Comparison of the Surgical Effect and the Long-term Clinical Outcomes Between Robotic and Sternotomy Mitral Valve Replacement

Robot-assisted mitral valve surgery has been increasingly used by surgeons to achieve better results. This study was to assess the safety and effectiveness of totally endoscopic robotic mitral valve replacement (TE-MVR) and to provide evidence that it is a reasonable surgical choice by analyzing the clinical experience, surgical ecacy, and follow-up outcomes of this procedure. Method Between 47 patients underwent TE-MVR. From 293 patients underwent conventional sternotomy mitral valve replacement (CS-MVR), of whom 47 patients were selected to match the TE-MVR group (1:1). We performed a retrospective study by collecting perioperative data and assessed TE-MVR ecacy by comparing clinical outcomes and echocardiography with CS-MVR in a 10-year follow-up period.


Introduction
With the continuous improvement in surgical quality and fewer complications, surgeons and patients are no longer satis ed with the traditional sternotomy cardiac surgery. Minimally invasive cardiac surgery techniques, including partial sternotomy, mini-anterolateral thoracotomy and thoracoscopic surgery, particularly with robot assistance, have dramatically improved the success rate of minimally invasive surgery. [1] From 2007 to 2015, totally over 900 cases of cardiac surgery have been completed at our center, including 157 cases of mitral valve (MV) surgery (110 cases MV repair [2] and 47 cases MV replacement) [3]. Because of the advantages including a well-expanded operative eld and operative dexterity, robotic MV repair has become widely accepted following continuous multicenter trials that have shown the same safety and effectiveness between robotic and traditional sternotomy surgery [4][5][6]. However, in patients with severe MV calci cation, especially with severe synechia after rheumatic heart disease, MV replacement (MVR) is the only surgical option [7], but comparative evidence of robotic MVR is still rare.
The purpose of our study is to report the patients selection and surgical safety, of robotic MVR. After comparison with sternotomy, the long-term outcomes of robotic MVR were evaluated. The results of this work may provide evidence for surgical decision-making. were matched as a conventional sternotomy MVR (CS-MVR) group. These criteria were used to match the participants to reduce discrepancies between two groups including age (difference within 10 years), sex, New York Heart Association (NYHA) classi cation, lesion type and extent, preoperative Euro-Score II score, left ventricular ejection fraction (LVEF), and type of prosthesis. We paired the participants strictly based on the degree of lesion type and preoperative heart function grade as the central pairing criteria and used the other factors as secondary features.

Robotic surgical techniques
The anesthesia and cannulation methods for robotic MVR was the same as we previously described [3].
In brief, after induction of general anesthesia, the patient underwent double-lumen endotracheal intubation.
After left lung ventilation, ports are created in the chest wall. After all ports were placed and cardiopulmonary bypass (CPB) was established through peripheral cannulation, the ascending aorta was cross-clamped with a Chitwood cross-clamp (Scanlon International, Minneapolis, Minn) from the fourth intercostal space at the midaxillary line. A trocar was used to enter the ascending aorta from the right chest wall, directly followed by perfusion of crystalloid cardioplegia.
After cardiac arrest, left atriotomy was performed with a parallel atrial septal incision. Mostly, the valve lea ets were stiff due to calci cation, and the paravalvular structures were sometimes affected by calci cation. In typical cases, calci ed valve tissue was xed with a grasper and removed with curved scissors. The anterior valve lea ets were removed, and the posterior valve tissue was retained. Perivalvular tissue with severe calci cation was also simultaneously excised. The attached papillary muscles and chordae tendineae were preserved and xed to the annulus with arti cial chordae.
A prosthetic valve generally needs 10 to 12-needle 2-0 double-armed Ti-Cron sutures (Covidien Inc., Dublin, Ireland) to be anchored to the native annulus with Te on pledgets ( Figure 1). First, the prosthetic valve was removed from the holder and positioned in the chest cavity via a 2.5-3cm working port. Then, it was xed vertically through a small incision. There are two methods for suturing. First, all the knots can be tied using a knotting device, and then the knot is sent through the incision. Prosthesis with a plasty ring is suitable for this method because it is effective. Additionally, the knot can be buried deeper into the soft plasty ring and tissue to prevent it from loosening easily. Alternatively, knotting can be accomplished using the COR-KNOT device (LSI SOLUTIONS, Victor, NY, USA). The advantage of this system is that it makes every knot very uniform, so each knot has the same tension and quality. Additionally, the replacement process for each prosthesis can be reduced by 10 to 15 min. The left atrial appendage was closed with 4-0 polytetra uoroethylene running suture (W.L. Gore and Assoc, Inc. Flagstaff, AZ, USA). After de-airing, the Chitwood cross-clamp was released. The pacing lead was routinely placed on the surface of the right ventricle. TEE was conducted to evaluate the function of the prosthetic valve after weaning from CPB. The operation was completed once no regurgitation or paravalvular leak con rming. A chest drainage tube was placed after meticulous hemostasis.
Post-operative treatments Antibiotics should be used in patients with infective endocarditis for a su cient amount of time. After the extubation patients received Vitamin K antagonists treatment. The international normalized ratio (INR) was adjusted according to the type of the arti cial valve. Mechanical valve needs long-term anticoagulation and timely detection of INR values while the biovalve need to be anticoagulated for six months. The predischarge thoracic echocardiography was performed to observe the function of the prosthetic valve.

