All patients that had undergone MV surgery at Massachusetts General Hospital between January 2013 and December 2020 were identified via interrogation of the Society of Thoracic Surgeons database. In addition, we included all patients that had received robotically assisted MVr at our institution in 2021 and 2022 to compare the initial outcomes of that innovative approach to previous outcomes. All perioperative notes, surgical reports, and imaging findings were reviewed. The planned operative strategy (repair vs. replacement, minimally invasive vs. conventional approach) was obtained from the perioperative notes, and intraoperative complications or changes in strategy were registered as reported by the individual surgeon. Standard preoperative transthoracic and intraoperative transesophageal echocardiography reports were utilized to evaluate MV pathology and procedural success. Successful repair was defined as MVr without requiring intraoperative conversion to MV replacement. The study was approved, and individual informed consent waived by the Institutional Review Board of Massachusetts General Hospital. Written informed consent was obtained from the individuals for the publication of any potentially identifiable images or data included in this article.

Patients referred for MVr undergo routine preoperative transthoracic echocardiography and either coronary angiography or ECG-gated CT angiography for coronary assessment. In addition, CT angiography of the entire aorta and iliac vessels is added if a patient is considered for a minimally invasive approach. Patients are then scheduled to have a full sternotomy approach if additional valve surgery or coronary artery bypass grafting (CABG) is indicated. The most common exclusion criteria for minimally invasive MVr include dilation of the ascending aorta > 45 mm, significant mitral annular calcification, presence of breast implants, significant pectus excavatum, left diaphragm paralysis, and previous right thoracotomy or sternotomy. Patients with low left ventricular ejection fraction (LVEF) or a predicted risk of mortality > 4% (moderate surgical risk category) by the Society of Thoracic Surgeons (STS PROM) risk score are also preferably operated on via a full sternotomy, to keep aortic cross-clamp times as short as possible. Furthermore, patients with significant calcifications or soft plaques in the descending aorta are not eligible for minimally invasive MVr at our institution, because of the need for retrograde aortic perfusion. Among our staff, 2 individuals perform minimally invasive MVr, while the other surgeons continue to use the sternotomy approach in all MVP patients referred to their practice.

To directly compare outcomes of conventional and minimally invasive MVr, patients undergoing isolated MVr were identified from the larger patient cohort excluding those that had major concomitant procedures prohibiting minimally invasive access at our institution i.e., CABG, surgery on aortic, tricuspid, or pulmonary valve, surgery on the ascending aorta or aortic arch, those with acute endocarditis, and those with severe mitral annular calcification requiring debridement and annular patch repair (Figure 1). Non-exclusionary concomitant procedures included closure of an atrial septal defect or a persistent foramen ovale (PFO), the Cox-Maze procedure, and occlusion of the left atrial appendage.

The extensive expertise gained throughout the years of successfully performing minimally invasive MVr was the essential foundation to launch a robotic MVr program in 2021, with the goal of obtaining similar excellent outcomes without compromising patient safety and facilitating certain steps of the procedure. Due to less restricted mobility and orientation of the surgical instruments during robotic surgery, exposure of and access to the papillary muscles for neochord implantation is superior to standard minimally-invasive approaches with long-shafted one-directional instruments. Despite initially expected longer procedural times, we believe that by introducing the robotic approach, we can further improve patient care and will eventually achieve equal or even shorter procedural times as with the established techniques.

At our institution, the most experienced minimally-invasive MV surgeon and team consisting of a co-surgeon, a scrub technician, a perfusionist, and an anesthesiologist, underwent comprehensive training via a society-supported formal training program including simulator and cadaver training. In-person proctoring was done for the first 5 cases, and remote proctoring for the sixth, seventh, and eighth cases. This stepwise approach (Figure 2) led to a successful launch of our robotic MVr program (9). For the initial cases included in this analysis, we added selection criteria to the ones mentioned above for minimally invasive surgery: Because we suspected initially prolonged procedural times, we included patients with very low perioperative risk (STS PROM <1%) and focused on pathologies deemed uncomplicated to repair based on preoperative echocardiography. Robotic MVr is currently available on a weekly basis at our institution and offered to all patients eligible for minimally invasive MVr, as described above. All patients that had undergone robotic MVr at our institution until August of 2022 were included in this analysis.

