Between January 2018 and June 2024, 2,266 patients underwent mitral valve surgery at the Department of Cardiac Surgery, University Hospital Frankfurt. Of these, 688 patients with isolated mitral valve disease were treated through a minimally invasive approach and were included in the study. Amongst them, 514 received a standard minimally invasive approach via right lateral minithoracotomy under direct vision, and 174 underwent a EMIC procedure. The decision on surgical access was made independently of the planned technique and followed institutional protocols. All cases were evaluated in the interdisciplinary heart team, taking age, comorbidities, and overall risk into account, without predefined allocation criteria for the surgical approach. Patients undergoing non-isolated mitral valve procedures, combined cardiac surgeries, or mitral valve surgery performed via alternative surgical approaches, as well as patients who refused consent for data use, were not included in the study population.

Both minimally invasive techniques have been previously described in the literature (11, 12). In summary, cardiopulmonary bypass was established via femoro-femoral cannulation, either percutaneously under ultrasound guidance using the Seldinger technique or by surgical exposure of the femoral vessels. Surgery was then performed through a right lateral mini-thoracotomy in the fourth intercostal space using a soft-tissue retractor. The Chitwood clamp and endoscope were introduced through small auxiliary ports in the third intercostal space (13).

In the standard-MIC cohort, the mitral valve was visualized directly under thoracotomy (Figure 1), whereas with the EMIC technique a three-dimensional camera system was used. The left atrium was opened through the interatrial groove (Waterston's groove) to provide optimal exposure of the mitral valve with both approaches.

The vast majority of procedures were performed by the same two experienced surgeons with established expertise in minimally invasive mitral valve surgery. Both surgeons were experienced with the standard minimally invasive technique. From the time of inclusion onwards, one of them performed all his operations (all comers) using an endoscopic approach with indirect visualization of the mitral valve through the camera, whereas the other performed mitral valve repair using direct vision, the camera was used in adion only. In the endoscopic cohort, access was achieved via a lateral right mini-thoracotomy (4–5 cm) or a periareolar incision (2–3 cm) through the fourth intercostal space, using a soft-tissue retractor (without rib spreading) and endoscopic visualization via a lateral fourth intercostal port with CO₂ insufflation. The choice of a periareolar (“nipple-cut”) incision was based on patient-specific anatomical factors, including body habitus, sex, areolar diameter, and breast size (Figure 2).

Mitral valve reconstruction consisted of restoring normal leaflet motion and achieving annular stabilization. Accordingly, annuloplasty was performed in all repairs using flexible or rigid rings, with the size determined according to the anterior leaflet dimensions (14). Leaflet repair was achieved using triangular or quadrangular resection, sliding plasty, and artificial chordal replacement with expanded polytetrafluoroethylene (ePTFE, Gore-Tex) sutures using the Loop technique. When reconstruction was not feasible, valve replacement was carried out using biological prostheses (Carpentier-Edwards, Edwards Lifesciences; Epic™, Abbott) or mechanical valves (Open Pivot™, Medtronic).

This single-center cohort study was conducted at the Department of Cardiac Surgery, University Hospital Frankfurt, Germany. All analyses were performed according to an intention-to-treat principle, with patients assigned to groups based on the initially planned minimally invasive access strategy, irrespective of whether mitral valve repair or replacement was ultimately performed. Both valve repair and replacement procedures, as well as minor concomitant interventions, were analyzed. Preoperative data, including baseline characteristics and echocardiographic findings, were obtained from referring cardiologists or from the institutional outpatient clinic. Intraoperative data were extracted from surgical reports, and postoperative information was collected from electronic patient records and discharge summaries. Data were retrieved using the hospital systems Orbis and MetaVision, transferred into a dedicated Microsoft Access database, and subsequently exported to Microsoft Excel for statistical evaluation.

This study was conducted in accordance with the Declaration of Helsinki (as revised in 2013). This study was approved by the local Ethic Committee of University Hospital Frankfurt (No. 79/13), and written informed consent was waived due to the retrospective nature of this study.

Normal distribution of continuous variables was assessed using the Shapiro–Wilk test. Continuous data are presented as mean ± standard deviation or median (interquartile range), as appropriate. Categorical variables are expressed as counts and percentages and were compared using the Chi-square test or Fisher's exact test when expected cell frequencies were <5. Univariate analyses were performed to identify potential predictors, and variables with a p-value < 0.05 were subsequently included in multivariate logistic regression analyses. Early mortality was assessed descriptively as 30-day all-cause mortality. All statistical analyses were conducted using IBM SPSS Statistics, version 29.0 (IBM Corp., Armonk, NY, USA). Figures and tables were generated with Microsoft Excel 2021 and Jamovi. A two-sided p-value < 0.05 was considered statistically significant. The primary endpoint of the study was a 30-day major adverse outcome, defined as the composite of all-cause mortality or stroke. All other perioperative and postoperative outcomes were considered secondary endpoints. Given the retrospective and non-randomized study design, the absence of propensity-based adjustment, and the analysis of multiple secondary endpoints, all analyses were performed in a descriptive and exploratory manner and were not intended to support inferential comparisons between surgical approach.

