Author + information
- Received May 21, 2018
- Revision received July 13, 2018
- Accepted July 19, 2018
- Published online January 16, 2019.
- Albree Tower-Rader, MDa,∗,
- Divyanshu Mohananey, MDa,∗,
- Andrew To, MDa,b,
- Harry M. Lever, MDa,
- Zoran B. Popovic, MD, PhDa and
- Milind Y. Desai, MDa,∗ ()
- aHypertrophic Cardiomyopathy Center, Cleveland Clinic, Cleveland, Ohio
- bNorthshore Hospital, Auckland, New Zealand
- ↵∗Address for correspondence:
Dr. Milind Y. Desai, Heart and Vascular Institute, Cleveland Clinic, 9500 Euclid Avenue, Desk J1-5, Cleveland, Ohio 44195.
Objectives The association of left ventricular global longitudinal strain (LV-GLS) with clinical outcomes in patients with hypertrophic cardiomyopathy (HCM) has been examined in multiple studies. The authors conducted a systematic review aimed at summarizing and critically appraising the current evidence.
Background HCM is a common genetic cardiovascular disease with an estimated prevalence of 1 in 500 patients. LV-GLS derived from speckle tracking echocardiography is a sensitive noninvasive method of assessing regional left ventricular function. Several studies have suggested association of abnormal LV-GLS with outcomes in HCM patients.
Methods A computerized literature search of all English language publications in the PubMed and EMBASE databases was made looking at all randomized and nonrandomized studies conducted on patients with HCM where association of LV-GLS with clinical outcomes was studied. We then manually searched the reference lists of included articles. The Preferred Reporting Items for Systematic reviews and Meta-Analyses statement (PRISMA) of reporting systematic reviews was used.
Results Our search yielded a total of 14 observational studies published between 2009 and 2017 with a total of 3154 patients with HCM. Eleven of the 14 studies included a composite cardiac outcome which included mortality as their primary outcome of interest and 3 of the 14 studies looked at association of LV-GLS with ventricular arrhythmias and/or implantable cardiac defibrillator discharge. We noted wide variability in inclusion, methodology, follow-up, and consequently effect estimates, which was not conducive to performing a meta-analysis. However, despite the variation, all studies revealed a degree of association of abnormal LV-GLS with poor cardiac outcomes.
Conclusions Our systematic review of more than 3000 HCM patients suggests an association of abnormal LV-GLS with adverse composite cardiac outcomes and ventricular arrhythmias.
Hypertrophic cardiomyopathy (HCM) is a common genetic cardiovascular disease with an estimated prevalence of 1 in 500 patients (1,2). The disease is characterized by myocyte hypertrophy and disarray, coronary intramural hyperplasia, and fibrosis on pathology; however, there is significant heterogeneity in phenotypes (3,4). Patients can be asymptomatic, or can present with heart failure, arrhythmias, or sudden cardiac death (SCD) (5,6). Management for patients with HCM has traditionally focused on the identification of patients at high risk for SCD with referral for primary prevention implantable cardioverter-defibrillator (ICD) placement, or referral for septal reduction, either by septal myectomy or alcohol septal ablation, for patients with obstructive HCM and symptoms refractory to medical therapy (1,2).
Left ventricular global longitudinal strain (LV-GLS) derived from speckle tracking echocardiography is a sensitive noninvasive method of assessing left ventricular (LV) function, especially in the setting of a normal LV ejection fraction (EF) (7). It is easily obtained at the time of image acquisition or may be performed offline (7). Abnormalities in LV-GLS have been shown to occur in many myocardial diseases and are often associated with a worse prognosis (7). As such, the volume of literature published regarding LV-GLS in patients with HCM and its association with adverse outcomes has increased in the past decade. This state-of-the-art systematic review assesses and seeks to critically summarize studies to date of patients with HCM who had LV-GLS assessment and were followed for long-term outcomes.
