Automatic Morphological Evaluation Using Three-Dimensional Transeophageal Echocardiography of Patients with Mitral Prolapse and Insufficiency: Comparison with Patients Without Cardiac Chambers Alterations

1. Introduction

Mitral valve regurgitation caused by mitral valve prolapse presents clinical relevance due to its high prevalence in clinical practice [1,2,3,4,5,6,7,8,9,10,11,12,13]. Mitral valve apparatus consists of a complex group of structures such as valve cusps, the set of chordae that fix and traction the cusps, the mitral valve annulus, the papillary muscles that insert into the ventricular wall, and the ultrastructural and contiguous relationship with the left ventricle and left atrium [1,2,3,4,5,6,7,8,9,10,11,12,13]. Each of these elements has its own characteristics and particular aspects which make the mitral valve apparatus very unique.

A better understanding of the structures that make up the mitral valve allows us to determine the best time for surgical or percutaneous intervention [1,2,3,4,5,6,7,8,9,10,11,12,13]. Imaging methods for the geometric and functional analysis of heart disease have evolved rapidly with the development of computing, nanotechnology, and artificial intelligence. For the analysis of valvular heart disease, echocardiography occupies a prominent position due to its ability to demonstrate cardiac anatomy with great accuracy and proximity to anatomical reality.[1,2,3,4,5,6,7,8,9,10,11,12,13]. In the 21st century, we have the opportunity to rapidly demonstrate cardiac structures using transesophageal echocardiography and artificial intelligence from multiple spatial planes, taking into account multiplanar three-dimensional analysis, in real time, with the aid of algorithms specifically designed for demonstrating the mitral valve apparatus

Thus, we planned a study to determine the different elements that make up the mitral valve apparatus, the morphofunctional analysis of the structures related to the valve apparatus (left atrium, left ventricle), using a three-dimensional transesophageal echocardiography automatic algorithm dedicated to the analysis of the mitral valve apparatus, (Figure 1 and Figure 2) when compared to patients without structural cardiac chamber changes.

2. Methods

2.1. Patients

2.1.1. Study Design

A single-center from a Brazilian tertiary hospital centre, prospective study from February 2023 to December 2024 evaluating transesophageal three-dimensional echocardiographic variables with a mitral valve automatic analysis algorithm was performed in 59 consecutive patients with mitral valve regurgitation (grade 2 and 3) due to mitral valve prolapse and 43 patients without structural cardiac chamber alterations who went to echocardiographic examination due to patent oval forame. The study was performed in accordance with the principles established by the Declaration of Helsinki, and the project was approved by the Research Ethics Committee of the hospital. “Informed consent was obtained from all subjects involved in the study.”

2.1.2. Inclusion Criteria

- Consecutive patients with moderate to severe mitral valve regurgitation, resulting from valve prolapse in analysis for structural treatment.

- Consecutive patients with clinical suspicion of PFO.

- Patients with age over 18 years of age.

-Signed informed consent form to participate in the study.

2.1.3. Exclusion Criteria

- Atrial fibrillation or tachyarrhythmia with hemodynamic instability.

- Presence of significant aortic insufficiency, significant aortic stenosis, or significant mitral stenosis.

2.2. Echocardiography

Imaging Acquisition

The images were acquired using a Philips Healthcare Epiq CVx echocardiography device by different investigators of the study, available in the Echocardiography Department of the Hospital. Images were acquired during a conventional transesophageal echocardiographic examination using conventional two-dimensional projections, with analysis of the complex mitral valve apparatus, including detailed analysis of all its components: annulus, cusps, commissures, chordae, papillary muscles, and left ventricle. The projections used for image acquisition were: midesophageal commissural, long-axis, two-chamber, and four-chamber views with and without anteroflexion of the probe; and transgastric short-axis and long-axis views. After the 2D evaluation, 3D acquisitions were obtained in 3D zoom and full-volume formats. 3D zoom images). Repeated 3D zoom images of the mitral valve complex were taken from the atrial aspect corresponding to the surgical view, reducing the overall gain and increasing the magnitude of the selected image.Such acquisitions consist of obtaining subvolumes with angle and temporal resolution greater than the 3D zoom format with a higher volume rate, higher than 25 Hz, coupled to the electrocardiogram in consecutive cardiac cycles, and digitally stored for later analysis of the processed images.

Conventional 2D echocardiographic and Doppler parameters were measured, as follow: left ventricle anteroposterior diameters; RV: right ventricle diastolic diameter short axis diameter; LVEF: Left ventricule ejection fraction (Teicholz method); LA: left atrium anteroposterior diameter and volume; PSAP: pulmonary artery systolic pressure, following previously published international recommendations[16].

2.3. Imaging Analysis

The images extracted from the three-dimensional transesophageal study were evaluated using Philips Health Care’s artificial automatic software dedicated to the analysis of the mitral valve apparatus (MVA), which is part of Philips Health Care’s QLAB software, version 15.0.0. The analysis was performed offline on a workstation by two researchers involved in the investigation.

