Hemodynamics and Clinical Implications of Occult Left Ventricular Dysfunction in Adults Undergoing Ebstein Anomaly Repair

Abstract

Background:

Left ventricular global longitudinal strain (LVGLS) can detect early phases of LV systolic dysfunction, but its application has not been studied in Ebstein anomaly. We hypothesized that LVGLS can detect early phases of LV systolic dysfunction, and that patients with occult LV systolic dysfunction will have worse hemodynamics, end-organ dysfunction, and suboptimal postoperative LV reverse remodeling after tricuspid valve surgery in comparison to patients with normal LV systolic function.

Methods:

In this retrospective cohort study, 371 Ebstein patients that underwent tricuspid valve surgery were divided into 3 groups: normal LV systolic function (normal LVGLS and LVEF; n=244, 77%), occult LV systolic dysfunction (abnormal LVGLS with normal LVEF; n=44, 14%), and overt LV systolic dysfunction (abnormal LVGLS and LVEF; n=27, 9%).

Results:

Compared to the normal LV function group, the occult group had smaller LV volume and cardiac output (2.1±0.4 vs 2.9±0.6 L/min/m², p<0.001), worse end-organ dysfunction (glomerular filtration rate [GFR] 78±14 vs 91±18 ml/min/1.73m², p=0.01), suboptimal postoperative LV reverse remodeling. Although both the occult and overt groups had similar degree of end-organ dysfunction (GFR 78±14 vs 82±16 ml/min/1.73m², p=0.3), the occult group was less likely to be on heart failure therapy (48% vs 96%, p<0.001).

Conclusions:

Abnormal LVGLS was associated with suboptimal postoperative LV reverse remodeling. These data suggest that LVGLS can potentially be used for risk stratification, and provides a foundation for further studies to determine whether optimal heart failure therapy or tricuspid valve intervention can improve outcomes for LV systolic dysfunction in patient with Ebstein anomaly.

INTRODUCTION

Ebstein anomaly is one of the most common causes of right heart failure and mortality in adults with congenital heart disease.^{1, 2} Although the primary hemodynamic lesion is right heart volume overload due to tricuspid regurgitation, it can also lead to left ventricular (LV) systolic dysfunction because of the complex structural and hemodynamic interactions between the right and left heart.^{3, 4} Tricuspid valve surgery is an effective therapy for right heart volume overload due to tricuspid regurgitation in Ebstein anomaly, and has been shown to improve exertional symptoms and RV hemodynamic performance, but not LV systolic function as measured by LV ejection fraction (LVEF).^5–9

LVEF is the most commonly used metric of LV systolic function in clinical practice, but it is very sensitive to loading conditions and LV geometry.¹⁰ Both loading conditions and LV geometry are abnormal in Ebstein anomaly because right heart volume overload and dysfunction leads to a reduction in LV preload (abnormal loading condition) and leftward bowing of the interventricular septum (distortion of LV geometry).¹¹ Additionally, the restoration of tricuspid valve competence after tricuspid valve surgery leads to changes in LV preload and geometry.^{5, 8, 9} All these factors make longitudinal monitoring of LV systolic function very challenging in this population.

LV global longitudinal strain (LVGLS) is less dependent on LV loading conditions and geometry, and it is more sensitive for detecting early phases of LV systolic dysfunction prior to a reduction in LVEF (occult LV systolic dysfunction) thereby allowing for better risk stratification, early interventions and in turn, improved outcomes.^{12, 13} The role of LVGLS for diagnosis and prognostication in patients with Ebstein anomaly undergoing tricuspid valve surgery has not been systematically studied. The purpose of this study was to determine whether LVGLS can detect occult LV systolic dysfunction, and the hemodynamic and clinical implications of occult LV systolic dysfunction in patients with Ebstein anomaly undergoing tricuspid valve surgery.

METHODS

Study Population

The data that support the findings of this study are available from the corresponding author upon reasonable request. This is a retrospective cohort study of adult patients with Ebstein anomaly (age ≥18 years) that underwent tricuspid valve surgery for severe tricuspid regurgitation at Mayo Clinic Rochester from January 1, 2003 to December 31, 2018. These patients were identified from the Mayo Adult Congenital Heart Disease (MACHD) registry. From this cohort, we selected consecutive patients that underwent preoperative transthoracic echocardiogram (within 3 months prior to surgery) and postoperative echocardiogram (between 12–24 months after surgery), Supplementary Figure I. We excluded patients with the following conditions: (1) >mild aortic valve disease (aortic valve peak velocity >2 m/sec or >mild aortic regurgitation); (2) >mild mitral valve disease (mitral valve mean gradient >3 mmHg or >mild mitral regurgitation); (3) atrial arrhythmia or ventricular paced rhythm at the time of echocardiogram; (4) concomitant bidirectional Glenn operation or pulmonary valve surgery at the time of tricuspid valve surgery; (5) suboptimal echocardiographic images for offline assessment LVEF and LVGLS. The Mayo Clinic Institutional Review Board approved the study, and waived informed consent for patients that provided research authorization.

