Tissue Doppler Imaging (E/e’) and Pulmonary Capillary Wedge Pressure in Patients with Severe Aortic Stenosis

Abstract

Objective:

While American and European consensus statements advocate employing the ratio of the transmitral E velocity and tissue Doppler early diastolic mitral annular velocity (E/e’) in the assessment of left heart filling pressures, recent reports have questioned the reliability of this ratio to predict left atrial pressures in a variety of disease states. We hypothesized that there is a clinically significant correlation between E/e’ and pulmonary capillary wedge pressure (PCWP) in patients with severe aortic stenosis.

Design:

Retrospective cohort study

Participants:

733 consecutive patients with severe aortic stenosis who underwent Transcatheter Aortic Valve Replacement (TAVR) for severe aortic stenosis

Interventions:

None

Measurements and Main Results:

PCWP and E/e’_ave (average of the lateral and medial annulus tissue Doppler velocities) were measured using pulmonary artery catheter and transthoracic echocardiography during pre-procedural evaluation. Patients were grouped by left ventricular ejection fraction (LVEF) ≥ 50% and LVEF < 50%. Spearman rank correlation, anova, t-tests and chi squared tests were used to analyze the data. 79 patients met the inclusion criteria. There was no significant correlation between E/e’_ave and PCWP (n=79, spearman, r=0.096 p=0.3994). This correlation did not improve when ventricular function was considered (LVEF < 50%: n=11, spearman, r=−0.097, p=0.776; LVEF≥ 50%: n=68, spearman, r=0.116, p=0.345). There was no statistically significant difference in mean PCWP between each range of E/e’_ave.

Conclusion

We did not find a clinically relevant relationship between E/e’ and PCWP in patients with severe aortic stenosis.

Introduction

The measurement of left ventricular filling pressure involves an invasive procedure including left and right heart catheterization for direct and indirect measurement. Both of these approaches are associated with procedural risks to the patient. Accordingly, a number of echocardiographic methods suggested to estimate left heart filling pressures have been published in guideline documents.¹ The use of E/e’ was originally proposed in the late 1990s as a means of correcting for the effect of ventricular relaxation on the value of E, thereby allowing for a better correlation with left atrial pressure.² The metric quickly gained popularity in the literature in the late 1990s as a means to non-invasively evaluate left sided filling pressures in a number of clinical scenarios.^3–7 The metric made its way into the guidelines and has remained a corner stone in the echocardiographic evaluation of diastology.¹ However, recent reports suggest limited value of E/e’ as a surrogate for left atrial pressures in multiple disease states. ^8–14 We hypothesized that there is a clinically significant correlation between E/e’ and pulmonary capillary wedge pressure (PCWP) in patients with severe aortic stenosis.

Methods

Study population

After Institutional Review Board approval, we retrospectively reviewed consecutive patients who underwent transcatheter aortic valve replacement (TAVR) for severe aortic stenosis at Tufts Medical Center from November 15, 2012, to March 23, 2020. All datapoints were obtained before TAVR deployment. Exclusion criteria included 1) patients with incomplete data; 2) atrial fibrillation at the time of transthoracic echocardiogram (TTE); 3) mitral valve prosthesis; 4) more than mild mitral annular calcification (MAC); 5) more than mild mitral regurgitation (MR). 6) Presence of pericardial disease. Patients were divided into two groups according to left ventricular ejection fraction (LVEF): LVEF ≥ 50% and LVEF <50%.

Pulmonary capillary wedge pressure (PCWP)

Right heart catheterization (RHC) was performed as a part of pre-procedural evaluation for TAVR with conscious sedation with fentanyl and midazolam. A Swan-Ganz catheter was floated under fluoroscopic guidance from the venous access site to the right atrium, right ventricle, pulmonary artery, and pulmonary capillary wedge position. Pressure waveforms were recorded in each chamber to confirm the catheter position. PCWP was measured after confirmation of the catheter placement in the wedge position by fluoroscopy and pressure waveforms. An average PCWP captured throughout the respiratory cycles was recorded.¹⁵ The transducer zero-point was set at the mid-chest.