Data sources and searches
We performed a computerized literature search of all English language publications in the PubMed and EMBASE databases. We then manually searched the reference lists of included articles. This was last assessed as up-to-date on September 1, 2017 (Figure 1). Our aim was to include all studies conducted on patients with HCM where association of LV-GLS with clinical outcomes was studied.
A search was conducted using 3 sets of keywords in combination. The first set included the terms “cardiomyopathy, familial” or “cardiomyopathy, hypertrophic.” The second set included the terms “two-dimensional strain” or “echocardiogram” or “global longitudinal strain” or “ultrasonography”; and the third set included the terms “outcome” or “sudden death” or “mortality” or “arrhythmia” or “ventricular tachycardia” or “implantable cardiac defibrillator” or “congestive heart failure” or “heart failure” or “transplant.”
The Preferred Reporting Items for Systematic reviews and Meta-Analyses statement (PRISMA) of reporting systematic reviews and meta-analysis was applied to the methods for this study (8).
The following inclusion criteria were used:
1. Studies in patients with HCM, obstructive, nonobstructive or both; and
2. Studies which provided information on LV-GLS;
The following exclusion criteria were used:
1. Non–English language literature;
2. Case reports and case series;
3. Conference abstracts;
4. Studies on cardiac magnetic resonance imaging or Doppler echocardiography if information on LV-GLS (measured on echocardiography) was unavailable;
5. Studies where circumferential or radial strain was reported without information on LV-GLS; and
6. Studies where cross-sectional assessment of outcomes was done, such as 24- to 48-hour Holter monitoring without longer-term follow-up.
Data abstraction and reporting
Two authors (D.M. and A.T.) independently abstracted information, which was verified by an additional 2 authors (A.T.R. and M.Y.D.). We abstracted information on name, author, publication year, country of origin, vendor used for echocardiography and strain analysis, population studied, LV-GLS values, outcome studied, and effect size estimates. Categorical data is expressed as a percentage and continuous variables as means with standard deviations or medians with interquartile range. Effect size estimates such as odds ratio, hazard ratio (HR), and relative risk are reported along with their 95% confidence intervals (CIs) where available. We report results of receiver operator curve (ROC) analysis wherever available as area under the curve (AUC), sensitivity, and specificity along with the cutoff used to obtain these measures. Also, C-statistic and net reclassification index (NRI) are reported where available. For studies where subgroup data were combined to obtain values for the whole cohort, we combined the means and standard deviation of these groups based on formulae specified by the Cochrane Collaboration. We defined “obstructive HCM” as documented outflow tract gradient (resting or provoked) ≥30 mm Hg (1,2). By convention, negative strain values refer to shortening and thus greater deformation is represented as a more negative value (7). To avoid variability in reporting and interpretation of the studies included in this review, we have used LV-GLS % with a negative sign regardless of whether absolute value or negative integer was provided by original study. In our report, a higher absolute value (more negative) of LV-GLS is referred to as “better,” and a lower absolute value (less negative, or closer to zero) of LV-GLS if referred to as “worse.”
Our search yielded 14 observational studies published between 2009 and 2017 with 3154 HCM patients (9–22). We have divided this report into the following parts: study description, risk factors, and outcomes. Outcomes have been further subdivided on the basis of studied primary outcome into 2 categories: 1) studies which evaluated a composite cardiac outcome including mortality; and 2) studies which evaluated ventricular arrhythmias and/or ICD discharge but not mortality.
