Author + information
- Received June 17, 2015
- Revision received September 11, 2015
- Accepted September 29, 2015
- Published online August 1, 2016.
- Makoto Saito, MD, PhDa,
- Faisal Khan, MBBSb,
- Ted Stoklosa, MBBSb,
- Andrea Iannaccone, MDa,
- Kazuaki Negishi, MD, PhDa and
- Thomas H. Marwick, MBBS, PhD, MPHa,∗ ()
- ↵∗Reprint requests and correspondence:
Prof. Thomas H. Marwick, Baker IDI Heart and Diabetes Institute, 75 Commercial Road, Melbourne VIC 3004, Australia.
Objectives This study sought to investigate the associations of left ventricular (LV) strain and its serial change with major adverse cardiac events (MACE) in hypertensive heart disease, independent of and incremental to clinical and LV geometric parameters.
Background In patients with hypertensive heart disease, MACE are associated with abnormal LV morphology, but their association with subclinical LV dysfunction is unclear.
Methods We retrospectively studied 388 asymptomatic nonischemic patients with hypertension who had abnormal LV geometry at a baseline echocardiogram between 2005 and 2014. Global longitudinal strain (GLS) was measured using speckle tracking. Patients were followed for MACE (death and admission because of heart failure, myocardial infarction, and strokes) over median of 4 years. A Cox proportional hazards model was used to assess the association of parameters with MACE.
Results MACE (n = 72; 19%) were associated with higher prevalence of concentric hypertrophy and impaired GLS (both, p < 0.01). The association of GLS with MACE was independent of and incremental to clinical parameters and concentric hypertrophy. Echocardiographic follow-up was performed in 55 patients (median duration, 3 years); deterioration in GLS was also associated with the 10 patients experiencing MACE after the second echo. A risk score was developed using age >70, atrial fibrillation, concentric hypertrophy, and baseline GLS >–16% from the derivation cohort (C statistic, 0.71), and a separate validation cohort showed it to have good discrimination for MACE (C statistic, 0.71).
Conclusions GLS and its deterioration are associated with MACE in asymptomatic hypertensive heart disease. A risk score incorporating strain was useful for predicting risk of MACE.
Hypertensive heart disease (HHD) represents a constellation of abnormalities that include altered left ventricular (LV) morphology, left ventricular hypertrophy (LVH), and systolic and diastolic dysfunction. LVH (1) and abnormal LV morphology (2,3) are important independent predictors of adverse prognosis in this setting. However, the myocardial consequences of chronic systolic and diastolic hypertension (HTN) include not only myocyte hypertrophy, but also perivascular and myocardial fibrosis and medial thickening of the intramyocardial coronary arteries (4). Consequently, HTN is a risk factor for the development of major adverse cardiac events (MACE) (4–8), and potentiating many complications of diabetes and cardiovascular disease (8–10). The resulting changes in myocardial composition may be important contributors to the pathogenesis of HHD.
Recently, LV strain has been shown to be a sensitive marker of LV function, and providing long-term prognostic assessment over traditional clinical or echocardiographic parameters in a variety of heart diseases (11). The use of this sensitive parameter may be particularly helpful in the evaluation of asymptomatic patients (12). Although LVH is associated with impaired myocardial strain (13), the independent prognostic implications of strain in patients with HHD have remained undefined. In addition, there are scant data about the natural history of LVH, LV morphology, and function, which could clarify the pathogenic mechanism of development of cardiac events in HHD (14). We hypothesized that functional markers would add to existing LVH and geometry markers to predict adverse outcome in asymptomatic HHD, and that the inclusion of all of these variables would permit the derivation of a risk score for prediction of MACE. Accordingly, we sought to investigate the association of LV strain with MACE, independent of and incremental to clinical and basic echocardiographic parameters and the association of its serial change and MACE.
