Author + information
- Received May 22, 2002
- Revision received July 26, 2002
- Accepted August 1, 2002
- Published online March 5, 2003.
- Mei Wang, MD*,
- Gabriel W.K Yip, MRCP*,
- Angela Y.M Wang, MB*,
- Yan Zhang, MD*,
- Pik Yuk Ho, BN*,
- Mui Kiu Tse, RN*,
- Peggo K.W Lam, MA† and
- John E Sanderson, MD, FRCP, FACC*,* ()
- ↵*Reprint requests and correspondence:
Dr. John E. Sanderson, Department of Medicine and Therapeutics, The Chinese University of Hong Kong, 9/F Clinical Sciences Building, Prince of Wales Hospital, Shatin, N. T., Hong Kong SAR, China.
Objectives The aim of this study was to ascertain if left ventricular mitral annulus velocities measured by tissue Doppler imaging (TDI) are more powerful predictors of outcome compared with clinical data and standard Doppler-echocardiographic parameters.
Background Tissue Doppler imaging of basal or mitral annulus velocities provides rapid assessment of ventricular long axis function. But it is not known if TDI-derived velocities in systole and diastole add incremental value and are superior to the standard Doppler-echocardiographic measurements as a predictor of outcome.
Methods The study population consisted of 518 subjects, 353 with cardiac disease and 165 normal subjects who had full Doppler two-dimensional–echocardiographic studies with measurement of mitral inflow velocities in early and late diastole, E-wave deceleration time (DT), peak systolic mitral annular velocity (Sm) early and late diastolic mitral annular velocity (Em and Am) by TDI, early diastolic flow propagation velocity, and standard chamber dimensions. All subjects were followed up for two years. The end point was cardiac death.
Results Tissue Doppler imaging mitral annulus systolic and diastolic velocities were all significantly lower in the non-survivors (all p < 0.05) as was DT (p = 0.024). In the Cox model the best predictors of mortality were Em, Sm, Am, left ventricular ejection fraction, left ventricular mass, and left atrial diameter in systole (LADs). By backward stepwise analysis Em and LADs were the strongest predictors. After forcing the TDI measurements into the covariate model with clinical and mitral DT <0.16 s, Em provided significant incremental value for predicting cardiac mortality (p = 0.004).
Conclusions Mitral annulus velocity measured by TDI in early diastole gives incremental predictive power for cardiac mortality compared to clinical data and standard echocardiographic measurements. This easily available measurement adds significant value in the clinical management of cardiac patients.
A variety of indices derived from Doppler-echocardiography have been used to predict outcome in patients with left ventricular (LV) dysfunction including LV cavity dimensions, ejection fraction, and mitral inflow velocities. Shortening of the early diastolic deceleration time (DT) of the mitral E-wave suggests impaired LV filling and increased left atrial (LA) pressure and it has been shown to be a strong predictor of an adverse outcome in symptomatic and asymptomatic individuals with LV dysfunction (1–3). However, the confounding effects of changes in loading conditions can significantly affect the measurements based on Doppler recordings of ventricular filling velocities. Recently, tissue Doppler imaging (TDI) which measures the velocity of the myocardium during the cardiac cycle has been used to assess systolic and diastolic function (4). Tissue Doppler imaging can be used to measure mitral or tricuspid annulus velocities that reflect ventricular function in the long axis (5,6). Several studies have shown that the early mitral annulus velocity is a relatively preload-independent assessment of LV relaxation (7,8), and the ratio of peak early diastolic mitral inflow velocity (E) over the myocardial velocity can be used to estimate LV filling pressure (8–10). However, these measurements based on myocardial velocities have not been compared directly with other standard Doppler-echocardiographic measurements, including DT, for predicting prognosis. Thus, the purpose of this study was to examine whether TDI derived parameters added incremental value to clinical and other standard Doppler-echocardiographic measurements to predict cardiac mortality in patients with a variety of cardiac diseases and ventricular function.
