Change in Right Ventricular Strain After Cone Reconstruction of Ebstein’s Anomaly: A Cardiovascular Magnetic Resonance-Feature Tracking Study

Abstract

Background: Cardiovascular magnetic resonance Feature Tracking (CMR-FT) is a well-established method to assess myocardial contraction with diagnostic and prognostic value in many diseases. We aimed to evaluate the role of right ventricular (RV) CMR-FT in the perioperative assessment of Ebstein patients undergoing Cone repair. Methods: We analyzed the CMR data of 18 Ebstein patients before and after Cone repair including CMR-FT-derived global radial (GRS), global circumferential (GCS) and global longitudinal strain (GLS). Results: Following Cone repair, tricuspid regurgitation decreased from 48% to 6%, p = 0.0001. RV ejection fraction (51% to 33%, p = 0.0002), indexed RV stroke volumes (74 mL/m² to 43 mL/m², p = 0.0013) and GLS (−15.01% to −14.53%, p = 0.0155) decreased postoperatively. Conversely, GRS (15.00% to 17.83%, p = 0.0202) and GCS (−8.82% to −13.02%, p = 0.0026) improved. Indexed RV end-diastolic volumes (RVEDVis) decreased, although not significantly, from 161 mL/m² to 122 mL/m², p = 0.3465. Eight patients exhibited a higher RVEDVi after surgery. Pulmonary artery and aortic flow and left ventricular (LV) functional parameters remained unchanged. Conclusions: RV GLS appears to be affected by the hemodynamic alterations caused by Cone repair. RV GCS and GRS might serve as more independent parameters of myocardial function.

1. Introduction

Ebstein’s anomaly (EA) is characterized by the apical displacement of the tricuspid valve (TV). It is considered a developmental disorder of the right ventricular (RV) myocardium, accounting for less than 1% of congenital heart diseases. During embryogenesis, the septal and posterior leaflets of the TV fail to delaminate from the underlying myocardium, resulting in apical displacement and dilation of the functional TV annulus. Consequently, the RV is divided into two portions: a functional and an atrialized ventricle. Tricuspid regurgitation (TR) occurs with consecutive volume overload of the right heart [1,2].

The latest surgical approach for EA is da Silva’s Cone repair, which, for the first time, enables the anatomical repair of the TV [3,4,5]. This procedure can be performed in various anatomical forms of the disease and significantly reduces TR, with a low mortality rate of only 2.5% [3,6]. In this technique, the enlarged anterior leaflet is detached from the ventricular wall and rotated to create a cone shape. The remnants of the posterior and septal leaflet may be incorporated into this cone-shaped valve, the base of which is then reattached to the anatomic annulus [3].

Several studies have reported positive clinical outcomes after Cone repair with improved postoperative NYHA class and reduced TR. To assess the precise impact of Cone repair on RV function, several echocardiographic and cardiovascular magnetic resonance (CMR) studies have been conducted. However, identifying a parameter that reflects actual myocardial changes is difficult and complex, as most functional parameters are influenced by altered postoperative hemodynamics [4,5,7,8].

CMR Feature Tracking (CMR-FT) is a relatively new method for quantifying myocardial deformation [9]. Initially developed for the left ventricle, CMR-FT may also apply to the RV despite its thin wall and trabeculations [10,11]. Several studies have demonstrated that this method can evaluate subtle alterations in myocardial contraction that routine cardiac imaging cannot detect [12,13]. We are the first to perform CMR-FT of the right ventricle (RV) before and after Cone repair with the objective of assessing its value for the perioperative assessment of myocardial function in Ebstein patients undergoing Cone repair.

2. Materials and Methods

2.1. Patient Population

We retrospectively included all patients diagnosed with EA and who had Cone repair between June 2017 and September 2021 at our center. All individuals who had undergone a complete standardized CMR protocol before and after surgery were included in this study. Of those 21 patients meeting these criteria, 3 had to be excluded, because preoperative CMR images were not compatible with the post-processing software CVI. We compared CMR data between the preoperative baseline and the postoperative follow-up examination. The study received approval from the local ethics committee (Approval Number: 213/21 S-KK/8 April 2021).

