Comparison, Agreement, and Significance Cutoffs of 18F-Flurpiridaz PET Myocardial Blood Flow Quantitation Methods and Software Packages

¹⁸F-flurpiridaz is a novel PET myocardial perfusion radiotracer with a high myocardial extraction fraction and low positron range, providing high-resolution images. Additionally, the linear relationship between ¹⁸F-flurpiridaz myocardial extraction and myocardial blood flow (MBF) over a wide range of flow values permits accurate MBF measurement. Several groups have developed strategies to quantitate ¹⁸F-flurpiridaz MBF with methodologic differences. These methods have not previously undergone head-to-head comparisons. Methods: Three methods of ¹⁸F-flurpiridaz MBF quantitation were compared using Emory Cardiac Toolbox (ECTb; Syntermed), 4DM (INVIA), and Quantitative PET (QPET; Cedars-Sinai) software. All evaluable pharmacologic stress patients from the phase 3 ¹⁸F-flurpiridaz PET trial (NCT03354273) of ¹⁸F-flurpiridaz were included (n = 405). We adopted Bland–Altman plots to determine absolute differences between MBF quantitative methods, Fleiss κ for agreement across software packages, and Cohen κ for agreement of pairwise comparisons. Diagnostic performances of stress MBF and myocardial flow reserve (MFR) were determined by performing receiver-operating-characteristic analysis for areas under the curve (AUCs), using quantitative invasive coronary angiography as the reference standard. Results: Differences in mean stress MBF and MFR between methods across coronary territory distributions ranged from 0.23 to 0.29 mL/min/g and from 0.37 to 0.40 mL/min/g for ECTb and 4DM, respectively; from 0.09 to 0.11 mL/min/g and from 0.36 to 0.38 mL/min/g for ECTb and QPET, respectively; and from 0.25 to 0.26 mL/min/g and from 0.39 to 0.40 mL/min/g for QPET and 4DM, respectively. There was substantial agreement across software packages, with the Fleiss κ ranging from 0.77 to 0.79 and 0.72 to 0.75 for stress MBF and MFR, respectively. Similar results were found in pairwise comparisons. For coronary artery territories with at least 70% stenosis, median stress MBF and MFR AUCs were 0.73–0.74 and 0.71–0.73, respectively, when evaluated on a per-patient basis, and 0.75–0.77 and 0.74–0.75, respectively, in pooled coronary distributions on a per-vessel basis across all methods. Conclusion: All software packages demonstrated high agreement and similar ¹⁸F-flurpiridaz MBF quantitation, paving the way for the interoperability of these platforms in clinical practice.