Echocardiographic variables to be analyzed with artificial intelligence software dedicated to the analysis with three-dimensional transesophageal echocardiography of the elements of the mitral valve apparatus in patients with mitral regurgitation and in patients with patent foramen ovale: mitral annulus anteroposterior diameter (AP Diameter), mitral annulus anterolateral-posteromedial diameter (AL-PM Diameter), sphericity Index (AP / AL-PM), intertrigonal distance, commissural diameter, saddle shaped annulus Area (3D), saddle shaped annulus perimeter (3D), D-shaped annulus area (2D), D-shaped annulus perimeter, annulus height, non-planar angle, tenting volume, coaptation depth, tenting area, angle aortic (AAo-AP), maximum prolapse height, maximal open coaptation gap, maximal open coaptation width, total open coaptation area (3D), anterior leaflet area, posterior leaflet area, distal anterior leaflet angle, posterior leaflet angle, anterior leaflet length, posterior leaflet length, annulus area (2D), anterior closure line length (2D), anterior closure line length (3D), posterior closure line length (2D), posterior closure line length (3D), C-shaped annulus, annular displacement (max), annular displacement velocity (max), tenting volume fraction, annulus area fraction (2D) (Figure 1 and Figure 2).

2.4. Statistics

Normally distributed quantitative variables were expressed as mean ± standard deviation or as median and interquartile range (IQR: P25–P75) when not normally distributed; qualitative variables were summarized as number and percentage. The comparison between etiologies (MI and PFO) was performed using Student’s t-test for quantitative variables that were normally distributed. For those that were not normally distributed, the Mann-Whitney test was used. For data from participants with MI, Pearson’s correlation coefficient (r) and its 95% confidence interval (95% CI) were calculated between the variables ERO and Regurgitant Volume, as well as a scatterplot between the two variables. The Pearson correlation coefficient and its 95% CI were also calculated between each of the variables in the new software and ERO and Regurgitant Volume. When correlation values were p<0.20, simple linear regression models were adjusted, thus obtaining the coefficient of explanation (R²) for each of the adjusted models. To adjust a multiple linear regression model, we first assessed the presence of multicollinearity among the variables obtained via the new software, and then selected which variables would be introduced into the multiple linear regression model.The ERO and Regurgitant Volume variables were categorized based on the 75th percentile of each distribution. Each of the variables obtained via the new software was compared according to the ERO and Regurgitant Volume variable groups. The Student’s t-test was applied, and the effect size was calculated using Cohen’s test when the variables were normally distributed; otherwise, the Mann-Whitney test was applied, and the effect size was calculated using Cliff’s δ. The effect size magnitudes obtained via Cohen’s d were interpreted as: trivial (<0.20), small (0.20–0.49), moderate (0.50–0.79), and large (≥0.80). The magnitudes obtained via Cliff’s δ were interpreted as: trivial (<0.15), small (0.15–0.32), moderate (0.33–0.46), and large (≥0.47). All analyses were performed in StataNow/MP 19.5 (StataCorp, College Station, TX, USA) and results with p-values less than 0.05 were considered significant. Intra- and inter-observer variability (ICC= Intraclass correlation coefficient) were tested in 15 patients from MVP and PFO groups by the same operator and by a second operator 10 months after the first acquisition.

3. Results

The distribution of demographic data according to etiology, echocardiogram measurements according to etiology, comparison of new automatic software measures between etiologies, descriptive measures of effective regurgitant orifice and regurgitant volume among patients with mitral valve prolapse, new automatic software measures according to 75 percentile of ERO, new automatic software measures according to 75 percentile of regurgitant volume are depicted in Table 1, Table 2, Table 3, Table 4, Table 5 and Table 6.

Intraclass correlation coefficient (ICC) for intra-observer variability was >0.96 for all variables for MVP and PFO patients with coefficient of variation < 1.2%. For inter-observer variability, ICC was >0.95, with coefficient of variation < 1.7%. for all variables for MVP and PFO patients.

4. Discussion

The main finding of this manuscript is that twenty-seven automatic parameters (79%) wefe different concerning MVP and patients without cardiac chambers alterations (PFO patients) (p<0.05): anteroposterior anular diameter, anterolateral-postromedial diameter, commissural diameter, saddle shaped annulus area (3D), saddle shaped annulus perimeter (3D), D-shaped annulus area (2D), D-shaped annulus perimeter, annulus height, non-planar angle, maximal open coaptation width, anterior leaflet area, posterior leaflet area, anterior leaflet lenght, posterior leaflt lenght, annulus area (2D), anterior closure line lenght (2D), anterior closure line lenght (3D), posterior closure line lenght (2D), posterior closure line lenght (3D), C-shaped annulus (p<0.05). Other two parameters (5%), tenting area and annular displacement velocity max, were marginally different concerning groups (both p: 0.052), The effect sizes (Hedges’ g between 1.0 and 2.0) indicate very large differences. with 95% CIs far from zero, reflecting annular remodeling and global enlargement of the valve apparatus in cases of MI. These differences suggest significant dilatation of the annulus and valve leaflets, consistent with the pathophysiology of degenerative mitral regurgitation.