This study was designed as retrospective cohort study. The source population comprised 440 consecutive adult patients with asymptomatic HTN (range 40 to 85 years of age) who underwent clinically indicated echocardiography between July 2005 and June 2014, and were found to have features of HHD in the absence of ischemic heart disease, more than mild valvular heart disease, cardiomyopathy, or pulmonary arterial HTN in the medical history. After exclusion of 52 patients with poor image quality, the study comprised 388 patients (Study 1). HTN was defined as systolic blood pressure (SBP) ≥140 mm Hg, diastolic blood pressure ≥90 mm Hg, or current pharmacological treatment for HTN. HHD was recognized on the basis of LVH (left ventricular mass index [LVMI]; >95 g/m2 for women and >115 g/m2 for men), or abnormal relative wall thickness (RWT) >0.42, in the presence of a normal ejection fraction (EF) >50%. Among them, 74 patients had a follow-up echocardiogram >1 year after the first examination. After excluding 19 patients who had MACE between the paired echograms, 55 patients were enrolled to this subanalysis (Study 2) (Online Figure 1). The study was approved by the Tasmanian Health Research Ethics Committee.
Clinical parameters at the time of the echocardiographic examination (comorbidity and medical history, medication, and serum markers) were gathered by an investigator blinded to the echocardiographic findings.
Transthoracic echocardiography was performed by experienced sonographers using a commercially available ultrasound system (Vivid 7, Vivid 9, or Vivid I, GE Vingmed, Horten, Norway). Echocardiographic images were digitally recorded and downloaded as DICOM files for offline analysis. Conventional echocardiographic parameters were measured according to the recommendations of the American Society of Echocardiography (15,16). LV mass was calculated according to the American Society of Echocardiography formula (15), and normalized to body surface area. RWT was computed as 2 × posterior wall thickness/LV end-diastolic diameter. RWT and LVMI were used to categorize patients as having either: 1) increased RWT but normal LVMI (concentric remodeling); 2) normal RWT, increased LVMI (eccentric hypertrophy); or 3) increased RWT and LVMI (concentric hypertrophy), according to the American Society of Echocardiography guidelines (15). LV volumes and LVEF were calculated by the biplane method of disks using 2-dimensional images and LV volumes were indexed to body surface area. Left atrial volume was calculated by the biplane modified Simpson method using 2-dimensional images and indexed to body surface area. The transmitral early diastolic velocity (E) and its deceleration time were acquired in the apical 4-chamber view using pulsed-wave Doppler at the level of the mitral valve tips during diastole. The early diastolic mitral annular tissue velocity (eʹ) was calculated as the average of septal and lateral mitral annular velocities, and E/eʹ was calculated. Systolic pulmonary artery pressure was estimated from the tricuspid regurgitation pressure gradient and the maximum inferior vena cava diameter and its collapsibility.
We measured the 2 myocardial strain parameters (global longitudinal strain [GLS] and global circumferential strain [GCS]) using standard methodologies for speckle tracking (Research Arena, TomTec, Unterschleissheim, Germany) (17). After manual tracing of the LV endocardial border, the dedicated software automatically tracked the myocardium throughout the cardiac cycle in the apical 4-chamber, 2-chamber, and long-axis views. GLS was obtained by averaging peak values of segmental strain in the apical views (18). The peak values of the 6 segmental circumferential strain curves were obtained from the short-axis view at the papillary muscle level and averaged to provide GCS. The mean frame rate was 42 ± 19 frames/s. In patients with atrial fibrillation, strain parameters were measured if the ratio of preceding and pre-preceding intervals was 1 (19).
The primary outcome was MACE (all-cause death or admission for heart failure [HF], acute myocardial infarction, and strokes). The secondary outcome was admission for HF. Outcomes were checked by data linkage to administrative databases from the Clinical Informatics and Business Intelligence Unit of the Department of Health and Human Services of Tasmania. This captures the death registry and all admission data in Tasmanian public hospitals. Patients were censored at the time of outcome or at the end of follow-up (June 30, 2014).
