OUP user menu

Will simultaneous measurement of E/e' index facilitate the non-invasive assessment of left ventricular filling pressure in patients with non-valvular atrial fibrillation?

Chen Li, Juqian Zhang, Can Zhou, Liwei Huang, Hong Tang, Li Rao
DOI: http://dx.doi.org/10.1093/ejechocard/jep218 296-301 First published online: 18 December 2009


Aims The ratio of early transmitral flow velocity (E) to early diastolic mitral annular velocity (e’) was applied as a non-invasive index to assess left ventricular filling pressure (LVFP). However, the reliability of E/e' was undermined in patients with atrial fibrillation (AF). Recently, a novel method entitled ‘dual Doppler’ was established, which allows simultaneous recording and display of E and e'. Our study investigates whether the dual-Doppler method improves the reliability of E/e' in AF patients.

Methods and results Forty-nine patients with non-valvular AF underwent conventional echocardiography, dual-Doppler echocardiography, and cardiac catheterization within 4 h. Of 22 patients (45%c) with increased LVFP (>15 mmHg), higher E/e' measured by both conventional tissue Doppler imaging (TDI) and dual-Doppler method was observed. Conventional echocardiographic variables were correlated with LVFP (E/e'sep, r = 0.404, E/e'lat, r = 0.487), but E/e' measured synchronously in the dual-Doppler mode yielded a better correlation (E/e'synchronous sep, r = 0.754, E/e'synchronous lat, r = 0.765). The intra- and interobserver variability of the dual-Doppler method was significantly lower than the conventional TDI method.

Conclusion Good correlations were found between E/e'' and LVFP in patients with AF, particularly in the dual-Doppler mode. E/e' measured by the dual-Doppler method can therefore be applied to assess diastolic dysfunction in AF patients.

Registry number: ChiCTR-DT-00000342.

  • Atrial fibrillation
  • Diastolic dysfunction
  • Echocardiography
  • Tissue Doppler imaging
  • Dual Doppler


Diastolic dysfunction (DD) is a common cause of congestive heart failure (CHF) in the elderly population.1 Patients with atrial fibrillation (AF) are more susceptible to develop CHF because of tachycardia-mediated cardiomyopathy and diastolic abnormality.28 Meanwhile, the presence of DD can increase both newly onset and recurrence of non-valvular AF.912 Identifying potential DD in AF patients is therefore of great importance.

Echocardiography is the non-invasive method of choice to assess DD in patients with sinus rhythm. However, in AF patients, the echocardiographic evaluation of diastolic function remains a clinical challenge owing to the absence of effective atrial contraction. Attempts have been made to evaluate DD in AF patients using echo parameters independent of atrial contraction, such as transmitral early peak flow velocity (E) divided by tissue Doppler mitral annular motion velocity (e’).1316 The reliability of E/e' index was doubted in AF patients because irregular cardiac rhythm contributes to varying values of E and e'.

A recently established echocardiographic method entitled ‘dual Doppler’ enables pulse wave (PW) Doppler imaging and tissue Doppler imaging (TDI) to be conducted simultaneously. Presumably, a more reliable E/e' value can be acquired using dual Doppler. In a recent study, Kusunose et al.17 have reported synchronous E/e' correlated with plasma BNP and pulmonary capillary wedge pressure in AF patients. However, this study failed to compare both synchronous and conventional E/e' with a gold standard such as left ventricular filling pressure (LVFP) to demonstrate the incremental value of synchronous E/e' in estimating DD.

Our aim was to assess whether the dual-Doppler method could provide a more reliable measurement of E/e' than the conventional method and hence facilitate the non-invasive estimation of DD in AF patients.


Patient population

The investigation conforms with the principles outlined in the Declaration of Helsinki.18 All patients gave written informed consent for invasive diagnostic procedures. The research protocol was approved by the local institutional review committee.

