Project Summary There is a gap in the knowledge for staging in patients with primary mitral regurgitation (PMR) because there is a 20% chance of having an LV ejection fraction (EF) < 50% after mitral valve repair despite having an LVEF > 60% prior to surgery. Current predictive models such as STS-PROM and EuroSCORE, do not incorporate individual responses to oxidative and inflammatory response to cardiopulmonary bypass (CPB). This is especially important in PMR because there is severe cardiomyocyte cytoskeletal breakdown, myofibrillar loss, oxidative stress and mitochondrial damage despite LVEF > 60%. Red blood cell hemolysis during CPB and resultant free hemoglobin (Hb) and heme produce oxidative stress and cell injury. Formation of exosomes (Exos) constitute a protected delivery system of Hb and heme through the circulation to cells throughout the body. Here we show that Exos Hb content peaks 30 minutes after aortic cross clamp release (XCR) is related to the extent of acute cardiac injury in PMR patients. Injection of XCR Exos into rats results in acute cardiac and kidney injury. Another important factor is that current guidelines incorporate Echo/Doppler imaging that underestimate the severity of LV remodeling in PMR. Cardiac magnetic resonance (CMR) imaging is the gold standard for cardiac volume and mass from serial short axis slices independent of geometric assumptions. We have shown that CMR is essential in PMR, where simple LV dimensions and geometry-based volumes belie true LV volumes. Taken together, we hypothesize that CPB through Exos Hb release subjects this already adversely remodeled LV and damaged myocardium to an acute injury from which it does not completely recover resulting in a decrease in post-surgical LVEF < 50%. This grant addresses this problem in a series of three connected Aims that 1) relate Exos Hb content to heart and kidney injury and LVEF post-surgery in PMR patients and in our in vivo rat model system, 2) develop treatment strategies that not only target Exos Hb but also block Exos uptake by cells. Compared with conventional methods, machine learning methods have advantages in improving prediction performance. 3) Develop machine learning models that use parameters including baseline CMR LV function and blood Exos Hb content, and post XCR Exos Hb content, free Hb, urine KIM-1 and NGAL, blood creatinine, and hsTroponin to predict the CMR LVEF six months post PMR surgery. Timing for surgical intervention in patients with PMR is a major question today because of the now well- accepted 20% risk of a decrease in LVEF post-surgery. The current study will not only establish LV functional guidelines for surgical risk but also intraoperative heart and kidney injury. The findings may identify new CMR guidelines for surgical intervention and potentially novel myocardial protection strategies that minimize further damage to the already compromised PMR LV.