Description of included literature
Of the 14 included studies, 8 were prospective (9,10,12,13,15,18,20,21) and 6 were retrospective (11,14,16,17,19,22). Table 1 contains a detailed description of included studies. The number of patients varied from 41 to 1019. The ultrasound equipment varied between studies (General Electric, Waukesha, Wisconsin, in 10 studies; Phillips, Bothell, Washington, in 3 studies; and in 1 study, either General Electric, Phillips, or Siemens, Malvern, Pennsylvania, was used, Table 1). In 12 of 14 studies, speckle tracking was performed using proprietary software specific to the ultrasound vendor used for image acquisition. This included EchoPac (GE Medical Systems, Horten, Norway; GE Healthcare, Waukesha, Wisconsin) in 10 studies, and QLab (Phillips Medical Systems) in 2 studies. One study used a vendor-neutral software, Velocity Vector Imaging (VVI; Siemens Medical Solutions, Mountain View, California) for LV-GLS analysis (22). These 13 studies use speckle tracking in apical 2-, 3-, and 4-chamber views to determine segmental and global longitudinal strain, obtained using proprietary software and either a 16-, 17-, or 18-segment model. In all cases, this represents Lagrangian strain, not natural strain. The final study manually measured Lagrangian strain in only the apical 4-chamber view (19). Four studies did not provide information on interobserver testing and/or variability (11,13,15,17), whereas 1 study referenced previously reported inter- and intraobserver variability for their group (14,23). Among the studies reporting interobserver and intraobserver variability, several different methods were used (correlation coefficients, coefficient of variation, and standard error of measurement). Thus, depending on the method used to assess inter- and intraobserver variability, the values vary from 0.91 to 0.98 (correlation coefficient) to 4.1% to 9.5% (coefficient of variation); however, these are not directly comparable due to methodological differences. One study used standard error of measurement showing excellent inter- and intraobserver variability of 0.8% and 1.3%, respectively (22). Recent data suggest that standard error of measurement is the most useful metric of observer variability as it allows for assessment of the CIs for an index measurement, calculation of the minimum detectable difference, and for comparison between methods (24). Follow-up varied among these studies from 12 months to 9.4 years (median) (Table 1). Most of the studies were conducted in either Europe or United States. The average LV-GLS values for HCM patients included in the review varied from mean LV-GLS of −9.89 ± 2.59% to −16.5 ± 3.6% (Table 1).
Risk factors for adverse outcomes
Details of risk factors are presented in Table 2. Patients ranged from young to middle-aged with a maximum mean age of 63.3 ± 12.1 years; ≥50% were men. There was variable representation of patients with or without LV outflow tract obstruction, ranging from 12% to 100% of the patients having obstructive HCM. Although all 14 studies reported a mean LVEF >50%, only 6 studies specifically reported preserved LVEF as an inclusion criteria. Seven studies did not clearly mention presence of preserved LVEF; whereas, Reant et al. (14) (2016) noted that 3% patients had low LVEF. Although presumably absent at the time of image acquisition, approximately half of the studies included patients with paroxysmal atrial fibrillation. Data regarding baseline New York Heart Association (NYHA) functional class was available for 10 studies. The majority of patients were NYHA functional class I or II, although patients who were NYHA functional class III or IV were included in 7 of 10 studies. There was significant variability in reporting the frequency of known risk factors for SCD. For instance, while Saito et al. (12) included 8% of patients with family history of SCD, >60% of the cohort in Candan et al. (18) study had this history. History of septal reduction therapy was specifically referenced in 7 studies, and patients with this history were excluded in 4 of 7 studies (Table 2).
Composite cardiac outcome including mortality
Eleven studies included death with their primary composite outcome. Although the definition of this outcome varied widely across the included literature, most studies used a combination of either cardiac or all-cause mortality with sudden death or appropriate ICD discharge. Nine studies also included congestive heart failure (CHF) readmission, worsening of CHF, cardiac transplantation, and all-cause readmission in their composite outcome. Details of composite outcome definitions for each study are presented in Figure 2 and Table 3. The incidence of primary outcome varied from 8% at a median duration of 4.3 years to 40% at a median duration of 1 year, respectively. The heterogeneity in event rates is partially explained by the heterogeneity in definitions. Paraskevaidis et al. (9) reported an event rate of 40%, which consisted of all-cause death and re-hospitalization. Of the 20 events in this study, 18 were re-hospitalizations whereas 2 were deaths. In contrast, the study by Reant et al. (14) (2016) reported an event rate of 8% which consisted of a composite of cardiovascular death, ICD shock, and CHF readmission. Of the 37 patients with events in this study, 21 had cardiovascular deaths, 4 experienced ICD shocks, and 13 had CHF readmissions. Annual event rates varied from 0.7%/year to 40%/year in the studies; however, the study with the lowest annual event rate used a composite of cardiac death and ICD discharge (22), whereas the study with the highest annual event rate used a composite of death and hospitalization for any cause (9).
