The Relationship Between Left Ventricular Wall Thickness, Myocardial Shortening, and Ejection Fraction in Hypertensive Heart Disease: Insights From Cardiac Magnetic Resonance Imaging

Abstract

Hypertensive heart disease is often associated with a preserved left ventricular ejection fraction despite impaired myocardial shortening. The authors investigated this paradox in 55 hypertensive patients (52±13 years, 58% male) and 32 age‐ and sex‐matched normotensive control patients (49±11 years, 56% male) who underwent cardiac magnetic resonance imaging at 1.5T. Long‐axis shortening (R=0.62), midwall fractional shortening (R=0.68), and radial strain (R=0.48) all decreased (P<.001) as end‐diastolic wall thickness increased. However, absolute wall thickening (defined as end‐systolic minus end‐diastolic wall thickness) was maintained, despite the reduced myocardial shortening. Absolute wall thickening correlated with ejection fraction (R=0.70, P<.0001). In multiple linear regression analysis, increasing wall thickness by 1 mm independently increased ejection fraction by 3.43 percentage points (adjusted β‐coefficient: 3.43 [2.60–4.26], P<.0001). Increasing end‐diastolic wall thickness augments ejection fraction through preservation of absolute wall thickening. Left ventricular ejection fraction should not be used in patients with hypertensive heart disease without correction for degree of hypertrophy.

In systemic hypertension, left ventricular (LV) hypertrophy (LVH) may occur in the face of increased afterload.1 The development of hypertensive LVH is pathological and is an independent predictor for sudden cardiac death,2 ventricular arrhythmias,3 coronary artery disease4 and heart failure,5 which is often in the context of a normal LV ejection fraction (LVEF). Hypertrophied hypertensive myocardium is associated with reduced long‐axis shortening (LAS), yet LVEF, which is a traditional marker of LV systolic function, remains in the normal range,6 leading some to believe that hypertensive heart disease is a diastolic disorder. The apparent paradox of global myocardial systolic long‐axis dysfunction but normal LVEF has previously been explained by a “compensatory increase in short‐axis shortening”.7 However, observational data do not support this hypothesis; for example, there are abnormalities of both midwall and longitudinal fractional shortening in patients with hypertensive hypertrophic LV disease.8, 9, 10

An alternative theoretical explanation argues that an increase in end‐diastolic LV wall thickness (EDWT) leads to increased end‐systolic LV wall thickness (ESWT) and a correspondingly augmented absolute wall thickening (AWT), where AWT is the difference between ESWT and EDWT. Assuming there is no significant change in the external diameter of the left ventricle during systole,11 the resultant absolute displacement of the endocardial border could be normal and therefore LVEF would be maintained. This concept has been demonstrated by mathematical modeling of concentric LVH.12, 13 Such mathematical modeling controls for, and therefore removes, the potential impact of other variables such as body surface area, heart rate, blood pressure (BP), ventricular‐arterial coupling, peripheral vascular resistance, and abnormalities in LV relaxation.

We sought to accurately determine the biophysical relationship between EDWT, longitudinal and midwall myocardial shortening, AWT, and LVEF in a human hypertensive cohort using cardiac magnetic resonance imaging (CMR) since it is the gold standard for noninvasively assessing LV mass and volume.14 A detailed assessment of the relationship between these variables will provide better understanding of the pathophysiology of hypertensive heart disease and may have important implications for refining our understanding of the mechanisms of heart failure with normal ejection fraction, with implications for its treatment.

Materials and Methods

Study Patients

Fifty‐five hypertensive patients (age 52±13 years; male sex, 58%; office systolic BP [SBP], 174±29 mm Hg, office diastolic BP [DBP], 98±16 mm Hg) were recruited from the Bristol Heart Institute tertiary hypertension clinic between 2011 and 2013. Baseline demographic and clinical characteristics were recorded. In order to investigate hypertensive LVH, exclusion criteria consisted of clinical or CMR evidence of any concomitant myocardial pathology that may confound the hypertrophic response (eg, previous myocardial infarction, moderate to severe valvular heart disease, dilated cardiomyopathy, suspected athlete's heart, and hypertrophic or infiltrative cardiomyopathy).