After assessment for normal distribution, continuous variables were expressed as median with interquartile range [25th—75th percentile], and categorical variables as numbers with percentages. Unmatched group comparisons were conducted using the Wilcoxon rank sum test or Fisher's exact test, as appropriate. For the subgroup analysis, propensity scores were calculated using logistic regression. A 1:1 propensity score analysis was conducted using the “nearest neighbor” algorithm, with a caliper setting of 0.1 standard deviations and without replacement. Before matching, multiple imputations was used to compensate for missing data. Standardized mean differences were used to evaluate the balancing of covariates after matching. Variables used to calculate the propensity score were age, sex, Barlow's valve, arterial hypertension, diabetes, peripheral arterial disease, previous stroke, chronic obstructive lung disease, preoperative LVEF, body mass index, and STS PROM.

After matching, continuous variables were reported as median with interquartile range and compared using Wilcoxon signed rank test; categorical variables were reported as numbers with percentages and compared using McNemar's test. Long-term freedom from re-operation and death was examined using Kaplan-Meier method with log-rank test. Data analysis was conducted using R software (R-Studio, version 3.4.1, Boston, MA, United States).

A total of 1407 patients underwent MV surgery via full sternotomy or right anterolateral mini-thoracotomy at our institution between January 2013 and December 2020. Of them, 830 patients were operated for mitral valve prolapse (MVP), of which 38 received a planned MV replacement (common reasons were too complex valve anatomy, severe mitral annular calcifications, or patient preference) and 31 were re-operations. Surgical repair of native MVP was thus performed in 761 patients, and those were included for further analysis (Figure 1). Table 1 shows the preoperative characteristics of all 761 patients. In summary, the median age was 64 [55–72] years, and 245 patients (32.2%) were female. Mitral regurgitation was severe in 720 (94.9%), moderate in 37 patients (4.9%), and less than moderate in 4 patients (0.2%) (MVP was not the primary indication for surgery in those latter two groups). Significant mitral annular calcification was present in 38 patients (5%). Moderate or more aortic stenosis was found in 12 patients (1.6%), moderate or more aortic regurgitation in 19 patients (2.8%), and moderate or more tricuspid regurgitation in 112 patients (15%). A concomitant diagnosis of coronary artery disease existed in 102 patients (13.4%), and of atrial fibrillation in 177 patients (23.3%). Twelve patients (1.6%) had a dilated ascending aorta. Table 2 provides an overview of mitral valve repair details in the unmatched sternotomy and mini-thoracotomy groups.

As expected, patients of the sternotomy group underwent significantly more concomitant procedures, as depicted in Supplementary Table S1. Briefly, in the sternotomy group, 98 patients (19.7%) underwent concomitant CABG, 5 patients (1%) aortic valve repair or replacement, and 8 patients (1.6%) ascending aortic replacement with or without root replacement. Regarding concomitant procedures that are feasible via sternotomy as well as via a mini-thoracotomy approach, LAA occlusion was performed more frequently in the sternotomy group (228 patients (45.8%) vs. 10 patients (3.8%) in the mini-thoracotomy group, p < 0.001), as well as the bi-atrial Cox-Maze procedure (85 patients (17.1%) vs. 2 patients (0.8%), p < 0.001). The left Cox-Maze procedure was performed in similar proportions in both groups (37 patients (7.4%) in the sternotomy group, 15 patients (5.7%) in the mini-thoracotomy group, p = 0.5). A PFO was closed in 66 patients of the sternotomy group (13.3%) and 41 patients of the mini-thoracotomy group (15.6%; p = 0.4).

