Introduction: Physiology-guided coronary revascularisation is associated with improved outcomes however it is unclear if physiological indices can reliably assess individual lesions in the commonlyencountered scenario of serial/diffuse coronary artery disease. Methods: 3-D printed serial disease phantoms were assessed within an in vitro model of circulation. FFR of a lesion was predicted from the size of step-up on pressure-wire pullback in the presence of serial lesions (FFRapp) and compared to phantoms with no accompanying lesion (FFRtrue). A mathematical model to minimise error in predicting FFRtrue (FFRpred) was developed in 32 phantoms and validated in another 20 phantoms. In 54 patients with serial disease, resting and hyperaemic pressure-wire pullbacks were performed, with FFR, iFR, Pd/Pa of each stenosis measured by the trans-lesional gradient in each index and FFRpred derived mathematically, as in the in vitro study. These indices were compared to the ‘true’ trans-lesional gradient of each index following stenosis isolation by PCI of one of the stenoses or using a disease-free sidebranch. In a subset of 27 patients, Doppler average peak flow velocity (APV) was also measured before and after stenosis isolation, to calculate hSR (hSR=DP/APV, where DP=trans-lesional pressure gradient). In a further subset of 24 patients, prospective CT coronary angiography was performed prior to stenosis isolation by PCI to assess the value of a novel non-invasive FFRCT-derived PCI planning tool in estimating FFRtrue of each stenosis. Results: Both in vitro and in the clinical cohort, individual stenoses are underestimated and misclassified in serial disease, proportional to total FFR. Mean errors for FFR, iFR and resting Pd/Pa were 7 33.3%, 19.8% and 23.4% respectively, and 14.3% for FFRpred (P<0.001). Stenosis misclassification rates based on 0.80, 0.89 and 0.91 thresholds were 17%, 24% and 20% for FFR, iFR and Pd/Pa respectively. With FFRpred only 11% of stenoses were misclassified. Apparent and true hSR correlated strongly (R=0.87, P<0.001), with 7% of stenoses misclassified, thus suggesting that serial stenoses largely behave as resistors in series (providing further validation of the principles upon which our mathematical solution was based). In the subset of patients that had a prospective CT scan, applying a novel interactive PCI planning tool to FFRCT outputs results in significant improvements in the estimation of stenosis severity with the absolute value of underestimation falling to 0.01+/-0.05 (7% error). Conclusion: Physiological assessment of stenoses in the presence of serial CAD is prone to significant underestimation, proportional to the cumulative burden of disease within the vessel. All invasive pressure-derived physiological indices, resting and hyperaemic, are prone to significant error and stenosis misclassification when used conventionally. Doppler-based resistance indices are less prone to this error but utilisation is limited by difficulties in obtaining Doppler traces. The finding that serial stenoses generally behave as resistors in series supports the use of an FFR correction equation, derived using our 3D-printed in vitro study, that does not require measurement of coronary occlusive pressure. Applying this correction equation to routine pressure wire pullback measurements significantly reduces error in estimating true stenosis significance and has significant clinical utility. In addition, we demonstrate and validate a noninvasive FFRCT-derived PCI planning tool that aids more accurate prediction of true stenosis significance in serial CAD compared to contemporary invasive and non-invasive methods. The next step is to establish multi-centre clinical utility and outcome studies of the improved physiological methods established in this thesis.
Optimising Physiology Guided Management of Serial Coronary Artery Disease
Modi, B. (Author). 2019
Student thesis: Doctoral Thesis › Doctor of Philosophy