Objective Late gadolinium enhancement (LGE) on cardiac magnetic resonance (CMR) imaging has been reported to be associated with unfavourable outcomes; however, few studies have addressed the prognostic value of left ventricular (LV) deformation parameter indicated by global longitudinal strain (GLS) in two-dimensional speckle-tracking (2DST) echocardiography in patients with non-ischaemic dilated cardiomyopathy (DCM). This study aims to investigate whether the combination of GLS and LGE is useful in stratifying the risk in patients with DCM.
Methods We studied 179 consecutive symptomatic patients with DCM (age, 61±15 years; 121 males; left ventricular ejection fraction (LVEF) 33%±9%; New York Heart Association (NYHA) class II: n=71, III: n=107, IV: n=1) who underwent CMR and echocardiography with conventional assessment and 2DST analysis.
Results There were 40 rehospitalisations for heart failure, including 7 cardiac deaths and 2 implantations of LV assist device during follow-up (3.8±2.5 years). Univariable Cox proportional hazard regression analysis showed that NYHA class, blood pressure, B-type natriuretic peptide, LV end-diastolic and end-systolic volumes, LVEF, left atrium volume, GLS and LGE were significantly associated with long-term outcome. Multivariable analysis revealed that GLS and LGE were independently associated with long-term outcome (p<0.05, both). In additional analyses, we found independent associations between GLS and LV reverse remodelling after the optimal medical therapy, and between LGE and life-threatening arrhythmias (p<0.05, both).
Conclusion Combining GLS and LGE could be useful for risk stratification and prognostic assessment in patients with DCM.
Statistics from Altmetric.com
Non-ischaemic dilated cardiomyopathy (DCM) is associated with significant morbidity and premature mortality due to progressive heart failure (HF) and sudden cardiac death.1 Although the prognosis of patients with DCM has improved owing to advances in medical therapy and the introduction of device therapy, the 5-year mortality rate remains as high as 20%.1 This reflects the difficulty in assessing individual risk in patients with DCM. Therefore, the best risk-stratification tools would be needed to allow earlier intervention in high-risk patients and improve both the quality of life and survival of patients with DCM.
Left ventricular (LV) myocardial contractility is a strong outcome predictor, with a major impact on medical decisions.2 ,3 The recently developed two-dimensional speckle-tracking (2DST) echocardiography is an innovative method providing multidimensional myocardial mechanics that involves longitudinal, radial and circumferential motion.4 It has been reported that global longitudinal strain (GLS) detected by using 2DST echocardiography offers additional benefit over left ventricular ejection fraction (LVEF) and other markers of longitudinal motion in predicting outcome in patients with HF;5–7 however, few studies have addressed this issue in patients with DCM. On the other hand, detection of myocardial fibrosis evaluated with late gadolinium enhancement (LGE) on cardiac magnetic resonance (CMR) imaging may assist with the risk stratification in DCM.8 The presence of myocardial fibrosis shows an important mechanism for the occurrence of arrhythmias and poor response to treatments of HF.9–11
Accordingly, the purpose of this study is to investigate whether GLS by 2DST echocardiography combined with LGE on CMR is useful for the risk stratification and prognostic assessment of patients with DCM.
We performed a retrospective, longitudinal study in a cohort of 459 consecutive patients with DCM who were referred to our hospital between January 2005 and December 2013. Of all patients, 262 with any contraindications to CMR (significant renal dysfunction (glomerular filtration rate of ≤30 mL/min/1.73 m2), or implanted devices such as pacemakers and/or defibrillators), significant organic valvular disease, atrial fibrillation or flutter, hypertensive heart disease, recent myocarditis, arrhythmogenic right ventricular cardiomyopathy, myocardial non-compaction, congenital heart disease or tachycardia-induced cardiomyopathy were excluded. The remaining 197 patients underwent echocardiography and CMR (figure 1).
The diagnosis of DCM was made according to the criteria of the World Health Organization/International Society and Federation of Cardiology.12 Invasive coronary angiography was performed in all patients to exclude significant coronary artery stenosis. Baseline clinical variables included New York Heart Association (NYHA) functional class, serum creatinine level, plasma B-type natriuretic peptide (BNP) level and prescribed medication for HF (table 1). The investigation conforms with the principles outlined in the Declaration of Helsinki. For this retrospective analysis of clinically acquired data, the institutional review board waived the need for patients' written informed consent.