Transthoracic echocardiography

Comprehensive transthoracic echocardiographic exam was performed as a pre-procedural evaluation for TAVR and reviewed by cardiologists boarded in advanced echocardiography in accordance with guidelines published by the American Society of Echocardiography.^16,17 Pulse wave Doppler was used for mitral inflow peak E (early diastolic) and peak A (late diastolic) velocities with the sample at the tip of the mitral valve leaflets. Mitral annular early diastolic velocity (e’) was measured using tissue Doppler imaging with the sample at lateral (e’_lat) and medial (e’_med) side of the mitral annulus in the apical four-chamber view. Other echocardiographic parameters including left ventricular end-diastolic volume, left ventricular ejection fraction, aortic valve mean gradient, and peak velocity through the aortic valve were measured and recorded. The severity of MR and MAC were obtained as well.

Reproducibility

Inter-observer variability was assessed in 25 randomly selected patients by repeating the measurement of e’ velocity on two occasions. The second observer was blinded to the results of the first examination. Inter-observer error for e’_med was 12% ± 12, e’_lat was 12% ± 17.

Statistical Analysis

Characteristics of the study sample were summarized using means and standard deviations or percentages. Spearman Rank Correlation analysis was used to describe bivariate relationships between two continuous variables. Student t-tests and one-way analysis of variance were used to compare means between two or three groups, respectively. An alpha level of p<0.05 was as the threshold for statistical significance. Data were analyzed using SAS Software v9.4_M6 through Enterprise Guide version 8.2 (Copyright © 2019, SAS Institute Inc., Cary, NC, USA).

Subgroup analysis

As part of a sub-group analysis, patients who underwent RHC and TTE on the same day were analyzed separately to exclude the effect on hemodynamics due to prolonged intervals between exams.

Results

Patient characteristics

A total of 733 patients were reviewed during the study period. Of these patients, 79 patients met the inclusion criteria (Figure 1). Of the 79 patients included in the study, the mean age was 79 ± 9; there were 51 men (65%) in the group; the mean weight was 84 ± 20 kg; 73 patients (92%) had a history of hypertension, 22 patients (28%) had a history of diabetes; 3 patients (4%) had a history of end stage renal disease requiring hemodialysis; and 11 patients (14%) had a prior history of myocardial infarction (Table 1).

Figure 1.

Box-plot distribution of E/ e’_ave in different LVEF groups.

Table 1.

Characteristics of study population, and stratified by LVEF (n=79). Data summarized as mean +/− standard deviation (n) or median <q1-q3> (n), or % (ratio). History of myocardial infarction was significantly higher among subjects with LVEF<50% compared to subjects with LVEF ≥50%.

Variable Label	N=79	LVEF>=50 (n=68)	LVEF <50 (n=11)	p-value comparing cases with LVEF >=50 to cases with LVEF <50	Statistical test used
Age	79.3 +/− 8.5 (79)	79.3 +/− 8.6 (68)	79.8 +/− 7.9 (11)	0.8381	t-test
Male gender	64.6% (51/79)	63.2% (43/68)	72.7% (8/11)	0.5415	chi-square
Weight in kg	83.5 +/− 20.2 (79)	83.6 +/− 21.0 (68)	82.9 +/− 15.4 (11)	0.9178	t-test
Body Mass Index	29.1 +/− 5.9 (79)	29.1 +/− 6.0 (68)	28.7 +/− 5.6 (11)	0.8068	t-test
Hypertension	92.4% (73/79)	92.6% (63/68)	90.9% (10/11)	0.8400	chi-square
Diabetes	28.2% (22/78)	25.0% (17/68)	50.0% (5/10)	0.1009	chi-square
End stage renal disease	3.8% (3/78)	4.4% (3/68)	0.0% (0/10)	0.4982	chi-square
Myocardial infarction	13.9% (11/79)	8.8% (6/68)	45.5% (5/11)	0.0011	chi-square
Coronary artery disease	54.4% (43/79)	52.9% (36/68)	63.6% (7/11)	0.5087	chi-square
Atrial fibrillation	11.4% (9/79)	13.2% (9/68)	0.0% (0/11)	0.1999	chi-square
Elapsed days between TTE and RHC	4.5 < 0 – 22> (74)	5 < 0.5 – 23> (64)	1.5 < 0 – 11> (10)	0.1839	K-W test

