Comparison of clinical non-commercial tools for automated quantification of myocardial blood flow using oxygen-15-labelled water PET/CT

Abstract

Aims: Absolute quantification of myocardial blood flow (MBF) with (15)O-water cardiac positron emission tomography (PET)/CT has recently demonstrated to hold promising diagnostic value for the detection of coronary artery disease (CAD). However, methodological differences in utilized analysis software packages (SP) could affect generated MBF values, potentially prohibiting widespread clinical applicability of obtained normal thresholds. The aim of this study was to compare two validated non-commercial SP, Carimas and Cardiac VUer, for the quantification of MBF using (15)O-water PET.

Methods and results: One hundred patients with intermediate likelihood of CAD and scanned in academic centres in Amsterdam (n = 50) and Turku (n = 50) were included in the study. Patients underwent a (15)O-water PET/CT scan during rest and vasodilator stress based on clinical indications. A single observer, blinded from clinical results and with no prior experience in either SP, analysed all patients twice with both SP. Reproducibility of each SP was assessed using intraclass correlation coefficients (ICC). Intersoftware agreement was assessed using paired t-tests and linear regression. ICC was excellent for each SP for both global and regional MBF (ICC >0.90). Global MBF was comparable between Carimas and Cardiac VUer during rest (1.02 ± 0.28 vs. 0.99 ± 0.23 mL min(-1)g(-1), respectively, P = 0.07), and slightly higher for Carimas during stress (2.73 ± 0.82 vs. 2.63 ± 0.84 mL min(-1)g(-1), respectively, P = 0.01). At a regional level, for resting conditions small (<10%) but significant discrepancies were noted in each vascular territory while for stress MBF, a significant difference was only observed for the LAD region. Differences between SP for the LAD territory were abolished after exclusion of the distal apical segment, which is susceptible to spillover artefacts. An excellent correlation between MBF values was found for global (r = 0.96) and regional MBF (r > 0.94 for all).

Conclusion: For global and regional MBF, Carimas and Cardiac VUer showed excellent agreement and intra-observer reproducibility. These results confirm that, for patients with intermediate likelihood of CAD, these validated SP are interchangeable and can be utilized for routine clinical practice of (15)O-water cardiac PET.

Keywords: 15O-water; Coronary artery disease; Myocardial blood flow; Myocardial perfusion imaging; Software reproducibility.