Comparison of Doppler Flow Velocity and Thermodilution Derived Indexes of Coronary Physiology

OBJECTIVES: The aim of this study was to compare Doppler flow velocity and thermodilution-derived indexes and to determine the optimal thermodilution-based diagnostic thresholds for coronary flow reserve (CFR). BACKGROUND: The majority of clinical data and diagnostic thresholds for flow-based indexes are derived from Doppler measurements, and correspondence with thermodilution-derived indices remain unclear. METHODS: An international multicenter registry was conducted among patients who had coronary flow measurements using both Doppler and thermodilution techniques in the same vessel and during the same procedure. RESULTS: Physiological data from 250 vessels (in 149 patients) were included in the study. A modest correlation was found between thermodilution-derived CFR (CFR(thermo)) and Doppler-derived CFR (CFR(Doppler)) (r(2) = 0.36; P < 0.0001). CFR(thermo) overestimated CFR(Doppler) (mean 2.59 ± 1.46 vs 2.05 ± 0.89; P < 0.0001; mean bias 0.59 ± 1.24 by Bland-Altman analysis), the relationship being described by the equation CFR(thermo) = 1.04 × CFR(Doppler) + 0.50. The commonly used dichotomous CFR(thermo) threshold of 2.0 had poor sensitivity at predicting a CFR(Doppler) value <2.5. The optimal CFR(thermo) threshold was 2.5 (sensitivity 75.54%, specificity 81.25%). There was only a weak correlation between hyperemic microvascular resistance and index of microvascular resistance (r(2) = 0.19; P < 0.0001), due largely to variation in the measurement of flow by each modality. Forty-four percent of patients were discordantly classified as having abnormal microvascular resistance by hyperemic microvascular resistance (≥2.5 mm Hg · cm(−1) · s) and index of microvascular resistance (≥25). CONCLUSIONS: CFR calculated by thermodilution overestimates Doppler-derived CFR, while both parameters show modest correlation. The commonly used CFR(thermo) threshold of 2.0 has poor sensitivity for identifying vessels with diminished CFR, but using the same binary diagnostic threshold as for Doppler (<2.5) yields reasonable diagnostic accuracy. There was only a weak correlation between microvascular resistance indexes assessed by the 2 modalities.