All studies showed a significant association between worse LV-GLS and increased composite cardiac outcomes on univariate analysis. Although all studies used some form of hazard calculation to determine the effect size, statistical methods varied and included Kaplan-Meier survival analysis, Cox regression, and Fine-gray proportional hazard. The univariate HR for %LV-GLS was reported in 5, and had a significant association with outcomes in 4 of the 11 studies, and ranged from 0.56 to 1.89 (9,14,17,19,21); however, in the 2 studies with HR <1.00, the absolute value of %LV-GLS was used in the analysis instead of the conventional negative value (14,19). With multivariate analysis there was a significant association between %LV-GLS and the composite cardiac outcomes in 4 of 5 studies with HR ranging from 0.90 to 2.14 (14,17,19,21).
A threshold value for LV-GLS was used in 9 of 11 studies and varied from −9.65% to −16%. These different thresholds were chosen based on either: 1) median value in the study (12,22); 2) ROC curve analysis (11,17); or 3) prior studies (14–16,20,21). Within the 2 studies which used ROC analysis to determine the optimal cutpoint for %LV-GLS, the values were −9.85% (AUC: 0.788, sensitivity 92.3%, and specificity 71%) and −9.65% (AUC: 0.807, sensitivity 100%, and specificity 65%) (11,17). Kaplan-Meier survival analysis showed a significantly worse event-free survival for patients with LV-GLS worse than these cutpoints in both analyses (11,17). A cutoff of LV-GLS ≥ −15% was initially selected by Reant et al. (15) (2015) based on prior research showing that LV-GLS worse than this value was predictive of worse outcomes in patients with asymptomatic severe aortic stenosis (25). ROC analysis was performed within this study for the prespecified LV-GLS cutoff of −15% (AUC: 0.754, sensitivity 92.3%, and specificity 71%) (15). Thus, in the 3 studies which specifically used ROC analysis, the use of LV-GLS cutoffs of −15%, −9.85%, and −9.65% were associated with a sensitivity of 67%, 92%, and 100%, and specificity of 77%, 71%, and 65%, respectively (11,15,17). Based on this initial study using a cutoff of LV-GLS >−15% in HCM patients, this same cutpoint was applied in 2 other studies. In all 3 studies, LV-GLS ≥−15% was associated with worse outcomes with variable effect size (HR: 3.84; 95% CI: 1.27 to 11.62; p = 0.017 by Multivariate Cox automated backward-entry selection versus HR: 2.53; 95% CI: 1.29 to 4.95; p = 0.007 by Univariate Fine-Grey proportional hazard) (14,15,21). Hiemstra et al. (21) found that the addition of LV-GLS ≥−15% improved prediction of events compared to a base model (C-statistic 0.73, likelihood ratio p < 0.001, NRI: 0.30; 95% CI: 0.15 to 0.42; p < 0.001) (21). In addition to a cutoff of LV-GLS >−15%, Reant et al. (14) also used a cutoff of LV-GLS >−14% based on a study by Debonnaire et al. (14) (see discussion below) and found that this was also associated with an increased risk of events. Similarly, Hartlage et al. (16) chose an LV-GLS >−16% as their cutoff based on a prior meta-analysis showing that the lower end of the CI for normal LV-GLS is −15.9% (26). They found that application of a cutoff of LV-GLS >−16% to an HCM population was associated with an increased risk of composite cardiac events (HR: 4.87; 95% CI: 1.07 to 22.2; p = 0.041) and that addition of LV-GLS >−16% to a model improved prediction of events (C-statistic 0.81; NRI: 0.75; 95% CI: 0.21 to 1.23; p < 0.0001) (16). Liu et al. (20) built upon this concept but chose an additional cutpoint of LV-GLS >−10% based on locally weighted scatterplot smoothing analysis to divide his cohort into 4 subgroups, of which patients with HCM and LV-GLS >10% had significantly worse event-free survival compared to the other groups. In this cohort an LV-GLS >−10% had a 19% sensitivity and 95% specificity for the prediction of events, compared to 72% sensitivity and 52% specificity for an LV-GLS >−16%. The largest study of 1019 patients, which was conducted by our group, used a median value of −13.7% as a cutoff and observed that the addition of LV-GLS >−13.7% to 2 separate models which used existing risk scores, the American College of Cardiology/American Heart Association risk factors and the European Society of Cardiology SCD risk score, improved prediction of events (22). Interestingly, in this study the authors also examined the effect of myectomy on long-term events and found that patients with LV-GLS worse than the median of 13.7% who had not undergone myectomy had worse outcome than those with LV-GLS > −13.7% who had undergone myectomy (22). Thus, although there was significant variation of cutoff values chosen as well as measured effect size, there was a consistent finding of worse strain being associated with worse outcomes (Table 3).
