The loss of mitral leaflet coaptation surface area caused by restrictive chordal tethering to dysfunctional myocardial wall segments is the well-recognized mechanism of ischemic mitral regurgitation (MR). An accurate characterization of the left ventricular (LV) distribution pattern, magnitude, and reversibility of the contractile injury substrates that predispose to the occurrence of ischemic MR may improve the accuracy of therapeutic intervention. Only recently have high-resolution LV regional contractile metrics become clinically available to map myocardial ischemic substrates (hibernating, infarcted) across patient-specific LV geometry. Application of MRI-based multiparametric strain analysis in our pilot ischemic MR study group suggested that high-resolution 3D topographical mapping of LV contractile injury may reveal a more complex array of associated regional contractile injury than is discernible from echocardiography. This initial study identified a ?sentinel? LV region (basilar and mid subregions of the posterior and posterolateral LV regions) in which the presence of severe contractile injury clearly predisposes to the development of ischemic MR. We will enroll ischemic coronary artery disease patients with (?3+ MR; n=90) and without (?1+ MR; n=90) ischemic MR who are scheduled for standardized surgery (ACC/AHA Clinical Guidelines). Preoperative MRI- based multiparametric strain analysis will provide high-resolution 3D LV topographical maps of regional contractile injury to statistically correlate to occurrence of ischemic MR and to postoperative studies obtained at 3-months and yearly. An independent core laboratory will catalogue all echocardiography-based metrics of ischemic MR for inclusion in Support Vector Machine analyses, along with all other identified clinical variables. MRI-based LV displacement datasets are obtained in <30 minutes using Navigator-gated Spiral Displacement ENcoding with Stimulated Echoes (DENSE). Patient-specific LV strain fields are calculated using the recently developed Radial Point Interpolation Method (RPIM). Regional contractile function is ?normalized? by comparing multiple patient-specific strain metric values (at each of 11,520 LV grid points) to their respective average +/- SD values from our normal human strain database, with z- score (SD) calculation (total computer analysis <20 seconds). Support Vector Machine analyses will search all metric variables (multiparametric strain, echo-based metrics, and all clinical variables) for patterns that predict ischemic MR recurrence. We will use high-resolution 3D topographical mapping of ?normalized? LV contractile function to characterize the distribution, magnitude, and reversibility of the regional contractile injury substrates (hibernating; infarcted) associated with ischemic MR. We will then test the hypothesis that the novel application of machine learning Support Vector Machine analyses can identify hybrid combinations of both regional contractile injury patterns and clinical variables that accurately predict post-repair recurrence of ischemic MR.
Since the loss of mitral leaflet coaptation surface area caused by restrictive chordal tethering to dysfunctional myocardial wall segments is the well-recognized mechanism of ischemic mitral regurgitation, an accurate characterization of the left ventricular distribution pattern, magnitude, and reversibility of the contractile injury substrates (hibernating/infarcted) that predispose to the occurrence of ischemic mitral regurgitation may improve the accuracy of therapeutic intervention. Only recently have high-resolution LV regional contractile metrics become clinically available to map these myocardial ischemic substrates across patient-specific LV geometry. We will test the hypothesis that in patients with ischemic CAD, the ischemic status of a 4-subregion ?sentinel? zone (basilar/mid subregions of posterior/posterolateral regions) is the primary determinant of the presence of significant ischemic MR and that three defining ischemic substrate (infarcted, hibernating) characteristics (location, magnitude, and reversibility) can be used in a hybrid model to predict recurrence of ischemic MR.