Average office SBP and DBP values were acquired from both arms in all patients after seated rest, assessed using standard automated sphygmomanometry with an appropriately sized cuff.15 Standard 24‐hour ambulatory BP monitoring (ABPM) was also performed.16

A cohort of 32 age‐ and sex‐matched normotensive healthy volunteers (age, 49±11 years; male sex, 56%; office SBP 126±12 mm Hg, office DBP 77±10 mm Hg) free from cardiovascular disease, with normal BP and electrocardiographic (ECG) findings, and taking no regular medication, were used as an age‐ and sex‐matched control group.

The study complied with the Declaration of Helsinki. The local research ethics committee confirmed that this study conformed to the governance arrangements for research ethics committees. Patients provided written consent for their CMR results to be used for research.

CMR Protocol

All CMR studies were performed at 1.5T (Avanto; Siemens, Erlangen, Germany) using a spine coil, a body array coil, and retrospective ECG triggering by specialist CMR technicians. Steady‐state free precession (SSFP) end‐expiratory breath‐hold cines were acquired in the standard four‐, two‐, and three‐chamber cardiac long‐axis planes and in the LV short‐axis view from the atrioventricular ring to the apex. Representative CMR parameters were as follows: repetition time (TR) 40.05 ms, echo time (TE) 1.13 ms, slice thickness 8 mm, no interslice gap, field of view 260×320mm, flip angle 75^o, in‐plane voxel size 2×2 mm.

CMR Analysis

All CMR analysis was performed by a single CMR reader, with >4 years of CMR experience, using cvi42 software (Circle Cardiovascular Imaging Solutions Incorporated, Calgary, AB, Canada). LV volume, LVEF, and LV mass were estimated using established clinical methods as previously described.17 Mass to volume ratio (M/V) was calculated as previously described.18 These data were acquired blinded to the myocardial fractional shortening measurements, and vice versa. Changes in lengths in the longitudinal and radial directions of the myocardium between end‐diastole and end‐systole have recently been demonstrated to represent a simple technique to quantify myocardial strain relative to both in vivo myocardial tagging and validated finite element computation modeling techniques.19 Consequently, we used a six‐point mitral annular plane systolic excursion indexed to the LV end‐diastolic length as a measure of global LAS as previously described.17 EDWT and ESWT were measured in the middle of each of the basal and mid‐LV myocardial segments on the long‐axis cines with measurements performed perpendicular to the LV wall. Measurements of the thickness of the apical segments were prone to partial volume averaging, particularly at end‐systole, as a result of the conical configuration of the left ventricle and therefore were not included in the analysis. EDWT has previously been demonstrated to have better levels of interobserver and intraobserver agreement when measured from long‐axis compared with short‐axis cines.20 Furthermore, it was easier to take into account the effect of through‐plane motion of the mitral valve during systole using long‐axis cines compared with the short‐axis cines using mitral valve plane tracking software (cvi42; Circle Cardiovascular Imaging Solutions Incorporated). Papillary muscles and trabeculae were excluded from wall thickness measurements. Relative wall thickness was defined as EDWT indexed to left end‐diastolic diameter. AWT was defined as the absolute difference between ESWT and EDWT. Radial strain (RS) was defined as the percentage increase in wall thickness (ie, engineering strain or relative wall thickening). LV internal diameters at end‐diastole (LVIDd) and at end‐systole (LVIDs) were measured at the LV basal and midlevels from the long‐axis cines. All measurements were repeated twice and the mean value used for subsequent analysis. Midwall circumferential fractional shortening (mFS) was estimated using the following established equation, which has been previously described10:

mFS (%) = ([LVIDd + EDWT] – [LVIDs + H])/(LVIDd + EDWT) × 100.

Where:

H = ([LVIDd + EDWT]³ – [LVIDd]³ + [LVIDs]³)^1/3 – LVIDs.