Study Design

We hypothesized that LVGLS can detect early phases of LV systolic dysfunction (occult LV systolic dysfunction defined as abnormal LVGLS with normal LVEF), and that the patients with occult LV systolic dysfunction will have worse preoperative hemodynamic profile, and suboptimal postoperative LV reverse remodeling after tricuspid valve surgery in comparison to patients with normal LV systolic function (normal LVGLS with normal LVEF). To test this hypothesis, we divided the cohort into 3 groups: (1) normal LV systolic function (normal LVGLS with normal LVEF); (2) occult LV systolic dysfunction (abnormal LVGLS with normal LVEF), and (3) overt LV. We assessed preoperative hemodynamic profile using the following indices: (1) LV structure and function (LV end-diastolic volume [LVEDV] indexed to body surface area, LV end-systolic volume [LVESV] indexed to body surface area, and LV stroke volume [LVSV] indexed to body surface area); (2) End-organ function and neurohormonal activation (estimated glomerular filtration rate [GFR] and N-terminal pro-brain natriuretic peptide [NT-proBNP]); (3) Aerobic capacity (%-predicted peak oxygen consumption [VO₂]). End-organ function and aerobic capacity were based on assessments performed within 6 months prior to tricuspid valve surgery.

Postoperative LV reverse remodeling was defined as the delta (Δ) of each metric of LV structure and function (ΔLVEDV, ΔLVESV, ΔLVSV, ΔLVLGLS, and ΔLVEF), and calculated as postoperative minus preoperative value. Exploratory analysis was performed to assess for temporal change in LV systolic function (LVEF and LVGLS) in the subset of patients that had echocardiograms at least 24 months prior to preoperative echocardiogram.

Echocardiography

LVGLS was assessed by speckle tracking echocardiography, and abnormal LVGLS was defined as LVGLS equal to or less negative than 18% which is the vendor-specific cut-off point for abnormal LVGLS.¹⁴ LVEF was assessed using the biplane Simpson’s method, and abnormal LVEF was defined as LVEF <52% in males and <54% in females.¹⁰ Diastolic LV eccentricity index, a measure of ventricular interdependence, was assessed from the parasternal short axis window using 2D echocardiography.^{15, 16}Two experienced sonographers (JW and KT) performed these offline analyses and calculations. Intra- and interobserver agreements for LVGLS and LVEF were assessed in 20 randomly selected patients. The severity of tricuspid valve regurgitation was based on qualitative Doppler echocardiography.¹⁷

LV speckle tracking strain imaging was performed using Vivid E9 and E95 (General Electric Co, Fairfield, Connecticut) with M5S and M5Sc-D transducers (1.5–4.6 MHz) at frame rate of 40 to 80 Hz as previously described.¹⁸ In brief, 3-beat cine-loop clips were obtained from 3 apical views (2-chamber, 3-chamber, and 4-chamber). These images were exported (DICOM) and then analyzed offline using TomTec (TomTec Imaging Systems version 4.6, Unterschleissheim, Germany). Offline analysis involved manual endocardial to mid-wall tracing of a single frame at end-systole by a point-click approach, with a region of interest that covers at least 90% of the myocardial wall thickness. The periodic displacement of the tracing was automatically tracked in subsequent frames. Tissue velocity was determined by the TomTec Imaging Systems version 4.6 software, according to a shift of the points divided by time between B-mode frames, and the software automatically calculated LVGLS as well as LV strain from the 3 individual apical windows, Supplementary Figure II and III.

Similarly, speckle tracking strain imaging was also used for the assessment of right ventricular (RV) systolic function, and RV systolic dysfunction was defined as RV global longitudinal strain (RVGLS) equal to or less negative than 20%.

