Total Thoracoscopic versus Robotic Surgery for Repair of Atrial Septum Defect: A Propensity Matching Score Analysis.

Robotic surgery can provide less surgical trauma than conventional surgery, but differences between robotic and thoracoscopic surgery for atrial septal defect (ASD) repair are not well documented. To explore whether ASD can be repaired by thoracoscopic surgery or robotic surgery, which procedure is less invasive, and the difference in outcomes between these two procedures, this article studies 160 patients undergoing ASD repair at our institution. Sixty-five patients underwent total thoracoscopic surgery and 95 patients underwent total endoscopic robotic surgery. Propensity score matching yielded 64 well-matched patient pairs. Surgical data and early postoperative outcomes between the two matched groups were analyzed and compared. The results show that thoracoscopic and robotic surgery to repair ASD are both safe and reliable, and the early curative effect is good. However, regardless of similar complication rates, robotic surgery has a shorter time, less postoperative drainage, and faster recovery than thoracoscopic surgery.

1. Introduction

Atrial septal defect (ASD) is the most common congenital heart disease (CHD) of adulthood [1], accounting for about 30–40% of adult CHD [2, 3]. The 2008 ACC/AHA and 2010 ESC guidelines for the management of adult CHD recommended that atrial septal defect repair should be performed either surgically or percutaneously if right atrial and ventricular enlargement occur, regardless of symptoms [4, 5]. Compared to conventional sternotomy, minimally invasive cardiac surgeries for closure of ASD have been proven to be safe and effective with less surgical trauma and faster recovery [3, 6, 7]. The last five years have seen a dramatic growth in the adoption of thoracoscopic and robotic technology, with the publication of several large series showing satisfactory early outcomes [8-10]. Some studies have reported the similar early outcomes when comparing thoracoscopic surgeries with open surgeries [7, 11]. However, research comparing total thoracoscopic ASD closure with totally endoscopic robotic surgery is relatively scarce. Therefore, in this study, we sought to compare the outcomes of ASD patients treated with total thoracoscopic surgeries and totally endoscopic robotic surgeries.

Conventional ASD repair via median sternotomy has been recognized as the standard approach with low morbidity and mortality since the 1950s [11]. However, conventional surgeries have the shortcomings of large incision, unaesthetic scar, and damaging the integrity of the sternum [12]. With the advancement of minimally invasive technology, total thoracoscopic ASD repair is being performed increasingly over the past decade, owing to its advantages in minimizing surgical trauma [13]. Ma reported 96 cases of total thoracoscopic ASD repair without robotic assistance and concluded that the procedure was associated with faster recovery and superior quality of life compared to conventional surgeries [14]. Robotic surgery system was first used in cardiac surgery in the late 1990s [8]. Since then, the safety and efficacy of robotic surgeries have also been demonstrated by many studies [6], with a mean 30-day mortality of 0.7% (0%–0.8%) reported by Doulamis et al. in 2019 [15]. Robotic surgeries have gradually become another effective alternative to conventional sternotomy. Thus far, only a few reports comparing robotic surgery to thoracoscopy for mitral disease have been published [16, 17]. However, for ASD repair, the comparison between the two techniques has rarely been reported. In this study, to minimize the selection bias, we closely matched patients' baseline characteristics between the two groups using the propensity-score matching method. To the best of our knowledge, this study is the first of its kind reported in the literature.

2. Materials and Methods

2.1. Patient Selection

A retrospective analysis was performed for 160 patients undergoing closure of secundum-type ASD through minimally invasive approach from November 2015 to January 2022. The average age of patients was 38.62 ± 15.58 years (range: 7–73 years). Of the 160 patients, 105 were female and 55 were male. Patients' baseline characteristics were listed in Table 1. The preoperative diagnosis, including size, anatomic type of the ASD, and some other associated anomalies were confirmed by transthoracic echocardiography (TTE). The selection criteria for both robotic and thoracoscopic surgeries were listed as follows: (1) confirmed secundum ASD with left-to-right shunt in the atrial level, age >5 years, body weight ≥15 kg; (2) without severe valvular, aortic, or congenital heart diseases that require concomitant repair; (3) without severe coronary artery disease requiring concomitant CABG; (4) without conditions unsuitable for endoscopic visualization (history of right thoracotomy, severe pleural adhesion, severe pericarditis, deformities of the thorax, and morbid obesity); (5) without severe asthma, emphysema, or COPD; (6) without severe peripheral vascular disease which may restricts femoral cannulation; and (7) without significant hepatic compromise, dialysis dependent renal failure, untreated cerebrovascular disease, and severe bleeding disorder.