Follow-up data
Follow-up of patients was conducted by regular outpatient review and telephone interview in a systematic manner. The follow-up outcomes assessed included postsurgical overall survival, stroke, reoperation, and readmission for worsening heart function and other major cardiovascular and cerebrovascular events, and postoperative TTE ndings. MACCEs include all medical events with cardiovascular and cerebral vascular origins that had caused death or required hospitalization and medical attention including stroke, thromboembolic events, deteriorating heart function, prosthetic dysfunction, infective endocarditis.

Statistical methods
Categorical data were statistically described as frequencies (%). The rates of the two samples were compared using the χ2 test, and ordered count data were compared using the rank-sum test. Continuous data were described as the mean ± standard deviation (SD), and continuous variables in the two groups were compared using Student's t-tests. The Kaplan-Meier plot and two-sample log-rank test were generated to reveal visual and statistical differences in late period complications between groups. CPB and ACC time were deal with a nonlinear regression analysis model (r2) to evaluate the effect of the learning curve. Euro-Score II was calculated with the special calculator online [8]. Signi cance was set at P<0.05. All data were statistically analyzed with SPSS 22.0 statistical software (IBM, Armonk, NY, USA).

Demographic information and valve status
There were 47 patients in both groups after 1:1 matching. No signi cant difference was found in patient demographics, MV pathology and co-morbidity of other diseases. Preoperative and echocardiography data of two groups are shown in Tables 1.

Operative data
Surgery was completed successfully and no conversion to sternotomy surgery was noted in both groups. The CPB and ACC times associated with robotic surgery were signi cantly longer than those associated with median sternotomy (P<0.001). The size distribution of the mechanical prosthesis or bioprosthesis is presented in Table2. Approximately 2.1% (1/47) and 29.8% (14/47) patients received concomitant cardiac procedures in two groups respectively, and left atrial appendage obliteration in CS-MVR were more likely to be performed (P=0.003). Intraoperative TTE con rmed satisfactory prosthesis function in all patients. The operative data are summarized in Table 2.

Postoperative data
No death, stroke, new onset of atrial brillation or severe infection occurred during hospitalization in TE-MVR. In CS-MVR, one patient (2.1%) underwent re-exploration for bleeding due to excessive chest drainage. One patient (2.1%) suffered multiple organ failure after septic shock and died on the sixth postoperative day. The other patients in CS-MVR recovered well. Differences between groups became indistinguishable in terms of postsurgical complications. Regarding the postoperative drainage volume, it was signi cantly higher in CS-MVR (P<0.001) and the difference in transfusion of red blood cells and plasma was statistically signi cant between two groups. The robotics group recovered well after surgery, and there was no prosthetic valve dysfunction in the predischarge TTE. The EF was slightly lower than before surgery in TE-MVR group, but this difference between two groups was not signi cant (Table 3).
Follow-up data Patients were followed up by outpatient and telephone interviews. Forty-two (89.4%) patients were followed up in TE-MVR group with an average duration of 71.68±23.62 months TE-MVR group(range from 28 to 110 months) . In CS-MVR group, 40 (87.0%) patients were followed up for 136.47±43.55 months (range from 69 to 218 months). During the long-term follow-up, 2.1% (1/47) of patient in TE-MVR had reoperation because of infective endocarditis three weeks after discharge and this patient died of acute pancreatitis after another two years. In CS-MVR group, 12.8% (6/47) of the patients died, 14.9% (7/47) of the patients had stroke, 4.3% of the patients had redo-MVR due to paravalvular leakage and infective endocarditis and 12.8% (6/47) of the patients were readmitted because of recurrent heart failure and 6.4% (3/47) with paravalvular leakage. Kaplan-Maier analysis showed equivalent 10-year rates of death, stroke, redo-operation and re-admission between TE-MVR and CS-MVR (all P>0.05). However, the cumulative major adverse cardiac and cerebrovascular events (MACCE) was 5.7fold more common in CS-MVR than in TE-MVR, although not reaching statistical signi cance (6.4% vs. 36.2%, odd ratio =5.7, P=0.10). (Figure 3).