Pulmonary capillary wedge pressure (PCWP)

In our study population, the mean PAWP was 14.4 mmHg ± 7.1 mmHg. The distribution of PCWP was as follows: 30 patients (38%) had a PCWP > 15 mmHg; 10 patients (13%) had a PCWP > 20 mmHg; and 3 patients (4%) had a PCWP > 25 mmHg. There was no significant difference in PCWP in patients with LVEF< 50% compared to LVEF ≥ 50% (16mmHg ± 11 vs. 14mmHg ± 6, p=0.365) (Table 2). The mean amount of Fentanyl was 86mcg ± 56 and Midazolam was 1.7mg ±1.0 intravenously.

Table 2:

Characteristics of study population, and stratified by LVEF (n=79). Data summarized as mean +/− standard deviation (n) or median <q1-q3> (n), or % (ratio). LVEF: Left ventricular ejection fraction. AV: Aortic valve. LVIDd: Left ventricular internal diameter in end-diastole. MAC: Mitral annular calcification. MR: Mitral regurgitation, AI: Aortic regurgitation. Subjects with LVEF<50% had significantly lower AV peak velocity, e’_med, and significantly higher mean LVIDd, E/ e’_lat, E/ e’_med, E/ e’_ave compared to subjects with LVEF ≥50%.

Variable Label	Whole cohort N=79	LVEF>=50% (n=68)	LVEF <50% (n=11)	p-value comparing cases with LVEF >=50% to cases with LVEF <50%	Statistical test used
Wedge Pressure [mmHg]	14.4 +/− 7.1 (79)	14.1 +/− 6.2 (68)	16.2 +/− 11.4 (11)	0.3646	t-test
LVEF [%]	56.5 +/− 10.5 (79)	60.0 +/− 4.6 (68)	34.5 +/− 10.6 (11)
AV peak velocity [m/s]	4.1 +/− 0.7 (79)	4.2 +/− 0.7 (68)	3.7 +/− 0.7 (11)	0.0446	t-test
AV mean gradient [mmHg]	41.4 +/− 15.5 (79)	42.3 +/− 15.9 (68)	35.6 +/− 12.4 (11)	0.1905	t-test
AV peak gradient [mmHg]	70.5 +/− 22.2 (78)	72.2 +/− 22.4 (67)	60.0 +/− 19.1 (11)	0.0913	t-test
Aortic valve area [cm^2]	0.8 +/− 0.2 (79)	0.8 +/− 0.2 (68)	0.7 +/− 0.1 (11)	0.4918	t-test
LVIDd [cm]	4.3 +/− 0.7 (74)	4.2 +/− 0.7 (65)	4.8 +/− 0.8 (9)	0.0256	t-test
E velocity [cm/sec]	78.8 +/− 22.8 (79)	76.8 +/− 21.3 (68)	91.2 +/− 28.3 (11)	0.0509	t-test
e’(lateral) [cm/sec]	6.3 +/− 1.9 (79)	6.4 +/− 1.9 (68)	5.6 +/− 2.0 (11)	0.2175	t-test
e’(medial) [cm/sec]	4.6 +/− 1.3 (79)	4.8 +/− 1.2 (68)	3.5 +/− 0.8 (11)	0.0011	t-test
E/e’(lateral)	13.8 +/− 6.3 (79)	13.0 +/− 5.3 (68)	18.5 +/− 9.6 (11)	0.0065	t-test
E/e’(medial)	18.4 +/− 8.0 (79)	17.0 +/− 6.6 (68)	2/.1 +/− 10.3 (11)	<.0001	t-test
E/e’(average)	15.4 +/− 6.1 (79)	14.5 +/− 5.4 (68)	21.1 +/− 7.6 (11)	0.0007	t-test
A velocity [cm/sec]	94.3 +/− 24.4 (79)	93.9 +/− 23.4 (68)	96.2 +/− 31.5 (11)	0.7818	t-test
E/A	0.9 +/− 0.6 (79)	0.9 +/− 0.5 (68)	1.1 +/− 0.6 (11)	0.3106	t-test
MAC +/1	N=79	N=68	N=11	0.3545	chi−square
0.no/non	41.8% (33)	39.7% (27)	54.5% (6)
2.mild	58.2% (46)	60.3% (41)	45.5% (5)
MR severity	N=79	N=68	N=11	0.2216	chi-square
0.no/none	3.8% (3)	4.4% (3)	. ( .)
1.trace	70..% (56)	73.5% (50)	54.5% (6)
2.mild	2 5.3% (20)	22.1% (15)	45.5% (5)
AI severity	N=79	N=68	N=11	0.0830	chi-square
0.no/none	30.4% (24)	26.5% (18)	54.5% (6)
1.trace	35.4% (28)	39.7% (27)	9.1% (1)
2.mild	34.2% (27)	33.8% (23)	36.4% (4)
AV mean gradient [mmHg]	39 < 30– 50> (79)	40 < 30.5– 50> (68)	35 < 25– 50> (11)	0.1480	K-W test
Aortic valve area [cm^2]	(79)	0.8 < 0.7– 0.9> (68)	0.8 < 0.7– 0.8> (11)	0.3634	K-W test
E/A	0.74 < 0.62– 1> (79)	0.74 < 0.63– 0.94> (68)	0.88 < 0.57– 1.68> (11)	0.4196	K-W test