CMR image acquisition
CMR imaging was performed using a 1.5-T Intera Achieva scanner (Philips Medical Systems, Best, the Netherlands) and a standardised protocol. Ten minutes after intravenous injection of 0.1 mmol/kg gadolinium-DTPA (diethylenetriamine penta-acetic acid) (Schering AG, Berlin, Germany), LGE images were obtained by using an inversion-recovery gradient echo sequence in identical long-axis and short-axis planes. Inversion times were optimised to null normal myocardium, and images were repeated in two separate phase-encoding directions to exclude artefacts.13 The LGE was only deemed to be present when the area of signal enhancement could be seen in both phase-swapped images and in a cross-cut long-axis image obtained by the independent observers (figure 2).
Comprehensive transthoracic echocardiography was performed by highly experienced research sonographers by using commercially available Aplio (Toshiba Medical Systems, Tokyo, Japan).
Two-dimensional and colour Doppler echocardiography were performed in standard parasternal and apical views. LV end-diastolic volume, end-systolic volume and ejection fraction (EF) were measured using a modified Simpson method. Left atrial volume measured by the prolate ellipse method.14 Mitral regurgitation was graded semiquantitatively as mild, moderate or severe according to current recommendations.15
LV strain measurements
The LV GLS, global circumferential strain (GCS) and global radial strain (GRS) values were obtained by using 2DST software (Toshiba Medical Systems). Images in three cardiac cycles were acquired from the parasternal short-axis view at the midpapillary level for GCS and GRS, and from apical four-chamber and two-chamber views for GLS. Peak strain was defined as the peak negative value on the strain curve during the entire cardiac cycle for GLS and GCS and as the peak positive value on the strain curve for GRS (figure 2). The interobserver and intraobserver variability for LV strain values were studied in a group of 30 randomly selected subjects by one observer, repeated twice, and by two observers who were unaware of each other's measurements and of the study time point.
Follow-up and end points
All patients were followed for non-fatal events by telephone and medical record review; however, two were lost. Cardiac deaths, rehospitalisation for HF, cardiac transplantation or implantation of an LV assist device were defined as adverse cardiac events building the primary composite end point. The second composite end point was sudden cardiac death and life-threatening ventricular arrhythmia events. Life-threatening arrhythmias were defined as ventricular fibrillation and sustained ventricular tachycardia, which lasts more than 30 s, at R–R intervals of ≤400 ms. LV reverse remodelling was assessed for midterm outcome and was defined as an increase in LVEF ≥10 U, combined with a decrease in left ventricular end-diastolic volume (LVEDV) ≥10% by echocardiography after the optical medical therapy for 12 months.16
Data are expressed as mean±standard deviation for continuous variables and as proportions for categorical variables. Data from two patient groups with and without cardiac events were compared by using Student's t-test for continuous variables, the Wilcoxon-Mann-Whitney test for non-normally distributed continuous variables and the χ2 test for categorical variables. A univariable and multivariable Cox proportional hazards model was used for to investigate which prognostic factors identified by using univariable analysis were significantly associated with cardiac events. Basically, we screened variables considering statistical significance in the univariable analysis. Screened variables were checked for mutual correlation by scatter plots to avoid colinearity in the multivariable model. When suggestive correlation among variables was detected, we selected one of them based on clinical relevance or interpretability. Event-free survival curves were calculated according to the Kaplan-Meier method, and the comparison among curves was carried out with the log-rank test. Univariable and multivariable logistic regression analyses were used to investigate the association of baseline variables with LV reverse remodelling after the optimal medical therapy. Two-tailed p values of <0.05 were considered statistically significant. We used MedCalc V.15 for all the analysis.
Of 197 patients with a diagnosis of DCM, eight with inappropriate CMR images, eight with inappropriate echocardiography images and two who were lost to follow-up were excluded, resulting in a final cohort of 179 patients (figure 1). LGE was observed in 100 (59%) of 179 patients. Of them, 81 patients (81%) had midwall LGE, 16 (16%) diffuse LGE, two patchy focal LGE and one epicardial LGE.17 Significant difference was found in GLS value between patients with and without LGE (−8.3%±3.7% vs −9.8%±4.9%, p<0.05; see online supplementary figure S1).