The variable maximum prolapse height was not normally distributed, but through the Mann-Whitney test it was possible to verify that patients with MVP had a statistically higher median Maximum Prolapse Height than that presented by patients with PFO (p<0.001), with a very large effect size (g=4.22; 95%CI 2.69–6.40), corroborating the greater degree of leaflet prolapse.

The combination of clinically relevant absolute differences (0.6–1.0 cm in principal diameters) and very large effect sizes (Hedges’ g > 1.5) supports the clinical relevance of the anatomical discrepancies identified by the automatic software, regardless of the absence of prior sample size calculation.

The mean ERO (0.42 cm²) and mean regurgitant volume (62.9 mL/beat) were close to levels considered to represent patients with important mitral valve regurgitation. In this sense, we choosed to correlate the use of the automatic software in the subgroup of patients in the 75 percentile of ERO and regurgitant volume, what could represent patients with more advanced MVP. In this sense we tested the software in patients with ERO >0.48 cm² and regurgitant volume > 70 ml/beat. According to Table 5, it can be observed that there was a statistically significant effect with a moderate to important effect size for the variables: posterior leaflet area, posterior leaflet length and tenting area. Posterior leaflet area showed statistically higher mean values among patients with ERO ≥ 0.48 cm² when compared to those with ERO < 0.48 cm² (mean difference = 2.39 ± 1.07; CI95: 0.23 to 4.54; p=0.031 and moderate to important effect size: d=0.74). Clinically, it can be interpreted that a larger posterior leaflet area was associated with elevated ERO. In the same way, posterior leaflet length showed a statistically higher mean in the ERO ≥ 0.48 cm² group when compared with those with ERO < 0.48 cm² (mean difference = 0.40 ± 0.18; CI95: 0.04 to 0.75; p=0.032 and moderate to important effect size: d=0.73). Thus, a longer posterior cusp was observed in patients with greater regurgitation. Also, tenting area presented an asymmetric distribution and was smaller among patients with ERO ≥ 0.48 cm² (median 1.02; P25=0.67 and P75=1.31) versus (median 1.53; P25=1.17 and P75=2.13); p=0.048; moderate effect size: =0.41). That is, smaller tenting area values were observed in the ERO ≥ 0.48 cm² group. Clinically, this pattern was compatible with a prolapsing/degenerative phenotype (smaller tenting area) in cases with higher ERO.

According to Table 6, it can be observed that there was no statistically significant effect on any of the variables of the new software according to P75 of the regurgitation volume (p>0.05). Even so, small effect sizes suggest a morphofunctional pattern compatible with a primary valve mechanism (degenerative/prolapsing) in the regurgitation volume group with values greater than or equal to P75. Also, the set of data consistently points in the direction (even without p<0.05) of a primary valve phenotype in cases with RV≥70 ml/btm: larger/longer posterior cusp and smaller tenting/coaptation depth. The slightly larger annulus appears as a trend, without statistical proof in this sample. In summary, the pattern favors a degenerative/prolapse mechanism as reason for mitral valve changes.

Previous studies have adressed tridimensional echocardiography softwares dedicated to mitral valve apparatus analysis 6,8, 16,17,18,19,20. They brought important information concerning mitral annulus 8, distance from posteromedial pappilary muscles to mitral valve edge what can bring implications to good results for surgical repair 11.

Mitral valve repair, first described by Carpentier 56 years ago 15, remains the gold standard for treating degenerative mitral regurgitation. Recent advances in minimally invasive and robotic cardiac surgery aim to enhance the safety and efficacy of mitral valve repair 13. In this setting, the development of automated echocardiographic software tools may further improve reproducibility and standardization of the procedure, supporting its wider adoption in clinical practice. We believe that our investigation could bring and add new information concerning automatic mitral valve analysis and eventual useful information for surgical treatment. The efficiency, accuracy and potencial use of the coming mitral valve apparatus analysis softwares are to be validaded in future investigations 15-20.

5. Limitations

This investigations concerns a small sample single center mitral valve prolapse patients with moderate and severe mitral regurgitation from a tertiary Brazilian hospital. Our results have to be validated in a higher population sample envolving multiple centers in different countries.

6. Conclusions

Automatic mitral valve parameters analysis were different concerning MVP and no structural cardiac chamber (PFO) patients. 75 percentile ERO MVP patients (ERO> 0.48 cm²) correlated to posterior leaflet parameters. This anatomic information could be useful to planning and surgical treatment of mitral valve prolapse patients.