Forty-nine patients with non-valvular AF with a stable haemodynamic status and preserved ejection fraction were recruited. Patients underwent conventional echocardiography and dual-Doppler echocardiography 4 h before the conductance catheterizations which aimed to measure left ventricular end-diastolic pressure (LVEDP). Medications remained unchanged for at least 3 days before catheterization. Blood pressure and heart rate were monitored repeatedly before echocardiography and catheterization.

Conventional echocardiography

Transmitral peak early flow velocities (E) and mitral annulus movement at the septal (e'sep) and lateral (e'lat) corner of the mitral annulus were measured in the apical four-chamber view as previously described19,20 using HI Vision 900 (Hitachi Medical Company, Limited, Tokyo, Japan) with a broadband phased array transducer (EUP-S50A, 2–4 MHz). The frame rate was set to >25 Hz while the Nyquist level was set to 0.8–1.4 m/s. E/e' was calculated using E and e' values as the mean of 10 consecutive Doppler signals.21 Left ventricular ejection fraction (LVEF) was measured using the biplane Simpson's method.

Dual-Doppler measurements

In the dual-Doppler method, two sample volumes can be placed simultaneously, and the transducer receives echo signals from two sample volumes in turn. Therefore, a simultaneous measurement of both Doppler spectra is allowed at the cost of halving the time resolution of Doppler spectra, which was usually at a megahertz level (Figure 1).

Figure 1

(A) The PW sample volume (1) was placed between the tips of two mitral leaflets, while the TDI sample volume (2) was placed at the septal corner of mitral annulus. There is a rapid switching between two sample volumes in dual-Doppler modality. (B) Echo signals of pulse wave 1, 3, 5, 7 … formed Doppler spectrum of transmitral velocity. (C) Signals of pulse wave 2, 4, 6, 8 … formed tissue Doppler imaging spectrum of mitral annulus. Thus, simultaneous measurement of both E and e' was realized.

E/e'synchronous sep was measured in the apical four-chamber view with the PW sample volume positioned between the tips of two mitral leaflets, and the TDI sample volume placed at the septal corner of the mitral annulus. E and e'sep were measured at precisely the same cardiac cycle, and E/e'synchronous sep was derived as E value divided by the corresponding e'sep value. E/e'synchronous lat was measured in the same way except that the TDI sample volume was placed at the lateral corner of the mitral annulus. Both E/e'synchronous sep and E/e'synchronous lat were derived from corresponding mean values of 10 continuous cardiac cycles.


All echo variables were measured repeatedly by a second sonographer to assess the intra- and interobserver reproducibility. To test whether the reproducibility of non-synchronous E/e' can be improved by involving more cardiac cycles, we calculated the non-synchronous E/e' using a mean E and e' value of 50 consecutive cardiac cycles instead of routinely 10 cycles.

Left ventricular filling pressure measurements by conductance catheter

High-fidelity, manometer-tipped catheters were placed into the left ventricle and calibrated as described previously.22 Measurements were taken from at least 20 consecutive cardiac cycles. Mean LVEDP of 10 consecutive cycles were measured to reflect the LVFP.

Statistical analysis

Descriptive characteristics of continuous variables were expressed as mean ± standard deviation (SD). Continuous variables were compared by Student's t-test while categorical data were compared by χ2 test using SPSS (version 15.0, SPSS Incorporated, Chicago, IL, USA). The correlation between echocardiography indices and catheter measurements was analysed using linear regression. Diagnostic accuracies of echo variables were compared by receiver operating characteristic (ROC) curve analysis using MedCalc (version, Mariakerke, Belgium). P < 0.05 was considered significant in all analyses.


Baseline and echocardiographic characteristics

Twenty-seven patients with normal LVEDP were classified as group A and 22 patients with increased LVEDP (>15 mmHg) were classified as group B. Patients in group B had a higher systolic blood pressure (SBP) and a larger dimension of the left atrium (LA) compared with group A. E/e' measured by conventional TDI or dual Doppler were significantly higher in group B patients (Table 1).