Statistical Analysis

Statistical analysis was generated using the Real Statistics Resource Pack software (release 3.2.1) and SPSS version 21 (IBM Corp, Armonk, NY). Categorical variables were analyzed using Fisher exact tests. Normally distributed continuous variables were expressed as mean±standard deviation and compared using unpaired Student t tests or one‐way analysis of variance with Bonferroni post hoc correction for between‐groups comparisons, as appropriate. Multiple linear regression analysis was performed to investigate independent determinates of EDWT, LAS, and mFS on LVEF. The impact of multicollinearity was excluded by acceptable values in variance inflation factor tests. Reproducibility was assessed by intraclass correlation coefficient (two‐way mixed, absolute agreement, average measures). The intraobserver intraclass correlation coefficient for EDWT was 0.962 (95% confidence interval [CI], 0.956–0.968), for ESWT was 0.966 (95% CI, 0.959–0.971), LV end‐diastolic length was 0.987 (95% CI, 0.984–0.989), and for LV end‐systolic length was 0.990 (95% CI, 0.987–0.992). Statistical significance was set at two‐tailed P<.05.

Results

Demographics

The demographic and baseline clinical data for hypertensive patients and normotensive controls are documented in Table 1. The study sample was stratified into tertiles by EDWT. There were significantly more women with EDWT <9 mm compared with both EDWT 9 m to 11 m and EDWT >11 mm, respectively (P<.05). No other demographic differences were demonstrated across the study population. There were no statistically significant differences in the antihypertensive medication regimens between the hypertensive cohorts. There were step‐wise increases in both M/V and RWT with increasing EDWT, consistent with increasing concentric LVH with increasing EDWT in our sample of patients. EDWT correlated with office SBP (R=0.43, P<.001) and office DBP (R=0.32, P<.005) but did not correlate with ABPM SBP (R=0.24, P=.12), ABPM DBP (R=0.18, P=.27), or ABPM mean arterial pressure (R=0.18, P=.27).

Table 1.

Demographics and Baseline Clinical Characteristics of Hypertensive Patients and Normotensive Controls

	Controls	EDWT <9 mm	EDWT 9 mm to 11 mm	EDWT >11 mm	P Value
	(n=32)	(n=16)	(n=21)	(n=18)
Demographics
Age, y	49±11	47±15	55±10	52±11	.112
Male sex, %	56	13	76	78	<.001 a
BMI, kg/m2	26±5	29±4	30±4	32±5	<.001 b
Blood pressure, mm Hg
Office SBP	126±12	174±27	172±30	176±32	<.001 c
Office DBP	77±10	97±16	100±17	98±17	<.001 c
ABPM SBP	–	154±28	150±12	168±23	.087
ABPM DBP	–	89±18	90±9	97±13	.309
ABPM MAP	–	108±21	105±10	115±13	.192
Anti‐HTN medication
Anti‐HTN medications, No.	–	3±2	4±2	4±2	.266
ACE inhibitor, %	–	38	71	39	.057
ARB, %	–	25	33	61	.074
Calcium channel blocker, %	–	44	2	61	.496
Thiazide diuretic, %	–	44	43	28	.555
Loop diuretic, %	–	13	5	17	.494
K+‐sparing diuretic, %	–	75	86	72	.856
β‐Blocker, %	–	44	38	44	.923

Abbreviations: ABPM, 24‐hour ambulatory blood pressure monitor; ACE, angiotensin‐converting enzyme; ARB, angiotensin II receptor blocker; BMI, body mass index; DBP, diastolic blood pressure; HTN, hypertension; MAP, mean arterial pressure; SBP, systolic blood pressure. ^aPatients with end‐diastolic wall thickness (EDWT) <9 mm vs all other subgroups. ^bControls vs patients with EDWT 9 mm to 11 mm and those with EDWT >11 mm, respectively. ^cControls vs all other subgroups.

Impact of Increasing EDWT on mFS, LAS, and RS

There was a strong positive significant correlation between mFS and LAS (R=0.73, P<.001) (Figure 1A). Both mFS (R=0.84, P<.001) and LAS (R=0.64, P<.001) significantly correlated with RS (Figure 1B and 1C). EDWT correlated with mFS (R=0.68, P<.001) (Figure 2A), LAS (R=0.62, P<.001) (Figure 2B), and RS (R=−0.48, P<.001) (Figure 2C). The hypertensive cohort with EDWT >11 mm had significantly lower LAS and mFS compared with hypertensive patients with EDWT 9 mm to 11 mm, hypertensive patients with EDWT <9 mm, and normotensive controls on pairwise comparisons with Bonferroni adjustment (Table 2).