2.2. CMR Acquisition Protocol

CMR acquisition followed a standardized protocol established at our center. A standard 1.5 Tesla CMR scanner (MAGNETOM Avanto^®, Siemens Healthineers, Erlangen, Germany) was employed using a 12-element cardiac phased array coil. For volumetric assessment, we obtained axial, multiphase, steady-state-free-precession (SSFP) cine images using retrospective ECG-gating. Axial slices were preferred over conventional short-axis slices due to their superior reproducibility and were obtained using coronal and sagittal localizers [14,15]. To minimize motion artifacts, breath holding was performed during expiration. Slice thickness was 4 to 8 mm depending on the patient’s body weight. We acquired one slice per breath-hold (8–12 s), capturing 25 phases per cardiac cycle for each slice. An acquisition matrix of 192 × 192 was used.

For the purpose of volumetric assessment, the endocardial contours of the functional RV and left ventricle (LV) were manually outlined in axial slices at both end-systole and end-diastole. Papillary muscles and trabeculations were excluded from these tracings as they are considered part of the cardiac mass. The atrialized RV was excluded from the measurements, as it does not contribute to right ventricular ejection fraction (RVEF) [15,16].

Measurement of aortic flow, pulmonary artery flow, and flow through the TV was conducted using two-dimensional phase contrast images. The regurgitant volume and fraction of the TV were obtained from phase contrast images en face to the tricuspid valve orifice. In cases where this measurement was not technically feasible, the regurgitant volume and fraction were calculated from the pulmonary artery flow and the right ventricular stroke volume as described by Fratz et al. [16].

CMR-FT analysis was executed utilizing short-axis and four-chamber SSFP cine images. Subsequently, all CMR data were transferred to a dedicated workstation for post-processing (Circle Cardiovascular Imaging Inc., CVI42^® 5.12.1 software, Calgary, AB, Canada).

2.3. CMR Feature Tracking

Two expert readers quantified RV myocardial deformation using Circle CVI42 5.12.1 software (Tissue Tracking, Circle Cardiovascular Imaging Inc., Calgary, AB, Canada). In both short-axis and four-chamber SSFP cine images, we manually delineated the endocardial and epicardial contours of the functional RV at the end-diastolic phase (Figure 1).

During the preoperative evaluation, the atrialized portion of the RV was excluded from the analysis, as it does not contribute to RV pump function [1]. Contours were then automatically propagated throughout the cardiac cycle using a dedicated algorithm. The accuracy of propagation was visually verified using an overlay mode. In cases of inadequate tracking, the contours were manually adjusted, and the algorithm was reapplied. Global longitudinal strain (GLS) was measured on four-chamber cine images, while global circumferential strain (GCS) and global radial strain (GRS) were derived from the right ventricular free-wall of short-axis images (basal, midventricular and apical slices). Free wall strain was favored because the preoperative assignment of the septum to the atrialized or functional ventricle can be challenging in short-axis slices. GLS and GRS measurements were possible in 17 out of 18 patients, and GCS could be measured in 15 patients before and after the surgery. This was due to the complex geometry of the RV in Ebstein patients. The reproducibility of the measurements was assessed in our previous study, where two blinded expert readers performed CMR-FT twice on 10 randomly selected patients [10].

2.4. Statistical Analysis

Continuous variables are presented as medians with ranges (minimum to maximum). Pre- and postoperative data were compared using the Wilcoxon signed-rank test. A p-value less than 0.05 was considered statistically significant. All statistical analyses were conducted using GraphPad Prism (version 9.2.0, GraphPad Software Inc., La Jolla, CA, USA).

3. Results

The study included 18 patients, comprising 7 males and 11 females. The median age at the time of the Cone repair was 14.9 years, ranging from 2.4 to 40.3 years. The global median follow-up time after Cone repair was 12.9 months, ranging from 3 to 19 months. None of the patients required reoperation during the follow-up period.

Table 1. Baseline characteristics of patients with Ebstein’s anomaly (n = 18).

Parameter	Min	Median	Max	Mean	SD
Age at Cone procedure (years)	2.37	14.90	40.27	18.97	10.33
Height pre Cone CMR (m)	0.77	1.69	1.83	1.58	0.27
Weight pre Cone CMR (kg)	9	59.5	77	54.78	20.2
Body surface area (cm/m2)	0.44	1.69	1.93	1.54	0.43
Time interval pre Cone CMR to Cone repair (days)	1	229	1156	280	294
Time interval Cone repair to post Cone CMR (days)	94	387	575	377	120

SD, standard deviation; CMR, cardiovascular magnetic resonance.

shows the baseline characteristics.