The study population consisted of 518 subjects (243 females and 275 males), mean age 57.5 years (range 19 to 90 years) who were referred to the Echocardiographic Laboratory at the Prince of Wales Hospital during 1999 to 2000. Patients with a variety of cardiac diseases and ventricular function were prospectively recruited as well a group of entirely normal subjects (n = 165) without any objective evidence of cardiac or other diseases were studied.
Echocardiograms were obtained using commercially available ultrasound equipment (GE-VingMed System FiVe with a 3.5 MHz transducer, General Electric-VingMed Sound AB, Horten, Norway). All patients were examined at rest in the left lateral decubitus position. All recordings were performed by one investigator (M. W.). The echocardiographic techniques and calculations of different cardiac dimension and volumes were performed according to the recommendations of the American Society of Echocardiography (11,12). Left ventricular ejection fraction (LVEF) by two-dimensional echocardiography was obtained by modified biplane Simpson’s method from apical four- and two-chamber views.
Left ventricular dimensions and wall thickness were made in parasternal long axis with M-mode cursor positioned just beyond the mitral leaflet tips, perpendicular to the long axis of the ventricle. Left ventricular diameter in diastole and systole, LV mass, and fracture shortening were measured. Left atrial dimension at the end of systole (LADs) was assessed by M-mode cursor through the aortic valve in parasternal long-axis view.
Mitral flow velocities
The mitral flow velocities were recorded with pulsed wave Doppler with the sample volume placed at the tip of the mitral valve tips from the apical four-chamber view. From the mitral valve inflow velocity curve the following measurements were made: peak E-wave velocity and its DT, peak A-wave velocity, and the isovolumic relaxation time was measured from aortic valve closure to the mitral valve opening.
Myocardial velocities were recorded using a standard pulse-wave Doppler technique as previously described (8,9,13). Color-coded tissue Doppler images were acquired over a predetermined two consecutive cardiac cycles for each of four mitral annular segments and were transferred to a workstation composed of a personal computer whose software package provides customized image visualization, processing, and analysis (Echopac, GE-VingMed, Norway). The sample volume was placed at the junction of the LV wall with the mitral annulus of the septal and lateral myocardial segments from the four-chamber view and inferior and anterior myocardial segments from the two-chamber view. Peak velocities during systole (Sm), early diastole (Em), and late diastole (Am) were measured. The final value represented the average of four sites.
Color M-mode Doppler propagation velocities of LV flow in early diastole was obtained from the apical four-chamber view. An M-mode cursor was placed as parallel as possible through the center of the mitral inflow, and position was adjusted to obtain the longest column of color flow from the mitral annulus to apex. Velocities were acquired by a rainbow color system, which was adjusted using pulse repetition frequency. Propagation velocity was measured by the slope along the bright yellow (aliasing) isovelocity line during early filling, from the mitral valve plane to 4 cm distally into the LV cavity as previously recommended (14).
All patients were prospectively followed for two years. Death and mode of death were identified from hospital records or telephone contact with relatives. Cardiac death was defined as death caused by heart disease including sudden death. Ischemic heart disease and heart failure were the most common cause.
Continuous data are expressed as mean and 95% confidence interval (CI). Comparison between groups of continuous variables were tested by unpaired ttest. Each variable was evaluated by using Cox proportional hazard survival analysis for cardiac death. The univariate variables that had significant difference with the occurrence of a cardiac death were identified. Multivariate analysis of the univariate variables was then performed and adjusted for age and LV/LA geometry to identify independent predictors of cardiac death with each model (p < 0.05). Each model was the variables identified to have significant relationship to cardiac mortality interpreted by age and LV/LA geometry. Patients who died from other causes of death or lost to follow-up were censored. The incremental value of TDI over clinical data and mitral inflow variables was assessed by a modified stepwise procedure in three modeling steps in the same order as in clinical practice. The first step consisted of fitting a multivariate model of clinical data used as baseline risk factors. Then mitral inflow variables were added in a stepwise backward selection manner to the clinical model. The third step consisted of adding TDI variables to the second model. A significant improvement in model prediction was based on the likelihood ratio statistic, which follows a chi-square distribution and the p value was based on the incremental value compared to the previous model.