Other Study Variables

The surgical notes were reviewed to determine type of tricuspid valve surgery (repair vs replacement) and concomitant procedures performed at the time of tricuspid valve surgery. The surgical techniques for performing tricuspid valve repair at this institution have been extensively described.^{7, 19} We also reviewed electronic health records and retrieved the following data: demographic characteristics, comorbidities and medications, laboratory data, patient-reported symptom status, and peak VO₂. All-cause mortality data were obtained from the electronic health records and Accurint database.²⁰ In order to assess appropriate use of medical therapy for LV systolic dysfunction, we defined guideline directed medical therapy (GDMT) as concomitant use of beta blocker therapy and angiotensin converting enzyme inhibitor/angiotensin-II receptor blocker.^{21, 22}

Statistical Analysis

Data were presented as mean ± standard deviation, median (interquartile range), estimates (95% confidence interval) and count (%). Between-group comparisons were performed using Fisher’s exact test, t-test, and Wilcoxon rank sum test as appropriate. Time-to-event analyses were performed using the Kaplan Meier’s method, comparisons performed using log-rank test. All statistical analyses were performed with JMP software versions 14.0 (SAS Institute Inc, Cary NC).

RESULTS

A total of 317 patients met the study inclusion criteria (Supplementary Figure I). The median age at the time of preoperative echocardiogram was 39 (27–59) years and 133 (42%) were men. The preoperative clinical and imaging data are shown in Table 1.

Table 1:

Baseline Characteristics (N=317)

	Normal LV systolic function (n=244, 77%)	Occult LV systolic dysfunction (n=44, 14%)	Overt LV systolic dysfunction (n=27, 6%)	p	p*
Age, years	37 (26–47)	38 (29–49)	39 (28–51)	0.7	0.9
Male	103 (42%)	19 (43%)	11 (41%)	0.6	0.8
Body mass index, kg/m2	26±5	27±6	28±3	0.1	0.3
Prior tricuspid valve surgery	49 (21%)	11 (16%)	9 (13%)	0.6	0.9
NYHA I	24 (11%) [n=215]	1 (3%) [n=38]	---	0.1	---
NYHA II	123 (57%) [n=215]	12 (32%) [n=38]	9 (26%) [n=26]	0.004	0.8
NYHA III-IV	68 (32%) [n=215]	25 (66%) [n=38]	17 (65%) [n=26]	0.001	0.9
Comorbidities
Atrial flutter/tachycardia	39 (16%)	14 (32%)	7 (26%)	0.01	0.3
Atrial fibrillation	34 (14%)	18 (41%)	13 (48%)	<0.001	0.4
Nonsustained ventricular tachycardia	7 (3%)	2 (5%)	1 (4%)	0.6	0.9
Hypertension	41 (17%)	7 (16%)	5 (19%)	0.5	0.8
Coronary artery disease	6 (3%)	2 (5%)	2 (8%)	0.6	0.7
Medications
Loop diuretics	39 (16%)	14 (32%)	10 (37%)	0.01	0.2
Beta blockers	47 (19%)	18 (41%)	21 (78%)	<0.001	<0.001
ACEI/ARB	24 (10%)	14 (32%)	20 (74%)	<0.001	<0.001
Aldosterone antagonist	---	7 (16%)	19 (33%)	---	0.08

NYHA: New York Heart Association; ACEI: Angiotensin converting enzyme inhibitor; ARB: Angiotensin II receptor blockers;

p value denotes comparison between Normal LV systolic function and Occult LV systolic dysfunction groups while p* denotes comparison between Occult LV systolic dysfunction and Overt LV systolic function groups

Preoperative LV Systolic Function

The mean LVGLS was −22±3% and the mean LVEF was 61±8%. There was excellent intraobserver correlation (intraclass correlation coefficient 0.91 [0.87–0.95]) and interobserver correlation (0.88 [0.83–0.93]) for LVGLS. There was also good intraobserver and interobserver correlations for LVEF (0.85 [0.79–0.91]) and LVGLS (0.80 [0.74–0.86]) respectively. There was a good correlation between LVGLS and LVEF (r=0.54, p<0.001).

Cardiac magnetic resonance imaging (MRI) data were available in 92 (29%) patients. There was a good correlation between MRI-derived LVEF and echo-derived LVEF (r=0.71 [0.66–0.76], p<0.001). We also observed a modest correlation between MRI-derived LVEF and echo-derived LVGLS (r=0.53 [0.48–0.59], p<0.001) which was almost identical to the correlation between echo-derived LVEF and echo-derived LVGLS (r=0.54 [0.51–0.57], p<0.001). When we applied the same sex-specific cut-off points for normal vs abnormal LVEF to the subset of 92 patients that had both echocardiogram and cardiac MRI data, we observed that there were 13 (14%) patients and 11 (12%) patients with reduced LVEF based on echocardiogram and cardiac MRI respectively. There was no significant difference in the proportion of patients with reduced LVEF based on echocardiogram vessels cardiac MRI (13/92 vs 11/92, p=0.7).