Table 1

Patient baseline characteristics before propensity matching.

Characteristics	Thoracoscopic group	Robotic group	P value
	(N = 65)	(N = 95)
Male (n (%))	23(35.4)	3233.7	0.824
Age (y, mean ± SD)	39.69 ± 14.82	37.88 ± 16.12	0.473
Weight (kg, mean ± SD)	59.33 ± 10.77	58.06 ± 10.79	0.464
Diabetes (n (%))	0(0.0)	8(8.4)	0.042
Hypertension (n (%))	5(7.7)	5(5.3)	0.771
Current smoke (n (%))	2(3.1)	5(5.3)	0.787
Congestive heart failure (n (%))	1(1.5)	6(6.3)	0.290
Atrial fibrillation (n (%))	4(6.2)	5(5.3)	1.000
Systemic embolism (n (%))	0(0.0)	0(0.0)	NA
Defect diameter (cm, mean ± SD)	2.88 ± 0.77	2.67 ± 0.89	0.128
Defect type (n (%))			0.945
Fossa ovalis type	38(58.5)	54(56.8)
SVC type	3(4.6)	5(5.3)
IVC type	21(32.3)	29(30.5)
Mixed type	3(4.6)	7(7.4)
LVEF (%, mean ± SD)	62.45 ± 4.66	61.74 ± 4.36	0.328
PASP (mmHg, mean ± SD)	51.02 ± 10.98	52.41 ± 11.82	0.452
Creatinine (μmol/L, mean ± SD)	63.69 ± 15.37	62.86 ± 11.90	0.714
With PAPVC (n (%))	2(3.1)	7(7.4)	0.419

SVC type: superior vena cava type; IVC type: inferior vena cava type; LVEF: left ventricular ejection fraction; PASP: pulmonary artery systolic pressure; and PAPVC: partial anomalous pulmonary venous connection.

2.2. Anesthesia and Position

After successful induction of general anesthesia, all the patients received double lumen tube intubation for left single-lung ventilation. A transesophageal echocardiography (TEE) probe was routinely placed to confirm the diagnosis of ASD and examine the repair quality intraoperatively. Patients were maintained at supine position with right hemithorax elevated to 20°–30°.

2.3. Robotic Surgery

After systemic heparinization, the right internal jugular vein was cannulated percutaneously (14–16F, Kangxin Medical, China) for superior vena cava drainage under the guidance of TEE. The right femoral artery was surgically cannulated for arterial return (14–20F, Kangxin Medical, China) and inferior vena cava drainage was from the femoral vein (16–24F, Kangxin Medical, China). After the initiation of left single-lung ventilation, a 3–4 cm thoracotomy was made through the fourth intercostal space on the anterior axillary line to place a 0‐degree endoscope. On the right anterior axillary line, two 8-mm endoscopic trocars were placed through the third and sixth intercostal spaces for the left and right arms of the robot, respectively. The atrial retractor arm was placed through the fifth intercostal space on the midclavicular line.

The pericardium was incised about 2 cm anteriorly to the phrenic nerve. After the initiation of cardiopulmonary bypass (CPB), superior and inferior vena cava were blocked. Ascending aorta clamping was not performed and the operation was performed on the beating heart. Depending on the size of the defect, ASD was closed by primary suturing or using a bovine pericardial patch. At the end of closing ASD, left atrial deairing was completed by holding positive pressure of the lungs to expel blood from the left atrium. The right atrium was closed using two layers of running 4–0 polypropylene suture. After the patient weaned from CPB, the effect of repairing was confirmed by TEE. A chest tube was inserted through the right arm port site following decannulation of the femoral vessels and meticulous hemostasis.

2.4. Thoracoscopic Surgery

The location of the three ports: port1 (1.5 cm) was made parasternally in the third intercostal space for entries of left-hand surgical instruments; port 2 (1.5 cm) was made in the fifth intercostal space between the midclavicular line and anterior axillary line for entries of right-hand surgical instruments; A thoracoscope was inserted through a small incision (port 3, 2 cm) in the fourth intercostal space on the right anterior axillary line. The establishment of the peripheral CPB and intrathoracic part of operation was the same as those of the robotic group.

