Detection of Late Gadolinium Enhancement on Cardiovascular Magnetic Resonance by Global Longitudinal Strain Using Speckle-Tracking Echocardiography in Patients with Nonischemic Cardiomyopathy

The presence of late gadolinium enhancement (LGE) using cardiovascular magnetic resonance (CMR), suggesting myocardial brosis is useful as a prognostic index in patients with nonischemic cardiomyopathy (NICM). The present study aimed to investigate whether left ventricular (LV) global longitudinal strain (GLS) using speckle-tracking echocardiography (STE) can be used as a surrogate marker for the detection of CMR-LGE in patients with NICM.

Left ventricular (LV) global longitudinal strain (GLS) using speckle-tracking echocardiography (STE) is reportedly a prognostic index in patients with reduced LV ejection fraction (LVEF), including NICM [11][12][13][14]. The presence of CMR-LGE may be an important factor associated with severely impaired STE-GLS. STE-GLS may be used as a surrogate marker for the detection of CMR-LGE in patients with NICM.
However, the signi cance of STE-GLS for detecting CMR-LGE has not yet been examined in patients with NICM. Therefore, we investigated whether STE-GLS can be used as surrogate marker for detecting CMR-LGE in patients with NICM.

Study patients
Using two-dimensional (2D) echocardiography, we retrospectively enrolled 50 patients with NICM having LVEF of < 50% in whom both STE and CMR were performed between January 2014 and June 2017 and who did not t the exclusion criteria. All patients had experienced chronic cardiac failure of at least 12 months, with the typical onset symptoms of this condition, including gradually progressive breathlessness, fatigue, and palpitation. None of the patients in the present study exhibited clinical symptoms or signs of ongoing myocarditis. The presence of severe coronary artery disease (CAD; > 50% diameter luminal stenosis in any coronary artery) was ruled out in all patients using coronary angiography. Moreover, patients with cardiogenic shock, unstable hemodynamic status, severe renal dysfunction (glomerular ltration rate < 30 ml/min/1.73 m 2 ), CMR contraindications (e.g., metal implants), moderate or severe mitral and aortic valve regurgitation or stenosis [15,16], hypertrophic cardiomyopathy, any in ltrative heart disease, atrial brillation, and postcardiac operation were carefully excluded. The present study was approved by the Committee for the Protection of Human Subjects in Research at Wakayama Medical University.

Echocardiographic measurements
Standard transthoracic echocardiographic examinations were performed in all patients using a Vivid E9 or Vivid 7 digital ultrasound system (GE Medical Systems, Horten, Norway). Echocardiography was performed within 1 week of performing CMR. Images were recorded with > 40 frames per second. Two cardiac cycles were stored in cineloop format for o ine analysis. LV end-diastolic volume index (LVEDVI), LV end-systolic volume index (LVESVI), LVEF, and left atrial volume index (LAVI) were measured using biplane Simpson's rule from the apical four-and two-chamber views, according to the criteria established by the American Society of Echocardiography [17]. The E and A waves were measured based on the mitral in ow pro le, assessed in the apical four-chamber view using pulsed-wave Doppler echocardiography, with the sample volume placed at the tips of mitral lea ets during diastole. The e' velocity from the septal and lateral mitral valve annuli was measured in the apical four-chamber view using Doppler tissue imaging of the mitral annulus. The systolic transtricuspid pressure gradient (TR-PG) was calculated from the maximal continuous-wave Doppler velocity of the tricuspid regurgitant jet, calculated using the modi ed Bernoulli equation. STE-GLS was determined using 2D speckle-tracking strain analysis using an EchoPAC version 113 workstation (GE Medical Systems, Horten, Norway), as described previously [18][19][20]. This speckletracking software tracks the frame-to-frame movement of speckles (natural acoustic markers) in standard 2D echocardiographic images. The percentage change in length/initial length of the speckle pattern over the cardiac cycle was calculated as the longitudinal strain.
Two-dimensional speckle-tracking strain analysis was performed using standard 2D images of the apical two-chamber, four-chamber, and long-axis view by a certi ed and experienced cardiologist blinded to the CMR data. After de ning the mitral annulus and apex at the end-systolic frame in each view, the software automatically traced the endocardial border, mid-myocardial layer, and epicardial border, including the entire myocardium. The width of the region of interest (ROI) was manually adjusted to ensure accurate tracking of the myocardial wall. The software automatically tracked speckles throughout the cardiac cycle; it accepted segments of good tracking quality and rejected poorly tracked segments. The tracking algorithm facilitates further manual adjustment of ROI for ensuring that all myocardial segments were included throughout the cardiac cycle. After completion of 2D speckle tracking in the three apical views, the results of the LV longitudinal strain analysis were automatically combined in a single bull's eye summary provided the peak systolic longitudinal strain for each LV segment, with the mean peak systolic longitudinal strain value for each view and mean global longitudinal peak systolic strain value for the entire left ventricle (STE-GLS).
A total of 15 patients was randomly selected for assessment of intra-and interobserver reliability of speckle-tracking echocardiographic measurements. The intraobserver analysis was performed by the same observer at two different time points; the observer was blinded to the clinical information and investigation results of the other time point. The interobserver analysis was performed by two observers who were blinded to the clinical information and investigation results. We evaluated the intra-and interobserver reliability of speckle-tracking echocardiographic measurements using intraclass correlation coe cient (ICC).

