Introduction

Hypertrophic cardiomyopathy (HCM) is the most prevalent inherited cardiomyopathy and is associated with increased cardiovascular morbidity and mortality. Particularly, patients with HCM experience more frequent sudden cardiac death (SCD) and death because of heart failure and show an increased risk of stroke-related mortality due to high prevalence of atrial fibrillation.^1,2 However, risk stratification in patients with HCM remains challenging, mainly because of a large heterogeneity of phenotypes with different prognosis, varying from asymptomatic mild cardiomyopathy throughout life to the occurrence of SCD at young age. Current approach for risk stratification in patients with HCM is mainly focused on SCD and advocates for combination of clinical and echocardiographic parameters.^3,4 However, those parameters are known to have limited sensitivity and specificity, particularly, in predicting cardiovascular events other than SCD.⁵ Therefore, research has focused on identifying potential additional prognosticators to optimize HCM patient management.^6–11

See Editorial by Maron

See Clinical Perspective

Two-dimensional speckle tracking strain analysis has been recently proposed as a new method to improve assessment of left ventricular (LV) function as compared with conventional echocardiography. In patients with HCM, in whom LV ejection fraction is mostly within the normal ranges, global longitudinal strain (GLS) has shown to be able to detect subtle myocardial systolic dysfunction, and initial studies have proposed this parameter as a potential novel prognostic factor.^12–16 Similarly, left atrial volume index (LAVI) has been shown to be associated with specific clinical outcomes, such as new onset of atrial fibrillation and SCD, probably reflecting not only LV diastolic dysfunction but also LV outflow tract (LVOT) obstruction, mitral regurgitation, and intrinsic atrial myopathy.^17–19 However, the prognostic value of the combination of GLS and LAVI has not been investigated thoroughly. Therefore, the objective of this study was to evaluate the incremental prognostic value of GLS and LAVI for hard adverse clinical outcomes in a large cohort of patients with HCM and with a long-term follow-up.

Methods

Patient Population

The population consisted of patients with HCM, defined according to current guidelines: a maximal LV hypertrophy (LVH) ≥15 mm, in absence of any other cardiac or systemic disease that could cause a similar degree of LVH.³ Patients were identified from an ongoing clinical registry and excluded if age was <16 years. Patient data were prospectively collected in the departmental cardiology information system (EPD-Vision; Leiden University Medical Center, Leiden, The Netherlands) and included the following information: demographic characteristics, New York Heart Association functional class, use of medications, comorbidities, and the currently adopted SCD risk factors, such as unexplained syncope, nonsustained ventricular tachycardia (nsVT) on 24 hours electrocardiographic Holter monitoring (≥3 beats at ≥120 bpm), and positive family history for SCD at young age (<50 years) in first- or second-degree relatives. Furthermore, an ECG and echocardiogram were performed in all patients at the moment of the first visit at the outpatient clinic. Interventions, such as percutaneous revascularization, septal alcohol ablation, myectomy, or other cardiac surgery during follow-up or before the first outpatient visit, were also recorded. The study complies with the Declaration of Helsinki and was approved by the institutional review board. Because of the retrospective design of this study, the Medical Ethical Committee waived the need of written informed consent.

Echocardiography

Standard transthoracic 2-dimensional echocardiographic studies were performed using commercially available ultrasound machines (Vivid 5, Vivid 7, and E9, GE-Vingmed, Milwaukee, WI). Images were digitally stored and analyzed offline (EchoPAC, version 112, GE Medical Systems, Horten, Norway). LV end-diastolic and end-systolic diameters were measured from the parasternal long-axis view. LV volumes, LV ejection fraction, and left atrial (LA) volumes were measured using Simpson method and indexed for body surface area.²⁰ LA volume was calculated at end systole, tracing the LA endocardium in the 4- and 2-chamber views. Measurements of septal and posterior wall thickness were obtained in the parasternal long-axis from an M-mode acquisition while maximal LV wall thickness was assessed from a short-axis view at 3 different levels (basal, mid, and apical). LV diastolic function was assessed mainly using the mitral inflow peak velocities of E divided by the peak early diastolic velocity (E′) of the lateral mitral annulus by tissue Doppler imaging, obtaining the E/E′ ratio.²¹ Assessment of the presence of systolic anterior movement of the mitral valve was performed from a parasternal long-axis view and from apical 3- and 5-chamber views. LVOT resting peak gradient was quantified by continuous wave Doppler. The presence and grade of mitral regurgitation were assessed according to a multiparametric approach as recommended.²²