Ventricular arrhythmias and/or ICD discharges
There were 3 studies which examined the association of LV-GLS with ventricular arrhythmias and/or appropriate ICD therapies but did not include mortality as an endpoint. Two studies evaluated the association of LV-GLS with primary outcome of appropriate ICD discharge (13,18). Debonnaire et al. (13) reported an adjusted HR of 1.15 (95% CI: 1.02 to 1.30; p = 0.03) with an event rate of 23% and Candan et al. (18) reported an adjusted odds ratio of 1.28 (95% CI: 1.01 to 1.6; p = 0.039) with an event rate of 27% for %LV-GLS. In both studies, ROC analysis was conducted and LV-GLS cutoffs of −14% (AUC: 0.65, sensitivity 86%, and specificity 45%) (13) and −10.7% (AUC: 0.68, sensitivity 70%, and specificity 74%) (18) were identified. Interestingly, in the study by Debonnaire et al. (13), assessing both left atrial volume index and LV-GLS on top of conventional SCD risk factors in primary prevention patients resulted in reclassification of approximately 1 of 5 primary prevention ICD recipients into a very low-risk group who experienced no appropriate ICD therapy during follow-up. Heimstra et al. (21) reported a secondary endpoint of aborted SCD and appropriate ICD therapy for which LV-GLS >−15% remained independently predictive of events after adjusting for gender, left ventricular hypertrophy, and nonsustained ventricular tachycardia (NSVT) (Table 4).
Only 1 study looked at the association of LV-GLS with NSVT as an isolated outcome. Di Salvo et al. (10) measured LV-GLS at baseline and NSVT on 24-hour Holter monitor conducted every 3 months over a 2-year follow-up period and found comparable mean LV-GLS values for those in the event and no-event groups. Although LV-GLS was not included in the multivariable analysis, they reported that their variable of >3 LV segments with longitudinal 2D strain ≥−10% was significantly associated with NSVT after adjusting for maximal LV thickness, E/Ea, family history of SCD, LV outflow tract obstruction, and N-Terminal pro-brain natriuretic peptide (p < 0.0001, coefficient = 3.044, standard error = 0.709) (10) (Table 4).
In our systematic review of 14 studies with more than 3000 HCM patients evaluating the association of LV-GLS and clinical outcomes, we make several observations. First, we report marked variability in both the risk profiles of the included cohorts and in definitions of studied primary clinical outcomes. Second, we observed several methodological short-falls such as small sample sizes, no clear mention of obstructive HCM or preserved LVEF in the inclusion or exclusion criteria, absence of interobserver testing, variable duration of follow-up, differing treatment of LV-GLS as a variable (negative versus absolute value, categorical or continuous variable, and cutoff values), and variable definition of end-points. Third, we note significant heterogeneity in effects which may be partially explained by variable statistical methods and effect size measures, follow-up durations, and different LV-GLS cutoffs. These shortcomings make it impossible to conduct a meta-analysis, and consequently we present this in the format of a systemic review. Despite all this, we noted that in the majority of studies abnormal LV-GLS was consistently associated with increased cardiac events.
