Relationship of Multidirectional Myocardial Strain with Radial Thickening and Ejection Fraction and Impact of Left Ventricular Hypertrophy. A Study in a Community-Based Cohort

Abstract

Background

Left ventricular (LV) systolic strain provides additional prognostic value to LV ejection fraction (LVEF) and wall motion analysis. However, the relationship between myocardial multidirectional strain and LVEF, and the effect of LV hypertrophy on this relationship, are not completely understood especially in unselected populations.

Methods

LV global longitudinal (ε_L) and circumferential (ε_C) systolic strain analysis was performed by two-dimensional speckle-tracking echocardiography in 215 participants from a community-based study. LV radial wall thickening was measured as global radial strain (ε_R), and LVEF was assessed by biplane Simpson’s method.

Results

ε_R was significantly associated with ε_C (β=−0.56, p<0.01) and with ε_L (β= −0.18, p<0.01). The contribution of ε_L to ε_R was especially evident in subjects with lower ε_C and in presence of LV hypertrophy (β= −0.30, p<0.01). ε_L and ε_C were significantly associated with LVEF (β= −0.36 and β=−0.49, both p<0.01) independent of LV mass and other confounders, and their interaction significantly improved the prediction of LVEF (R-square change=0.14) but not of ε_R (R-square change=0.002).

Conclusions

ε_R is mainly related to ε_C with a smaller contribution of ε_L, which becomes especially evident in subjects with lower ε_C and in presence of LV hypertrophy. Therefore, radial thickening may not detect subclinical LV longitudinal function reduction in normal ventricles and when ε_C is preserved. While a reduction in ε_L has a limited impact on ε_R, it exerts a greater effect on global LVEF, therefore for a more accurate LVEF prediction both ε_L and ε_C need to be considered.

In clinical practice, left ventricular (LV) systolic function is commonly measured at the endocardial level (LV chamber function), and is usually assessed by echocardiographically determined ejection fraction (LVEF). The analysis of LV wall motion, which is often performed by visual estimation of systolic myocardial thickening, is also a widely used method for assessing LV systolic function.^1,2 More recently, speckle-tracking echocardiography has made possible the non-invasive assessment of myocardial deformation (strain) over different spatial axes.^3,4 Myocardial strain analysis has provided new insights in the understanding of LV mechanics far beyond the information provided by LVEF. Past studies showed that, even in the presence of a normal LVEF, LV strain may be selectively affected in the longitudinal or circumferential direction depending on the pathophysiology and the extension of the damage to the myocardial wall.^5–7 Furthermore, the analysis of LV myocardial strain appears to carry prognostic value that is greater than, and additional to, the traditional LV systolic function evaluation by LVEF and wall motion analysis.⁸

In previous studies including a wide range of LV conditions, from normal to ischemic heart disease and heart failure, good correlations were found between LVEF and LV longitudinal strain,^1,9 to the point that the latter has been proposed as an alternative method to quantify global LVEF.¹⁰ However, theoretical considerations and the evidence of the additional prognostic value of longitudinal strain over LVEF, along with its ability to uncover systolic dysfunction when LVEF is still normal, are reasons to believe that these two parameters express different aspects of LV function and should not be considered as equivalent.^11,12 LVEF and LV wall thickening are, in fact, the final result of the contraction of a complex system of fibers in the myocardial wall,¹³ in which the subendocardial and subepicardial fibers have prevalent longitudinal direction and are arranged in counter-directional helical loops, and the midwall fibers have prevalent circumferential orientation.¹⁴ This architectural structure determines the systolic longitudinal shortening and torsion of the LV, and the thickening of the myocardium in the radial direction, which in turn causes the shift of the endocardium towards the LV cavity.^15,16

Since the relationships of longitudinal and circumferential LV wall systolic components with radial thickening and LVEF have not been completely characterized especially in unselected populations without overt cardiac disease, we aimed at analyzing the association of the longitudinal and circumferential components of LV wall systolic function with LV radial thickening and global LVEF in participants from a community-based cohort study. Furthermore, we evaluated the impact of LV hypertrophy on these associations. The study of these relationships would provide a better understanding of LV mechanics and potentially clarify the basis for the additional prognostic information provided by the multidirectional assessment of myocardial deformation over the assessment of wall motion and global LVEF.