Discussion
Carpentier completed the world's rst robot-assisted MV surgery in 1998 [9]. After two decades of development, robotic technology has rapidly evolved and aided in the remarkable advancements witnessed in mitral valve surgery with some apparent advantages which outperform thoracoscopic techniques [10,11]. From our experience with TE-MVR, robotic surgical system can provide a good view of the surgical eld. With 10-fold magni cation, subtle lesions can be accurately located and removed during the operation; accurate localization is vital in the removal and repair of calci ed or adherent valve annulus structures. Meanwhile, the glass-free three-dimensional visual effects can aid the surgeon in accurately distinguishing tissue structures and performing procedures, including excision and suturing.
Likewise, the LA approach can obviate the right atrial-atrial septal approach used in the traditional way. By minimizing the retraction of the atrial tissue, mitral valve structure is exposed more precisely. Nonetheless, robotic technology is more reliable intraoperatively given that the serpentine structure of the front part of the robotic arm can exibly perform seven-dimensional activities, greatly expanding its range of movement. Therefore, this technique makes it possible to accomplish complex actions, in addition to other unique features. The gradual elimination of micro tremors in the hands also leads to smoother and more accurate surgical procedures.
Compared with CS-MVR, the CPB and ACC times of TE-MVR were signi cantly longer than CS-MVR, which was reported previously [12,13] and the process of operative time shortening was similar to the learning curve reported in Kuo's study [14]. However, shorter CPB and ACC durations were seen in our study than in In our series of consecutive TE-MVR cases, no perioperative complications such as reoperation, postoperative new onset of atrial brillation, low cardiac output syndrome, cerebrovascular events, and postsurgical renal failure was observed. In CS-MVR group, only one patient experienced re-exploration for bleeding, and one patient suffered from renal failure. Notably, no new cases of atrial brillation were observed in TE-MVR group, while 4 cases were observed in the CS-MVR group (0% vs. 8.52%, P=0.117). Although no signi cant differences were noted herein, the results of previous studies have suggested that the incidence of AF is lower than that associated with traditional surgery [15] . Similar results were also shown by Suri et al. in patients undergoing MV repair [16] . Two main reasons can explain this nding. First, robotic MVR is performed from the interatrial sulcus into the left atrium. Because the mitral annulus is parallel to the sagittal plane of the human body, the manipulator can be conveniently directed through the right chest wall to the MV tissue. Atrial tissue retraction is avoided with an atrial retractor. A median thoracotomy often requires a right atrial and an atrial septal incision. During the process, the right atrium and atrial septum undergo forceful retraction, which injures atrial structures. Second, robotic surgery can be performed under 10 times magni cation in 3D to diminish the possibility of inappropriate operation of normal tissues, resulting in lower arrhythmia rates.
The tracheal intubation time, length of ICU stay, and postsurgical hospital stay were signi cantly shorter in TE-MVR than in CS-MVR, which is consistent with the ndings reported by Suri et al. [17]. Hawkins reported a shorter ICU stay and postoperative length of hospital stay and obtained similar conclusions when comparing robotic MVR with minimally invasive and conventional MVR [10]. Considering the obvious clinical effects, the main reason for these ndings is that robotic surgery obviates the need to disrupt the sternum and reduces complications associated with sternotomy. In addition to pain relief during postoperative recovery, robotic surgery is associated with considerable reduced blood loss and blood product transfusion, which can also minimize the possibility of transfusion-related acute lung injury, infection and allergy [18]. In principle, the use of blood products should be minimized to ensure clinical safety. Our center adopts strict indications for blood transfusion. In addition, the reason for the lower number of robotic blood transfusions is attributed to three phrases. First, sternum-sparing approach with punctures in the lateral chest wall prevents possible blood loss from the marrow cavity in the sternum. Second, the adoption of closed-type extracorporeal circulation technology decreases the need for intubation and the perturbation of great vessels and the heart so that perioperative bleeding is reduced. Third, the surgical eld in robotic surgery is clear, and the magni cation is su cient for convenient and comprehensive hemostasis. In addition, accumulated robotic surgical experience would further facilitated reduction of surgical blood loss, which was in agreement with results from Coyan and colleagues [19].
