Mid-term Outcome Of Surgical Treatment In Pediatric Patients With Ebstein's Anomaly: A Single-center Cohort Study.

Background: Ebstein’s anomaly is a malformation of the tricuspid valve and myopathy of the right ventricle. Surgery is now the main treatment for the defect. To summarize our surgical results and experience based on patients with Ebstein’s anomaly who were under 7 years of age and treated with different surgical treatments. enrolled Results: The median age of the 80 patients at the time of surgery was 3.63 years. Sixty-four (80.00%) patients underwent biventricular repair while 13 (16.25%) underwent 1.5-ventricle repair. With the median follow-up 27.50 months, the long-term survival of the total cohort, 1.5-ventricular repair and biventricular repair was 82.35%, 91.67% and 100%, respectively. The long-term freedom from reoperation rate was 97.50%, 92.31% and 98.44%, respectively. Mild, moderate and severe TR before surgery occurred in 6 (7.50%), 18 (22.50%) and 56 (70.00%), respectively. The early outcomes of 78 patients were 65 (83.33%), 11 (14.11%) and 2 (2.56%); the mid-term outcomes of 72 patients were 49 (68.06%), 19 (26.38%) and 4 (5.56%). Both early and long-term valve regurgitation were signicantly decreased (p< 0.001) compared with preoperative condition. No more severe regurgitation occurred (p=0.404), though some early mild regurgitation became acceptable moderate regurgitation during long-term follow-up (p=0.036). Compared with Carpentier procedure, cone procedure had better long-term effect, while the effect of whole-valve technique needed more operation and long-term follow-up. Conclusion: The reoperation rate and mid-term mortality of surgical treatment for Ebstein’s anomaly were both low,


Introduction
Ebstein's anomaly (EA) is a malformation of the tricuspid valve (TV) and myopathy of the right ventricle (RV) [1] . The incidence is 1 per 200,000 live births and it accounts for 1% of congenital heart disease [2] .
The core and critical distinguishing characteristic of EA from other congenital regurgitant lesions is the degree of apical displacement of the septal lea et (≥ 8 mm/m 2 body surface area) [3] . The pathological characteristics of each EA patient is different, which makes the surgical management complex and individualized, especially when it is treated with tricuspid valvuloplasty. A number of univentricular palliation and biventricular repair methods, including various modi cations of techniques, have been described [4][5] . Studies on surgical outcomes in patients with EA often include a majority of adults; consequently, long-term outcomes of biventricular repair are generally good. However, there are few studies on neonates and young children with large sample sizes. We therefore conducted a retrospective study on patients with EA who were under 7 years of age and treated with different surgical treatments based on the recent 10-year data of our institution and summarized our surgical experience.

Patient selection and data collection
Eighty patients under 7 years old who were diagnosed with EA and underwent different surgical treatments were consecutively enrolled in our research from January 2010 to December 2019. The diagnosis mainly relied on echocardiography and transesophageal ultrasound and was con rmed during the intraoperative probe. The diagnostic criterion was apical displacement of the TV from the atrioventricular ring >0.8 cm/m 2 body surface area [6] . Children with other complex congenital heart diseases, including pulmonary atresia, tetralogy of Fallot and corrected transposition of the great arteries were excluded. Demographic data and clinical characteristics included sex, age, weight, body length, cardiac functional class, presence of cyanosis, Carpentier type, associated cardiac anomalies, the presence of preoperative arrhythmias, prior cardiac operations and preoperative degree of tricuspid regurgitation (TR). Surgical procedures and postoperative parameters, including cardiopulmonary bypass (CPB) time, aortic cross clamp time, mechanical ventilation time, intensive care unit (ICU) stay, complications, mortality, cause of death, and TV and RV function, when available, were also comprehensively evaluated. Patients' records and postoperative follow-up were conducted through our outpatient service and over the telephone.

Surgical methods standard
In general, surgical procedures can be divided into three categories, biventricular repair, 1.5-ventricle repair or univentricular palliation. The decision for the type of repair was based on the ratio of the atrialized right ventricle (aRV) to the functional right ventricle (fRV), the size of the TV annulus and the size of the pulmonary artery in addition to the severity of TR. Biventricular repair was performed in patients with a retained RV function and an acceptable anatomy and fRV in the absence of severe pulmonary valve stenosis or atresia. Due to the lack of preoperative quantitative evaluation, most of the decisions for 1.5ventricle repair were made by the surgeon during the operation. In patients with extensive RV failure, severe TR and/or the presence of other relevant cardiac comorbidities, e.g., PA, a univentricular palliation was the intended surgical strategy. All operations were analyzed as separate events.