2.5. Study Endpoints

The primary endpoint of this study was in-hospital mortality. The secondary endpoints included operation time, CPB time, ICU stay time, mechanical ventilation time, postoperative hospital stay time, postoperative drainage, total RBC usage, in-hospital complications, and 6-month MACCE. MACCE was defined as major cardiovascular and cerebrovascular events during 6-month follow-up since the day of operation, including events of death, new-onset arrhythmia, myocardial infarction, stroke, and peripheral vascular embolism.

2.6. Statistical Analysis

Quantitative variables were expressed as mean ± standard deviation if normally distributed; otherwise, as the median and 25–75 percentile. Categorical variables were expressed as frequencies and percentages. To compare the two groups before matching, we use the independent-sample T test for continuous data with normal distribution, nonparametric test (the Mann–Whitney U test) for continuous data with skewed distribution, and χ2 analysis or Fisher's exact tests for categorical variables. After matching, continuous variables with normal distribution were compared by paired T test and continuous variables with skewed distribution by the Wilcoxon rank-sum test. Categorical variables were compared by McNemar's test between two matched groups. The survival data were analyzed using the Kaplan–Meier method and compared by the log-rank test. The matching criteria included eleven fixed-effect variables (gender, age, weight, defect size, diabetes, hypertension, history of atrial fibrillation, history of congestive heart failure, left ventricular ejection fraction, pulmonary artery systolic pressure, and blood creatinine). Patients in the thoracoscopic group were then matched in a 1 : 1 nearest neighbour fashion to patients in the robotic group with similar propensity scores; a propensity score difference of less than 0.1 was required for each match. All statistical tests above were two sided, with significance set at P < 0.05. Statistical analysis was performed using SPSS Statistics, version 25.0 (IBM, Chicago, Illinois).

3. Results

Baseline characteristics of the two groups before matching were summarized in Table 1. Table 1 demonstrates the baseline characteristics of the two groups were not entirely balanced before PSM. Patients who were in the robotic group were more likely to have a higher rate of diabetes (8.4% vs. 0.0%, P < 0.05). During the process of PSM, 64 patients in the thoracoscopic group were successfully matched with 64 patients of the robotic group in a 1 : 1 fashion. After propensity matching, all the baseline characteristics were balanced between the two groups. Baseline characteristics after matching are summarized in Table 2.

Table 2

Patient baseline characteristics after propensity matching.

Characteristics	Thoracoscopic group	Robotic group	P value
	(N = 64)	(N = 64)
Male (n (%))	23(35.9)	21(32.8)	0.710
Age (y, mean ± SD)	39.64 ± 14.93	36.86 ± 15.91	0.310
Weight (kg, mean ± SD)	59.40 ± 10.84	57.91 ± 9.42	0.407
Diabetes (n (%))	0 (0.0)	0 (0.0)	NA
Hypertension (n (%))	5 (7.8)	3 (4.0)	0.715
Current smoke (n (%))	2 (3.1)	4 (6.3)	0.676
Congestive heart failure (n (%))	1 (1.6)	2 (3.1)	1.000
Atrial fibrillation (n (%))	4 (6.3)	2 (3.1)	0.676
Systemic embolism (n (%))	0 (0.0)	0 (0.0)	NA
Defect diameter (cm, mean ± SD)	2.87 ± 0.77	2.65 ± 0.80	0.111
Defect type (n (%))			0.780
Fossa ovalis type	38 (59.4)	39 (60.9)
SVC type	3 (4.7)	4 (6.3)
IVC type	20 (31.3)	16 (25.0)
Mixed type	3 (4.7)	5 (7.8)
LVEF (%, mean ± SD)	62.17 ± 4.13	62.06 ± 4.55	0.889
PASP (mmHg,mean ± SD)	50.95 ± 11.05	50.81 ± 10.82	0.942
Creatinine (μmol/L, mean ± SD)	63.92 ± 15.38	62.83 ± 11.32	0.650
With PAPVC (n (%))	2 (3.1)	5 (7.8)	0.437

SVC type: superior vena cava type; IVC type: inferior vena cava type; LVEF: left ventricular ejection fraction; PASP: pulmonary artery systolic pressure; and PAPVC: partial anomalous pulmonary venous connection.

Perioperative data for two matched groups were presented in Table 3. After propensity matching, the robotic surgery had shorter operation time (P < 0.001), shorter CPB time (P < 0.001), shorter length of ICU stay (P=0.001), shorter mechanical ventilation time (P=0.02), and shorter postoperative hospital stay time (P=0.025). In addition, the robotic group had statistically significant less thoracic drainage (P=0.001) and less total RBC usage (P < 0.001). The total hospital costs showed no significant difference between the groups (P=0.097).

