the mitral valve and apical level). Since the distribution of LV twist and hypertrophy in the left ventricular is not uniform in HCM, it is possible that LV twist did not completely reflect the extent of LV myocardial fibrosis. The further development of three-dimensional (3D) speckle tracking might resolve this issue, as suggested by other authors [33]. Thirdly, the acquisition of orbicular LV short-axis images from base to apex is often technically difficult and may be settled by continued advances in 3D speckle tracking as previously suggested [34]. Fourthly, all the systolic parameters (apical rotation, basal rotation, LV twist) were heart-rate-independent because we measured peak values and did not measure timing. Finally, the ROI width at the hypertrophic segment might not cover the entire wall, because we adjusted the ROI to the thinner side of the myocardial wall. However, in the present study, LV twist did not depend on the LGE location in the myocardial wall. The development of software that can precisely trace the epicardial border might resolve this issue [33]. More clinical studies with larger patient populations are needed to confirm the validity of LV twist as a marker for myocardial fibrosis and as a predictor of major cardiac events. Future studies with larger sample-size will include patient follow-up to further confirm the relationship between left ventricular torsion angle and the severity of myocardial fibrosis. We would aim to elucidate whether the left ventricular torsion angle may be used to evaluate myocardial fibrosis or is independent of myocardial fibrosis, and whether it may predict prognosis for HCM patients. If so, LV twist analysis may serve as a simple, convenient indicator for clinical risk stratification and evaluation of therapeutic efficacy.