Figure 1.

Scatter graphs for hypertensive patients and normotensive controls showing the relationship of midwall fractional shortening to long‐axis shortening (A), radial strain to midwall fractional shortening (B), and radial strain to long‐axis shortening (C).

Figure 2.

Scatter graphs for hypertensive patients and normotensive controls showing the relationship of midwall fractional shortening to mean end‐diastolic wall thickness (A), long‐axis shortening to mean end‐diastolic wall thickness (B), and radial strain to mean end‐diastolic wall thickness (C).

Table 2.

LV Volumetric, Myocardial Thickness, and Myocardial Shortening Data for Hypertensive Patients and Normotensive Controls

	Controls	EDWT <9 mm	EDWT 9 mm to 11 mm	EDWT >11 mm	P Value
	(n=32)	(n=16)	(n=21)	(n=18)
LV volumetrics
LVEF, %	64±7	64±6	67±8	67±11	.422
Indexed EDV, mL/m2	77±18	90±11	83±17	81±17	.109
Indexed ESV, mL/m2	29±11	32±8	28±12	27±11	.586
Indexed SV, mL/m2	48±12	56±5	55±10	55±13	.146
Myocardial mass and thickness
LV mass index, g/m2	58±11	81±20	88±11	118±16	<.001 a , b
RWT, mm/mL	0.09±0.02	0.11±0.02	0.13±0.03	0.14±0.03	<.01 c , d
M/V, g/mL	0.76±0.13	0.91±0.14	1.08±0.20	1.45±0.27	<.05 e , f
AWT, mm	4.4±0.9	5.4±1.4	6.5±1.2	6.8±1.3	<.05 a , g
Myocardial shortening
Long‐axis shortening, –%	16±2	13±3	11±2	8±2	<.005 b
Radial strain, %	62±15	68±15	65±12	52±12	<.05 h
mFS, –%	18±2	20±2	18±3	15±3	<.001 a

Abbreviations: AWT, absolute wall thickening; EDV, end‐diastolic volume; ESV, end‐systolic volume; LV, left ventricular; LVEF, left ventricular ejection fraction; M/V, mass to volume ratio; mFS, midwall fractional shortening; RWT, relative wall thickness; SV, stroke volume. ^aControls vs all other subgroups. ^bPatients with end‐diastolic wall thickness (EDWT) >11 mm vs all other subgroups. ^cControls vs patients with EDWT 9 mm to 11 mm and EDWT >11 mm, respectively. ^dPatients with EDWT >11 mm vs those with EDWT <9 mm. ^ePatients with EDWT >11 mm vs all other subgroups. ^fPatients with EDWT 9 mm to 11 mm vs all other subgroups. ^gPatients with EDWT <9 mm vs all other subgroups. ^hPatients with EDWT >11 mm vs EDWT 9 mm to 11mm and those with EDWT <9 mm, respectively.

Impact of Increasing EDWT on ESWT and AWT

As EDWT increased, there was an increase in end‐systolic wall thickness (R=0.92, P<.001) (Figure 3A) and an increase in AWT (R=0.43, P<.005) (Figure 3B). There were significant increases in AWT in all hypertensive subgroups compared with normotensive controls (Table 2) and there were significant increases in AWT in hypertensive patients with EDWT 9 m to 11 m and EDWT >11 m compared with EDWT <9 mm, respectively (Table 2).

Figure 3.

Scatter graphs for hypertensive patients and normotensive controls showing the relationship of end‐diastolic wall thickness (EDWT) to end‐systolic wall thickness (ESWT) (A), EDWT to absolute wall thickening (AWT) (B), EDWT to left ventricular ejection fraction (LVEF) (C), and AWT to LVEF (D).

Impact of Increasing EDWT and AWT on LVEF

Despite reductions in mFS, LAS, and RS with increasing EDWT, there was a borderline significantly weak correlation between EDWT and LVEF (R=0.26, P=.05) (Figure 3C). Absolute wall thickening significantly correlated with LVEF (R=0.70, P<.0001).