GLS was measured using speckle tracking analysis on standard apical views (2-, 3-, and 4-chamber views), acquired with a frame rate of 40 to 90 Hz (mean, 60 fps). The region of interest was automatically created and manually adjusted when necessary to fit the entire wall thickness. GLS was then calculated by averaging the peak longitudinal strain in 17 segments from the 3 different views.

Clinical Outcome

The primary end point was a combined end point of all-cause mortality, heart transplantation, aborted SCD, and appropriate implantable cardioverter defibrillator (ICD) therapy. Aborted SCD was defined as a successful resuscitation from cardiac arrest with documented ventricular tachycardia and ventricular fibrillation. Appropriate ICD therapy was defined as antitachycardia pacing and shock for ventricular tachycardia or ventricular fibrillation. The occurrence of events during follow-up was obtained by medical charts review, retrieval of survival status through the municipal civil registries, and by contact with the general practitioner of the patient. The secondary end point included (aborted) SCD and appropriate ICD therapy.

Statistical Analysis

Continuous variables are expressed as mean±SD, when normally distributed, and as median (interquartile range), when not normally distributed. Categorical variables are presented as absolute numbers and percentages. Differences in baseline characteristics between patients with and without the primary end point were assessed using Student t test, Mann–Whitney U test, or χ² when appropriate. Univariable Cox regression analysis was performed for all clinical and echocardiographic variables, and the variables with a P<0.10 were included in a multivariable Cox regression analysis (selecting among the ones highly correlated with each other) to identify independent predictors of the primary and secondary end points: hazard ratio (HR) and 95% confidence intervals were calculated. Kaplan–Meier curves were constructed to estimate the cumulative event-free survival for the primary end point and compared by the log-rank test. Cut-off value for LAVI (34 mL/m²) was defined based on guidelines recommendations, whereas for GLS (−15%) it was chosen based on the median value of GLS in the current population and on previously suggested cut-off value from the literature in patients with HCM.^12,14,15 To evaluate the incremental value of GLS and LAVI on top of clinical and standard echocardiographic parameters, likelihood ratio testing was performed, as well as calculation of the overall C-statistic as proposed by Harrell et al²³ as an analogue of the area under the receiver operating characteristic curve for survival analysis for both primary and secondary end points. Furthermore, we assessed the impact of adding GLS and LAVI to a basic model using the continuous net reclassification improvement. A P<0.05 was considered significant. Statistical analysis was performed with the SPSS software package and the R-package survINDRI (version 20, IBM Corp, Armonk, NY).

Results

Patient Population

A total of 427 patients with HCM (52±15 years, 66% men) were included (Table 1). A pathogenic or likely pathogenic gene mutation was found in 167 (63%) of the patients who underwent genetic testing (n=264). Mean LV ejection fraction was normal in this HCM patient population (65±9%), but mean GLS was impaired (−15±4%) and median LAVI was increased (36 [28–47] mL/m²).

Long-Term Clinical Outcome

During a median follow-up of 6.7 (interquartile range, 3.3–10.0) years, 103 patients reached the primary end point: 53 patients experienced aborted SCD or appropriate ICD therapy, 2 patients underwent heart transplantation, and 48 patients died. Cause of death was of cardiac origin in 22 patients (11 heart failure, 10 SCD, and 1 other cardiovascular cause), noncardiac in 10 patients (3 sepsis, 6 malignancy, and 1 suicide), and unknown in 16 patients. As shown in Table 1, there were no significant differences in demographics, cardiovascular risk factors, and symptoms between patients who reached the primary end point and those who did not. However, patients who reached the primary end point were more likely to undergo a septal intervention, used more frequently β-blockers, and showed a higher incidence of nsVT at 24-hour ECG Holter monitoring. Furthermore, patients who reached the primary end point had a significantly larger maximum wall thickness, worse LV diastolic function (E/E′), more prevalence of systolic anterior movement, more impaired (less negative) GLS, and a larger LAVI (Table 1).