Cumulative survival curves were performed by Kaplan-Meier method. Subgroups of life-table curves were compared using Cox regression model. A value of p < 0.05 was considered significant.
The study population was followed-up for a mean of 23 months (range 0.3 to 36 months). Forty-six patients (8.88%) died and 33 patients (6.37%) died due to a cardiac cause (1 subject was lost to follow-up). Causes of cardiac death were an ischemic event (acute coronary syndrome with or without myocardial infarction), heart failure, or sudden death. Thirteen patients died due to other causes: 4 died from severe peritonitis associated with peritoneal dialysis, 2 from cerebral vascular accidents, 2 from cancer, 1 from pneumonia, 1 from postoperative complications (non-cardiac), and 3 deaths were unknown.
Standard clinical and echocardiographic measurements and outcome
The relationship between etiology and subsequent mortality rates are shown in Table 1. The association between the clinical and echocardiographic variables and the end point was assessed by univariate analysis as shown in Table 2. Hazard ratios for these variables are also shown. The differences in echocardiographic parameters between those alive or dead are shown in Table 3. There was a significant difference in age and LV/LA geometry between the survivors and non-survivors except for LV diameter in systole. After all geometric variables were entered into Cox regression analysis, LADs was the only geometric variable predicting the cardiac mortality rather than LV diameter in diastole and systole (hazard ratio [HR]: 2.96, 95% CI: 1.13 to 7.78). There was also a significant difference in LVEF by two-dimensional echocardiography and fractional shortening between those alive and cardiac deaths as shown in Table 3. After each of these significant variables was entered into Cox regression model interacted by age and LV/LA geometry, LVEF remained associated with cardiac deaths (HR: 0.84, 95% CI: 0.73 to 0.97).
Left ventricular mass was significantly lower in the survivors compared to those who died (293.38 ± 135.47 vs. 397.39 ± 134.63, p < 0.0005). Univariate analysis showed that LV mass was significantly associated with cardiac death (HR: 1.004, 95% CI: 1.002 to 1.006). After LV mass was entered into Cox regression model interacted by age and LV/LA geometry, it was still strongly related to cardiac mortality (HR: 1.005, 95% CI: 1.001 to 1.009).
After the clinical features were adjusted by stepwise backward manner in the multivariate analysis, significant clinical risk factors for cardiac death were ischemic heart disease, diabetes, heart failure, and chronic renal failure (p < 0.05) (Table 4).
Doppler mitral inflow parameters
Early diastolic mitral inflow velocity and DT <0.16 s were associated with cardiac death by using univariate Cox regression analysis (Table 2). In the multivariate analysis DT <0.16 s (HR: 2.43, 95% CI: 1.13 to 5.24) was the most important predictor of cardiac death compared to other Doppler mitral inflow parameters. The DT <0.16 s also improved predictive cardiac mortality compared to clinical data (p = 0.025) as shown in Figure 1.
There was significant difference between the survivors and non-surviors for Sm, Em, and Am (Table 3). There was no difference in E/Am. Univariate associations with cardiac mortality was shown in Table 2. The Sm, Em, Am remained significant amongst clinical data and DT <0.16 s after adjustment for age and LV/LA geometry in the Cox regression analysis. Using the stepwise incremental model, TDI variables provided incremental value for risk stratification, additional to clinical data and DT <0.16 s as indicated by the increase of the chi-square of the incremental model by the use of TDI echocardiographic data (p = 0.008) in Figure 1. However, in the multivariate (backward stepwise) analysis Em had a strongest impact on cardiac mortality amongst TDI variables as shown in Table 4.
The parameters of E/Em and E/left ventricular flow propagation velocity (Vp) (both indirect measures of LV filling pressures) were also associated with cardiac mortality. After both measurements were entered into the covariate model with clinical data and DT <0.16 s, E/Em (HR: 1.02, 95% CI: 0.995 to 1.052) was a stronger predictor compared to E/Vp but it did not provide incremental value to clinical and DT <0.16 s (p = 0.13).