We observed a continuous growth in the overall institutional volume of surgery for MVP with increasing proportions of minimally invasive operations since 2014, when the mini-thoracotomy approach was first introduced at our institution (Figure 3). Overall native MVP case volume was n = 69 in 2013 and n = 127 in 2020. With the start of our dedicated minimally invasive MVr program, institutional repair success rates improved markedly over the years included in this analysis, with 95.4% in 2013 (before the introduction of minimally invasive MVr) vs. 99.2% in 2020 (Figure 4). Residual MR was mild or more in 9.2% of patients in 2013 vs. 2% in 2020, with a continuous reduction over the years (Figure 5).

Since the start of our minimally invasive MVr program, there has been a continuous growth in the proportion of surgically more complex patients [i.e., anterior mitral leaflet (AML) or bileaflet prolapse], with proportions similar to before 2014 (Figure 6). In addition, a steady increase in the utilization of chordal reconstruction with GoreTex neochords and a combination of limited leaflet resection and neochord insertion could be observed. At the same time, a purely resectional technique was utilized less commonly in recent years (Figure 7).

The repair success rate with the minimally invasive approach was 100%, and only 3/263 patients (1.1%) required intraoperative conversion to full sternotomy. Reasons for conversion were bleeding from the aortic root in 2 and an intraoperative localized aortic dissection at the insertion point of the cardioplegia cannula in 1 patient. All 3 patients were discharged home with no significant increase in their hospital stay or further complications.

For the subgroup analysis of isolated MVr via sternotomy vs. mini-thoracotomy, we identified 599 patients that received MVr without major concomitant procedures (Figure 1). Of those, 346 were operated via sternotomy and 261 via right anterolateral mini-thoracotomy. Propensity matching resulted in 214 matched pairs with comparable baseline characteristics (Table 3 and Figure 8). Outcomes of the unmatched groups are summarized in Supplementary Table S2. Between the matched groups, there were significant differences in repair techniques: The neochords + annuloplasty technique was used more frequently in the mini-thoracotomy group (81.8% vs. 38.3% in the sternotomy group, OR 2.1, 95%-CI 1.6–2.8, p < 0.001) while the resection + annuloplasty technique was used less often (4.7% vs. 43.5% in the sternotomy group, OR 0.1, 95%-CI 0.05–0.2, p < 0.001). Occlusion of the left atrial appendage was also done less frequently in the mini-thoracotomy group (3.3% vs. 40.7% in the sternotomy group, OR 0.09, 95%-CI 0.04–0.2, p < 0.001) as well as the bi-atrial Cox-Maze procedure (0.9% vs. 11.7% in the sternotomy group, OR 0.08, 95%-CI 0.009–0.3, p = 0.01). There were no significant differences in the rates of PFO closure or the left Cox-Maze procedure. One patient in the sternotomy group required emergency CABG to the circumflex artery, and 1 patient each of the sternotomy and minithoracotomy groups required emergency replacement of the ascending aorta because of an intraoperative localized aortic dissection. Patients operated via minithoracotomy had significantly increased cardiopulmonary bypass times of 161 min (IQR 147–185) vs. 116 (98–145, p < 0.001); and increased aortic cross-clamp times of 94 min (84–113.8) vs. 88 (70–109, p = 0.001). Information on operative details between the matched groups is displayed in Table 4. MVr was successful in 210 patients (98.1%) of the sternotomy and 214 patients (100%) of the mini-thoracotomy group (p = 0.9), with no significant differences in residual MR and mean mitral valve pressure gradient (see Table 5).

Patients operated via mini-thoracotomy had a slightly higher postoperative LVEF than those operated via sternotomy (62% [57–66] vs. 62% [55–66], p = 0.03), required significantly shorter time on mechanical ventilation (4.4 h [2.4–6.3] vs. 4.8 h [3.3–7.5], p = 0.002) and were hospitalized significantly shorter (5 days [4–5] vs. 6 days [5–7], p < 0.001). There were no significant differences between the matched groups regarding postoperative complications. Operative mortality was 0% in both groups, while 30-day mortality was 0.9% in the sternotomy and 0% in the minithoracotomy group (p = 0.5). Kaplan-Meier method with log-rank test revealed no difference regarding long-term freedom from re-operation and death (p = 0.19, Supplementary Figure S1). Postoperative outcomes of the matched groups are summarized in Table 6.