Baseline parameters associated with adverse cardiac events
During the mean follow-up duration of 3.8±2.5 years (IQR, 464–1984 days), adverse cardiac events occurred in 40 patients (22%), including seven cardiac deaths, two implantations of LV assist device and 31 rehospitalisations including 13 life-threatening arrhythmic events. The clinical and echocardiographic parameters are summarised in table 1.
A higher incidence in cardiac events was observed in patients with LGE compared with patients without LGE (29% vs 14%; HR 3.01; 95% CI 1.48 to 6.18; p=0.002). The presence of midwall LGE was significantly associated with adverse outcome among patients with LGE (p<0.01).
We evaluated the following parameters to determine the association of their baseline value with adverse cardiac events using Cox proportional hazard regression analyses: age, sex, NYHA class III or IV, systolic and diastolic blood pressure, BNP level, LVEDV, left ventricular end-systolic volume (LVESV), LVEF, left atrial volume, mitral regurgitation≥moderate, GLS and LGE. All these parameters except for age, sex and mitral regurgitation were significant univariable correlates of adverse cardiac events. On multivariable analysis, GLS and the presence of LGE were independently associated with adverse cardiac events (table 2). LVEF was not independently associated with adverse cardiac event, although LVEF had a significant correlation with GLS (r=0.57, p<0.01; see online supplementary figure S2).
Dividing all 179 patients into two groups with the median value (−8.3%) of GLS, log-rank test revealed reduced GLS was significantly associated with adverse cardiac events (p<0.01; figure 3A). Dividing all 179 patients into four groups with the median of GLS (−8.3%) and the presence of LGE, log-rank test showed preserved GLS (≤−8.3%) was associated with a favourable outcome independently of LGE, and furthermore, even in patients with reduced GLS (>−8.3%), the presence of LGE was significantly associated with a worse outcome (p<0.05; figure 3B). Dividing all patients with the optimal cut-off from ROC analyses (−8.0%) of GLS, log-rank test also showed reduced GLS was significantly associated with adverse cardiac events (p<0.01).
Predictors of life-threatening arrhythmia events
In all 179 patients with DCM, life-threatening arrhythmia events occurred in 13 patients (7%) including 13 ventricular tachycardia events and six ventricular fibrillation events; however, sudden cardiac death was not found in the follow-up period.
Univariable Cox proportional hazards regression analysis revealed that the presence of LGE, LVEDV and LVESV were significantly associated with life-threatening arrhythmic events (p<0.001, p=0.004 and p=0.02, respectively), although GLS and LV reverse remodelling were not (p=0.30 and p=0.39, respectively). In multivariable Cox proportional hazards regression analysis, only the presence of LGE was independently associated with life-threatening arrhythmic events (HR 2.41, 95% CI 1.01 to 5.36, p<0.05). GLS was still not associated with life-threatening arrhythmic events even after adjustment for LVEF and LGE in this cohort, although GLS has been reported to be significantly related to life-threatening arrhythmic events in patients with acute myocardial infarction.7 ,18
Kaplan-Meier curves also showed that patients with LGE had a higher probability of life-threatening arrhythmic events (p<0.01; figure 3C). No life-threatening ventricular arrhythmia events occurred in patients without LGE, in contrast with all 13 events found in patients with LGE (figure 3C).
Predictors of LV reverse remodelling
Of all 179 patients with DCM, 50 patients had not undergone 1-year follow-up echocardiography. Of the remaining 129 patients, 77 patients (60%) had a significant LV reverse remodelling. LV reverse remodelling was significantly associated with the absence of LGE, left atrial volume and GLS at baseline (all p<0.05), and strongly associated with a favourable long-term outcome (p<0.01; figure 4A). There were no differences in medical therapy between patients with and without LV reverse remodelling. Multivariable logistic analysis showed that only GLS was independently associated with LV reverse remodelling (OR 0.76, 95% CI 0.67 to 0.86). Dividing all patients into four groups with the median of GLS (−8.5%) and LGE, we found a significant difference in the proportion of patients with LV reverse remodelling (p<0.01; figure 4B). In patients with preserved GLS (≤−8.5%), the absence of LGE was significantly associated with a higher proportion of patients with LV reverse remodelling after the optimal medical therapy (p<0.05).
Interobserver and intraobserver variability
The coefficients of variation of interobserver variability for GRS, GCS and GLS were 17%, 12% and 8%, respectively. The coefficients of variation of intraobserver variability for GRS, GCS and GLS were 5%, 4% and 3%, respectively.