Evaluation standard
There are some evaluation criteria that can help us understand the preoperative condition and postoperative recovery of patients. Cyanosis was determined when the increasing level of deoxyhemoglobin in the blood resulted in a purplish appearance of the skin mucous membrane. The severity of EA was evaluated by echocardiography and transesophageal echocardiography and determined by the internationally recognized Carpentier type. Early outcome mainly referred to death or reoperation within 30 days after surgery or before hospital discharge. Mid-term outcome mainly referred to death or reoperation within one to ve years after surgery.

Statistical analysis
Descriptive statistics for the 80 patients are reported. Data are presented as means ± standard deviations (SDs) for continuous data with a normal distribution, as medians (25th percentile, 75th percentile) for continuous data with a nonnormal distribution, or as numbers and percentages for categorical variables, as appropriate. factors for early mortality. Freedom from reoperation and cumulative survival rates were analyzed according to the Kaplan-Meier test. There was insu cient incidence of deaths and reoperations to perform an analysis to identify risk. Differences in the frequency of TR at preoperative, early and midterm follow-up examinations were performed using the chi-square test and Fisher test. Statistical signi cance was de ned as P<0.05 with two-tailed distributions. All statistical analyses were performed with SPSS version 22.0 software (SPSS, Inc., Chicago, IL, USA) and GraphPad Prism version 5.0 software (GraphPad Software, Inc., La Jolla, CA, USA).

Patient pro les
The demographic data and clinical characteristics of the patients are summarized in Table 1.

Surgical procedures and postoperative data
The surgical procedures performed and postoperative data are presented in Table 2. Of the 80 patients, two (2.50%) patients underwent the Glenn procedure because of tricuspid ori ce stenosis and right ventricle dysplasia. Tricuspid valve replacement (TVR, 29 # , Medtronic, Bio-valve) was performed for one (1.25%) patient because of severe tricuspid valve dysplasia. Two (2.50%) patients underwent unexpected reoperation, and 1 patient underwent rerepair because of recurrent TR on the 6 th day postoperatively; another patient received the Glenn procedure because of acute right heart failure on the 4 th day postoperatively. The early mortality was 1.25%, and 1 patient's venous pressure increased and blood pressure decreased after biventricular repair. The Glenn procedure and extracorporeal membrane oxygenation were then performed, but the patient consistently could not maintain spontaneous breathing after surgery because of poor right heart function and passed away on the 38th day after the operation.

Surgical technique
The detailed surgical strategies are shown in Table 3. Regular methods of EA surgical correction included the Danielson procedure, Devage procedure, Carpentier procedure, cone procedure, and whole-valve technique. The main procedures applied in our center were as follows. 1) Danielson procedure: lateral aRV plication, and the lobes were raised to the level of the normal valve ring. 2) Devage procedure: plication with 1 or 2 mattress sutures from the junction of the anterior and posterior lobe to the junction of the septal and posterior lobe, whereby the apical aspects of the septal and anterior lea ets would effectively create a bicuspid valve. 3) Carpentier procedure: most of the anterior valve and adjacent part of posterior valve were detached and sutured at the level of the normal tricuspid valve ring, longitudinal aRV plication, such that the TV became a single valve structure dominated by the anterior valve. 4) Cone procedure: all three lobes were cut off at the root of the lobe, the adhesion of the subvalve tissue was released, and the three lobes were then structured to be a cone-like body, clockwise rotated and sutured into the normal valve ring. 5) Whole-valve technique: an optimized and improved technique based on the cone procedure in our center. In the case of tricuspid dysplasia, the free margin of each lobe is actively widened with fresh autologous pericardium to ensure su cient alignment between the lobes. Among the 13 patients, 4 received the intended bidirectional cavopulmonary shunt (BCPS) as a preoperative plan, the other 9 received unplanned BCPS because of poor RV function during the operation. Other modi ed techniques based on regular methods, including right ventricle plication, TV annuloplasty plication and autologous lea et augmentation, prosthetic ring or arti cial chordae, were not used in the enrolled patients. ASD subtotal closure was also considered in the research.

Follow-up and mid-term outcome
The median follow-up of the 80 patients was 27.50 (10.25, 56.75) months, with 3 patients loss to followup after discharge, and the effective follow-up rate was 96.25%. One early death occurred after 1.5 ventricular repair on the 38th day after the operation, as mentioned above. One late death occurred in the 6th year after the Glenn procedure because of RV dysplasia. As shown by the survival curve in Figure 1a, the overall survival was 98.71% at 4 years (50 months) and 82.35% at 8 years (100 months). The survival of patients with 1.5-ventricular repair was 91.67% at 4 years and at 8 years, and no patients who received biventricular repair have died to date. The freedom from reoperation curve is shown in Figure 1b. The long-term freedom from reoperation rate of overall, 1.5-ventricular repair and biventricular repair was 97.50%, 92.31% and 98.44%, respectively. Except for 2 patients who received tricuspid valve rerepair and the Glenn procedure on the 6th day and the 4th day after surgery, no patients received reoperation to date. Of the 6 cases with arrhythmia in the early postoperative period, 1 patient had atrioventricular block of the second degree, 3 had untreated transient ventricular tachycardia, and 2 improved spontaneously after discharge. Other clinically important complications, including right ventricular out ow tract obstruction, left ventricular out ow tract obstruction or TV stenosis, were not observed either before or after surgical repair.