Clinical outcomes were defined as follows: low-cardiac-output syndrome (LCOS) was defined as postoperative hemodynamic instability requiring inotropic/vasopressor therapy or mechanical circulatory support (ECLS or ECMO). Postoperative myocardial infarction was defined based on standard clinical criteria documented in the medical records. Neurological events were defined as any new postoperative neurological complication and categorized as transient ischemic attack, minor or major stroke, coma, or other neurological complications (e.g., delirium).

Baseline characteristics are provided in Table 1. Patients had a mean age of 61.9 ± 11.9 years and were predominantly male, with 36% being female. Patients in the EMIC group more frequently presented with advanced heart failure (NYHA ≥ III, p = 0.023). Non-sinus rhythm was defined as the presence of paroxysmal or chronic atrial fibrillation, including cases with atrial fibrillation despite pacemaker rhythm, with no significant difference between groups. Previous relevant cardiac surgery was defined as any prior mitral valve reconstruction or valvuloplasty procedure. Apart from NYHA class and previous cardiac surgery, baseline characteristics were comparable between groups.

Intraoperative data are summarized in Table 2. Cardiopulmonary bypass and aortic cross-clamp times were significantly longer in the endoscopic group (169 ± 47 min and 88 ± 25 min, respectively) compared with the ST group (132 ± 39 min and 73 ± 23 min; both p < 0.001). Among 688 patients treated with MIMVS, mitral valve repair was performed in 480 cases, corresponding to an overall repair rate of 69.8%, while 208 patients (30.2%) underwent valve replacement. Percutaneous cannulation was used in 96.7% of all cases, with no significant difference between groups.

Postoperative outcomes are summarized in Table 3. A total of 59 patients (8.6%) required re-thoracotomy. Most re-explorations were performed via re-mini thoracotomy for postoperative bleeding or hematoma evacuation. Other indications included pericardial effusion and, in isolated cases, pericardial tamponade. Four patients required multiple re-operations. There was no significant difference between the two groups (p = 0.449). Gastrointestinal complications occurred in 9 patients (1.3%), mainly due to postoperative bleeding, while two cases were caused by intestinal ischemia. Neurological complications were observed in 6 patients (0.9%), including two transient ischemic attacks, two strokes, one case of stroke-related postoperative coma, and one case of postoperative coma unrelated to cerebrovascular events. Pericardial effusion was significantly more frequent in the EMIC group (11.7%; p = 0.008), and the need for renal replacement therapy was also higher in this group (10.9%; p = 0.003). Residual mitral regurgitation was defined as grade ≥ 2 and did not differ significantly between groups. Low-cardiac-output syndrome occurred in 13 patients (1.9%), of whom two required ECLS therapy and eight could be treated conservatively. Ventilation time differed significantly between groups (29 ± 69 h vs. 22 ± 52 h; p < 0.001). Causes of early mortality included low-cardiac-output syndrome in two patients, sepsis and multi-organ failure in three patients, and one case of severe neurological injury. Thirty-day mortality was 2.9% in the EMIC group and 1.0% in the ST group and did not differ significantly between groups (p = 0.133).

This study has several limitations that should be considered when interpreting the results. The retrospective design carries an inherent risk of bias, particularly due to the absence of randomization. A relevant temporal bias must be acknowledged, as the endoscopic approach was introduced later and includes the early adoption phase, whereas the standard approach represents a longer-established technique. Baseline differences between both groups, including a higher NYHA functional class and a prevalence of redo surgery in the EMIC group, further limit direct comparability between both approaches. A systematic long-term follow-up beyond the 30-day mortality period could not be performed within the given study framework and should be addressed in future investigations. Potential examiner bias in preoperative echocardiographic assessment must also be taken into account, as examinations were performed under varying conditions and influenced by individual operator experiences. In some cases, discrepancies between preoperative findings and intraoperative assessment of valve etiology were observed. Another limitation is the unequal distribution of patients between groups, which reduces comparability and increases the potential for distortion. Due to the very low number of early mortality events, time-to-event analyses were not performed. The absence of formal learning curve adjustment represents an additional limitation of the present study. Future studies could apply statistical methods such as propensity score matching or inverse probability weighting to improve the balance for even better comparison. Finally, differences in surgical experience between operators may lead to an additional source of variability.

The present study demonstrates the feasibility of the EMIC approach within a high-volume centre performing routine minimally invasive mitral valve surgery. The observed perioperative and early postoperative outcomes provide descriptive insight into its application during institutional implementation. Future prospective studies are warranted to evaluate long-term outcomes and patient-centred parameters such as recovery and quality of life.