This study is, as far as the authors are aware, the first to demonstrate the association between GLS by 2DST echocardiography combined with LGE on CMR and the prognosis and the reverse remodelling after medical therapy in patients with DCM. The main results are as follows: (1) GLS by 2DST echocardiography is independently related to the primary composite end point along with LGE, and these two markers have prognostic value over other well-established clinical and echocardiographic parameters; (2) GLS is strongly associated with LV reverse remodelling after the optimal medical therapy in patients with DCM and (3) LGE is independently associated with life-threatening arrhythmia events, and decreases the possibility of LV reverse remodelling in patients who had DCM with preserved GLS. Thus, our findings have important implications for the clinical management of symptomatic patients with DCM, and for understanding the relation among myocardial fibrosis according to LGE, LV function according to GLS and adverse cardiac events.
LGE on CMR
In this study, LGE was found in 56% of patients with DCM, a figure comparable to that of recent meta-analysis of nine studies yielded a total of 1488 patients that reported a range of prevalence of 17%–71%,19 and similar to that of pathological studies on DCM hearts in which macroscopic LV scarring was detected in up to half of the cases.20
Assomull et al13 reported that in 101 patients with DCM, myocardial fibrosis determined by using CMR is a predictor of the combined end point of all-cause mortality and cardiovascular hospitalisation, which is independent of ventricular remodelling. Gulati et al8 also documented the prognostic significance of myocardial fibrosis in 472 patients with DCM. However, neither of these studies included QRS duration, echocardiographic indices of LV deformation or serum BNP level, despite their reported prognostic value. In this study, we included a variety of novel and established indices of long-term outcome after an optimal medical therapy, and demonstrated that LGE at initial presentation was independently associated with adverse cardiac events (p<0.05) and was the only independent indicator of life-threatening arrhythmia events (p=0.04) even accounting for these clinical and biochemical variables.
LV myocardial deformation assessed by using GLS
2DST echocardiography is emerging as a novel technique to allow the assessment of LV mechanics through the quantification of active myocardial deformation in multiple directions (radial, circumferential and longitudinal).21 ,22 Particularly, the measurement of LV GLS, which is a measure of the active shortening of the LV in the longitudinal direction, is more reproducible than LVEF and wall motion score index, and does not rely on geometric assumptions.23 ,24
Kalam et al25 showed with 16 published articles (n=5721 adults) that there is a strong evidence of the prognostic value of GLS, which appears to have superior prognostic value to LVEF for predicting major adverse cardiac events. Mignot et al26 reported, with a multicentre study including 147 patients with HF and LVEF ≤45%, that strain assessment is highly feasible and reliable in patients with LV dysfunction and allows for the cardiovascular risk stratification in patients with HF with greater accuracy than LVEF. Sengeløv et al27 also demonstrated with 1065 HF patients with reduced EF that GLS is a superior predictor compared with all other echocardiographic parameters according to all-cause mortality. They gave one possible explanation stemming from the technique itself, which directly investigates the local myocardial contraction, segment-by-segment, before calculating the strain average. In contrast, other validated parameters measure close or remote consequences of contraction impairment rather than direct myocardial dysfunction. A better link to prognosis is obtained by targeting the origin of contraction abnormalities by using 2DST strain. Buss et al28 prospectively studied 210 patients who had DCM with standard CMR including measurement of LGE during the median follow-up period of 5.3 years, and demonstrated that LV feature tracking longitudinal strain assessed with CMR is an independent predictor of survival in patients with DCM. These strongly support our results, although feature tracking longitudinal strain based on CMR have some problems to overcome regarding the agreement with longitudinal strain based on tagging CMR, interobserver reproducibility and temporal resolution.29 ,30
In this study, GCS and GRS were not significantly associated with outcome. Although GRS has few evidence as a prognostic predictor, GCS has been reported as one of remarkable predictors in patients with HF. Cho et al31 demonstrated that GCS is a powerful predictor of cardiac events and appears to be a better parameter than GLS in patients with acute HF. One of reasons is the fact that GCS and GRS were measured only at the midpapillary level in the short-axis view, so it may have underestimated overall LV dysfunction. Furthermore, the lower reproducibility of GCS and GRS than of GLS might explain its limited association with outcome. Cho et al31 also allowed the lower reproducibility of GCS in their article. Several other studies support the superiority of GLS to GCS and GRS as an indicator of adverse events or LV dysfunction.32 ,33