METHODS

Study population

The present study was performed in participants from the Cardiac Abnormalities and Brain Lesions (CABL) study, a community-based cohort study whose primary aim was to investigate cardiovascular predictors of subclinical cerebrovascular disease. CABL parent study is the Northern Manhattan study (NOMAS). NOMAS and CABL study design and methodological details have been described previously.^17,18 Subjects were contacted by using random digit dialing of both published and unpublished telephone numbers. Briefly, patients were enrolled if they: 1) were at least 50 years of age; 2) lived in a pre-defined geographic area of northern Manhattan for at least 3 months in a household with a telephone; and 3) did not have a prior diagnosis of stroke. Study participants for whom speckle-tracking strain analysis over the three dimensional axes was available constitute the sample of the present study. The study was performed in accordance with the ethical standards of the Declaration of Helsinki and was approved by the Institutional Review Board of Columbia University Medical Center. Written informed consent was obtained from all study participants.

Cardiovascular risk factors were ascertained through direct examination and interview by trained research assistants. Hypertension was defined as systolic blood pressure ≥140 mm Hg or diastolic blood pressure ≥90 mm Hg at the time of the visit (mean of two readings), or history of hypertension or use of anti-hypertensive medications. Diabetes mellitus was defined as fasting blood glucose ≥126 mg/dL or history of diabetes or of use diabetes medications. Hypercholesterolemia was defined as total serum cholesterol >240 mg/dL, history of hypercholesterolemia or of use of lipid-lowering treatment. Coronary artery disease (CAD) was defined as prior myocardial infarction, coronary artery bypass grafting, or percutaneous coronary intervention.

Echocardiographic Assessment

Transthoracic echocardiography evaluation was performed using a commercially available system (iE 33, Philips, Andover, MA) by a trained registered cardiac sonographer. LV diameters and wall thickness were measured from a parasternal long-axis view according to the recommendations of the American Society of Echocardiography (ASE).¹⁹ LV mass was calculated with a validated formula²⁰ and indexed by body surface area (LV mass index). LV hypertrophy was defined using the ASE guidelines sex-specific cutoffs.¹⁹ Left ventricular relative wall thickness, an index of LV geometry, was calculated with the formula: (2×posterior wall thickness)/end-diastolic diameter.²¹ LV ejection fraction (LVEF) was calculated using the biplane modified Simpson’s rule. Speckle-tracking analysis of LV systolic function was performed offline (QLAB Advanced Quantification Software version 8.1, Philips) as previously described.²² The software utilizes an image-processing algorithm that analyzes myocardial motion by frame-to-frame tracking of stable patterns of pixels (“speckles”) in two-dimensional gray-scale image loops. Global longitudinal peak systolic strain (ε_L) was calculated averaging longitudinal strain over 12 segments from the apical four and two-chamber views (Fig. 1, panels a and b). From manually selected landmark points in apical views, endocardial borders are automatically detected by the software, and each wall is divided in to three segments (basal, mid and apical). Manual correction was performed when LV wall was not correctly sampled. Subsequently, automatic tracking of myocardial speckles was performed throughout the cardiac cycle. Strain curves were visualized for each segment, and the average peak systolic strain was calculated. LV circumferential and radial functions were assessed from a parasternal short axis view at the papillary muscle level (Fig. 1, panels c and d). From manually selected anatomical landmark points (circumference delimiting the outer LV myocardial border, center of the circumference, limit of the inferior and anterior interventricular septum), the software identifies LV endocardial and epicardial borders and divides the LV in six segments. Then, tracking of myocardial speckles is performed throughout the entire cardiac cycle. Manual adjustments were performed when border recognition was not optimal. Peak systolic circumferential strain (ε_C) and peak systolic radial strain (ε_R) were calculated averaging the peak systolic strain values from the analyzed ventricular segments. Segments with poor tracking quality (either visually estimated or based on internal software algorithms) were excluded from the calculation of the average global strains. Three cycles were recorded at a frame rate > 45 fps, and were averaged for strain analysis. Aortic and mitral valve opening and closing times were measured from the LV outflow Doppler profile and were incorporated in the speckle-tracking strain profile in order to measure systolic strain and to exclude post-systolic components.

Fig. 1.

Speckle-tracking analysis of LV strain. Measurement of longitudinal systolic strain from the apical views (a and b). Measurement of systolic circumferential (c) and radial strain (d) from the parasternal mid-ventricular short-axis view.