We included all patients (n = 16) that underwent robotically assisted MVr at our institution. Repair success rate was 100%, with one patient requiring a second CPB run for residual mild-moderate MR after the initial repair attempt. Repair techniques included resection + annuloplasty, neochord implantation + annuloplasty as well as the combined technique. Three patients had no residual MR and 13 had trace residual MR after robotic MVr. Concomitant PFO closure was done in 4 patients (25%). Mean CPB and aortic cross-clamp times were 239 ± 59 min and 153 ± 48 min respectively, which is substantially longer than in the matched minimally invasive and full sternotomy groups of isolated MVr. Mean hospitalization time was 5.4 ± 1.8 days and there were no relevant postoperative complications during the hospitalization. Preoperative characteristics and results of the robotic cohort are displayed in Table 7.

Since the first reported case of minimally invasive mitral surgery in 1996 (10), minimally invasive MVr has undergone a significant and continuous expansion (11), supported by large series demonstrating the safety and benefit of this type of procedure (3–6). However, full sternotomy remains the most used approach for MVr. In this retrospective analysis, we observed that setting up a program dedicated to the incorporation of mini-thoracotomy for MVr has made it possible to change the management standard in an efficient and controlled manner without significant increases in complication rates. As shown by many groups, the benefits of minimally invasive MVr can be shorter postoperative ventilation times and shorter overall hospital stay without increased postoperative complications, and at similar procedural costs compared to the sternotomy approach (12–15). Our results are in line with these previous studies, showing significantly decreased postoperative ventilation time and hospital length of stay. A possible reason for shorter ventilation time, which we were not able to explore based on the presented data, may be less postoperative pain due to the smaller incision and less intraoperative spreading with quicker recovery to sufficient breathing patterns triggering earlier extubation. Another important benefit of the minimally invasive approach lies in an expedited recovery of these patients from an open-heart surgery. Without sternotomy-related precautions with regards to mobility, these otherwise healthy adults enjoy an earlier resumption of their active lifestyle as well as an earlier return to their work environment.

An important observation was the increase in overall MVr volume at our institution after starting the mini-thoracotomy program. This resulted in regular interaction of a team of experienced MV surgeons and echocardiographers for preoperative case planning and intraoperative consultation which we believe significantly contributes to patient safety and improved repair success rates (Figures 4, 5). This is in line with several prior reports that detected a strong relationship between individual surgeon and institutional case volumes with repair success rates. We believe that this benefit has a wider beneficial effect as well. Chikwe et al. found that even surgeons with lower MVr volumes (<25/year) have higher repair success rates when operating at the same institutions as higher-volume surgeons (>50/year) when compared to working at other institutions. Newell et al. also recently showed that at an institution with high MVr volumes, outcomes of low- and high-volume surgeons were comparable (16–19). Higher volumes also improve educational experiences with regular trainee exposure to complex MVr, allowing for stepwise teaching opportunities.

It should be emphasized that the quality of MVr was not negatively affected by choosing a minimally invasive approach, with a repair success rate of 100% in this cohort and no differences in residual MR compared to patients operated via sternotomy. In addition, neochord insertion, which might have superior long-term success rates (1, 20, 21), was more frequently used in mini-thoracotomy patients of our cohort. Longer lines of coaptation found after the neochord technique may translate into improved long-term outcomes (22, 23). As displayed in Figure 7, increasing expertise in complex MVr via mini-thoracotomy has changed our institutional approaches to MVP and a combination of resection and neochord insertion is not uncommon. In our opinion, it is crucial to be able to provide a wide range of repair techniques tailored to individual valve anatomy, without necessarily restricting the approach to either “resecting” or “respecting”.