Transthoracic echocardiography

For the 79 patients who met inclusion criteria, overall the mean LVEF was 57 ± 11 %. The majority (68 patients, 86%) had LVEF ≥ 50%. The mean E velocity was 79 cm/s ± 23; the mean e’_lat was 6.3 cm/s ± 1.9; and e’_med was 4.6 cm/s ± 1.3. The mean E/e’_lat was 13.8 ± 6.3; the mean E/e’_med was 18.4 ± 8.0; and the mean E/e’_ave was 15.4 ± 6.1. As expected, the peak velocities and mean gradients were statistically significantly lower in the LVEF <50% group. Notably, while the e’_lat velocities were not significantly different between patients with LVEF < 50% and LVEF ≥ 50%, the e’_med was significantly lower in patients with LVEF < 50% (Table 2). Among all 79 patients, 40 patients (51%) had E/e’_ave > 14 while only five patients (6%) had E/e’_ave <8 in all LVEF groups. Among 11 patients with LVEF < 50%, 9 patients (82%) had E/e’_ave >14. There was only one patient with E/e’_ave <8 in this group. (Figure 1).

Relationship between E/e’ and PCWP

There was no significant correlation between E/e’_ave and PCWP (n=79, spearman, r=0.096 p=0.3994, spearman) (Figure 2). The correlation did not improve when patients with depressed systolic ventricular function were studied independently from those with preserved ejection function (LVEF < 50%: n=11, spearman, r=−0.097, p=0.778; LVEF≥ 50%: n=68, spearman, r=0.116, p=0.3448, spearman). Of the 40 patients with an E/e’_ave > 14, 16 patients (40%) had a PCWP > 15 mmHg; 6 patients (15%) had a PCWP > 20 mmHg; and 3 patients (8%) had a PCWP > 25 mmHg. The positive predictive value for E/e’_ave > 14 for detecting a PCWP > 15 was 40%; for a PCWP > 20 mmHg was 15%; and for a PCWP > 25 was 7.5%. Notably, only five patients had E/e’_ave < 8, and one patient (20%) had a PCWP > 20 mmHg, 2 patients (40%) had a PCWP > 15 mmHg, 4 patients (80%) had a PCWP > 10 mmHg. There was not enough sample size to calculate the negative predictive value of E/e’_ave < 8 predicting PCWP < 15. The correlation was not statistically significant between E/e’_med and PCWP (r=0.087, p=0.4474), and was also not significant between E/e’_lat and PCWP (r=0.116, p=0.3091).

Figure 2.