Besides, the negative impact of transfusions is also curial and detrimental to long-term survival. Although nonsigni cant difference was found between the two groups regarding the long-term survival and serious late complications such as thromboembolic events, redo MVR, and readmission for worsening heart function, TE-MVR has a tendency of lower rates of serious events. Nevertheless, patients are more satis ed with the surgical wounds and more likely to collaborate with physicians during postoperative rehabilitation, which could further speed recovery [17,20].
We have performed more than 150 robotic MV operations, and 30% of them are MVRs. The ratio of replacement is higher than that in centers in developed countries [10,21]. It is believed that the etiological difference between China and developed countries may explain the higher proportion of MVR in MV surgery. The living and nutrition condition of most Chinese residents was unfavorable three to four decades before, and rheumatic fever was common among children, which could lead to rheumatic calci cation of the heart valves. Mitral valve repair is quite tricky for severe rheumatic valvular pathology [22] and MVR is inevitable among these rheumatic patients. Moreover, the majority of patients opted for a mechanical valve rather than a bioprosthetic valve [14]. But surgeons and patients in developed countries are more willing to choose biological valves. We believe two main reasons may be contributing to this. First, with the decreasing technical di culty and more mature robotic surgical skills, the condition for redo robotic surgery turns relatively mature. Spiller and colleagues [23] even reported the third-time redo robotic MVR recently. Second, due to more favorable economic condition in developed regions, increasing patients incline to receive a biological valve rather than slightly cheaper mechanical prosthesis with longterm intaking anticoagulants and blood tests [24].
However, robotic MV surgery is still an evolving technology. Therefore, with consideration for its safety and effectiveness, we should be more rigorous in selecting patients for the best possible long-term outcomes. Patients requiring isolated MV surgery are generally preferred. However, with continual advancements in technology, these contraindications are gradually being eliminated. Onan reported robotic MVR in a patient with pectus excavatum [25], con rming the possibility of using robots in patients with thoracic deformity. Although infective endocarditis is not a contraindication for robotic surgery, valves with complex structural lesions can cause many di culties intraoperatively. Gillinov and colleagues [11] reported the outcomes of patients with infective endocarditis receiving robotic MV surgery and showed the results of the operation were satisfactory. In our TE-MVR group, 3 of 47 patients had infective endocarditis and recovered well after robotic MVR. A history of previous cardiac surgery or severe pulmonary hypertension was also a contraindication for robotic surgery, but Murphy [21] also reported patients who underwent a redo robotic MVR, as well as patients with preoperative COPD. In his report, no complication was noted, such as paravalvular leakage or obstruction of the left ventricular out ow tract postoperatively. The follow-up outcomes were acceptable, and the cardiac function was class 1 and 2. Recently, the range of patient selection has been gradually expanded, re ecting these improvements.

Limitations
This retrospective study has some limitations, including a relatively small population from a single center. It is largely since TE-MVR is more expensive, with an average increase of $5,000 per case (unpublished data). Thus, during the same period, most patients in China selected the CS-MVR rather than TE-MVR.
Considering that this is a nonrandomized control study, surgeon bias is still a challenge. All TE-MVRs were performed by a single surgeon, while CS-MVRs were performed by several surgeons. Although all patients were paired, we believe that the patients in TE-MVR were milder, highly selected cases given that a large LA diameter was noted in CS-MVR.

Conclusion
Robotic MVR is safe and reliable, and it is associated with less trauma, bleeding, and blood product transfusion than median sternotomy. In addition, the long-term outcomes are excellent and bene cial to selected patients; thus, robotic MVR can be considered an alternative option of minimally invasive surgery.