View this table:
Table 1

Baseline and echocardiographic characteristics of included patients

Group A (LVEDP ≤ 15 mmHg, n = 27)Group B (LVEDP > 15 mmHg, n = 22)P-value
Age, y (mean ± SD)56.7 ± 10.960.2 ± 11.40.485
Male gender, n (%)13/27(48.1)14 /22(63.6)0.168
BSA (m2)1.70 ± 0.141.71 ± 0.160.861
SBP, mmHg (mean ± SD)126.9 ± 11.4139.3 ± 14.10.017
DBP, mmHg (mean ± SD)78.7 ± 13.881.9 ± 12.90.097
Heart rate (mean ± SD)75.8 ± 9.175.3 ± 10.20.803
CAD, n (%)14/27 (51.8)14/22 (63.6)0.416
LVEDD, mm (mean ± SD)52.5 ± 12.952.7 ± 11.30.904
LVESD, mm (mean ± SD)37.1 ± 8.736.7 ± 10.70.661
IVS, mm (mean ± SD)9.9 ± 1.111.0 ± 1.80.224
LA, mm, (mean ± SD)38.4 ± 8.146.5 ± 10.40.042
LVEF, % (mean ± SD)59.2 ± 7.959.0 ± 6.80.957
E/e'synchronous sep (mean ± SD)9.3 ± 2.714.9 ± 3.2<0.001
E/e'synchronous lat (mean ± SD)7.7 ± 1.812.0 ± 2.6<0.001
E/e'sep (mean ± SD)10.0 ± 3.614.1 ± 3.8<0.001
E/e'lat (mean ± SD)8.1 ± 2.813.5 ± 4.0<0.001
LVEDP (mean ± SD)10.6 ± 3.422.5 ± 4.2<0.001
  • BSA indicates body surface area; SBP, DBP, systolic/diastolic blood pressure; CAD, coronary artery disease; LVEDD, left ventricular end-diastolic diameter; LVESD, left ventricular end-systolic diameter; IVS, interventricular septum; LA, left atrium; LVEF, left ventricular ejection fraction; LVEDP, end-diastolic LV pressure.

The heart rate (75.5 ± 5.8 vs. 73.8 ± 8.1, P = 0.268), systolic pressure (133.0 ± 8.1 vs. 134.3 ± 10.4, P = 0.636) and diastolic pressure (80.5 ± 9.2 vs. 81.7 ± 7.9, P = 0.167) measured before echocardiography and before catheterization had no significant differences.

Predictive value of echocardiographic parameters

E/e'sep and E/e'lat demonstrated significant linear correlation with LVFP (r=0.404 and 0.487, respectively) (Figure 2A and B). As demonstrated by E/e'synchronous sep (r = 0.754, P < 0.001), E/e' measured synchronously in the dual-Doppler mode correlated better with LVFP than in the conventional mode (Figure 2C). An even closer correlation was found between E/e'synchronous lat and LVFP (r = 0.765, P < 0.001) (Figure 2D).

Figure 2

Both E/e'sep and E/e'lat were correlated with left ventricular filling pressure (A and B) while synchronous E/e' measured in the dual-Doppler mode yielded a better correlation (C and D).

E/e'synchronous sep and E/e'synchronous lat showed significantly higher sensitivity and specificity in diagnosing elevated LVFP compared with conventional E/e'sep and E/e'lat (E/e'sep vs. E/e'synchronous sep, z = 2.309, P = 0.017; E/e' lat vs. E/e'synchronous lat, z = 2.702, P = 0.007) (Figure 3).

Figure 3

Receiver operating characteristic curves of four echocardiographic parameters in diagnosing elevated left ventricular filling pressure. Below are area under curves, criterion value, sensitivity, specificity, positive predictive value, and negative predictive value of four echocardiographic variables. Criterion value here refers to the threshold value with the highest specificity and sensitivity for each echo variable, which corresponds to the farthest point on the receiver operating characteristic curve from the diagonal line.