Cardiac mortality was compared by Kaplan-Meier analysis according to tertiles of Em or Sm: ≤3, 3 to 5, and >5 cm/s (Figs. 2 and 3). ⇓When Em or Sm was >3 but ≤5 cm/s, the HR of cardiac death was significantly ⇓increased compared with Em or Sm >5 cm/s (HR of Em: 12.79, 95% CI: 2.92 to 56.01; HR of Sm: 5.73, 95% CI: 1.87 to 17.60) Furthermore, when Em or Sm was ≤3 cm/s, the HR of cardiac death was significantly increased compared with Em or Sm >5 cm/s (HR of Em: 28.47, 95% CI: 6.50 to 124.63; HR of Sm: 20.55, 95% CI: 6.86 to 61.62).
Similarly, cardiac deaths were compared by Kaplan-Meier analysis according to tertiles of Am: ≤4, 4 to 7, and >7 cm/s. When Am was >4 but ≤7 cm/s, the HR of cardiac death was increased compared with Am >7 cm/s (HR: 4.28, 95% CI: 1.54 to 11.87). When Am was ≤4 cm/s, the HR of cardiac death was significantly increased compared with Am >7 cm/s (HR: 11.53, 95% CI: 4.10 to 32.39).
The E/Em was also analyzed with respect to cardiac deaths according to tertiles: ≤15, 15 to 20, and >20. When E/Em was >15 but ≤20, the HR of cardiac death was increased but not statistically significantly compared with E/Em ≤15 (HR: 4.75, 95% CI: 0.96 to 23.58). When E/Em was >20, the HR of cardiac death was significantly increased compared with E/Em ≤15 (HR: 20.00, 95% CI: 5.95 to 67.23).
In this study in patients with a variety of cardiac diseases TDI parameters, especially Em, were the most powerful predictors of cardiac death in the subsequent two years, and provide significant incremental prognostic value compared with clinical information and variables derived from mitral inflow velocities such as a mitral E-wave DT <160 ms. Although the ratio E/Em did not give incremental prognostic value, it was another powerful predictor of cardiac death. Those patients with an Em ≤3 cm/s, Sm ≤3 cm/s, Am ≤4 cm/s, and E/Em >20 were at most at risk of cardiac death in the following two years.
TDI variables and prognosis
Several studies that have assessed the prognostic significance of echocardiographic derived measurements, such as LVEF (15,16), LADs (17), LV mass (18), systolic and diastolic time intervals (19), and particularly those derived from Doppler studies of mitral inflow velocities such as DT (1–3). Studies that have used TDI, however, are limited. Our results show for the first time the value of Em for determining cardiac mortality in a large number of subjects with a variety of cardiac vascular etiologies. The final mechanism in these different diseases is probably the same, that is, ischemia of subendocardial fibers. Tissue Doppler imaging enables regional and global myocardial systolic and diastolic velocities to be measured. The velocities derived from the annulus or LV base primarily reflect longitudinal motion, due to the longitudinally directed fibers, which are found in the subendocardium (5,20). This may explain why these measurements are so useful for the assessment of the consequences of ischemia, to which the subendocardium is particularly sensitive (20–22). When the annulus/basal velocity is averaged from four sites— septal, lateral, inferior, and anterior—it reflects global function. The Sm has been shown to be a good measurement of global systolic function (6)and can detect abnormal systolic function in patients with heart failure and a normal ejection fraction (diastolic heart failure) (23). The Em appears to be a good indicator of diastolic function and correlates well with the time constant of isovolumic relaxation (Tau) (6,24,25). Nagueh et al. (26)also demonstrated that load increases on average raised the transmitral E velocity by 70%, whereas the same manipulations produced only a 13% change in Em. Therefore, low Em values are indicative of abnormal LV relaxation even when LV filling pressures are increased. However, age does affect TDI variables especially Em (27). Our results show that TDI parameters were still associated with cardiac mortality even after adjusting for age-related changes based on a large number of normal subjects. Therefore, both Sm and Em appear to be good independent indices of systolic and diastolic function respectively, and although Em is marginally superior as a prognosticator they are intrinsically linked as systolic function determines to some extent LV relaxation in early diastole (27).