Overall, <5% of observations were missing, with the exception of eʹ and E/eʹ (13%). Missing eʹ was imputed from propensity models using parameters without missing data (age, gender, body mass index, SBP, heart rate, LVMI, left atrial volume index, GLS, systolic pulmonary artery pressure, diabetes mellitus, serum creatinine, atrial fibrillation, and hemoglobin). E/eʹ was calculated using actual E velocity and interpolated eʹ. Other continuous variables (<5%) were imputed using the corresponding mean value. The percent change in study parameters were calculated as absolute differences divided by absolute values at baseline. The significance of differences between the groups was assessed using Student t test or Mann-Whitney U test according to the distribution of the study parameters. For categorical variables, the chi-square test or Fisher exact test were used, as appropriate. For multiple comparisons, a 1-way analysis of variance or chi-square test were used, followed by Bonferroni correction. Comparison of the repeated measurements was carried out using a paired t test or Wilcoxon signed-rank test or McNemar test, as appropriate. Univariable and multivariable linear regression analyses were used to evaluate the associations between LV strain parameters and study parameters. Univariable and multivariable Cox proportional hazards models were used to determine the features associated with MACE. The optimum model candidates were selected based on the significant variables in the univariable analysis (p < 0.10) and clinically relevant parameters. Two models (a clinical and an echocardiographic model) were created to avoid overfitting. The incremental value of GLS was also assessed in 3 modeling steps, using nested models. The first step consisted of fitting a multivariable model of age, gender, SBP, and comorbidities. Then, concentric hypertrophy was included in the second step. Finally, GLS was included in the third step. The change in overall log-likelihood ratio was used to assess the increase in predictive power. Harrell’s C statistic was used to evaluate model and score performance (20).
Patients were randomly divided into 2 groups: a derivation group (n = 194; MACE = 36) and validation group (n = 194; MACE = 36) with permuted-block randomization to create and validate a risk score for predicting MACE in HHD. Age, atrial fibrillation, and concentric hypertrophy were selected because they are traditionally strong risk factors (8). Continuous variables were made binary by the use of generally accepted external cutoff points (age >70 years [21,22] and GLS >–16% ) to mimic clinical categorization and construct a simple, general-purpose, easily implemented scoring system. The parameter with the lowest regression coefficient among the 4 variables in the multivariable Cox model was assigned a numeric value of 1, and the other 3 variables were assigned scores based on the values of their regression coefficients relative to that of the lowest value, and rounded to the nearest integer. The score was derived by summation of the assigned numeric score. The developed score was validated in the validation sample. The plots of observed versus predicted outcomes were used to evaluate model performance on calibration. Comparisons of area under the curves was performed with the method suggested by DeLong et al. (24).
Interobserver and intraobserver variabilities were examined for GLS and GCS using the intraclass and interclass correlation coefficients. Measurements were performed in a group of 15 randomly selected subjects by 1 observer, and then repeated in 2 weeks apart by 2 observers. Statistical analysis was performed using standard statistical software packages (SPSS software 20.0, SPSS Inc., Chicago, Illinois; and R software version 3.0.2, R Foundation for Statistical Computing, Vienna, Austria), and statistical significance was defined by p < 0.05 (2-tailed).
The 388 patients enrolled in this study mostly had mild HTN (median SBP, 145 mm Hg), with an expected profile of comorbid diseases (Table 1). All were taking antihypertensive treatment, the most prevalent being angiotensin-converting enzyme inhibitors or angiotensin receptor blockers in 56%. The median LVMI was 100 g/m2, and 47% of patients had concentric remodeling, 22% had eccentric hypertrophy, and 31% had concentric hypertrophy. Although LVEF (median EF, 68.0%) and GCS (median, –28.0%) were normal, GLS (median, –16.6%) was abnormal in many patients (Table 1).
Associations between strain parameters and comorbidities
Comorbid diseases were associated with myocardial strain parameters. GLS was significantly associated with heart rate (β = 0.05 [95% confidence interval (CI): 0.03 to 0.07]; p < 0.01), diabetes (β = 1.08 [95% CI: 0.30 to 1.86]; p < 0.01), and atrial fibrillation (β = 0.85 [95% CI: 0.11 to 1.59]; p = 0.02), but not age, male sex, body mass index, SBP, chronic obstructive pulmonary disease, cerebrovascular disease, renal function, and anemia. When the patients were divided into groups with (n = 102) and without (n = 286) antihypertensive treatment, there were no significant associations between SBP and GLS in either group. Similarly, GCS was significantly associated with heart rate (β = 0.09 [95% CI: 0.06 to 0.13]; p < 0.01) and atrial fibrillation (β = 1.98 [95% CI: 0.67 to 3.30]; p < 0.01), but not age, male sex, body mass index, SBP, diabetes, chronic obstructive pulmonary disease, cerebrovascular disease, renal function, or anemia.
