Coronary microvascular disease (CMD) is notoriously difficult to diagnose non-invasively, and current methods of assessing CMD utilize only the peak velocity of the coronary flow pattern. While new imaging techniques such as cardiac magnetic resonance imaging (MRI) have improved the assessment coronary perfusion, there are currently no non-invasive methods that incorporate the coronary flow pattern over a complete cardiac cycle to definitively assess and predict the development of CMD. Coronary blood flow (CBF) reflects the summation of flow in the coronary microcirculation, and our lab has begun to harness the full CBF pattern under varying flow and disease conditions (e.g. type 2 diabetes) to determine whether it might harbor novel clues leading to the early detection of CMD. Our past and preliminary data indicate an early onset of CMD in both type 2 diabetes mellitus (T2DM) and metabolic syndrome (MetS) that occurs prior to the onset of macrovascular complications and that are characterized by blood flow impairments and alterations in coronary resistance microvessel (CRM) structure, function, and biomechanics. Our data also uncovered innovative correlations between CRM structure/biomechanics and our newly-defined features of the coronary flow pattern, some of which were unique to normal or diabetic mice. We have initially utilized these CBF features, in the presence and absence of other factors such as cardiac function, to develop a mathematical model in collaboration with Drs. Christopher Bartlett and William Ray that to date demonstrated that 6 simple factors can predict a normal vs. diabetic coronary flow pattern with 85% predictive accuracy. Utilizing a multidisciplinary approach, these preliminary data strongly suggest that the coronary flow pattern and physiological modulators of it (e.g. coronary micovascular structure/function/biomechanics, cardiac function, etc), may be useful in directly diagnosing early CMD. Therefore, we hypothesize that dissecting the elements that influence coronary flow patterning will be critical determinants in the direct assessment of coronary microvascular disease using computational modeling. Using our previous publications and our preliminary data as guides, the hypothesis will be tested by addressing two specific aims: 1) Determine whether unique time-dependent CBF patterning in normal and T2DM is dictated by a combination of CRM remodeling and biomechanics, coronary flow pattern dynamics, and cardiac function, permitting the development of a computational model to accurately predict CMD; 2) Determine the reproducibility and robustness of the machine learning model in predicting CMD in a diet-induced obesity/diabetes mouse model. If successful, these studies will be the first to simultaneously examine the influence of CRMs, CBF, and cardiac structure/function on the distinct pattern of coronary flow, and it will determine whether a mathematical model may be useful in establishing a direct assessment of CMD to eventually enable clinicians to conduct a more direct non-invasive diagnosis of CMD for the prevention and/or treatment of heart disease.
Coronary Artery Disease (CAD) is the leading cause of heart disease and is associated with hypertension, diabetes, and metabolic syndrome. Coronary Microvascular Disease (CMD) is comprised of structural and functional deficits of the tiny coronary arteries that may be an earlier indicator of disease prior to the onset of overt CAD. The proposed multidisciplinary research aims to develop a computational artificial intelligence model that will accurately predict CMD based on a non-invasive coronary flow pattern obtained by Doppler echocardiography.