Of the 317 patients, 244 (77%) had normal LV systolic function, 44 (14%) had occult LV systolic dysfunction, and 27 (9%) had overt LV systolic dysfunction. Of the 317 patients, 69 (21%) had prior tricuspid valve surgery, and among these 69 patients with prior tricuspid valve surgery, 49 (71%) had normal LV systolic function, 11 (16%) had occult LV systolic dysfunction, and 9 (13%) had overt LV systolic dysfunction. There was no significant difference in the distribution of LV function subgroups between the patients with vs without prior tricuspid valve surgeries.

In our study cohort of 317 patients, we observed significant between-group differences in the hemodynamic profiles of the 3 subgroups (Table 2). Compared to the group with normal LV systolic function, the patients with occult LV systolic dysfunction had smaller LVEDV, LVESV, LVSV and lower LV cardiac index despite having a higher resting heart rate. Similarly, the occult LV systolic dysfunction group had worse renal function (estimated GFR), neurohormonal activation (NT-proBNP) and aerobic capacity (peak VO₂). The occult LV systolic dysfunction group also had worse right heart function and hemodynamics, and higher LV eccentricity index, suggestive of worse ventricular interdependence in this group (Table 2).

Table 2:

Comparison of Clinical and Hemodynamic Indices

	Normal LV systolic function (n=244, 77%)	Occult LV systolic dysfunction (n=44, 14%)	Overt LV systolic dysfunction (n=27, 6%)	p	p*
Echocardiography
LVGLS, %	−24±2	−16±2	−15±3	<0.001	0.1
LVEF, %	63±6	56±3	49±4	<0.001	0.008
LVEDV index, ml/m2	46±8	39±5	48±7	<0.001	<0.001
LVESV index, ml/m2	17±6	16±5	25±5	0.2	0.003
LV SV index, ml/m2	40±5	27±4	30±5	<0.001	0.04
Heart rate, bpm	71±8	79±6	73±4	0.008	0.1
LV CI, L/min/m2	2.9±0.6	2.1±0.4	2.3±0.4	<0.001	0.08
LV eccentricity index	1.29±0.22	1.35±0.14	1.28±0.13	0.009	0.03
RA reservoir strain, %	−31±7	−23±6	−28±5	<0.001	0.02
RVGLS, %	−19±4	−16±3	−18±3	0.03	0.09
TAPSE, mm	25±6	23±5	24±4	0.04	0.5
RV s’, cm/s	14±6	11±3	12±3	0.02	0.3
RV end-diastolic area, cm2	46±12	57±10	59±9	<0.001	0.3
RV end-systolic area, cm2	28±7	42±9	39±10	<0.001	0.2
RV FAC, %	39±7	26±7	33±8	<0.001	0.002
IVC collapsibility, %	48±11	31±13	38±9	0.008	0.03
† RA pressure, mmHg	8±3	11±3	8±4	0.04	0.08
TR velocity, m/s	2.4±0.3	2.2±0.2	2.3±0.3	0.2	0.7
Cardiac MRI (n=92)	(n=67)	(n=14)	(n=11)
RVEDV index, ml/m2	139 (106–171)	163 (129–221)	166 (134–198)	<0.001	0.3
RVESV index, ml/m2	78 (57–114)	97 (66–126)	91 (66–126)	<0.001	0.1
RV ejection fraction, %	43±11	39±12	40±9	0.08	0.7
LVEDV index, ml/m2	57±10	47±8	59±11	<0.001	<0.001
LVESV index, ml/m2	15±5	22±6	27±8	0.03	0.09
LVSV index, ml/m2	39±6	28±5	32±6	<0.001	0.1
LV ejection fraction, %	64±8	58±6	45±5	0.07	0.004
Clinical data
Age, years	37 (26–47)	38 (29–49)	39 (28–51)	0.7	0.9
Atrial septal defect	160 (66%)	32 (73%)	10 (70%)	0.4	0.9
Systemic O2 saturation	96±3	95±2	95±1	0.6	0.9
Peak VO2, %	69±7 [n=183]	56±12 [n=36]	61±11 [n=21]	<0.001	0.3
GFR, ml/min/1.73m2	91±18	78±14	82±16	0.01	0.3
NT-proBNP, pg/ml	96 (51–216)	181 (104–529)	198 (94–616)	<0.001	0.8
GDMT	24 (10%)	21 (48%)	26 (96%)	<0.001	<0.001