Table 2. Comparison of pre- and postoperative CMR data.

CMR Parameters	Preoperative (n = 18 *)	Postoperative (n = 18 *)	p-Value
RV global radial strain (%)	15.00 (3.47–24.37)	17.83 (3.46–35.88)	0.0202
RV global circumferential strain (%)	−8.82 ((−14.99)–(−2.58))	−13.02 ((−20.22)–(−7.62))	0.0026
RV global longitudinal strain (%)	−15.01 ((−30.01)–(−8.53))	−14.53 ((−18.67)–(−3.00))	0.0155
RVEF (%)	51 (21–61)	33 (13–50)	0.0002
RVEDVi (mL/m2)	161 (62–292)	122 (87–253)	0.3465
RVESVi (mL/m2)	74 (42–152)	75 (47–196)	0.3520
RVSVi (mL/m2)	74 (17–161)	43 (19–57)	0.0003
LVEF (%)	56 (28–67)	61 (43–70)	0.0859
LVEDVi (mL/m2)	63 (40–198)	65 (38–91)	0.4878
LVESVi (mL/m2)	28 (19–142)	26 (14–39)	0.4747
LVSVi (mL/m2)	35 (19–56)	37 (23–52)	0.2383
TI direct (%)	48 (16–90)	6 (0–19)	0.0001
TI indirect (%)	55 (0–80)	8 (0–20)	0.0039
PA flow netto indexed (mL/m2)	38 (18–52)	41 (20–52)	0.2897
PA flow antegrade indexed (mL/m2)	39 (20–54)	42 (21–55)	0.2480
Cardiac index PA (L/min/m2)	2.8 (1.7–4.3)	2.8 (1.8–3.9)	0.9678
Aorta flow netto indexed (mL/m2)	38 (23–48)	40 (20–54)	0.1066
Aorta flow antegrade indexed (mL/m2)	38 (23–48)	41 (20–54)	0.1058
Cardiac index aorta (L/min/m2)	2.8 (1.8–4.7)	2.8 (1.6–5.0)	0.5619
Heart rate	82 (56–98)	73 (59–98)	0.1099

Data are presented as median (min, max). CMR, cardiovascular magnetic resonance; RV, right ventricle; EF, ejection fraction; EDVi, indexed end-diastolic volume; ESVi, indexed end-systolic volume; SVi, indexed stroke volume; LV, left ventricle; TI, tricuspid insufficiency; PA, pulmonary artery. * for RVGRS and RVGLS n = 17 and RVGCS n = 15. SD, standard deviation; CMR, cardiovascular magnetic resonance.

compares pre- and postoperative CMR data.

Following Cone repair, we observed a significant decrease in TR (48% vs. 6%, p = 0.0001). We did not observe a significant change in indexed right ventricular end-diastolic volumes (RVEDVis) after surgery (161 mL/m² vs. 122 mL/m², p = 0.3465); however, there was a trend towards postoperative reduction. Notably, eight patients (44%) exhibited a higher RVEDVi after surgical repair. Both right ventricular ejection fraction (RVEF) (51% vs. 33%, p = 0.0002) and indexed right ventricular stroke volume (RVSVi) (74 mL/m² vs. 43, p < 0.0003) showed significant reductions after surgery. During surgery, a reduction in the size of the atrialized RV was performed in fifteen out of eighteen patients (83%). In the group of the eight patients with increased RVEDVi after Cone repair, six of them had a reduction in the atrialized RV during surgery.

No significant differences were observed in the pulmonary artery and aortic flow, nor in the left ventricular functional parameters, including the left ventricular ejection fraction (LVEF), left ventricular end-diastolic volume indexed (LVEDVi), left ventricular end-systolic volume indexed (LVESVi), and left ventricular stroke volume indexed (LVSVi).

Regarding CMR-FT, there was a significant decrease in GLS after Cone repair (−15.01% vs. −14.53%, p = 0.0155), while significant improvements were noted in GRS and GCS (15.00% vs. 17.83%, p = 0.0202 and −8.82% vs. −13.02%, p = 0.0026), as illustrated in Figure 2.