Impact of Increasing EDWT on Indexed EDV, Indexed ESV, and Indexed SV

Increasing EDWT negatively correlated with both indexed end‐diastolic volume (EDV; R=−0.37, P<.05) and indexed end‐systolic volume (ESV; R=−0.30, P<.05). However, the reduction in both indexed EDV and indexed ESV with increasing EDWT resulted in no significant difference in indexed stroke volume (SV; R=−0.081, P=.55).

Multiple Linear Regression Analysis

Regression analysis was performed using variables that have a biophysically plausible and direct influence on LVEF. LVEF was set as the dependent variable and EDWT, LAS, and mFS as independent variables. All variables were continuous. EDWT, LAS, and mFS were all independently and positively correlated with LVEF (Table 3). Essentially, a 1 mm increase in EDWT would independently account for an increase in LVEF by an absolute value of 3.43%. The increase in EDWT compensates for the independent reduction in LVEF by an absolute value of 2.01% and 1.05% for a 1.00% absolute reduction in LAS and mFS, respectively.

Table 3.

Multiple Linear Regression Analysis Assessing the Independent Effects of EDWT, LAS, and mFS on LV Ejection Fraction

Variable	Univariate Coefficient of Regression (95% CI)	P Value	Multivariate Coefficient of Regression (95% CI)	P Value
EDWT	0.91 (−0.01 to 1.82)	.051	3.43 (2.60–4.26)	<.0001
LAS	0.86 (0.06–1.66)	.035	2.01 (1.29–2.74)	<.0001
mFS	0.906 (0.23–1.59)	<.01	1.05 (0.26–1.84)	<.01

Abbreviations: CI, confidence interval; EDWT, end‐diastolic wall thickness; LAS, long‐axis fractional shortening; LV, left ventricular; mFS, midwall fractional shortening.

Discussion

Our study investigated the impact of EDWT on LVEF in hypertensive heart disease using segmental engineering strain measurements derived from CMR, the gold‐standard noninvasive cardiac imaging modality for LV wall thickness and function, in a sample of 55 hypertensive and 32 normotensive patients.

The pathophysiology of LVH and its functional consequences are incompletely understood. However, the concept that LV wall geometry affects the LVEF is longstanding.21 De Dumesnil and colleagues22 proposed that wall thickening is the direct reflection of shortening that occurs in the circumferential and longitudinal directions. Subsequent work by Rademakers and colleagues23 demonstrated the importance of cross‐fiber shortening in determining wall thickening in a canine model using tagged cardiovascular magnetic resonance imaging. More recently, the notion that LV geometry impacts ejection fraction (EF) has been reaffirmed by an echocardiographic study by Aurigemma and colleagues.24 This showed that elderly patients with high relative wall thickness maintained their LVEF despite depressed mFS. Palmon and colleagues25 demonstrated that both longitudinal and circumferential shortening were reduced in hypertensive patients with LVH and a normal EF, using tagging CMR. Furthermore, Vinch and colleagues26 demonstrated significantly lower mFS in patients with hypertensive heart disease compared with normal controls, despite unchanged mean endocardial shortening and EF. Similar findings were demonstrated by Koh and colleagues.27 More recently, Mizuguchi and colleagues6 showed reduced longitudinal, circumferential, and radial strain in hypertensive patients with concentric hypertrophy in an echocardiographic study.

It is therefore clear that LVH is associated with abnormalities of both midwall and longitudinal fractional shortening. However, the reason LVEF, a traditional marker of LV systolic function, usually remains in the normal range despite these abnormalities is unclear. A theoretical explanation argues that an increase in EDWT leads to increased ESWT and a correspondingly augmented AWT, where AWT is the difference between ESWT and EDWT. Assuming there was no significant change in the external diameter of the left ventricle during systole,11 the resultant absolute displacement of the endocardial border could be normal and therefore maintain LVEF. This concept has been demonstrated by mathematical modeling of concentric LVH.12 Furthermore, in a transesophageal echocardiographic study of 15 patients with hypertension, Frielingsdorf and colleagues28 demonstrated that absolute and fractional wall thickening was inversely related to EDWT, but they did not investigate the relationship between these variables and LVEF.