LVGLS: Left ventricular global longitudinal strain; LVEF: Left ventricular ejection fraction; LVEDV: Left ventricular end-diastolic volume; LVESV: Left ventricular end-systolic volume; LVSV: Left ventricular stroke volume: CI: Cardiac index; RVGLS: Right ventricular global longitudinal strain; RA; right atrium; TR: Tricuspid regurgitation; NT-proBNP: N-terminal pro-brain natriuretic peptide; GFR: Glomerular filtration rate; VO₂: Oxygen consumption symmetry GDMT: Guideline directed medical therapy for heart failure. RVEF: Right ventricular ejection fraction; RVEDV: Right ventricular end-diastolic volume; RVESV: Right ventricular end-systolic volume; s’: annular tissue Doppler systolic velocity; IVC: Inferior vena cava; TAPSE: Tricuspid annular plane systolic excursion; MRI: Magnetic resonance imaging

^†RA pressure was based on inferior vena cava size and collapsibility

p value denotes comparison between Normal LV systolic function and Occult LV systolic dysfunction groups while p* denotes comparison between Occult LV systolic dysfunction and Overt LV systolic function groups

On the other hand, both the occult and overt LV systolic dysfunction groups had similar severity of renal function, neurohormonal activation, and impaired aerobic capacity. However, in spite of having similar degree of end-organ dysfunction, the patients in the occult LV systolic dysfunction group were less likely to be on GDMT compared to the overt LV systolic dysfunction group (48% vs 96%, p<0.001), Table 2.

Relationship between RV and LV Systolic Function

The mean RVGLS was −18±5%, and 206 (65%) patients had RV systolic dysfunction. There was a correlation between RVGLS and LV eccentricity index (r= −0.56, p<0.001), and between RVGLS and LVGLS (r=0.48, p<0.001). These data suggest a relationship between RV systolic function, LV systolic function and LV eccentricity index (which is a measure of ventricular interdependence). Consistent with the above postulate, we observed a significant difference in RV systolic function across the 3 subgroups (Table 2). As compared to the normal LV systolic function group, the patients with occult LV systolic dysfunction had lower RVGLS (−16±3 vs −19±4%, p=0.03) and higher LV eccentricity index (1.35±0.14 vs 1.29±0.33, p=0.009).

Preoperative change in LV systolic function

Of the 317 patients, 144 (45%) had echocardiograms performed at least 24 months prior to the preoperative echocardiogram. The LVGLS and LVEF at the time of first echocardiogram was −23±3% and 62±9% respectively; and of the 144 patients, 124 (86%) had normal LV systolic function, 18 (41%) had occult LV systolic dysfunction, and 2 (1.3%) had overt LV systolic dysfunction. The median interval between the first echocardiogram and the preoperative echocardiogram was 27 (25–31) months. There was a significant decrease in LVGLS from the first echocardiogram to the preoperative echocardiogram (−23±3% vs −21±4%, p=0.01), but minimal decrease in LVEF (62±9% vs 60±8%, p=0.2).

Surgical Data

Of the 317 patients, 114 (36%) patients had at least 1 prior cardiac surgery, of which 69 (21%) were tricuspid valve surgeries (repair n=61 and replacement n=8) and 45 (14%) were for closure of atrial septal defects. The indications for the current surgery were severe tricuspid regurgitation (inclusion criteria), exertional dyspnea or fatigue, RV volume overload, progressive RV dysfunction, and paroxysmal atrial arrhythmias.

Of the 317 patients, 219 (69%) had tricuspid valve repair while 98 (31%) had tricuspid valve replacement (97 bioprostheses and 1 mechanical prosthesis). The following concomitant procedures were performed at the time of tricuspid valve surgery: isolated cavotricuspid isthmus ablation (n=51, 16%), right atrial Maze (n=69, 22%), pulmonary vein isolation (n=25, 8%), biatrial Maze (n=12, 4%), right atrioplasty (n=281, 89%), RV plication (n=92, 29%), and closure of atrial shunt (n=211, 66%). Tricuspid valve repair was performed in 117 (73%) vs 26 (59%) vs 16 (59%) (p=0.1) of the patients with normal LV systolic function vs occult LV systolic dysfunction vs and overt LV systolic dysfunction respectively. Hence there were no significant between-group differences in the type of tricuspid valve surgery.

At the time of hospital discharge, 148 (47%) had none/trivial tricuspid regurgitation, 168 (53%) had mild or mild/moderate tricuspid regurgitation, and 1 (0.3%) patient had moderate tricuspid regurgitation. There were no significant between-group differences in the severity of residual tricuspid regurgitation across the 3 LV systolic function subgroups.

