Jørgen Frøkiær

Objectives Due to incomplete extraction, corrections are required when using 82Rb for absolute MBF measurements. Previous studies have established these corrections based on healthy subjects, however, the performance in patients with pathologically reduced MBF has only been studied in few individuals. In this study, we compared 82Rb cardiac PET with the non-invasive gold standard for absolute blood flow quantification, 15O-water, in patients with known ischemic heart disease (IHD). Methods Thirty-nine patients with known IHD referred to 82Rb PET were included in the study. Each patient underwent rest and adenosine-induced stress imaging with 82Rb and 15O-water PET per standard protocol within 2 weeks. Clinical assessment of 82Rb was performed in QPET (Cedar-Sinai), whereas dynamic scans for both 15O-water and 82Rb was analyzed in aQuant. Both softwares apply a 1 tissue compartment model for both tracers and the extraction correction by Lortie et al for 82Rb. Comparisons included 1) correlation of absolute MBF for 82Rb versus 15O-water, 2) standard clinical practice for 82Rb using assessment of static, gated and dynamic scans by 2 experienced physicians (consensus), 3) in-house established cut-offs for normal 82Rb stress MBF (>2.0 mL/min/g) (1) and 4) flow reserve (CFR>2). For 15O-water, a previously established cut-off (2.3 mL/min/g) (2) for stress MBF validated against invasive FFR was used. Results Mean global MBF (mL/min/g) at rest was significantly higher for 82Rb (82Rb: 1.04 [range 0.59-1.9] vs 15O-water: 0.88 [0.54-1.37], p<0.001), whereas at stress the difference was insignificant (2.33 [0.79-3.97] vs 2.43 [0.71-4.33], p=0.27). Linear correlation between MBFs was good both at a global level (R2=0.78) and on per vessel level (R2=0.76). For CFR the linear correlation was less clear (R2=0.38). Analysis of 82Rb K1 vs 15O-water MBF suggests a lower extraction of 82Rb at higher flow compared to Lortie et al. 15O-water PET identified IHD in 17 patients and 49 vessels. In the standard 82Rb clinical assessment using all available information, 82Rb on the patient level was false positive in seven and false negative in zero, yielding a sensitivity of 100%, specificity 64%, PPV 68 % and NPV 100%. Comparing stress MBF, 82Rb on the patient-level was false positive in one and false negative in five, yielding a sensitivity of 71%, specificity 95%, PPV 92% and NPV 81%. On the vessel-level there were six false positive and 24 false negative, yielding a sensitivity of 51%, specificity 91%, PPV 81% and NPV 72%. Global CFR (patient-level) identified five false positive and 10 false negative with a sensitivity of 41%, specificity 77%, PPV 58% and NPV 63%. Conclusion Standard clinical assessment of 82Rb with compiled information from relative distribution in late images, gated images and absolute MBF provided a higher sensitivity and NPV than stress MBF alone. However, absolute stress MBF provided the highest specificity and PPV, suggesting that the use of stress MBF in the diagnostic algorithm could improve the standard clinical assessment. CFR performance was inferior to all other measures. The extraction correction implemented by Lortie et al overestimates resting MBF in patients with IHD. The low extraction of 82Rb results in large uncertainties on especially stress MBF and care must be taken when utilizing absolute MBF. (1) CM. Hoff et al. Eur Heart J Cardiovasc Imaging. 2015 May;16 Suppl 1:i38-i55 (2) Danad I et al. Coll Cardiol. 2014 Oct 7;64(14):1464-75