Reproducibility of speckle-tracking measurements

Intra-and inter-observer reproducibility of speckle-tracking measurements were assessed in a random sample of 20 study participants. Intra-observer intraclass correlation coefficients (ICC) were: 0.82 for ε_L (95% CI: 0.60–0.93, p<0.01), 0.89 for ε_C (95% CI: 0.74–0.95, p<0.01), 0.92 for ε_R (95% CI: 0.82–0.97). Mean difference between measurements was 0.07±2.3% for ε_L, 0.37±2.2% for ε_C, and 0.47±3.8%for ε_R (p values non-significant for all comparisons); coefficients of variation (standard deviation/mean) were 8.4% for ε_L, 8.0% for ε_C, and 9.6% for ε_R. Inter-observer ICCs were 0.85, 0.82 and 0.83 respectively for ε_L, ε_C and ε_R. Mean difference between measurements was 0.08±2.4% for ε_L, 0.88±2.6% for ε_C, 1.15±5.2% for ε_R (p values non-significant for all comparisons); coefficients of variation were 9.2% for ε_L, 9.7% for ε_C, and 11.2% for ε_R.

Statistical Analysis

Data are presented as mean ± standard deviation for continuous variables and as proportions for categorical variables. One-way ANOVA was performed to test differences in ε_R and LVEF between tertiles of ε_L and ε_C. Univariate linear regressions were used to assess the correlates of ε_L and ε_C. Multivariate linear regressions were performed to assess the independent association of LV deformation parameters and LVEF. Predictors and outcome variables were standardized with corresponding standard deviations and parameter estimates (β) and relative standard errors were reported. A stepwise selection of the covariates was utilized, with entry and removal criteria set at a probability level of 0.1 and 0.2 respectively. For all statistical analyses, a two-tailed p<0.05 was considered significant. Statistical analyses were performed using SPSS software version 17.0 (SPSS Inc., Chicago, IL).

RESULTS

Clinical, demographic, and echocardiographic characteristics of the 215 study participants are shown in Table 1. Mean LVEF was 61.5±9.0%. 33 subjects (15.3%) had LVEF <55%, of these 24 had LVEF between 45–54%, 4 had LVEF between 30–44% and 5 had LVEF <30%.

Table 1.

Clinical and echocardiographic characteristics of the study sample

Clinical/demographics	N=215
Age, years	68.6 ± 10.3
Body mass index, kg/m2	27.3 ± 4.4
Women, n (%)	123 (57.2)
Hypertension, n (%)	150 (69.8)
Diabetes, n (%)	55 (25.6)
Hypercholesterolemia, n (%)	129 (60.0)
CAD, n (%)	5 (2.3)
Echocardiographic data
LV end-diastolic diameter, mm/m2	26.1 ± 3.3
LV mass index, g/m2	104.6 ± 26.6
LV hypertrophy, n (%)	87 (40.5)
Relative wall thickness	0.48 ± 0.08
LVEF, %	62.5 ± 8.6
LVEF < 55%, n (%)	33 (15.3)
Speckle-tracking data
εL, %	−17.2 ± 3.6
εC, %	−18.6 ± 4.4
εR, %	27.5 ± 8.1

LVEF: Left ventricular ejection fraction. CAD: Coronary artery disease ε_L: Global systolic longitudinal strain. ε_C: Global systolic circumferential strain. ε_R: Global systolic radial strain.

Correlates of ε_L and ε_C

Univariate correlates of longitudinal and circumferential function were assessed in linear regression analyses. ε_L was significantly correlated with age (β=0.15, p<0.05), LV mass (β =0.24, p<0.01), male sex (β=0.14, P<0.05), hypertension (β=0.24, p<0.01), and diabetes (β =0.15, p<0.05). ε_C was significantly correlated with LV mass (β=0.33, p<0.01) and hypertension (β=0.14, p<0.05).