Table 2 summarizes the intra- and interobserver variability of different methods. Using the mean value of 50 continuous cardiac cycles could only slightly improve the reproducibility of non-synchronous E/e'. The dual-Doppler method, however, yielded much better reproducibility than the conventional method.

View this table:
Table 2

Interobserver variability of E/e' index measured in the conventional and dual-Doppler mode

Intraobserver variabilityInterobserver variability
Conventional method (10 cycles)Conventional method (50 cycles)Dual-Doppler methodConventional method (10 cycles)Conventional method (50 cycles)Dual-Doppler method
Septal E/e' (%)**13.412.97.1**
Lateral E/e' (%)**14.911.4*7.9**
  • *P < 0.05 comparing with the conventional method using the mean value of 10 consecutive cardiac cycles.

  • **P < 0.005 comparing with the conventional method using the mean of 10 cycles.


The loss of effective atrial contraction in AF patients prevented a valid application of transmitral and pulmonary vein Doppler echo in patients with AF. E/e', a TDI-derived index of LV filling that is independent of atrial contraction, has been shown to be useful for DD assessment and non-invasive LVFP estimation.2328 However, the accuracy and reproducibility of E/e' index is debated in patients with AF because of the cycle variation. The variation of E/e' index in AF patients is assumed to be caused mainly by the following three elements: (i) the genuine fluctuation of E/e' index, which reflects the changing hemodynamic status in cardiac cycles with different R–R interval; (ii) differences in personal preference between observers; (iii) variation derived from mismatching of E and e' in the conventional Doppler method, where E and e' were measured in different cardiac cycles.

The first element is inherent in patients with AF. The second element can be avoided, or at least minimized by standardizing the procedure of E/e' measurement and proper training of sonographers. Our study focused mainly on the last element. In AF patients with a changing value of E and e', it would be difficult for the observer to match the E value with its corresponding e' so as to calculate the E/e' index accurately. Using a mean value of repeated measurements may to some degree eliminate the random influence of mismatching. E/e' index is usually obtained from 5 to 10 cardiac cycles in clinical setting to balance the feasibility and reliability. Our study demonstrated that using the mean value of as many as 50 consecutive cycles failed to further improve the reproducibility of non-synchronous E/e'.

However, the mismatching can be avoided in the dual-Doppler method. Therefore, a more accurate measurement of E/e' index can be achieved in AF patients. In our study, we found that the intraobserver variability of both the septal and lateral E/e' index was significantly improved in the dual-Doppler mode (9.7 and 10.3% vs. 6.4 and 6.7%), so as the interobserver variability (13.4 and 14.9% vs. 7.1 and 7.9%). Partly because of the better reproducibility, E/e' measured synchronously in the dual-Doppler method had a closer correlation with LVFP than the non-synchronous E/e' index measured in the conventional method. Dual Doppler also improved the sensitivity and specificity of E/e' index in diagnosing elevated LVFP.

Collectively, we believe that the synchronous E/e' index measured by the dual-Doppler method could be more reliable than conventional E/e' in AF patients.

Study limitation

Overall, the patient population was elderly with non-valvular AF and preserved LVEF. Further study should be done before extrapolating our findings to other clinical settings.

E and e' were measured simultaneously in the dual-Doppler mode, but catheterization and echocardiography were done at different times. Beat-to-beat variation can therefore affect the correlation of echocardiographic parameters and LVFP.

Serial changes and the prognostic value of echocardiographic parameters merit further investigation.

Conflict of interest: none declared.


This work was supported by grant from the National Science and Technology Pillar Programme of China in the 11th Five-year Plan (grant number 2007BAI05B01) and Science and Technology Pillar Programme of Sichuan Province (grant number 2009SZ0134).


  • The authors contributed equally to this study.


View Abstract