The secondary end point included 63 events, of which 53 aborted SCD or appropriate ICD therapy and 10 SCD.

Survival Analysis

Univariable Cox proportional hazard regression analysis showed that atrial fibrillation, New York Heart Association class, nsVT on 24-hour Holter monitoring, use of β-blockers, GLS, LV ejection fraction, LAVI, maximal LVH, E/E′, LA diameter, LVOT resting peak gradient, systolic anterior movement, mitral regurgitation >2, and septal intervention during follow-up all had a significant association with the primary end point. However, multivariable analysis for the primary end point revealed only GLS (HR, 1.10 [1.03–1.19]; P=0.007) and LAVI (HR, 4.27 [2.35–7.74]; P<0.001) as independent predictors (Table 2).

When dividing the population according to the pre-specified cut-off value of LAVI, patients with LAVI ≥34 mL/m² had worse outcome as compared with patients with LAVI <34 mL/m². The cumulative event-free survival at, respectively, 2 and 6 years was 98% and 93% versus 94% and 81%, respectively (log-rank, 19.7; P<0.001; Figure 1A). When dividing the population according to the pre-specified cut-off value of GLS, patients with GLS ≥−15% had worse outcome as compared with patients with GLS <−15% (Figure 1B). The cumulative event-free survival at, respectively, 2 and 6 years was 98% and 91% versus 92% and 76%, respectively (log-rank, 27.1; P<0.001).

• Download figure
• Download PowerPoint

When dividing the population in 4 groups based on the pre-specified GLS and LAVI cut-off values, the group of patients with both GLS <−15% and LAVI <34 mL/m² had the best outcome, whereas patients with both GLS ≥−15% and LAVI ≥34 mL/m² had the worst outcome. The cumulative event-free survival at 6 years was 99% for GLS <−15% and LAVI <34 mL/m² versus 63% for patients with GLS ≥−15% and LAVI ≥34 mL/m² (log-rank, 49.3; P<0.001; Figure 2).

• Download figure
• Download PowerPoint

When considering the secondary end point, GLS and LAVI showed a significant association with this outcome (HR, 1.12 [1.06–1.19]; P<0.001 for GLS and HR, 3.94 [2.33–6.66]; P<0.001 for LAVI) together with sex, nsVT, LA diameter, and maximum LVH at the univariable analysis. When corrected for sex, LVH, and nsVT, GLS and LAVI remained independently associated with the secondary end point at multivariable analysis (HR, 1.08 [1.01–1.16]; P=0.023 for GLS and HR, 3.70 [2.08–6.60]; P<0.001 for LAVI).

Incremental Value of GLS and LAVI

Figure 3 shows the results of the likelihood ratio test and the Harrell C-statistic for LAVI and GLS on top of clinical and standard echocardiographic parameters associated with the primary end point at the univariable Cox regression analysis. The addition of LAVI ≥34 mL/m² to a basic model provided a significant improvement (P<0.001) with an increase of the C-statistic from 0.68 to 0.71. The sequential addition of GLS ≥−15% further improved the model (likelihood ratio test P=0.008). Overall, the combined addition of LAVI and GLS to the clinical and standard echocardiographic risk factors provided the best model (likelihood ratio test P<0.001, C-statistic=0.73). The incremental value of this model was also demonstrated by a net reclassification improvement of 0.30 (95% confidence interval, 0.15–0.42; P<0.001).

• Download figure
• Download PowerPoint

Similarly, when considering the secondary end point, the addition of LAVI ≥34 mL/m² and GLS ≥−15% to a basic model, including sex, nsVT, and maximum LVH, also provided incremental prognostic value: χ² increased from 25 to 29 adding GLS (P=0.046) with an improvement of C-statistic from 0.68 to 0.70. More importantly, adding LAVI increased the χ² to 39 (P<0.001) with an improvement of C-statistic to 0.79. The net reclassification improvement was 0.26 (95% confidence interval, 0.11–0.41; P<0.001).