Table 3

Perioperative details (propensity-matched groups).

	Thoracoscopic group	Robotic group	P value
	(N = 64)	(N = 64)
Operation time (min, median (IQR))	300.00 (250.00, 373.75)	210.00(180.00, 240.00)	<0.001
CPB time (min, median (IQR))	123.50 (95.25, 164.50)	74.50 (64.25, 92.00)	<0.001
Surgical closure technique			0.244
Pericardial patch	61 (95.3)	64 (100.0)
Primary suturing	3 (4.7)	0 (0.0)
Concomitant PAPVC repair	2 (3.1)	5 (7.8)	0.453
ICU stay (h, median (IQR))	22.00 (20.00, 46.00)	19.00 (18.00, 24.00)	0.001
Mechanical ventilation time (h, median (IQR))	6.75 (4.50, 15.00)	4.00 (3.00, 6.75)	0.02
Postoperative hospital-stay (d, median (IQR))	9.00 (7.00, 13.00)	8.00 (7.00, 9.00)	0.025
Thoracic drainagea (ml, median (IQR))	277.50 (151.25, 460.75)	207.50 (101.25, 300.00)	0.001
Total RBC usage (U, median (IQR))	3.00 (0.00, 4.00)	0.00 (0.00, 0.00)	<0.001
Total hospital costs	1.33 (1.13, 1.76)	1.49 (1.34, 1.69)	0.097
(×104$, median (IQR))

a:the volume of thoracic drainage in the first 24 hours after operation; CPB: cardiopulmonary bypass; ICU: intensive care unit; and PAPVC: partial anomalous pulmonary venous connection.

The results of in-hospital complications showed that delayed mechanical ventilation (DMV), new-onset arrhythmia, and systemic embolism were the most common postoperative complications in this study. 4 cases with early new-onset arrhythmia (<14 days) were observed in the thoracoscopic group, all of which were atrial fibrillation; whereas in the robotic group, there was 2 patients with early new-onset arrhythmia (<14 days), including 1 case of atrial fibrillation and 1 case of supraventricular tachycardia. Early systemic embolism (<14 days) includes stroke and peripheral embolism. Before discharge, two patients (3.1%) of the thoracoscopic group and one patient (1.6%) of the robotic group presented with a stroke, while two patients (3.1%) of the thoracoscopic group had peripheral arterial embolism. In addition, there was no significant difference between the two groups on in-hospital mortality, new-onset arrhythmia (<14 days), systemic embolism (<14 days), delayed mechanical ventilation, renal failure, residual shunt, reoperation for bleeding, pneumonia, and pneumothorax (Table 4).

Table 4

In-hospital complications (propensity-matched groups).

	Thoracoscopic group	Robotic group	P value
	(N = 64)	(N = 64)
In-hospital mortality (n (%))	0 (0.0)	0 (0.0)	NA
New-onset arrhythmia (n (%))	4 (6.3)	2 (3.1)	0.687
Atrial fibrillation	4 (6.3)	1 (1.6)
Supraventricular tachycardia	0 (0.0)	1 (1.6)
Systemic embolism (n (%))	4 (6.3)	1 (1.6)	0.375
Stroke	2 (3.1)	1 (1.6)
Peripheral embolism	2 (3.1)	0 (0.0)
Reoperation for bleeding (n (%))	1 (1.6)	0 (0.0)	NA
DMV (n (%))	7 (10.9)	3 (4.7)	0.344
Residual shunt	1 (1.6)	1 (1.6)	1.000
Intraoperative (TEE)	0 (0.0)	0 (0.0)	NA
Before discharge (TTE)	1 (1.6)	1 (1.6)	1.000
Pneumonia (n (%))	3 (4.7)	1 (1.6)	0.625
Renal failure (n (%))	0 (0.0)	1 (1.6)	NA
Pneumothorax (n (%))	1 (1.6)	1 (1.6)	1.000

DMV: delayed mechanical ventilation (>24 h).