Postoperative LV Reverse Remodeling

The interval between tricuspid valve surgery and postoperative echocardiogram was 15±2 months, and the postoperative LVGLS and LVEF was −23±4% and 59±13% respectively. There was no postoperative improvement in LVEF (61±9% vs 58±13%, p=0.1) or LVGLS (−22±3% vs −23±4%, p=0.2) for the entire cohort. However the patterns of LV reverse remodeling was significantly different in the 3 subgroups (Table 3).

Table 3:

Postoperative Left and Right Ventricular Reverse Remodeling

	Normal LV systolic function (n=244, 77%)	Occult LV systolic dysfunction (n=44, 14%)	Overt LV systolic dysfunction (n=27, 6%)
ΔLVGLS, %	1 (−1 − 3)	2 (−1 − 3)	2 (−2 − 6)
ΔLVEF, %	3 (−2 − 6)	−4 (−7 − -2)	2 (−2 − 7)
ΔLVEDV index, ml/m2	7 (4 − 10)	10 (6 − 14)	2 (−2 − 7)
ΔLVESV index, ml/m2	5 (2 − 8)	9 (5 − 13)	1 (−1 − 3)
ΔLV SV index, ml/m2	5 (1 − 9)	3 (−1 − 7)	2 (−1 − 5)
ΔLV eccentricity index	−0.11 (−0.19 − -0.04)	−0.23 (−0.31 − -0.14)	−0.07 (−0.19 − 0.03)
ΔRV EDA, cm2	−11 (−15 - -7)	−9 (−16 - - 2)	−10 (−14 - -6)
ΔRV FAC, cm2	4 (2 − 7)	3 (−2 − 8)	2 (−3 − 7)
ΔRVGLS, %	3 (2 − 4)	3 (0 − 6)	2 (−1 − 5)

LVGLS: Left ventricular global longitudinal strain; LVEF: Left ventricular ejection fraction; LVEDV: Left ventricular end-diastolic volume; LVESV: Left ventricular end-systolic volume; LVSV: Left ventricular stroke volume: RVGLS: Right ventricular global longitudinal strain; EDA: End-diastolic area; FAC: Fractional area change

Δ signifies delta, calculated as postoperative value minus preoperative value for each metric. ( ) signifying 95% confidence interval

There was a postoperative increase in LVEDV, LVESV and LVSV, but no significant change in LVEF or LVGLS in the normal LV systolic function group (Table 3, Figure 1). In the group with occult LV systolic dysfunction, the LVEDV increased, LVSV was unchanged, and the LVEF actually decreased postoperatively (Table 3, Figure 1). We did not observe any significant postoperative change in LVEDV, LVSV or LVEF in the overt LV systolic dysfunction group (Table 3, Figure 1). All the subgroups had a reduction in RV size and LV eccentricity index consistent with effective RV unloading, but no postoperative change in LVGLS.

Figure 1:

(A & C) There were no significant between-group differences in the postoperative change in left ventricular global longitudinal strain (LVGLS) and LV stroke volume (LVSV) respectively. (B) Compared to the normal LV systolic function group and the overt LV systolic dysfunction group, the occult LV systolic dysfunction group had significant reduction in LV ejection fraction (LVEF); (D) All-cause mortality was higher in the occult LV systolic dysfunction group as compared to the normal LV systolic function group, but comparable to the overt LV systolic dysfunction group

p denotes comparison across the 3 groups while p* denotes pairwise comparison between occult and overt LV dysfunction

LV Systolic Dysfunction and Mortality

The patients were followed for 68 (37–106) months, yielding a total follow-up 1,236 patient-years from the time of surgery. During this period, 11 (4%) patients died and 1 (0.3%) patient underwent heart transplant. The cause of death was end-stage heart failure (n=4), arrhythmia/sudden cardiac death (n=1), stroke/intracranial hemorrhage (n=1), malignancy (n=1), sepsis/multi-organ failure (n=2), end-stage renal failure (n=1), and unknown etiology (n=1).

The unadjusted annual risk of all-cause mortality was 0.93% (95% confidence interval 0.76–1.12) per year for the entire cohort. Compared to the normal LV systolic function group, the groups with occult LV systolic dysfunction and overt LV systolic dysfunction had significantly higher risk of mortality (0.61 [0.43–0.79] vs 2.33 [1.85–2.90], p<0.001) and (0.61 [0.43–0.79] vs 1.96 [1.14–2.82], p<0.001) respectively, Figure 2. However, there was no significant difference in the risk of mortality between the occult LV dysfunction and overt LV systolic dysfunction groups (2.33 [1.85–2.90] vs 1.96 [1.14–2.82], p=0.7), Figure 2.