Current role of LV-GLS in HCM
Multiple studies have noted a clear association between impaired regional strain and GLS with histopathologic changes, myocardial fibrosis, and myocardial performance in patients with HCM (27–29). Additionally, there are studies suggesting LV-GLS may be more sensitive than late gadolinium enhancement by cardiac magnetic resonance imaging to detect fibrosis by histopathology (30). As described in this review, studies examining the association of LV-GLS with outcomes in HCM have been limited to retrospective or prospective cohort studies. Although the majority of the studies showed that the presence of worse LV-GLS is associated with worse outcomes, a consensus regarding at what value of LV-GLS a clinician should be concerned and alter care has not been determined. There were 3 studies of 11 in which the association between worse strain and worse outcomes was not statistically significant. In 2, the association was statistically significant on univariate but not on multivariate analysis, which may be related to small sample size (9,19). Additionally, caution should be taken in extrapolating the negative results of the Moneghetti et al. (19) study because they only performed manual Lagrangian strain analysis on the apical 4-chamber view. This may not be entirely reflective of LV-GLS calculated from apical 2-, 3-, and 4-chamber apical views which incorporate all LV segments, especially in a disease process such as HCM in which regional segmental longitudinal strain is worse in regions of hypertrophy and/or fibrosis and is not affected equally in all segments, a fact which has been used to help differentiate HCM from infiltrative cardiomyopathies (31,32). Although segmental strain patterns are useful for distinguishing HCM from infiltrative cardiomyopathies, only 1 study examined the association of abnormal segmental longitudinal strain with outcomes (10). Thus, at this time, the data are not sufficient to determine whether abnormal global or segmental longitudinal strain is more predictive of adverse events in HCM. The third negative study reviewed NSVT on intermittent Holter monitoring and did not find an association between NSVT and impaired LV-GLS. This could be due to the patients being potentially at lower risk, or simply that the outcome picked is often asymptomatic when it occurs and might have just been missed due to the timing of monitoring (10). Additionally, as noted in prior studies, depending on the vendor software and version used for speckle tracking, variation may exist in LV-GLS values obtained, although the more recent studies note that the variation observed is similar to that of LVEF (33,34).
Correspondingly, application of LV-GLS in the clinical practice of HCM is exploratory at this time. It has been proposed to be used as a potential marker of increased risk, perhaps tipping shared decision-making in favor of more aggressive intervention, but not dictating management decisions at this time. For instance, 1 such potential application would be as a risk modifier in assessment of risk for SCD. However, whereas many of the studies showed a higher burden of ventricular arrhythmias and appropriate ICD discharge with worse strain (13,18), there is no established cutoff at the current time which could guide such a decision. Another potential application would be in the identification of patients with obstructive HCM and worse strain who might benefit from surgical myectomy because these patients were shown to have improved outcomes with myectomy as opposed to those who did not have surgery (22).
As the data regarding LV-GLS continue to emerge in HCM, it has become clear that although abnormal LV-GLS is predictive of adverse events, the large degree of heterogeneity between studies has resulted in additional questions which must be addressed in future studies. Future studies should seek to fully define patient baseline characteristics and outcomes including a breakdown of individual events. Because of the myriad of known HCM phenotypes, current studies have adopted 1 of 2 strategies for patient inclusion: including patients with all phenotypes which introduces additional variables that must be accounted for in the analysis, or narrowing the inclusion criteria which can result in a smaller cohort with limited statistical power. One strategy that should be considered is the use of vendor-neutral software for strain analysis as this may allow for the inclusion of more patients, particularly in retrospective studies conducted at centers which obtain echocardiographic images using multiple ultrasound equipment vendors. Although these are potential future applications, they are currently more thought-provoking and require further investigation through carefully designed studies, either retrospective with larger more homogenous populations or, and perhaps more importantly, prospective studies with the addition of a cutpoint value at which intervention should be undertaken.