LV reverse remodelling after the optimal medication therapy
Merlo et al34 reported that LV reverse remodelling allowed a more accurate long-term prognostic stratification of patients with DCM receiving optimal medical therapy. This agrees with our result (figure 4A). Bhat et al35 studied 568 patients with a newly documented LVEF of ≤35% receiving optimal medical therapy, and reported that the LV end-systolic diameter index at diagnosis was a strong predictor of LV reverse remodelling. This supports our results that LV systolic function expressed by using GLS at baseline is significantly associated with LV reverse remodelling. However, their definitions of LV reverse remodelling as the improvement in LVEF of >35% in the follow-up period is not enough. We used the definition as an absolute increase in LVEF ≥10 U, combined with a relative decrease in LVEDV ≥10%. Masci et al16 concluded with this definition that the absence of LGE at baseline is a strong independent predictor of LV reverse remodelling at 2-year follow-up in 58 consecutive patients with DCM irrespective of the initial clinical status and the severity of ventricular dilatation and dysfunction. Our result also showed that the absence of LGE was significantly associated with LV reverse remodelling, but was inferior to GLS. For predicting LV reverse remodelling, LGE might be very useful in patients with preserved GLS but not in patients with reduced GLS (figure 4B). We supposed that GLS could be a reliable prognostic predictor in patients with DCM because GLS is significantly and independently associated with LV reverse remodelling after the optimal medical therapy.
This study is limited by the fact that it was a single centre, retrospective study and had a small number of events, which might diminish the power of the drawn statistical inference. LGE was assessed qualitatively, because a sufficient digital data for the quantitative assessment had not been saved. An additional limitation to this study is that inappropriate cases of enhanced CMR could not be included in this study; thus, our conclusions should not be extrapolated to all patients with DCM. A further study with a prospective and multicentre design is thus required to confirm our results.
In patients with DCM, we showed that LV GLS by 2DST echocardiography is independently associated with LV reverse remodelling, and that LGE is independently associated with life-threatening arrhythmias. These two markers, GLS and LGE, which have different properties, were demonstrated to be independently associated with long-term outcome, and might be useful for the risk stratification and prognostic assessment of patients with DCM.
What is already known on this subject?
The presence of late gadolinium enhancement on cardiac magnetic resonance is a significant prognostic indicator in patients with dilated cardiomyopathy.
Global longitudinal strain assessed by two-dimensional speckle-tracking echocardiography is a useful prognostic parameter in patients with heart failure.
What might this study add?
We demonstrated that the combining global longitudinal strain (GLS) by two-dimensional speckle-tracking strain echocardiography and late gadolinium enhancement (LGE) on cardiac magnetic resonance improves the risk stratification in patients with dilated cardiomyopathy.
The left ventricular reverse remodelling as short-term outcome and the adverse cardiac event as long-term outcome may be predicted with the combination of GLS and LGE.
How might this impact on clinical practice?
For the patients with preserved global longitudinal strain (GLS), the absence of late gadolinium enhancement (LGE) might more definitely predict left ventricular (LV) reverse remodelling as a short-term outcome. For the patients with reduced GLS, the presence of LGE might predict a worse long-term outcome. Risk stratification with the cardiac function evaluated by two-dimensional speckle-tracking (2DST) strain echocardiography and the myocardial fibrosis assessed by LGE on cardiac magnetic resonance (CMR) will serve a very useful purpose in clinical situation for the patients with dilated cardiomyopathy (DCM).
In clinical situations, patients with DCM are first evaluated for cardiac function with the conventional and 2DST strain echocardiography, and then assessed for the presence of myocardial fibrosis by using LGE on CMR. It is very reasonable to stratify the patients at risk for the absence of LV reverse remodelling as short-term outcome and the adverse cardiac event as long-term outcome with the combination of GLS and LGE.
The authors greatly thank the doctors of the Department of Cardiology and the sonographers of Echo Lab at Himeji Cardiovascular Center for their help in this study.
Contributors MC and TO: conception or design of the work and drafting the article. YT, TS and KK: data collection. MC, TO, AS, KO, SY and YT: data analysis and interpretation. MC, TO and HK: critical revision of the article. YY and HK: final approval of the version to be published.
Competing interests None.
Patient consent Obtained.
Ethics approval The Institutional Review Board and the Ethics Committees of the Himeji Cardiovascular Center.
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.