Figure 2:

Kaplan Meier curves comparing survival between different LV function groups.

As compared to the normal LV systolic function group, the occult LV systolic dysfunction group and the overt LV systolic dysfunction groups had significantly lower survival (p<0.001 for each pairwise comparisons). However, there was no difference in survival between the occult LV systolic dysfunction group and the overt LV systolic dysfunction groups (p=0.3).

p denotes comparison across the 3 groups while p* denotes pairwise comparison between occult and overt LV dysfunction

DISCUSSION

In this retrospective cohort study, we observed that the use of LVGLS for the assessment of LV systolic function can identify patients with occult LV systolic dysfunction (abnormal LVGLS with preserved LVEF), and that these patients had worse preoperative hemodynamics, end-organ dysfunction and aerobic capacity as compared to patients with normal LVGLS. We also observed that patients with abnormal LVGLS (regardless of LVEF) had suboptimal LV reverse remodeling without improvement in cardiac output even after successful tricuspid valve surgery and effective RV unloading.

Previous studies have assessed LV reverse remodeling after tricuspid valve repair in patients with Ebstein anomaly.^{5, 8, 9} In one of these studies, Ibrahim et al⁵ evaluated 27 patients with Ebstein anomaly undergoing cone repair, and reported an increase in LVEDV which they postulated was related to an improvement in effective RV stroke volume (enhanced pulmonary artery flow). Two other studies by Beroukhim et al⁹ and Kuhn et al⁸ reported similar postoperative improvement in LV preload and stroke volume after tricuspid valve surgery in 20 and 16 patients respectively. Consistently, all these studies showed no postoperative improvement in LVEF. However, none of these studies performed subgroup analyses to determine whether patients with LV systolic dysfunction prior to surgery, had similar postoperative LV reverse remodeling as compared to those with normal LV systolic function. Hence an important knowledge gap still exists regarding the prevalence, hemodynamics, and clinical implications of LV systolic dysfunction in patients with Ebstein anomaly, and the results of the current study addressed some of these gaps.

Pathophysiologic Mechanisms of LV Dysfunction

We observed 2 distinct hemodynamic phenotypes of LV systolic dysfunction in our cohort. The occult LV systolic dysfunction group had small LV with low stroke volume while the overt LV systolic dysfunction group had normal LV size with low stroke volume. Both groups had worse right heart hemodynamics, aerobic capacity, renal function, and neurohormonal activation as compared to the group with normal LV systolic function. It is unclear why one group had preserved LVEF while the other had reduced LVEF. We postulate that the observed difference may be related to right heart compliance and function. The patients with occult LV systolic dysfunction had worse right heart dysfunction (lower RA and RV strain, and higher RA pressure), leading to higher diastolic LV eccentricity index (a measure of diastolic ventricular interdependence), and in turn, under filling of the LV and reduced LV output (Table 2). The reduced LVEDV, which is the denominator for LVEF calculation, therefore provides an erroneously normal LVEF in the setting of abnormal LVGLS (surrogate for LV contractility). The unloading of the RV after tricuspid valve surgery attenuates the pericardial restraint and ventricular interdependence leading to normalization of LV preload (the denominator of LVEF calculation), and in turn a reduction in LVEF, and thus unmasking LV dysfunction that has been present all along. This concept of enhanced ventricular interdependence leading to reduced LV preload and stroke volume has been previously described in patients with Ebstein anomaly by Fujioka et al.¹¹ In that study, the investigators observed a leftward bowing of the ventricular septum in early diastole, which was more pronounced in patients with more tricuspid regurgitation and right heart volume overload.¹¹ Similarly, the patients with more leftward bowing of the ventricular septum (a correlate of diastolic LV eccentricity index in the current study) also had lower LV preload, stroke volume, and peak VO₂ similar to our study. Although LVGLS is also sensitive to leading changes in conditions, it is less dependent on LV geometry, and has been shown to correlate with end-systolic pressure volume relationship which is a load-independent measure of LV contractility.^{14, 21} This is supported by data showing good reproducibility of LVGLS in the clinical setting.^{14, 21}

LVGLS has been shown to detect early phases of systolic dysfunction prior to a decline in LVEF based on data from the cardio-oncology literature.²³ Hence, another potential explanation for the different LV systolic dysfunction phenotypes observed in our cohort is that both phenotypes are a continuum of the same disease process with the occult group representing an earlier phase of the disease while the overt group represents a more advanced form of the disease. However, if that was the case, we will expect a better LV output, end-organ function, and aerobic capacity in the occult LV dysfunction group as compared to the overt LV dysfunction group.