Relationship between E/ e’_ave and PCWP (With different plot color for LVEF ≥50% and LVEF <50%). The regression line on figure 2 represents the linear relationship between E/e’ average for the whole cohort There was no significant linear correlation between E/e’_ave and PCWP. Correlation coefficient = 0.13 (p=0.27). The correlation did not improve when patients with depressed systolic ventricular function were studied independently from those with preserved ejection function. (1) EF ≥50% (n=68), correlation coefficient =0.09 (p=0.48), (2) EF<50% (n=11), correlation coefficient = 0.12 (p=0.73).

The mean PCWP for patients with E/e’_ave <8 was 15.2 mmHg ± 6.0; for 8≤ E/e’_ave ≤ 14 was 12.9 mmHg ± 5.0; and for E/e’_ave > 14 was 15.5 mmHg ± 8.5 (Figure 4). There was no statistically significant difference in mean PCWP between each range of E/e’_ave.

Figure 4.

Relationship between E/e’_ave and PCWP for the same procedure date group. The correlation between E/e’_ave and PCWP in this group was still poor. (n=21, correlation coefficient = 0.21, p=0.35)

Subgroup analysis

The average time interval between right heart catheterization and TTE was 17±28 days. 22 patients had RHC and TTE performed on the same day. The correlation between E/e’_ave and PCWP in this group was still poor (n=21, spearman r=0.257, p=0.2607) (Figure 4).

Discussion

Our data does not support our hypothesis that there is a clinically relevant correlation between E/e’_ave and pulmonary capillary wedge pressure (PCWP) in patients with severe aortic stenosis. There was no significant linear correlation between E/e’_ave measured with the annulus and PCWP in patients with severe aortic stenosis. There was no difference in the mean PCWP between patients with E/e’_ave <8, 8≤ E/e’_ave ≤14, and E/e’ > 14 groups. The positive predictive value of E/e’_ave was poor.

Correlation between E/e’ and PCWP was first reported in 1997 by Nagueh SF et al. ² This novel idea of estimating left ventricular filling pressures by non-invasive methods using tissue Doppler E/e’ became widespread and serial studies that were conducted successfully revealed a relationship between the two parameters in a variety of disease states. These included atrial fibrillation, sinus tachycardia, hypertrophic cardiomyopathy, mitral valve disease and cardiac transplantation. ^3–7 This led to the inclusion of tissue Doppler E/e’ in the guidelines for recommendations for the evaluation of left ventricular diastolic function by echocardiography in 2009.^18,19 However, in 2000, Ommen SR et al. published data suggesting correlation between E/e’ and mean left ventricular diastolic pressures to be weak (r=0.62) and scattered. This was followed by other reports that failed to validate the use of E/e’ as a surrogate for left heart filling pressures in patients with significant primary MR, hypertrophic cardiomyopathy, decompensated advanced systolic heart failure, and following heart transplantation.^8–13 A 2016 meta-analysis of 24 studies using invasive measurements concluded there was insufficient evidence to support that E/e’ can reliably estimate left ventricular filling pressures in preserved LVEF.¹⁴

In our literature review, we found relatively few manuscripts that investigated E/e’ relationship to PCWP specifically in the setting of aortic stenosis. Bruch et al. published data in 2004 from a series of 23 patients in which E/e’_med had a modest correlation with PCWP (r=0.62, p<0.001).²⁰ Biner et al. in 2015 studied E/e’_ave in 113 TAVR patients and found an identically modest correlation between E/e’ and PCWP in patients with severe aortic stenosis (r=0.62, p<0.001). In reviewing their scatter plot, much of their signal comes from patients with PCWPs over 25 mmHg. The authors concluded that using a single echo Doppler parameter among patients with aortic stenosis to estimate PCWP is insufficient for clinical application. The prediction of PCWP required using a formula employing three different variables including E/e’, LVEF, and tricuspid regurgitation velocity time integral.²¹ In contrast, our study focused on the single relationship between E/e’_ave and PCWP among patients with severe aortic stenosis. Compared to these earlier published studies that found only a modest correlation, we did not find a statistically significant linear correlation. Our study is an important addition to these recent reports as this finding is not consistent with current guidelines recommending the use of E/e’ as a surrogate for left heart filling pressures in patients with aortic stenosis.^{17, 18} This result is relevant for anesthesiologists taking care of patients with severe aortic stenosis with no RHC data.