Inter- and Intraobserver Variability

Interobserver reproducibility for GLS and LAVI was assessed by 2 independent operators in 15 randomly selected patients. The intraclass correlation coefficient between 2 observers was 0.94 (P<0.001) for LAVI and 0.94 (P<0.001) for GLS. The intraclass correlation coefficient for intraobserver agreement was 0.95 (P<0.001) for LAVI and 0.91 (P<0.001) for GLS.

Discussion

The main findings of the current study can be summarized as follows: (1) in a large cohort of patients with HCM, GLS and LAVI demonstrated to be independently associated with the primary end point of all-cause mortality, heart transplantation, and aborted SCD, as well as with the secondary end point of (aborted) SCD or appropriate ICD therapy; (2) the presence of both preserved GLS and LAVI showed the highest cumulative event-free survival as compared with patients with impaired GLS or LAVI; and (3) the addition of GLS and LAVI provided incremental prognostic value on top of other clinical and standard echocardiographic parameters.

Risk Stratification in HCM

In patients with HCM, risk stratification is a clinical challenge and has been mainly focused on prevention of SCD, for which several risk markers have been proposed, such as family history for SCD, unexplained syncope, nsVT, LV thickness, and LVOT gradient.²⁴ Recently, O’Mahony et al⁴ developed a new risk prediction model to predict SCD in patients with HCM, which included the use of continuous variables instead of dichotomized variables and which was implemented in the current European Society of Cardiology guidelines.³ Although the new risk model improves risk stratification for SCD and subsequently identification of patients who can benefit from an ICD,^25,26 there are no recommendations in current guidelines for risk stratification for other adverse events, such as heart failure–related mortality and other cardiovascular deaths that may occur in patients with HCM.²⁷ Therefore, several studies have tried to identify additional prognostic markers to optimize clinical management of patients with HCM.

Among these, N-terminal pro-B-type natriuretic peptide, atrial fibrillation, New York Heart Association class, and functional exercise capacity were shown to be associated with worse overall prognosis in patients with HCM.^9–11 Furthermore, the presence of myocardial fibrosis, as assessed by cardiovascular magnetic imaging with late gadolinium enhancement, has been proposed as an important risk marker and showed to be associated not only with SCD but also with adverse cardiovascular events in patients with HCM .^6–8,28 However, in clinical practice, ideal prognosticators would be simple and readily available parameters, which should reflect structural abnormalities, such as myocardial fibrosis together with myocardial systolic and diastolic dysfunctions.

GLS as Risk Marker

Several studies have shown that GLS, measured by speckle tracking echocardiography, is able to detect subtle myocardial dysfunction in patients with HCM probably reflecting the characteristics myocardial fiber disarray, myocardial fibrosis, and microvascular dysfunction.¹⁶ A study of Serri et al²⁹ showed that GLS can be measured with good reproducibility and is significantly impaired in patients with HCM as compared with healthy controls. In a cohort of 32 patients with HCM who underwent septal myectomy, GLS significantly correlated with fibrosis in the myocardium samples and could predict arrhythmias better than cardiovascular magnetic imaging with late gadolinium enhancement.³⁰

Initial studies have also assessed the value of GLS to predict adverse events in patients with HCM.^12–15 Hartlage et al,¹² using a cut-off value for GLS of −16% in 79 patients with HCM, found an abnormal GLS to be predictive for a combined end point of heart failure hospitalizations, sustained ventricular arrhythmias, and all-cause mortality. In a population of 92 high-risk patients with HCM who received an ICD, Debonnaire et al¹⁴ demonstrated the value of GLS (with a cut-off of −14%) as potential marker in the prediction of appropriate ICD therapy, which is confirmed for a larger and more heterogeneous HCM population by the current study. Recently, a study by Reant et al¹⁵ showed the association between GLS and the combined end point of cardiac death, heart failure admission, and appropriate ICD therapy in a large cohort of 472 patients with HCM where patients with atrial fibrillation were excluded. Particularly, patients with a GLS >−15.6% showed to have higher risk for cardiac events. In the present study with a similarly large patient population, the prognostic value of GLS was also demonstrated for the hard end point of all-cause mortality and appropriate ICD therapy. Particularly, the same cut-off value of −15% for GLS showed significant incremental value over clinical and standard echocardiographic parameters. Furthermore, in the current study, patients in atrial fibrillation were not excluded and the multivariable Cox regression analysis corrected for atrial fibrillation, increasing the clinical application of these results considering the potential prognostic value of atrial fibrillation in patients with HCM.³¹ Finally, the current study evaluated the prognostic value of GLS in combination with LAVI, another potentially important prognosticator.