Discussion
EA's anatomical features [2] include 1) displaced and dysplastic septal and posterior leaflets; 2) an elongated, fenestrated, or dysplastic anterior leaflet; 3) a markedly dilated atrialized right ventricle that becomes thin and dyskinetic; and 4) a displaced functional annulus and a markedly expanded anatomic annulus. Many procedures have been reported based on this anatomical theory. In general, surgical methods can be divided into three categories: biventricular repair, 1.5 ventricle repair and univentricular palliation. In this research, different surgical methods and relative long-term prognosis of children with EA in our institution were systematically summarized for the rst time. It was a large-sample sized cohort study for Chinese EA children, and it was of great signi cance to understand the status quo of diagnosis and treatment for such disease in China.

Outcome summary
Among the patients diagnosed with EA under 7 years old in our study, 6 (7.50%) were younger than 1 year old, and 19 (23.75%) were between 1 and 2 years old. Twelve (15.00%) cases were serious Carpentier type III-IV. There were 64 biventricular repairs, 13 1.5-ventricle repairs, 2 univentricular palliations and 1 TVR. Compared to previously reported studies, the proportion of biventricular repair was high but the proportion of 1.5-ventricle repair and univentricular palliation rather low in our center. To date, we have not performed the Starnes procedure for neonates who must undergo emergency surgery. During the last 10 years, our concept of biventricular repair has constantly been updated. The Carpentier procedure was the conventional method in the early stage, and the cone and its modi ed methods gradually dominated in the late stage. The overall outcome of the operations was worthy of recognition. The rates of long-term mortality and reoperation were both very low. TR was signi cantly improved, and the proportion of severe regurgitation decreased from 70% before surgery to 5.56% during long-term follow-up. Serious long-term complications did not occur except for 6 cases of untreated arrhythmia.

Indications for surgery
Traditional indications for surgery included symptoms of cyanosis, fatigue, paradoxical embolism, decreasing exercise performance (preferably documented by exercise testing), progressive RV enlargement, progressive RV dilation or reduction of systolic function by echocardiography and/or appearance of atrial or ventricular arrhythmias [7] . The timing of surgery for EA continued to be controversial, particularly when confronted with an asymptomatic patient with normal exercise tolerance.
In fact, the vast majority of patients with EA require surgery in their lifetime. Dearani indicated [7] that the operation should occur in early childhood as long as the surgical team can provide a reproducible and predictable satisfactory tricuspid repair. Their repair rate in children approached 100%, and the risk of operation was <1%, as reported in a large sample-size retrospective study [8] . Similar results could be drawn in our research. The overall survival was 98.71% at 4 years and 82.35% at 8 years, and none of the patients who received biventricular repair have died to date. Among these patients, only 22.50% presented with symptoms of cyanosis. Therefore, we may delay the timing of the operation based only on symptoms.

