Quantification of myocardial blood flow with 82Rb positron emission tomography: clinical validation with 15O-water

Abstract

Purpose: Quantification of myocardial blood flow (MBF) with generator-produced (82)Rb is an attractive alternative for centres without an on-site cyclotron. Our aim was to validate (82)Rb-measured MBF in relation to that measured using (15)O-water, as a tracer 100% of which can be extracted from the circulation even at high flow rates, in healthy control subject and patients with mild coronary artery disease (CAD).

Methods: MBF was measured at rest and during adenosine-induced hyperaemia with (82)Rb and (15)O-water PET in 33 participants (22 control subjects, aged 30 ± 13 years; 11 CAD patients without transmural infarction, aged 60 ± 13 years). A one-tissue compartment (82)Rb model with ventricular spillover correction was used. The (82)Rb flow-dependent extraction rate was derived from (15)O-water measurements in a subset of 11 control subjects. Myocardial flow reserve (MFR) was defined as the hyperaemic/rest MBF. Pearson's correlation r, Bland-Altman 95% limits of agreement (LoA), and Lin's concordance correlation ρ (c) (measuring both precision and accuracy) were used.

Results: Over the entire MBF range (0.66-4.7 ml/min/g), concordance was excellent for MBF (r = 0.90, [(82)Rb-(15)O-water] mean difference ± SD = 0.04 ± 0.66 ml/min/g, LoA = -1.26 to 1.33 ml/min/g, ρ(c) = 0.88) and MFR (range 1.79-5.81, r = 0.83, mean difference = 0.14 ± 0.58, LoA = -0.99 to 1.28, ρ(c) = 0.82). Hyperaemic MBF was reduced in CAD patients compared with the subset of 11 control subjects (2.53 ± 0.74 vs. 3.62 ± 0.68 ml/min/g, p = 0.002, for (15)O-water; 2.53 ± 1.01 vs. 3.82 ± 1.21 ml/min/g, p = 0.013, for (82)Rb) and this was paralleled by a lower MFR (2.65 ± 0.62 vs. 3.79 ± 0.98, p = 0.004, for (15)O-water; 2.85 ± 0.91 vs. 3.88 ± 0.91, p = 0.012, for (82)Rb). Myocardial perfusion was homogeneous in 1,114 of 1,122 segments (99.3%) and there were no differences in MBF among the coronary artery territories (p > 0.31).

Conclusion: Quantification of MBF with (82)Rb with a newly derived correction for the nonlinear extraction function was validated against MBF measured using (15)O-water in control subjects and patients with mild CAD, where it was found to be accurate at high flow rates. (82)Rb-derived MBF estimates seem robust for clinical research, advancing a step further towards its implementation in clinical routine.

Fig. 1

Scatter plot of ⁸²Rb K ₁ rate constant vs. ¹⁵O-water MBF measurements and the fitted generalized Renkin-Crone function K ₁ = MBF × (1 − a × e^−b/MBF) (solid line) derived from the first group of 11 control subjects allowing the a and b parameters to be estimated (a = 0.80, b = 0.59, R ² = 0.97, RMSE = 0.145) to convert K ₁ rate constant estimates to ⁸²Rb MBF (dotted lines ±95% CI)

Fig. 2

a Scatter plot shows concordance between MBF measurements by ⁸²Rb and ¹⁵O-water in the second group of 11 control subjects and in a group of 11 CAD patients with a reduced major axis close to the line of identity. b Corresponding Bland-Altman plot

Fig. 3

a Scatter plot shows concordance between MFR by ⁸²Rb and ¹⁵O-water in 11 control subjects and CAD 11 patients with a reduced major axis very close to the line of identity. b Corresponding Bland-Altman plot

Fig. 4

Plot of the function relating extraction (E) for ⁸²Rb to flow estimates (F) derived from the present study in control subjects using ¹⁵O-water in comparison to previous studies using ¹³N-ammonia PET (Lortie et al. [14]) or the argon inert gas method (computed by Schelbert [39] based on a study by Glatting et al. [44])