Limitations

Our study is subject to the usual limitations of a retrospective study. Only 79 of 733 patients were included in the study. Aside from patients without data for right heart catheterization or tissue Doppler imaging, atrial fibrillation and severe MAC were the two conditions that excluded majority of patients in the study. This is explained by the high prevalence of atrial fibrillation in general, which is reported to be 9% of the overall age > 65 years old population, and similar risk factors for atrial fibrillation and severe degenerative aortic stenosis.^22,23 Similarly, MAC increases with age in the general population. Calcified aortic stenosis and MAC have similar etiologies and pathophysiological mechanisms, and therefore severe MAC and severe aortic stenosis often coexist.²⁴ Indeed, this high exclusion rate speaks to the limited generalized applicability of E/e’ among patients with severe AS.

We have to acknowledge the also the obvious limitation of e’ to function as a surrogate marker of global myocardial function in patients with regional wall motion abnormalities. Additionally, there are some reports showing discrepancies between PCWP and left atrial pressure, as well as left ventricular end diastolic pressure.^25–30 Indeed, in the study by Bruch et al. the author reported a slightly better r value between E/e’ and the left ventricular end diastolic pressure compared to the PCWP.⁹ Importantly, echocardiographic exam and right heart catheterization were not performed simultaneously; the average time interval between RHC and TTE was 17±28 days. However, both procedures were done in the elective setting without general anesthesia during the pre-procedural evaluation for TAVR, and therefore hemodynamic conditions were likely similar between the two exam Although we must acknowledge that sedation could have impacted PCWP, one might expect at least a weak correlation between PCWP and E/e’ if a biologic signal were truly present. However, we did not observe one. In fact, the correlation between E/e’_ave and PCWP was remained clinically insignificant when patients who underwent RHC and TTE on the same day were analyzed separately as a subgroup analysis. Our findings are consistent with other published literature which failed to validate the use of E/e’ as a surrogate for left heart filling pressures despite simultaneous measurement of TTE and RHC.^8,11 Finally, there were only five patients with E/e’_ave < 8 cm/s. This skewed distribution in patients with severe aortic stenosis being evaluated for TAVR might account for some of our inability to find a correlation with wedge pressure. We were unable to evaluate the negative predictive value of E/e’. That being said, this skewedness represents a real-world experience and within this group there was no correlation PCWP despite a wide range of PCWP values again challenging the routine clinical utility of E/e’ in severe AS patients. Obviously these findings may not hold for patients with mild and moderate AS.

We were not able to provide data concerning the serum level of brain natriuretic peptide and echocardiography, because it is not routinely measured in our institution for TAVR patients.

Conclusion

We did not find a clinically significant relationship between E/e’ and PCWP in patients with aortic stenosis presenting for TAVR.

Figure 3.

Distribution of PCWP in each E/ e’_ave groups. There was no statistically significant difference in mean PCWP between each range of E/e’_ave.

Table 3.:

Correlation of Wedge Pressure (1) with E/ e’_lat, (2) E/ e’_med, (3) and E/ e’_ave. Overall and stratified by LVEF. Used p<0.0167 to declare statistical significance (Bonferroni corrected p value). PCWP was not significantly correlated with any of the three E/e’ ratios (lateral, medial, or average).

Variable correlated with Wedge Pressure	result (n=79 subjects)	Whole cohort (n=79)		LVEF <50% (n=11)		LVEF ≥50% (n=68)
		Pearson	Spearman Rank	Pearson	Spearman Rank	Pearson	Spearman Rank
E/e’ (lateral)	correlation	0.065	0.116	−0.032	−0.161	0.068	0.140
	p-value	0.5699	0.3091	0.9266	0.6364	0.5800	0.2548
E/e’ (medial)	correlation	0.227	0.087	0.444	−0.060	0.101	0.100
	p-value	0.0442	0.4474	0.1709	0.8614	0.4121	0.4159
E/e’ (average)	correlation	0.122	0.096	0.119	−0.097	0.080	0.116
	p-value	0.2857	0.3994	0.7279	0.7776	0.5179	0.3448