Experience of surgical intervention
The original method from the 1960s was to uplift the descending valve, including the Hardy technique and Danielson technique [9] . In the 1990s, the Carpentier technique [10] and subsequent Quaegebeur technique [11] incorporated surgical delamination of the anterior lea et with annular reattachment to optimize monocusp mobility for ventricular septal coaptation. Since 2007, the cone reconstruction [12] technique has become one of the most widely recognized surgical procedures for EA. It is an extension of the previous technique in that septal and inferior lea et tissues were mobilized, after which the sides of all mobilized lea ets were connected in a manner to create 360 degrees of lea et tissue. Near-anatomic repair could be achieved when the newly constructed "cone" was reattached to the true annulus. After that, many other studies have reported their modi ed cone technique and achieved good results, including ringed annuloplasty, closure of lea et fenestration, Sebening stitch, and arti cial chordae among others [8] . According to our research, the long-term effect of cone surgery was signi cantly better than that of Carpentier surgery. However, during our clinical treatment and subsequent re-examination, we found that with the cone technique, some patients still showed different degrees of TR and RV enlargement at the later stage, even though the initial postoperative effect was good, especially for those with severe tricuspid dysplasia. From January 2018, we began our modi ed method named the "whole-valve technique". Our experiences are as follows: 1) all the clingy lea ets of the tricuspid valve su ciently dissociated from the right ventricular wall to guarantee effective closure of the detached part of the anterior and posterior lea ets; 2) in the cases of tricuspid valve dysplasia, fresh pericardial tablets were actively used for splicing the detached part of each lea et; and 3) new lea ets were sutured to the plication of the true tricuspid annulus. To date, a total of 8 cases of this technique have been performed, and the 2-year follow-up was satisfactory, including 2 Carpentier type IV cases. No additional Glenn procedure has been performed, and more cases and longer follow-up times will receive further attention.
In addition, the modi ed Devage procedure, also named the "play it where it lies" technique [13] , was performed in 2 patients, both of whom underwent the intended Glenn procedure and had moderate TR at the latest reexamination. In our opinion, with regard to the choice of surgical procedure, the "play it where it lies" technique should be chosen with caution. Due to the small number of cases, we were unable to summarize the effective TVR experience and indications for young children.
When should we perform BCPS? BCPS is applied selectively in patients to unload RV volume in the setting of severe RV dysfunction. Studies have shown that BDPS decreases the volume of the enlarged, dysfunctional right ventricle by 35-45% (depending on the patient's age), providing substantial preload to the under lled left ventricle and reducing the hemodynamic stress on a complex TV repair [7] . A multicenter retrospective study [13] from the Netherlands reported that between 1980 and 2013, 63 children (0-112 months) underwent 109 operations. Two of 37 biventricular repair patients received 1.5-ventricle repair at 11 and 105 months after surgery. Five 1.5-ventricle repair patients received BCPS in the rst surgery. The other 21 patients underwent univentricular palliation surgery due to extensive RV failure, severe TR or the presence of signi cant other cardiac abnormalities. In terms of this study, 7 1.5-ventricle repair cases were performed in 109 operations (6.4%), and the rate of 1.5-ventricle repair was rather low compared with our research. In our center, the rate of 1.5-ventricular repair was approximately 16.9%, which was higher than that reported in the Netherlands, yet it was signi cantly lower compared with the 30% at the Mayo Clinic, USA [8] and the more than 50% at Shanghai Children's Medical Center, China [15] . The study from Shanghai Children's Medical Center also suggested that the BCPS procedure should be presented when a preoperatively severely dilated RV (aRV>50%) with the anterior lea et causes signi cant RV out ow tract obstruction, outstanding TV morphologic abnormity, and hemodynamic instability after separation from cardiopulmonary bypass, regardless of the presence of RV failure or severe cyanosis. According to Dr Dearani's latest study [1] , BCPS is generally feasible, as pulmonary hypertension is rare in EA, but should be used with caution because it may increase left ventricular (LV) end-diastolic pressure in the setting of moderate to severe LV dysfunction. Indications for BCPS in adults include both heart failure indications (RV end-diastolic volume > 250 mL/m, RV ejection fraction < 25%, D-shaped left ventricle, postrepair right atrial pressure to left atrial pressure ratio > 1.5:1, and postrepair cardiac output depression) and no indications of heart failure (complex TV repair tension reduction and postrepair TV stenosis with a mean gradient > 8-10 mmHg). BCPS contraindications related to pulmonary hypertension include a mean pulmonary artery pressure > 20 mmHg, pulmonary arteriolar resistance > 4 Woods units, and elevated left heart pressure (LV end diastolic pressure or left atrial pressure > 12 mmHg). In our opinion, BCPS should be actively prepared when the tricuspid valve ori ce is smaller than normal or the proportion of aRV is larger than that of fRV. In addition, ASD subtotal closure is a good method to unload RV. Although saturation was sacri ced to some extent, the long-term prognosis of the TV and RV was not affected according to our follow-up.

Study limitations
This study is limited by its retrospective nature. The comparison between cone procedure and Carpentier procedure could be biased because of the age of operations. More arguments of RV function should be provided to assess the patient's preoperative condition and postoperative recovery. Due to the low number of deaths and reoperations, reliable analysis to identify risk factors could not be performed.

Conclusion
The reoperation rate and mid-term mortality of surgical treatment for Ebstein's anomaly were both low, tricuspid regurgitation was signi cantly improved during mid-term follow up. Cone procedure had the best mid-term effect among anatomic repair.

Declarations
Ethics approval and consent to participate The study protocol was approved by the Institutional Review Board of Fuwai Hospital. The requirement to obtain informed consent was waived because of the retrospective nature of the study.

Consent for publication
Consent was obtained from the patients or their relatives.

Availability of data and materials
The datasets used or analyzed during the current study are available from the corresponding author on reasonable request.