Relation of ε_L and ε_C with radial thickening and LVEF

To analyze the relationship of ε_L and ε_C with ε_R and LVEF, the study population was divided into tertiles of ε_L and ε_C. Subsequently, two matrix tables were built crossing ε_L and ε_C tertiles in order to obtain nine groups of subjects with different ε_L and ε_C combinations. In Table 2, ε_R was calculated for each group. The group in the lowest tertile of ε_L and ε_C had the lowest ε_R (19.7±6.9%), whereas the group with highest ε_L and ε_C had the highest ε_R (34.4±6.7%, p<0.01). ε_R significantly increased with increasing values of ε_C in every tertile of ε_L (all p<0.01). On the other hand, ε_R significantly increased with increasing ε_L only in the group with lowest ε_C (p<0.01). A similar analysis was then performed with LVEF (Table 3). The lowest LVEF was found in the group with lowest ε_L and ε_C (51.3±14.9%), and the highest LVEF was found in the highest ε_L/ε_C tertile combination (67.6±3.6%, p<0.01). LVEF increased with increasing ε_C in all ε_L tertiles (all p<0.01). LVEF, however, followed a different pattern than ε_R, also increasing significantly with increasing ε_L in all ε_C tertiles (all p<0.05). The frequency of a LVEF < 55% in each group also followed this pattern.

Table 2.

ε_R values in ε_L and ε_C tertiles

		εL
		1st tertile N=69	2nd tertile N=73	3rd tertile N=73	P value*
εC	1st tertile N=78	19.7 ± 6.9	23.6 ± 5.3	25.1 ± 5.7	<0.01
	2nd tertile N=64	25.9 ± 4.6	27.6 ± 6.5	28.8 ± 4.1	0.30
	3rd tertile N=73	32.6 ± 10.2	31.5 ±6.7	34.4 ± 6.7	0.35
	P value*	<0.01	<0.01	<0.01

Values in table are mean ε_R (%)± standard deviation. ε_L: Global systolic longitudinal strain. ε_C: Global systolic circumferential strain. ε_R: Global systolic radial strain.

^*One-way ANOVA between tertiles.

Table 3.

LVEF in ε_L and ε_C tertiles

		εL
		1st tertile N=69	2nd tertile N=73	3rd tertile N=73	P value*
εC	1st tertile N=78	51.3 ± 14.9 (55.6%)	60.6 ± 4.2 (16.0%)	61.1 ± 5.3 (5.9%)	<0.01
	2nd tertile N=64	59.1 ± 6.3 (22.2%)	62.9 ± 5.1 (10.0%)	64.1± 4.1 (0%)	<0.05
	3rd tertile N=73	64.5 ± 5.4 (6.7%)	65.3 ± 3.2 (0%)	67.6 ± 3.6 (0%)	<0.05
	P value*	<0.01	<0.01	<0.01

Values in table are mean LVEF (%)± standard deviation. Values in parentheses indicate the % of subjects with a LVEF <55% in each group. ε_L: Global systolic longitudinal strain. ε_C: Global systolic circumferential strain.

^*One-way ANOVA between tertiles.

The different impact of strain components on radial thickening and LVEF was further assessed by linear regression models (Table 4). In unadjusted analyses, both ε_L and ε_C were significantly associated with ε_R and with LVEF; however, the relative contribution of ε_L toward LVEF (β= −0.36) was higher than that toward ε_R (β= −0.18), whereas the parameter estimate of ε_C towards LVEF was slightly lower than that towards ε_R (β= −0.49 and −0.56, respectively). These results were confirmed also after adjusting for age, sex, body mass index, hypertension, diabetes and relative wall thickness. Subsequently, we tested whether an interaction was present between ε_L and ε_C in the relationship with ε_R and LVEF (Table 4, model 3). No interaction was found between ε_L and ε_C in relation to ε_R, whereas a significant interaction was found towards LVEF (β for the interaction term=−0.41, p<0.01). The analysis was performed separately in participants with and without LV hypertrophy to assess differences in LV mechanics due to LV mass increase (Table 4). Interestingly, ε_L was significantly associated with ε_R only in participants with LV hypertrophy, whereas ε_C was significantly and strongly associated with ε_R in both groups (all p<0.01). The significant association of ε_L and ε_C with LVEF was confirmed in both groups (all p<0.01); parameters estimates, however, were higher for ε_C than for ε_L in participants without LV hypertrophy, and were of similar magnitude in those with LV hypertrophy. R-square values of linear models predicting ε_R and LVEF (Figure 2) showed how the model including ε_L and ε_C predicted LVEF better than either term alone, and also better than ε_R; the inclusion of the interaction between ε_L and ε_C significantly improved the global fit of the model in predicting LVEF (R²-change=0.14) but not ε_R (R²-change=0.002).

Table 4.