Our findings that increasing EDWT was accompanied by reduced myocardial longitudinal and circumferential shortening are consistent with the existing literature. However, we additionally demonstrate, for the first time, the independent and significant relationships between both longitudinal and circumferential shortening and EDWT on LVEF using multiple linear regression analysis. While an increase in EDWT independently results in a significant increase in LVEF, it is associated with a reduction in LAS, mFS, and RS, which, in turn, significantly and independently lowers LVEF.

The Importance of AWT

In spite of the worsening strain abnormalities with increasing EDWT in our cohort, both LVEF and indexed SV remained in the normal range for all EDWT values. Our results offer further insights to help explain this apparent paradox. As EDWT increased, we also demonstrated a corresponding increase in wall thickness at end‐systole. As a result, there was a maintained AWT over the range of end‐diastolic wall thicknesses investigated. Consequently, the absolute endocardial displacement remained normal with increased EDWT, despite reduced myocardial shortening. The maintenance of AWT with increasing EDWT is not caused by compensatory radial thickening because there is a concomitant reduction in RS. We provide evidence that the preservation of AWT, and therefore LVEF, is secondary to the degree of LVH defined by EDWT. Of note, the influence of a given amount of AWT in smaller left ventricles will have a greater impact on LVEF. Our study provides in vivo validation of this mechanism described in mathematical modeling experiments of concentric LVH.12 In summary, a ventricle with normal EDWT, LAS, and mFS will result in similar AWT as a ventricle with concentric LVH, reduced LAS, mFS, and RS. The AWT is an important determinant of LVEF and indexed SV in both of these hypothetical scenarios. Our results show how there can be important abnormalities of systolic function, defined by reduced myocardial shortening, and yet normal LVEF and hence explain why LVEF is a poor marker of systolic dysfunction in the context of LVH.

Hypertensive Remodeling

LV remodeling is postulated to be a constructive adaptive physiological response resulting in a normalization of SV.29 We have demonstrated that indexed EDV and indexed ESV reduce with increasing EDWT but indexed SV remains within the normal range as EDWT increases, consistent with previous mathematical modeling.30 This lends weight to the hypothesis that hypertensive LVH occurs concurrently or before the abnormality causing contractile dysfunction.30 In the context of hypertension, reduced contractility of the hypertrophied myocardium is postulated to occur, at least in animal models, as a consequence of production of different myosin heavy chain isoforms mediated by changes in expression of myocardial contractile and metabolic proteins through secondary messenger cell signaling pathways such as phosphoinositides and proto‐oncogenes.31

Implications for Heart Failure With Preserved EF

Our findings may have important implications for understanding the pathophysiology of heart failure with preserved EF. The term heart failure with preserved EF describes patients with an LVEF >45% to 50% but clinical features of heart failure.7 Heart failure with preserved EF is a common disease, affecting approximately 50% of patients with heart failure.32 Myocardial shortening abnormalities are common in heart failure with preserved EF33 and patients often have hypertension and concentric LVH.34 As AWT compensates for myocardial shortening impairment in patients with hypertensive heart disease with normal LVEF and normal indexed SV, it is perplexing why some patients develop clinical symptoms of heart failure in the context of hypertensive heart disease. This is best explained by the blunted cardiovascular response to exercise with failure to augment SV and cardiac index on exertion35 and reduced contractile reserve.36

LV mass is an important determinant of risk in hypertensive heart disease with differing risk dependent on the remodeling pattern.37 We have shown the detailed biophysical relationships between EDWT and myocardial shortening, LVEF, and ventricular volumes. We suggest that geometric patterns seen in hypertension may be explained by the combination of wall thickness and myocardial shortening. A combination of both EDWT and myocardial shortening may give incremental prognostic information over the traditional marker of LVEF alone.