4. Discussion

4.1. Impact of Cone Repair on Right Ventricular Volumes

Cone repair has been shown to effectively alleviate RV volume overload by reducing TR, as evidenced by previous studies [4,5,8]. Consistent with these findings, we observed minimal to no residual TR and a decrease in RVEDVi postoperatively. However, it is worth noting that some patients presented with higher RV volumes after surgery. This phenomenon can be attributed to the fact that, in these cases, the functional annulus of the TV was severely displaced, resulting in a significantly larger atrialized portion than the functional RV (Figure 3). With Cone repair, the atrialized RV is reintegrated into the functional RV, which may lead to an increase in RV net volume despite reduced TR. This might depend on the extent of the reduction in the atrialized RV during surgery.

4.2. Role of CMR-FT in the Perioperative Assessment of Right Ventricular Function

It has been of clinical interest to assess the precise impact of Cone repair on RV function; however, identifying a parameter that reflects myocardial changes remains challenging. Several studies have evaluated conventional parameters of RV function before and after Cone repair using echocardiography and CMR, many of them reporting decreased postoperative RVEF and RVSVi [4,5,7,8], consistent with our findings.

However, it is crucial to recognize that the loading conditions strongly influence these parameters of RV function, and the regurgitant volume of the insufficient tricuspid valve may contribute to a ‘false’ impression of normal RVEF and RVSVi before surgery. With TV repair, the regurgitant volume is retained in the ventricle and the postoperative RVEF and RVSVi reflect only the volume ejected forward through the pulmonary valve. Hence, in the postoperative assessment, the impaired RVEF and RVSVi do not solely indicate postoperative RV deterioration; rather, they reveal a pre-existing myocardial dysfunction that becomes more apparent with a competent TV. Additionally, they reflect the differing volume status of the RV before and after Cone repair, making these parameters unsuitable overall for perioperative assessment of myocardial function [8].

Parameters used by other authors to evaluate the actual perioperative change in RV function were pulmonary artery flow and left ventricular end-diastolic volumes, reflecting the ability of the RV to pump the blood forward [4,5,8].

RV strain has been proven to be valuable in detecting myocardial dysfunction in several cardiac diseases, even when functional parameters such as RVEF remained within the normal range [12,13].

In the current study, we noted a deterioration of GLS alongside improvements in GRS and GCS following Cone repair (Figure 2). These findings align with a recent echocardiographic study by O’Leary et al., encompassing 257 EA patients. Their study reported similar results, wherein post Cone repair, the RV fractional area change (FAC), as an indicator of longitudinal RV function, demonstrated a significant decline. At the same time, there was a slight enhancement in the RV short-axis FAC [17].

The deterioration of GLS, similar to RVEF and RVSVi, observed in this study, suggests that GLS might also be affected by the preoperatively reduced afterload, thus masking the actual preoperative myocardial function, since pushing the blood volume backward through the TV is easy for the RV in the preoperative state. Postoperatively sustained pulmonary artery flows and left ventricular volumes reflect the maintained effective function of the RV, suggesting that postoperative impairment of these parameters is due to increased postoperative afterload rather than true myocardial impairment after Cone repair.

Our previous study showed that in unoperated EA patients, GLS did not differ from that of healthy individuals whereas significant impairment was observed in GCS and GRS, suggesting that in EA there is a predominant impairment of short-axis contraction, while the longitudinal function remains preserved [10].

Contrarily, a recent CMR study conducted by Kresoja et al. demonstrated that in unoperated patients with TR unrelated to EA, specific individuals exhibited a reduction in RV GLS while maintaining preserved RVEF. In these cases, the values of compensatory GRS and GCS were above normal levels [18].

This observation may support the hypothesis that in EA, longitudinal and circumferential fibers contribute differently to RV systolic function than normal hearts. In normal hearts, RV systolic function is primarily governed by the contraction of longitudinal fibers, leading to a peristaltic movement from the inflow to the outflow tract [19]. In the context of EA, the RV function appears to depend more on the shortening of circumferential fibers [20].

Our hypothesis suggested that in EA, compared to other heart diseases with TR, longitudinal fibers have a propensity to favor backflow. This could elucidate the initially normal GLS values in EA before surgical intervention and the subsequent decline in longitudinal function following the restoration of valve competency.

Despite sustained pulmonary artery flow indicating that the postoperative deterioration of GLS is primarily due to increased postoperative afterload, other factors might also contribute to this observation.