Online Table 1 shows the comparison of study parameters based on the geometric pattern of LVH. Patients with concentric hypertrophy had significantly impaired renal function, and diastolic function compared with those with concentric remodeling. However, GCS and LVEF were impaired in the patients with eccentric hypertrophy, but not significantly different between concentric remodeling and concentric hypertrophy groups.
During follow-up (median, 4.0 years [interquartile range (IQR): 2.3 to 6.4 years]), 72 patients (19%) suffered MACE (35 deaths and 37 hospital admissions caused by MACE, including 12 admissions with HF, 13 with acute myocardial infarction, and 12 with stroke).
Associations between baseline study parameters and MACE
Table 2 compares the clinical and echocardiographic characteristics in patients with HHD with and without MACE. MACE was significantly associated with age, higher heart rate, lower SBP, higher prevalence of atrial fibrillation, higher use of beta-blockers, higher prevalence of concentric hypertrophy, higher systolic pulmonary artery pressure, and greater impairment of diastolic function and GLS. Figure 1 shows representative cases comparing GLS traces in 1 patient with MACE with 1 patient without MACE.
Effects of HTN treatment on the associations between GLS, SBP, and MACE
To confirm the effect of HTN treatment on the associations between GLS and MACE, and between SBP and MACE, subgroup analyses were performed in patients with and without antihypertensive treatment, and with and without beta-blockers. GLS in the patients with beta-blockers were significantly lower than those without beta-blockers (–15.8 [IQR: –13.8 to –17.8] vs. –16.8 [IQR: –15.1 to –18.6]; p < 0.01). In this subgroup analysis, the associations between GLS and MACE were not significant in both patients with beta-blockers (hazard ratio [HR]: 1.10 [95% CI: 0.99 to 1.24]; p = 0.06) and those without beta-blockers (HR: 1.09 [95% CI: 0.99 to 1.19]; p = 0.08), but their effect sizes were similar. In addition, the association in patients without antihypertensive treatment showed similar results (HR: 1.18 [95% CI: 0.99 to 1.42]; p = 0.07). However, although a negative association between SBP and MACE was found in patients with antihypertensive treatment (HR: 0.98 [95% CI: 0.97 to 1.00]; p = 0.01), this association was not found in the patients without antihypertensive treatment (HR: 1.00 [95% CI: 0.96 to 1.03]; p = 0.42).
Independence and incremental value of GLS
In the multivariable Cox regression analyses (Table 3), age, SBP, atrial fibrillation, and concentric hypertrophy were significant predictors of MACE. GLS was independently associated with MACE in both clinical and echo models, with similar HRs. In sequential Cox models, the model based on clinical variables was significantly improved by the addition of concentric hypertrophy, and furthermore improved by adding GLS (Figure 2).
Score for predicting MACE in HHD
We selected 4 variables as explained in the methods section. After the multivariable analysis in the prediction of MACE with 4 parameters using the derivation cohort, each was assigned a numeric value based on its relative effect (Table 4). A score was constructed by adding the numeric values of the factors identified in each patient, and the score ranged from 0 to 6. The accuracy of the score was investigated in the validation cohort. Using receiver operating characteristic curve analysis for the prediction of MACE, the score showed good discriminative ability in overall and subgroups, which was significantly better than concentric hypertrophy alone (Figure 3). In overall and subgroups, a total score of ≥3 showed similar sensitivity (58%) and specificity (71% to 72%). Model calibration was retained when testing the models (Online Figure 2). Interestingly, the discriminative ability of this score for HF (area under the curve: 0.87; 95% CI: 0.79 to 0.95) was better than that for MACE. A total score of ≥3 showed good discrimination for admission for HF (sensitivity, 92%; specificity, 68%).