Limitations

The study population size of our investigation was modest. However, our correlations with AWT and our multiple linear regression results were highly significant, suggesting that the sample size was adequate. The latter is likely related to the improved accuracy and reproducibility of CMR compared with echocardiography, which affords a marked reduction in sample size for the same statistical power.38 Our study was limited to hypertensive patients who attended a tertiary hypertension clinic. Most patients are likely to have moderate to severe hypertension, which may preclude extrapolation beyond this particular cohort. However, the study was primarily designed to assess the interaction of EDWT on LVEF and not the impact of severity of hypertension. Nonetheless, there were no significant differences in office SBP or DBP across the subgroups with EDWT <9 mm, 9 mm to 11 mm and >11 mm, respectively (Table 1). Furthermore, modeling data would suggest that only a mild increase in EDWT (eg, >12 mm) is necessary to significantly increase LVEF.12, 39 Obesity can affect LV remodeling and hypertrophy40 and while obesity was common in our sample, there were no significant differences in mean body mass index across patients (Table 1).

Engineer strain was directly calculated by manually measuring LAS and RS. Similar techniques quantifying changes in myocardial length and thickness at end‐diastole and end‐systole have been validated against myocardial tagging and finite element models.19 However, we elected to calculate mFS using a widely used and previously validated equation10 because there is a large circumferential strain gradient (approximately 3%–36%) across the wall of the myocardium, with the subendocardial myocardium being displaced more than the epicardial myocardium.41 Consequently, the midwall myocytes cannot be easily tracked through the cardiac cycle as their position relative to the endocardial and epicardial borders changes continually during LV contraction. As a result, tracking‐derived values for circumferential strain may not be representative of mFS. Nevertheless, it should be noted that the equation used to estimate mFS makes assumptions about LV geometry, as it presumes the external and internal contours are ellipsoidal in systole and diastole. However, there are currently no alternative validated methods. In addition, assumptions are made in the calculations that there is no loss in myocardial muscle volume during contraction even though there is likely to be a small reduction in volume as a result of vascular compression. Finally, through‐plane motion was visually taken into account using our method, but this phenomenon may degrade the accuracy of strain values generated from strain software in the short axis. Our relatively simple measuring method for wall thickness, although using a gold‐standard imaging technique, was reproducible. Further work is required to confirm whether similar techniques can be applied to other settings, potentially introducing the possibility to explore preexisting large cardiac imaging databases.

In light of the limitations, our calculations should be considered only as improved approximations and, in our view, the trends observed and concepts described are likely to be valid and concordant with recent modeling studies.42

Clinical Implications

Hypertensive heart disease is often associated with a normal EF with the supposition that ventricular systolic function is also normal. We, however, have shown that myocardial shortening and strain (and, therefore, cardiomyocyte shortening) gradually decrease as wall thickness increases despite a maintained EF. We have shown that the AWT is a major determinant of EF and that, in turn, AWT is determined by both myocardial shortening and wall thickness. Myocardial shortening (and therefore function) and RS are reduced in the presence of LVH and normal EF. There is no compensatory increase in radial function that normalizes the EF when LAS is abnormal, as previously thought. EF is a poor measure of systolic function, particularly in the setting of hypertrophic ventricles. Our findings have important clinical, physiological, and prognostic implications in both hypertensive heart disease and heart failure with preserved EF.

Conclusions

Our study quantified, for the first time, the relationships between LV myocardial shortening (longitudinal shortening, mFS, and RS), EDWT, AWT, and EF. Our analyses provide additional novel insights into the mechanism by which hypertensive patients can have significant contractile dysfunction and a normal EF. We confirm previous work showing reduced LAS and mFS in the setting of normal LVEF and indexed SV. We demonstrate for the first time, using multiple linear regression analysis, that LAS, mFS, and EDWT are each significantly and independently correlated with LVEF. As EDWT increases, AWT is maintained, which preserves LVEF and indexed SV despite falls in both long‐axis and midwall fractional shortening. The maintenance of AWT is simply a result of increased EDWT and decreased myocardial fractional shortening. Importantly, LVEF and the term systolic function are not synonymous and LVEF should not be used as an accurate index of LV function in the presence of LVH, without correction for the degree of LV end‐diastolic wall thickness.

Conflict of Interest

Dr Bucciarelli‐Ducci is a consultant for Circle Cardiovascular Imaging Inc., Calgary, Canada.

J Clin Hypertens (Greenwich). 2016;18:1119–1127. DOI: 10.1111/jch.12849. © 2016 Wiley Periodicals, Inc.