Relationship of ε_L and ε_C with ε_R and LVEF in all participants and in those with and without LV hypertrophy

	All				LVH −				LVH +
	εR		LVEF		εR		LVEF		εR		LVEF
	β	P value	β	P value	β	P value	β	P value	β	P value	β	P value
Model 1
εL	−0.18	<0.01	−0.36	<0.01	−0.10	0.16	−0.26	<0.01	−0.30	<0.01	−0.44	<0.01
εC	−0.56	<0.01	−0.49	<0.01	−0.57	<0.01	−0.46	<0.01	−0.50	<0.01	−0.49	<0.01
Model 2*
εL	−0.20	<0.01	−0.36	<0.01	−0.14	0.07	−0.30	<0.01	−0.34	<0.01	−0.40	<0.01
εC	−0.56	<0.01	−0.48	<0.01	−0.58	<0.01	−0.49	<0.01	−0.46	<0.01	−0.44	<0.01
Model 3
εL − εC interaction	−0.05	0.34	−0.41	<0.01	−0.06	0.38	−0.36	<0.01	−0.04	0.68	−0.39	<0.01

Values in table are standardized correlation coefficients (β).ε_L: Peak systolic longitudinal strain. ε_C: Peak systolic circumferential strain. ε_R: Peak systolic radial strain. LVH: Left ventricular hypertrophy. LVEF: Left ventricular ejection fraction.

^*Adjusted for age, sex, body mass index, hypertension, diabetes, and relative wall thickness.

Fig. 2.

R-square values for linear models predicting ε_R and LVEF. On the x-axis are shown the independent variables included in each model: ε_R in the first model, ε_L in the second model, ε_C in the third model, ε_L and ε_C in the fourth model, ε_L and ε_C plus their interaction (ε_L*ε_C) in the fifth model.

DISCUSSION

In this study, we analyzed the LV systolic function measuring myocardial deformation by two-dimensional speckle-tracking echocardiography, in the attempt to elucidate the relationship between the multidirectional components of LV deformation and global LV function (LVEF). The main findings of our study are: (1) ε_C is the strongest contributor to LV radial thickening, with a smaller but also significant contribution from ε_L which becomes especially evident when ε_C is reduced and in hypertrophic ventricles; (2) ε_L and ε_C are significantly associated with LVEF and show significant interaction in their effect; and (3) a reduction in ε_L may not significantly affect radial thickening if ε_C is preserved, but it has a greater impact on LVEF. Because LVEF is correlated to both ε_L and ε_C, an important consequence of our findings is that, given a specific value of strain over one axis, LVEF distribution can vary significantly depending on the value of the orthogonally oriented strain. Those findings were confirmed after controlling for factors such as age, sex, body mass index, hypertension, diabetes, and LV relative wall thickness.

Relation of LV longitudinal and circumferential strain with radial thickening

As a result of a force applied in a given direction, an elastic body experiences two opposite deformations: one on the same direction of the force (linear or principal strain) and one on the perpendicular direction (lateral strain).²³ In the contracting heart, the shortening of longitudinal and circumferential fibers causes a deformation of LV segments both in the longitudinal and circumferential direction (negative principal strains), therefore concentrating all the lateral (positive) strain in the radial direction, resulting in wall thickening. This “cross-fiber shortening”, allows the heart to efficiently maximize LV fiber contraction in order to obtain high radial thickening values from the relatively limited shortening (10–15%) at the level of the sarcomere.^15,24,25 Consistently with this model, our study showed how both ε_L and ε_C are significant contributors to radial thickening. This contribution, however, is not equal. ε_C showed to be a greater contributor to ε_R than ε_L especially when LV hypertrophy is absent. This finding may be due at least partly to the fact that the circumferentially-oriented fibers constitute the majority of fibers in the normal LV wall,²⁶ and that in hypertrophic ventricles the relative contribution of longitudinal fibers increases thus giving a greater contribution to wall thickening. Also, we found that the contribution of ε_L to wall thickening was significant only in the lowest ε_C tertile group, when the LV circumferential function is reduced.