Clinical Implications

The results of the study suggest that LVEF underestimates the prevalence of LV systolic dysfunction in patients with Ebstein anomaly (a prevalence of 9% for LVEF vs 22% for LVGLS). This has two important clinical implications. First, the patients with occult LV systolic dysfunction were less likely to be on GDMT compared to the overt LV systolic dysfunction (48% vs 96%) even though both groups had similar degree of hemodynamic derangement and end-organ dysfunction. Although we cannot determine whether optimal use of GDMT would have altered the clinical course in this cohort, this at least deserves further investigation since these therapies have been shown to improve clinical outcomes for LV cardiomyopathy in the acquired heart disease population.^24–26 The second clinical implication is with regards to the timing of surgical intervention. Since abnormal LVGLS was associated with suboptimal postoperative LV reverse remodeling, it will be important to determine whether performing surgical intervention prior to the onset of occult LV systolic dysfunction will prevent incident LV dysfunction.

Limitations

We used echocardiography for the assessment of LVEF instead of cardiac MRI which is the gold standard for volumetric assessment. However there was a good correlation between echo- and MRI-derived LVEF suggesting that echo-derived LVEF is a robust surrogate of MRI-derived LVEF. Although LVGLS is also load-dependent, it is a good approximation of LV end-systolic pressure volume relationship which is a load-independent measure of LV contractility.¹² The higher risk of mortality noted in the patients with LV systolic dysfunction might have been influenced by other factors since we did not perform multivariate adjustments because of low event rates. Finally, the study design does not provide insight into the possible etiologies of LV systolic dysfunction or the potential benefit of (or lack thereof) of medical therapy for LV dysfunction in patients with Ebstein anomaly.

Conclusions

LVGLS identifies 2 distinct hemodynamic phenotypes of LV systolic dysfunction in patients with Ebstein anomaly undergoing tricuspid valve surgery. Although both phenotypes had similar severity of hemodynamic derangement and end-organ function, the patients with occult LV systolic dysfunction were less likely to be on medical therapy for heart failure. Furthermore, both phenotypes had suboptimal postoperative LV reverse remodeling. These data highlight the limitations of using LVEF as the primary metric of LV systolic function assessment considering the abnormal LV loading conditions and geometry in this population. Consistent with robust data from the acquired heart disease literature regarding superiority of LVGLS to LVEF for diagnosis and monitoring of LV systolic dysfunction, the current study supports the use of LVGLS as the primary metric for LV function assessment in patients with Ebstein anomaly. This is a single center retrospective study with inherent limitations, and hence the results of the study need to be validated in a different population. Notwithstanding, these results provide the scientific premise for further studies to determine whether optimal medical therapy and/or earlier surgical intervention can improve outcomes for LV systolic dysfunction in patient with Ebstein anomaly.

CLINICAL PERSPECTIVE.

The results of the study suggest that LVGLS can identify patients with occult LV systolic dysfunction, and these patients had worse hemodynamics and outcomes compared to patients with normal LVGLS. Additionally LVEF underestimates the prevalence of LV systolic dysfunction in this population (a prevalence of 9% for LVEF vs 22% for LVGLS). This has two important clinical implications. First, the patients with occult LV systolic dysfunction were less likely to be on GDMT compared to the overt LV systolic dysfunction (48% vs 96%) even though both groups had similar degree of hemodynamic derangement and end-organ dysfunction. Although we cannot determine whether optimal use of GDMT would have altered the clinical course in this cohort, this at least deserves further investigation since these therapies have been shown to improve clinical outcomes for LV cardiomyopathy in the acquired heart disease population. The second clinical implication is with regards to the timing of surgical intervention. Since tricuspid valve surgery does not reverse LV systolic dysfunction once it has occurred, it will be important to determine whether performing surgical intervention prior to the onset of occult LV systolic dysfunction will prevent incident LV dysfunction.

Abbreviations:

LVEF

Left ventricular ejection fraction

LVGLS

Left ventricular global longitudinal strain

LVEDV

Left ventricular end-diastolic volume

LVESV

Left ventricular end systolic volume

LVSV

Left ventricular stroke volume

GFR

Glomerular filtration rate

NT-proBNP

N-terminal pro-brain natriuretic peptide

VO₂

oxygen consumption

MRI

Magnetic resonance imaging

GDMT

guideline directed medical therapy

RVGLS

Right ventricular systolic function