LAVI as Risk Marker

Enlargement of LA occurs frequently in patients with HCM, reflecting significant LV diastolic dysfunction, LVOT obstruction, presence of mitral regurgitation, and intrinsic atrial myopathy.³² Increased LA diameter is currently implemented in the HCM risk model for SCD.⁴ However, LAVI is considered superior as an estimate of LA size¹⁷ and was suggested by initial studies to be of prognostic value for general risk stratification in patients with HCM.^18,19 In the study performed by Yang et al,¹⁸ LAVI was found to be an independent predictor of cardiovascular events in a population of 81 patients with nonapical HCM. Similar results were presented by Losi et al,¹⁹ who evaluated LAVI in 140 patients with HCM at baseline and during follow-up and showed worse prognosis in patients with an enlarged LAVI or a rapid increase in LAVI during follow-up. Debonnaire et al¹⁴ showed that LAVI (with a cut-off of 34 mL/m²) was independently associated with appropriate ICD therapy. Our study not only confirms the association of LAVI with the risk of appropriate ICD therapy or SCD but also shows for the first time the prognostic value of LAVI for a hard mortality outcome, including a large HCM population and with long-term follow-up.

Clinical Implications

The present study demonstrated that the combination of GLS and LAVI may improve risk stratification of patients with HCM; for SCD or appropriate ICD therapy, but also for the more general end point of all-cause mortality. The prediction model, including clinical and standard echocardiographic risk factors, showed a C-statistic of 0.68, which is in line with previous literature²⁶; the addition of GLS and LAVI increased the C-statistics to 0.73 for the primary end point and to 0.79 for the secondary end point, suggesting the improvement in predictive value. Such parameters, readily available from a standard echocardiographic screening, might, therefore, be of great value to improve risk stratification and, therefore, potentially to be included in future studies for a more comprehensive risk score for all-cause mortality on top of conventional parameters.

Although the identification of strict cut-off values for these parameters might be debatable, clinical application of GLS and LAVI in the standard management of patients with HCM might be stimulated by showing the clinicians how specific values perform in predicting the outcome. In this cohort, a cumulative event-free survival of 99% after 6 years was demonstrated for patients with both preserved GLS and LAVI (as defined using −15% and 34 mL/m²),^12–15,18 whereas event-free survival was only 63% after 6 years in patients with impaired GLS and LAVI (Figure 4). Therefore, the identification of patients who are considered at low risk could be improved using GLS and LAVI, with important implications for the timing of starting medical therapy, planning follow-up of outpatient visits, and SCD screening, as well as decision making over ICD implantation.

• Download figure
• Download PowerPoint

Limitations

This study has several limitations that should be mentioned. In this single-center study, only echocardiographic equipment of GE was used; therefore, the results (and the cut-off value proposed) should be interpreted with caution when compared with other vendors. The European Association of CardioVascular Imaging and the American Society of Echocardiography recently set up a task force to evaluate the intervendor variability. From this evaluation, GLS showed a variability <10% between different vendors, which is comparable to standard echocardiographic measurements currently used.³³ Furthermore, it is known that in patients with HCM, appropriate ICD therapy may overestimate the event rate when antitachycardia pacing for ventricular tachycardia that could have been self-terminating are included. In the Data Supplement, the results of the Cox analysis are provided when antitachycardia pacing was removed as an outcome, which showed similar results (Tables I and II in the Data Supplement). Other potential prognostic markers, such as cardiovascular magnetic imaging with late gadolinium enhancement or NT-pro-BNP, were not systematically assessed. Importantly, further prospective studies in large patient population are needed to validate these data, especially to determine the most appropriate cut-off value for GLS in patients with HCM and how these measurements could be implemented in daily clinical practice.

Conclusions

GLS and LAVI are both independently associated with adverse outcome in patients with HCM. The combination of these 2 parameters has incremental value on top of standard clinical and echocardiographic parameters for predicting adverse events and could be considered in a more comprehensive risk score assessment.

Footnotes