Relation of LV deformation with LVEF

While radial function is determined by LV wall fiber contraction and thickening, LVEF is influenced by the radial wall thickening and by the reduction of the LV long and short axes.^27,28 LV longitudinal strain by speckle-tracking provides a good estimate of the LV long axis reduction and of the systolic motion of the mitral annulus,²⁹ and this is likely the reason of the greater impact of ε_L on LVEF than on ε_R (the longitudinally oriented fibers are less represented in the LV wall, thus impacting less radial thickening, however the long axis reduction will translate in a greater effecton LVEF). We demonstrated that, besides their significant main effect on LVEF, a significant statistical interaction was present between ε_L and ε_C in determining LVEF. A possible explanation of this interaction may be found in the duplex action of longitudinal and circumferential shortening, which on one hand cause the LV long-axis shortening and the short axis radius reduction and on the other hand contribute to the radial thickening. It is known, in fact, that LV long and short axes reduction have greater impact than radial wall thickening on LVEF, as elegantly demonstrated in a pioneer study by Bogaert and Rademakers.³⁰ In that study, the authors showed how in the normal human left ventricle, regions with higher wall thickening did not necessarily produce a higher regional EF, whereas regions with higher longitudinal and circumferential shortening were associated with higher regional EF. Our data suggest that this may be particularly true for ε_L, which showed limited impact on radial thickening (probably due to the lesser representation of longitudinal fibers in the LV wall) but had a more significant effect on LVEF. Finally, the longitudinal fibers are also major determinants of the LV systolic twist, further contributing to LV ejection.³¹

Clinical implications

Our study showed that the measurement of only one strain component, either longitudinal or circumferential, does not provide an accurate estimate of LVEF or radial thickening in an unselected population with mostly normal LVEF. We showed that the best LVEF prediction was reached when both ε_L and ε_C, and their interaction, were taken into account. Previous studies that showed stronger correlations between LVEF and ε_L were performed in small selected populations often including both normal subjects and patients with very depressed LVEF due to ischemic heart disease, dilated cardiomyopathy and heart failure.^1,10,11 However, when LVEF is severely depressed, ε_L will also be very likely depressed, and their overall correlation may result excellent especially if normal or supernormal LVEFs are included in the distribution. We are the first to show that in an unselected population, which in large part included participants with normal LVEF but with a wide range of risk factors, global systolic function depends on the interaction of both ε_L and ε_C, and that this interaction is very heterogeneous. In fact, we showed that LVEF, even when in the normal range, may be associated to a heterogeneous array of wall mechanics. This heterogeneity is at least in part associated with some cardiovascular risk factors and cardiac structure features. Hypertension, diabetes, CAD and a higher LV mass were associated with a reduced longitudinal function, whereas an increased LV mass and hypertension were associated with a reduced circumferential function. In this respect, we confirm findings from other studies in hypertensive patients that reported a reduction in absolute values of LV longitudinal, circumferential or radial function.^32–37 Another consequence of our findings is that traditional wall motion analysis may not reveal subclinical LV dysfunction in presence of a reduction of longitudinal systolic function if the circumferential function is normal or increased. This is also in line with previous studies showing that myocardial wall deformation provides prognostic information additional to the traditional wall motion assessment.⁸

Study limitations

Although the main three-dimensional axes were explored, other components of the LV wall that contribute to determine LV systolic function, such as LV torsion, were not available for all subjects in our study. The interaction between differently oriented fibers inside the LV wall generates complex secondary strain patterns (shear strain) that could not be taken into account in our study. Also, the interaction between the three principal strains is a complex three-dimensional phenomenon, therefore the use of two-dimensional speckle-tracking might not fully describe all the components of this interactions. These additional components may have improved the prediction of LVEF allowed by the individual myocardial deformation components. Finally, our population was an unselected sample from the community, mostly including subjects with significant representation of cardiovascular risk factors; therefore our data may not apply to subjects with LV structure and function significantly affected by disease processes not represented in our cohort (acute ischemia, cardiomyopathies, post-ischemic remodeling, etc.).

Conclusions

In an unselected population from the community, neither ε_L nor ε_C, when considered alone, provide a good estimate of LVEF; therefore, for an accurate LVEF prediction, both ε_L and ε_C need to be considered. ε_R is mostly influenced by ε_C and less by ε_L; thus, radial thickening, often used to visually estimate LV systolic function, may not be sensitive enough to detect subclinical reductions in LV longitudinal function especially when circumferential function is preserved and in absence of LV hypertrophy. These observations may at least in part explain the additional prognostic value of ε_L over traditional parameters of LV systolic function observed in previous studies.