During the 6-month follow-up after operation, 24 patients were lost, and the follow-up rate was 81.25%. 1 patient of the thoracoscopic group and 1 patient of the robotic group presented with new-onset atrial fibrillation 5 months and 4 months after operation, respectively. In addition, 1 patient of the thoracoscopic group presented with a stroke 2 months after operation, and he was rehospitalized and cured. There was no significant difference in the incidence of 6-month MACCE between the two groups (log-rank χ2 = 2.822, P=0.093, Figure 1). There was no in-hospital death or follow-up death in both the matched groups, and the survival rate within 6 months were both 100%.

Figure 1

The Kaplan–Meier survival curve. Over 6 months of follow-up, freedom from MACCEs for matched thoracoscopic group and robotic group were 83.6% and 93.4%, respectively (P=0.093).

4. Discussion

In this study, we found the operation time and CPB time of the robotic group were both significantly shorter than the thoracoscopic group. To analyze the reason, we believe it is just because that robotic devices have many advantages over thoracoscopy, such as wrist-like articulated instruments that can move at six degrees of freedom and lack of tremors. This helps improve the speed and accuracy of suturing, thus shortening the operation and CPB time. On the contrary, the long-shafted instrument of thoracoscopy, to some extent, limited the surgeon's mobility and caused fatigue easily. Besides, robotic surgeries allowed maximum visualization of intracardiac structures and bleeding point by three-dimension (3D) high-definition imaging [6], which facilitate the process of surgical hemostasis intraoperatively. Accordingly, the 24-hour thoracic drainage volume of the robotic group decreased significantly in this study, reflecting less tissue damage and better hemostatic effect of robotic surgery. This is slightly different from the results reported by Wei et al. in 2020 [17]. In Wei's study, 24-hour drainage was higher in the robotic MVP group when compared to the thoracoscopic MVP group despite no statistical difference (P > 0.05).

There is a paucity of data comparing postoperative outcomes between robotic and thoracoscopic surgery for ASD repair in previous studies. In this study, we observed that the length of ICU-stay, postoperative mechanical ventilation time, and the length of postoperative hospital decreased significantly in the robotic group. This result suggests that patients in the robotic group recovered faster than their counterparts in the thoracoscopic group. In addition, our study also indicated robotic surgeries significantly reduced the usage of RBC. In this study, there was no significant difference between the two groups in the total hospital cost. However, cost‐benefit analysis of the robotic surgeries was beyond the scope of this study. Thus, the factors such as the institutional cost of the robot, the inflation rate, and per capita income increase were not taken into account. Further studies are still necessary to weigh the benefit of patients and the real economic cost.

In this study, none of the patients who underwent robotic surgery had mortality, as done in thoracoscopic surgery. This indicates that both of the two techniques are safe approaches for ASD repair. Balkhy et al. reported a single-center experience of 1103 patients who underwent robotic surgeries over 7 years and found that atrial fibrillation was the most common early postoperative complications, with the incidence of 12% [8]. In robotic surgeries, new-onset atrial fibrillation is thought to result from the atriotomy incision [1]. However, the incidence of new-onset atrial fibrillation in this study (1.6% for the robotic group, 6.3% for thoracoscopic group) was lower than that reported by Balkhy et al. Embolism events can be seen after both robotic and thoracoscopic surgeries in this study. Cerny et al. conducted a multicenter registry of 2563 robotic cardiac surgeries and reported a perioperative stroke rate of 0.2% in 2021 [18]. However, in this study, of the 4 patients (3 before discharge and 1 in follow-up) who experienced stroke, 2(50.0%) patients had underlying atrial fibrillation. In addition, in terms of early postoperative embolism events (<14 days), no statistically significant difference was found between the two groups. Besides, there was also no significant difference in the incidence of 6-months MACCEs after operation. However, longer follow-up time and greater sample sizes are warranted for validation of these findings.

5. Limitations

This study has several limitations. First, our study is a single‐centered, nonrandomized study. The lack of a prospective and randomized nature is a limitation. Observed indicators can be influenced by some subjective factors. Limited sample size may reduce the power to detect significant difference. Furthermore, we only compared the in-hospital and early follow-up outcomes. However, midterm and long-term follow-up results need to be further investigated. In addition, the hospital cost in this study cannot be expected to be reproduced in other centers due to the factors such as the inflation rate and different economic level.

6. Conclusion

This study indicates the early outcomes in totally endoscopic robotic surgery for ASD repair are similar to the total thoracoscopic surgery. Both these two procedures are proven to be safe and reliable for ASD closure. However, robotic surgery can provide shorter operation time, less postoperative drainage, and quicker recovery than thoracoscopic surgery.