With growing institutional and individual experience regarding minimally invasive MVr, a growth in proportion of more complex MVP cases was observed at MGH, such as anterior leaflet and bileaflet prolapse, undergoing MVr (Figure 6). While isolated posterior leaflet prolapse is the predominant lesion in MVP and associated with excellent repair rates, anterior and bileaflet prolapse present a more challenging surgical task. Yet, repairing these pathologies is feasible with similar long-term outcomes as MVr for isolated posterior leaflet prolapse and should be the primary goal for treatment of MVP (20, 24).

Especially in recent years, cardiac surgeons are confronted with the utilization of robotic support within a growing range of surgical specialties, sparking interest of patients and referring cardiologists. Perceived as an advance of established minimally invasive procedure, introducing robotically assisted cardiac surgery faces similar challenges as surgical training in general (9, 25). Robotic MV surgery was first reported on in 1998 (10, 26), with subsequent encouraging multicenter studies on the success of robotic MVr (27). However, robotic deployment, in addition to the difficulties inherent to a new procedure, is even more limited by its cost and low availability of equipment. Nevertheless, the learning curves appear to be acceptable, with reported rapid reductions in cross-clamp times and stabilization after 20–30 cases (28, 29). As mentioned above, facilitation of papillary muscle and mitral valve exposure and free multi-dimensional movement of surgical instruments are advantages of robotic MV surgery compared to the established methods.

At MGH, a stepwise approach is now followed to shift from standard minimally invasive to robotic MVr. Badhwar et al. proposed that institutional cardiac surgery case volumes be >250 cases/year, that an experienced team of anesthesiologists and perfusionists be present, and that the individual surgeon have at least 15 minimally invasive MVr cases in his or her record, to allow for safe initiation of a robotic MVr program. These criteria were by far exceeded in our case: Institutional yearly volume = ca. 1900 cases with an established team of cardiac anesthesiologists and perfusionists; individual volume of minimally invasive MVr >250 cases. Use of robotic platform for MVr was introduced only when minimally invasive MVr outcomes were excellent and comparable to those of sternotomy MVr, demonstrated by our subgroup analysis. Through guidance of experienced robotic MVr surgeons at all stages of training (theoretical, simulator, and cadaver training) as well as in-person and remote proctoring (Figure 2), we were able to launch a successful robotic MVr program that thus far has yielded satisfying results with a 100% repair success rate. As previously published, a decrease in aortic cross-clamp time can be expected over the next phase of robotic MVr cases. Especially through development of advanced simulation settings, adaptation of robotic surgery can be facilitated for surgeons already competent in complex minimally invasive MVr (30–32).

In an era of the rapid growth of transcatheter techniques, new percutaneous approaches to MVP repair are being proposed (33). However, the possibility of repairing all aspects of MVP (leaflet, annular and subvalvular abnormalities) and especially the availability of durable and safe annuloplasty currently favors surgical MVr. By reducing invasiveness and tissue trauma using minimally invasive and robotic approaches, earlier discharge and reduced use of rehabilitation facilities are possible, with reduced scarring to improve quality of life and patient satisfaction (34, 35).

It is our hope that the very good results of the presented cohort at an institution that fairly recently had started a minimally invasive MVr program might encourage centers to enable experienced surgeons and trainees to gain competence in minimally invasive MVr via formal training programs and interinstitutional proctoring support.

This work presents a retrospective analysis limited to a tertiary care center. Although propensity score matching was used to compare outcomes of minimally invasive and conventional MVr, remaining selection bias as well as surgeon bias need to be considered for the evaluation of the presented results. However, careful application of exclusion criteria and matching served to compare contemporary patient cohorts at a large-volume center. Regarding training aspects and institutional program, the surgical team at an institution like ours is large enough to have a reference surgeon in this type of procedure, and the training of residents is well-structured and standardized. The presented results and educational aspects might therefore not be entirely applicable to non-academic and lower volume centers.