Associations between temporal change of study parameters and MACE (study 2)
Among 55 patients who did not have MACE between the paired echocardiograms (median interval, 3.0 years [IQR: 2.0 to 4.8 years]), 10 patients had MACE after the second echo. The median follow-up duration after the second echo was 1.8 years (IQR: 0.9 to 3.5 years). Table 5 shows the serial change of study parameters and the associations between percent change of study parameters and MACE. The deterioration of left atrial volume index, E/eʹ, and GLS were significantly associated with MACE. Increased LVMI, LV volume, delta GCS, and RWT were not associated with MACE. Additionally delta GLS was significantly correlated with delta LVMI (β = –0.23 [95% CI: –0.43 to –0.03]; p = 0.02) and delta left atrial volume index (β = –0.17 [95% CI: –0.28 to –0.06]; p < 0.01), but not delta RWT (β = –0.03 [95% CI: –0.20 to 0.15]; p = 0.76) (Online Figure 3). Although multivariable models were limited by the sparse number of events, the significant associations between delta GLS and MACE were still observed even after adjustment for baseline age, delta LVMI, and delta left atrial volume index (HR: 0.29 to 0.37; all, p < 0.01).
The intraobserver and interobserver intraclass correlation coefficients were 0.96 (95% CI: 0.89 to 0.99) and 0.93 (95% CI: 0.78 to 0.98) for GLS, and 0.93 (95% CI: 0.83 to 0.98) and 0.88 (95% CI: 0.64 to 0.96) for GCS.
This study of HHD has several principal findings. First, LV strain (specifically, GLS), was associated with MACE, independent of and incremental to clinical parameters and concentric hypertrophy. Second, a multiparametric score including GLS could predict MACE with good accuracy, better than concentric hypertrophy alone. Finally, the deterioration in LV longitudinal function during midterm follow-up (but not the changes of LV morphology, and LV circumferential function) was also associated with MACE.
Echocardiographic predictors of MACE
HF (4–8), myocardial infarction, and stroke are the most frequent cardiovascular consequences of HTN (8,10). To date, echocardiographic features predicting these adverse outcomes have been limited to structural changes (LVH and abnormal LV geometric pattern) (1–3). Subclinical LV systolic dysfunction has recently been recognized in patients with HTN (13,25–29), with LV strain being recognized as improving prognostic assessment over the conventional clinical or echocardiographic parameters in a variety of heart diseases with preserved LVEF (11). The present study demonstrated that GLS was associated with MACE, independent of clinical parameters and concentric hypertrophy in HHD. Furthermore, its deterioration was also associated with MACE. Although chronic HTN and its comorbidities likely promote the occurrence of atherosclerotic events, the independent association of GLS with MACE suggests that the mechanism of this association is based on the detection of subtle myocardial damage induced by chronic HTN. Moreover, the association of MACE with deterioration of GLS, in the absence of such an association with LVH and GCS, suggests that LV longitudinal function might be more easily damaged by chronic pressure overload and several comorbidities rather than LV geometry and LV circumferential function.
Although the associations between GLS and MACE were not significant in subgroups without HTN treatment or beta-blockers, their effect sizes were similar with that in all patients. These results suggest that the effect of HTN treatment on the association between GLS and MACE might be small.
The role of changes in LV circumferential function has been controversial, relative to the LV longitudinal function in HHD (13,25–29). In the present study, GCS was not associated with MACE, and its temporal change was minor. This result suggests relative preservation of GCS in comparison with LV longitudinal function in HHD, as previously reported (26,28,29). However, there is a possibility that circumferential strain might not correctly reflect LV mid-wall shortening because of endocardial border tracking.
In the present study, there was no significant association between SBP and GLS even in patients without antihypertensive treatment. This might be because most patients had moderate HTN and were clustered around 145 mm Hg, resulting in a small effect of SBP (afterload) on GLS (LV myocardial deformation) compared with the effects of the HTN-related comorbidities, such as diabetes and atrial fibrillation.