Although our study is the most comprehensive review of literature on prognostic value of LV-GLS in patients with HCM, it suffers from certain limitations. First, the heterogeneity in study methodologies and reporting prevented use of meta-analysis techniques. Second, it is possible that restriction to English language literature and electronic databases may have introduced a publication bias, although formal assessment of such is not possible in a narrative review. Additionally, these individual studies may themselves suffer from bias including but not limited to selection bias as outlined above. However, despite these limitations, we believe that our robust methodology and systematic approach in bringing together data on this topic will help outline the current literature and set the stage for future investigations on this topic.
Our review of more than 3,000 patients with HCM suggests an association of abnormal LV-GLS with adverse composite cardiac outcomes and ventricular arrhythmias. Future studies should seek to provide uniform methods in data collection and reporting of outcomes as they relate to LV-GLS, and prospective studies are necessary to determine whether the use of a specific cutoff would be beneficial to aid in clinical decision-making regarding therapy.
COMPETENCY IN MEDICAL KNOWLEDGE: Despite significant heterogeneity in study methods, risk factors reported and the outcomes assessed, abnormal LV-GLS appears to be associated with adverse outcomes including death and ventricular arrhythmias.
TRANSLATIONAL OUTLOOK: Future studies should seek to fully describe methods and outcomes related to LV-GLS, and prospective studies are necessary to evaluate a threshold at which abnormal strain may be used to alter clinical management of patients with HCM.
↵∗ Drs. Tower-Rader and Mohananey contributed equally to this work and are joint first authors.
Dr. Desai is supported by the Haslam Family Endowed Chair in Cardiovascular Medicine. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- Abbreviations and Acronyms
- area under the curve
- congestive heart failure
- ejection fraction
- hypertrophic cardiomyopathy
- implantable cardioverter-defibrillator
- left ventricular global longitudinal strain
- net reclassification improvement
- nonsustained ventricular tachycardia
- New York Heart Association functional class
- receiver operator curve
- sudden cardiac death
- Received May 21, 2018.
- Revision received July 13, 2018.
- Accepted July 19, 2018.
- 2019 American College of Cardiology Foundation
- Gersh B.J.,
- Maron B.J.,
- Bonow R.O.,
- et al.
- Elliott P.M.,
- Anastasakis A.,
- Borger M.A.,
- et al.
- Teare D.
- Collier P.,
- Phelan D.,
- Klein A.
- Funabashi N.,
- Takaoka H.,
- Horie S.,
- et al.
- Reant P.,
- Mirabel M.,
- Lloyd G.,
- et al.
- Ozawa K.,
- Funabashi N.,
- Takaoka H.,
- Kobayashi Y.
- Candan O.,
- Gecmen C.,
- Bayam E.,
- Guner A.,
- Celik M.,
- Dogan C.
- Moneghetti K.J.,
- Stolfo D.,
- Christle J.W.,
- et al.
- Liu H.,
- Pozios I.,
- Haileselassie B.,
- et al.
- Hiemstra Y.L.,
- Debonnaire P.,
- Bootsma M.,
- et al.
- Tower-Rader A.,
- Betancor J.,
- Popovic Z.B.,
- et al.
- Serri K.,
- Reant P.,
- Lafitte M.,
- et al.
- Popovic Z.B.,
- Thomas J.D.
- Yingchoncharoen T.,
- Gibby C.,
- Rodriguez L.L.,
- Grimm R.A.,
- Marwick T.H.
- Popovic Z.B.,
- Kwon D.H.,
- Mishra M.,
- et al.
- Dhillon A.,
- Sweet W.,
- Popovic Z.B.,
- et al.
- Almaas V.M.,
- Haugaa K.H.,
- Strom E.H.,
- et al.
- Urbano-Moral J.A.,
- Rowin E.J.,
- Maron M.S.,
- Crean A.,
- Pandian N.G.