LV filling pressure and prognosis
Recently, E/Vp and E/Em have been reported to be good noninvasive correlates of pulmonary capillary wedge pressure (28)and LV filling pressure (8–10). Moller et al. (29)found them to be powerful predictors of the composite end point of cardiac death and readmission due to heart failure after a first myocardial infarction during a median follow-up of 13 months . By contrast, in our study we found that E/Em was a more powerful predictor of future cardiac mortality than E/Vp. The difference between our study and the one by Moller et al. (29)may be that in their study they used a composite end point which included heart failure readmissions increasing bias towards those with an already high filling pressure.
Our results showed that heart rate also correlated with cardiac death in the univariate analysis, but it was excluded in the multivariate stepwise analysis. Some studies have shown that tachycardia may increase the late mitral inflow velocity, but it does not appear to influence the early DT (28,29). Secondly, LVEF was another important predictor related to outcome. But it was interacted by DT and was not included in the multivariate analysis. The main aim of the study was to determine if TDI parameters provide incremental value compared with mitral diastolic velocities and DT, which are commonly used. Thirdly, our study end point was cardiac mortality including sudden death. Sudden death is contentious and although it is assumed this is due to a cardiac arrhythmia this cannot be proved. Not all these patients had an autopsy to exclude other possibilities, such as pulmonary embolism or cerebral hemorrhage, although it is unlikely these would constitute more than a fraction of cases.
The TDI derived parameters Sm, Em, and Am are powerful predictors of cardiac mortality. An Em <3 cm/s, Sm <3 cm/s, Am <4 cm/s, and E/Em >20 can identify patients at very high risk of cardiac death in the subsequent two years.
- late diastolic mitral annular velocity
- confidence interval
- mitral deceleration time
- early diastolic mitral inflow velocity
- early diastolic mitral annular velocity
- hazard ratio
- left atrial
- left atrial diameter in systole
- left ventricular
- left ventricular ejection fraction
- systolic mitral annular velocity
- tissue Doppler imaging
- left ventricular flow propagation velocity
- Received May 22, 2002.
- Revision received July 26, 2002.
- Accepted August 1, 2002.
- American College of Cardiology Foundation
- Giannuzzi P.,
- Temporelli P.L.,
- Bosimini E.,
- et al.
- Xie G.Y.,
- Berk M.R.,
- Smith M.D.,
- et al.
- Henein M.Y.,
- Gibson D.G.
- Sohn D.W.,
- Chai I.H.,
- Lee D.J.,
- et al.
- Nagueh S.F.,
- Middleton K.J.,
- Kopelen H.A.,
- et al.
- Nagueh S.F.,
- Mikati I.,
- Koplen H.A.,
- et al.
- Schiller N.B.,
- Shah P.M.,
- Crawford M.,
- et al.
- Schiller N.B.
- Nagueh S.F.,
- Lakkis N.M.,
- Middleton K.J.,
- et al.
- Nelson G.R.,
- Cohn P.F.,
- Gorlin R.
- Kelly M.J.,
- Thompson P.L.,
- Quinlan M.F.
- Eriksson S.V.,
- Caidahl K.,
- Hamsten A.,
- et al.
- Simpson I.A.
- Alam M.,
- Hogherd C.,
- Thorstrand C.,
- et al.
- Henein M.Y.,
- Priestley K.,
- Davarashvili T.,
- et al.
- Yip G.,
- Wang M.,
- Zhang Y.,
- Sanderson J.E.
- Ohte N.,
- Narita H.,
- Hashimoto T.,
- et al.
- Nagueh S.F.,
- Sun H.B.,
- Kopelen H.A.,
- et al.
- Garcia M.J.,
- Ares M.A.,
- Asher C.,
- et al.