CMR
All CMR examinations were performed using a 1.5-T clinical scanner (Intera Achieva; Philips Medical Systems, Best, the Netherlands), equipped with a 32-element cardiac phased-array coil for signal reception, as previously described. During the examination, patients were continuously monitored using single-lead electrocardiography, repeated blood pressure measurements, and pulse oximetry. With the patient in the supine position, contiguous short-axis cine images encompassing the left ventricle from base to apex were acquired using a standard steady-state free precession sequence. LGE imaging covering the entire ventricle was performed 10-15 min after the intravenous injection of 0.1 mmol/kg gadolinium diethylenetriamine penta-acetic acid (Gd-DTPA; Magnevist, Schering, Berlin, Germany). We used a three-dimensional inversion recovery turbo gradient echo sequence, and images were obtained during an end-expiratory breath hold. Scan parameters were as follows: TR, 4.1 ms; TE, 1.25 ms; ip angle, 15°; FOV, 350 × 350 mm; partial echo; matrix, 224 × 256; and spatial resolution, 1.56 × 2.24 × 10 mm 3 reconstructed to 0.68 × 0.68 × 5 mm 3 . The inversion time was adjusted to nullify the signal from viable myocardium [21].
All analyses were performed by consensus of the certi ed and experienced cardiologists blinded to the echocardiographic data on an o ine workstation (View Forum, Philips Medical System). According to the previous studies [22,23], the area of LGE was de ned as the area with a signal intensity of 5 SDs above the mean signal obtained in the normal myocardium on LGE images.

Statistical analysis
All statistical analyses were conducted using JMP Pro 13 (SAS Institute Inc., Cary, NC, USA). Categorical variables, presented as frequency counts and percentages, were compared using the Fsher's exact test.
Distribution of the continuous data was assessed using the Shapiro-Wilk test. Normally distributed variables were analyzed using the Student's t-test, whereas abnormally distributed variables were analyzed using the Wilcoxon test. Receiver operating characteristic (ROC) curve analysis was performed to establish STE-GLS as a predictor of the presence of CMR-LGE. The resulting sensitivity, speci city, and the area under the curve (AUC) were calculated. The best threshold value was determined by the maximum sum of sensitivity and speci city. A p value of < 0.05 was considered statistically signi cant.

Patient clinical characteristics
The presence, or absence of CMR-LGE was successfully analyzed in all patients with NICM. However, the assessment of STE-GLS in 9 (18%) of 50 patients was di cult because of poor tracking of the myocardium in > 3 LV segments. Following their exclusion, the nal study population comprised 41 patients, for whom both CMR-LGE and STE-GLS analyses were successful. We divided the nal study populations into two groups: 18 patients with CMR-LGE (Group A) and 23 without CMR-LGE (Group B). Patients' clinical characteristics of the nal cohort are summarized in Table 1. Hypertension and diabetes mellitus were present in 23 (56%) and 5 (12%) patients, respectively. Most patients were treated with angiotensin-converting enzyme inhibitors or angiotensin receptor blockers (98%), beta blockers (78%), and furosemide (68%). No signi cant differences in patient clinical characteristics were observed between both groups.

Comparison of echocardiographic ndings between both groups
The echocardiographic parameters of the nal study population are summarized in Table 2 vs. 49 ± 13 mL/m 2 , and 29 ± 10 vs. 24 ± 7 mmHg, respectively).