First, it is important to note that GLS calculations encompassed the previously atrialized RV wall after Cone repair, typically characterized by thinness and fibrotic changes [7]. As a result, the impaired movement in this region may also contribute to the altered postoperative RV longitudinal performance. Even though the atrialized portion of the RV also contributes to radial and circumferential strain measurements, we observed improvements in these parameters. This could be explained by the fact that GRS and GCS are given by strain analysis at basal, mid, and apical levels. Therefore, a postoperative deterioration in basal strain could, in the case of improved mid and apical strain, still result in overall improved strain measurements. The precise role of the former atrialized segment of the RV in postoperative systolic RV function remains uncertain. Therefore, conducting an FT analysis of the basal (septal) area before and after surgery would be valuable in gaining a deeper understanding.

It is also worth noting that the longitudinal fibers constitute the innermost layer of the RV wall [19] and that this layer is particularly susceptible to ischemia [21]. Thus, potential hypoxic damage through aortic cross clamping during Cone repair might also contribute to postoperatively impaired GLS.

Furthermore, it is conceivable that the delamination of the leaflets from the underlying myocardium during Cone repair could damage the inner myocardial layer, contributing to the reduced postoperative GLS [3].

While GLS was demonstrated to be effective in detecting early myocardial dysfunction in other diseases [12,13], our previous study evidenced that this does not seem to be the case in EA [10]. The results of this study might explain the limited diagnostic value of this parameter in EA.

Conversely, GRS and GCS do not seem to be much affected by the hemodynamic changes caused by Cone repair. We hypothesized that, unlike GLS, GRS and GCS might favor forward flow more than backflow and, therefore, could be more independent of afterload changes caused by the restored TV function. Hence, GCS and GRS might reflect true myocardial changes. The observed postoperative improvement could therefore indicate an actual improvement in myocardial function after Cone repair. Whether this is due to direct effects on myocardial function from the postoperatively reduced volume overload or due to myocardial remodeling over the time remains unclear, as our postoperative follow-up times vary too widely to provide a clear explanation for the time course of the suspected improvement.

However, it is important to note that minimal improvements in strain most likely do not always directly translate into increased forward stroke volume, especially given the reduced ventricular volume after surgery. This can be explained by the fact that strain is defined as the percentage of shortening (or, in the case of radial strain, the percentage of thickening) of the myocardium between end-diastole and end-systole measured in a certain slice. The effectively ejected stroke volume given a certain percentage of shortening, therefore, depends on the RVEDV. In cases of higher RVEDV, the same strain values mean higher ejected volumes than in cases where EDV is low. This may explain why improvements in GCS and GRS are not always accompanied by corresponding changes in pulmonary artery flow, as observed in this study.

4.3. Limitations

The study comprised a limited cohort of patients with a wide age range, primarily owing to the rarity of the disease and the specific use of Cone repair. Additionally, it is essential to note that the study’s retrospective nature resulted in varying and often limited patient follow-up durations. Possible effects depending on the follow-up time or the patient’s age could, therefore, remain undiscovered. Furthermore, we did not consider correlating strain parameters with clinical outcomes, such as adverse events after surgery, as this would not be expected to yield meaningful insights. Further investigations with a larger sample size are warranted.

5. Conclusions

In contrast to previous studies [8], we did not observe a significant reduction in RVEDVi (161 mL/m² vs. 122 mL/m², p = 0.3465) after Cone repair. Interestingly, 8 out of 18 patients exhibited higher postoperative RVEDVi. This could be attributed to the fact that, in Ebstein patients with a severely displaced tricuspid valve, postoperative RVEDVi might be increased due to the incorporation of the atrialized RV to the functional RV.

We observed a postoperative deterioration of RV GLS (−15.01% vs. −14.53%, p = 0.0155), similar to conventional CMR parameters. Pulmonary artery flows remained unchanged, indirectly suggesting maintained RV systolic function. This suggests that GLS may also depend on the altered postoperative afterload conditions caused by the restored valve function. Preoperative GLS would therefore be a similarly confounded parameter of myocardial function, making it unreliable for assessing perioperative changes in RV function in Ebstein’s anomaly. In contrast, GRS and GCS did not deteriorate—in fact, they even improved (15.00% vs. 17.83%, p = 0.0202 and −8.82% vs. −13.02%, p = 0.0026)—suggesting that they are less affected by postoperative hemodynamic alterations and therefore might have potential for the pre- and postoperative assessment of myocardial function in Ebstein’s anomaly.