Impact Of Plaque Components And Volume On Fractional Flow Reserve Derived From Optical Coherence Technology

Background:The coronary fractional flow reserve(FFR)can guide the revascularization of simple and complex coronary artery diseases and has better clinical outcome,however,which application is limited due to the costly pressure wire and complicated procedure.Computational flow reserves is a more simple and more practical clinical index which is based on cardiovascular imaging and artificial intelligence(AI)machine learning algorithms to evaluate coronary artery function.Both quantitative blood flow reserve(QFR)based on invasive coronary angiography(ICA)and optical flow ratio(OFR)based on optical coherence tomography(OCT)have good diagnostic consistency with FFR,but their algorithms do not include plaque parameters.At present,the intrinsic relationship between plaque parameters and the computational flow reserves remains unclear.Purpose:This study aims to explore the relationship between plaque components and volume in coronary artery and FFR and two computational fractional flow reserves,also compare the impact of plaque components and volume in coronary artery on the difference between the two different computational fractional flow reserve and FFR by building a computational model divided plaque compositions automatically using the deep learning method.Materials and Methods:First,a number of 391 patients from 5 clinical centers including Australia,the United States,Japan,Spain and China with a total amount of509 OCT retracement images in coronary artery were enrolled for plaque components labeling,and they were divided into training set and test set at a ratio of 9:1 to establish automatic plaque segmentation model.The constructed model derived from OCT image was applied to analysis the OCT retracement images in patients who had underwent FFR measurement,OFR and QFR analysis from August,2011 to October,2018,and the total plaque volume(TPV),the total proportion of calcific plaque(%),the total proportion of lipidic plaque(%)and the total proportion of fibrous plaque(%)were automatically calculated by the model.Then,we recalculated the data for obtaining the total calcific plaque volume(CPV),total lipidic plaque volume(LPV)and total fibrous plaque volume(FPV).Moreover,the interrogated vessels were divided into three subgroups according to the difference between OFR and FFR values:the underestimation group with(OFR-FFR)value<-0.05,the control group with-0.05≤(OFR-FFR)value≤0.05,and the overestimation group with(OFR-FFR)value>0.05.Results:A total of 212 vessels from 181 patients were enrolled in this study.The results showed that CPV and LPV had weakly negative correlation with the FFR value(r=-0.261,p<0.001 and r=-0.195,p=0.004,respectively).In addition,CPV and LPV were also had a negative correlation with OFR values(r=-0.303,p<0.001 and r=-0.315,p<0.001,respectively).Also,CPV and LPV had a negative correlation with QFR values(r=-0.233,p<0.001 and r=-0.206,p=0.003,respectively).LPV and(OFR-FFR)value had a weak negative correlation(r=-0.214,p=0.002),but not correlate with(QFR-FFR)value(p=0.084).CPV were not significant correlation with both(OFR-FFR)value and(QFR-FFR)value(p=0.403 and p=0.657,respectively).The regression analysis results also showed that for different plaque components,only LPV had a weak correlation with(OFR-FFR)values(r=-0.214,p=0.002).The analysis based on vessel-level showed that in interrogated vessels,28,164,and 20vessels were in the underestimation,control,and overestimation group,respectively.LPV was significantly higher in the underestimated group than the control group(85.96mm~3 vs 58.46mm~3,p=0.007)and the overestimated group was lower than control group(34.01mm~3 vs.58.46mm~3,p=0.017).Conclusion:The total volume of different plaque components may present different effects on computational flow reserves(OFR or QFR)measurements.Compared with FFR,the increase of lipidic plaque volume will lead to underestimation of OFR measurement.