Although events in HHD are associated with several clinical risk factors and LVH and LV geometry (4,8), there is a paucity of risk models applicable to HHD to date. A multiparametric risk score may be of value in strengthening cardioprotective therapy in addition to a focus on BP alone. In the present study, the combined score including clinical risks and GLS predicted MACE with good discrimination ability (C statistic, 0.71) and it demonstrated significantly better prediction than concentric hypertrophy, which is the current echocardiographic risk marker. Furthermore, this score seemed to be more effective for predicting HF-specific outcome, a finding that is concordant with the implied association between asymptomatic LV dysfunction and the onset of HF.
Our data should be interpreted in the context of their limitations. First, because the group comprised asymptomatic patients who had a relatively mild clinical condition, the patients who undertook follow-up echocardiography and the number of MACE in these patients were small. More patients are necessary to confirm the independent association between delta GLS and MACE. Second, although conventional clinical risks and treatments were assessed, there is a risk of unmeasured confounders. Third, 3-dimensional measurement of LV mass is considered to have greater reliability than 2-dimensional measurement (30), but it was not available in this study. Fourth, the specific cause of death was unavailable. Chronic HTN and several comorbidities related to reduced GLS could induce systemic organ damage (8), which could lead to noncardiovascular death. In addition, previous papers have shown that LV mass was associated with both cardiovascular mortality and all-cause mortality in patients with HTN (1,31). Finally, there is a vendor variability of GLS measurement, which could affect its optimal cutoff value for predicting MACE.
LV longitudinal function is associated with MACE, independent of and incremental to clinical parameters and LVMI in HHD. Deterioration in strain was more strongly associated with MACE than were LVH and LV circumferential function, reflecting a higher susceptibility to chronic HTN and higher sensitivity to MACE of GLS. A multiparametric risk score including GLS may lead to better appreciation of a risk of MACE in HHD.
COMPETENCY IN MEDICAL KNOWLEDGE: The myocardial consequences of chronic systolic and diastolic hypertension include not only myocyte hypertrophy, but also perivascular and myocardial fibrosis and medial thickening of the intramyocardial coronary arteries. All of these features make hypertension a risk factor for the development of MACE, but the extent to which functional evidence of hypertensive heart disease might predict MACE is undefined.
TRANSLATIONAL OUTLOOK: This study of hypertensive heart disease found that LV strain (specifically, GLS) was associated with MACE, independent of and incremental to clinical parameters and concentric hypertrophy. In addition, a multiparametric score including GLS predicted MACE better than concentric hypertrophy alone. Deterioration in LV longitudinal function during midterm follow-up was also associated with MACE.
For a supplemental table and figures, please see the online version of this article.
Dr. Marwick reported research collaboration with GE Medical Systems and Siemens. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose. Theodore Abraham, MD, served as Guest Editor for this paper.
- Abbreviations and Acronyms
- confidence interval
- early diastolic mitral annular velocity
- transmitral early diastolic velocity
- ejection fraction
- global circumferential strain
- global longitudinal strain
- heart failure
- hypertensive heart disease
- hazard ratio
- interquartile range
- left ventricular
- left ventricular hypertrophy
- left ventricular mass index
- major adverse cardiac events
- relative wall thickness
- systolic blood pressure
- Received June 17, 2015.
- Revision received September 11, 2015.
- Accepted September 29, 2015.
- American College of Cardiology Foundation
- Verdecchia P.,
- Schillaci G.,
- Borgioni C.,
- et al.
- Drazner M.H.
- Kalam K.,
- Otahal P.,
- Marwick T.H.
- Holland D.J.,
- Marwick T.H.,
- Haluska B.A.,
- et al.
- Lang R.M.,
- Bierig M.,
- Devereux R.B.,
- et al.
- Mor-Avi V.,
- Lang R.M.,
- Badano L.P.,
- et al.
- Stanton T.,
- Leano R.,
- Marwick T.H.
- Narayanan A.,
- Aurigemma G.P.,
- Chinali M.,
- Hill J.C.,
- Meyer T.E.,
- Tighe D.A.
- Mizuguchi Y.,
- Oishi Y.,
- Miyoshi H.,
- Iuchi A.,
- Nagase N.,
- Oki T.
- Kraigher-Krainer E.,
- Shah A.M.,
- Gupta D.K.,
- et al.