Discussion
In the present study, we investigated the signi cance of STE-GLS for the detection of CMR-LGE in patients with NICM. The major study ndings were as follows: 1) STE-GLS was worse in patients with NICM who have CMR-LGE than those without CMR-LGE, whereas no signi cant differences were observed in conventional echocardiographic measurements, such as LVESVI, LVEDVI and LVEF, between both groups. 2) STE-GLS of − 7.9% was found to be the best threshold value for identifying CMR-LGE, and the ROCrelated AUC of 0.74 con rmed a moderate performance of the test for the detection of LGE. To the best of our knowledge, this is the rst study investigating the threshold value of STE-GLS for the detection of CMR-LGE in patients with NICM.
LGE-determined myocardial brosis and prognosis in patients with NICM Areas of CMR-LGE correspond to those of myocardial brosis on histology, and approximately 30% of patients with NICM have a characteristic linear CMR-LGE in the mid-wall of the septum [3,24]. Patients with NICM who have CMR-LGE in the present study exhibited typical mid-wall LGE. Myocardial brosis provides a substrate for ventricular re-entrant arrhythmia [6] and is independently associated with an increased risk of mortality and cardiac failure morbidity in patients with NICM [4, 5, 7-10]. Moreover, its presence and extent in LV myocardium assessed by CMR-LGE in patients with NICM, substantially determines the likelihood of LV reverse remodeling in response to pharmacological therapy [25,26] and cardiac resynchronization therapy [27].

Detection of CMR-LGE in patients with NICM using STE-GLS
STE is emerging as a novel index for the assessment of LV mechanics via the quanti cation of active myocardial longitudinal deformation (STE-GLS). Severely impaired STE-GLS is reportedly a prognostic index in patients with reduced LVEF, including those with NICM [11][12][13][14]. However, the reason for severely impaired STE-GLS being a prognostic predictor has not yet been clearly explained. We hypothesized that myocardial brosis contributes to severely impaired STE-GLS. The best threshold value of STE-GLS of − 7.9% for identifying CMR-LGE, which suggests myocardial brosis, corresponds to the STE-GLS threshold values for identifying worse prognosis reported in previous studies [11,12]. This nding suggests that severely impaired STE-GLS as a prognostic predictor is associated with the presence of myocardial brosis.
LVEF was generally reported as a strong prognostic predictor, however there was no signi cant difference in LVEF between the patients with CMR-LGE and those without CMR-LGE. Moreover, according to the ROC curve analysis for identifying CMR-LGE by LVEF, the AUC of 0.53 showed a low performance of the test for the detection of CMR-LGE. These results suggest that it is di cult to identify CMR-LGE by measurement of LVEF, and that LVEF cannot be used as a surrogate marker for the detection of CMR-LGE in patients with NICM.
In this study, there were six patients with severely impaired STE-GLS (STE-GLS > -7.9%) in whom CMR-LGE was not found. CMR-LGE cannot depict diffuse myocardial brosis because CMR-LGE is based on the null point method. In patients without CMR-LGE, there might be some patients with diffuse myocardial brosis in whom STE-GLS might be severely impaired.

Advantages of STE-GLS in NICM compared with CMR
First, STE-GLS can be used in patients with NICM having renal dysfunction. The use of gadolinium contrast is limited in patients with renal dysfunction in CMR-LGE examinations. Therefore, another modality that does not require contrast agents for identifying myocardial brosis is required. Noninvasive STE-GLS without contrast agents may be used as an ideal modality for detecting myocardial brosis. Second, it is possible to estimate STE-GLS in patients in whom CMR-LGE examinations could not be completed because of the patient's intolerance to staying in the CMR examination room or di culty in holding their breath during CMR-LGE examinations. Finally, the present results contribute to the daily clinical practices associated with the management of NICM, even in the institutes where CMR is not available. STE-GLS, which is an established index and is measured rather straightforwardly in echocardiographic laboratories, would be helpful for decision-making in the management of NICM.

Study limitations
This study has several limitations. First, this was a retrospective observational study, with a relatively small sample size. A prospective multicenter study with large sample size should be conducted in future. Second, STE-GLS measurements are dependent on echocardiographic image quality. Accordingly, STE-GLS measurement was unsuccessful in 18% of the initial study cohort. ROI of the speckle-tracking software may be challenging to t strain analyses in patients with NICM patients owing to the thin wall, which could have in uenced our data. In addition, the de nition of successful measurements of STE-GLS may be too strict in this study. In some previous studies evaluating STE-GLS, the overall STE-GLS was calculated in only two of three apical images when only three apical images could be assessed because of poor tracking of ROI [13,28,29]. This alternative method should be considered for the clinical application of STE-GLS in patients with NICM.