Introduction
Traditionally, CBF, OEF and CMRO2 are obtained using the oxygen-15 steady state technique and PET. This technique requires 3 scans (H2O, CO and O2) during continuous administration of tracers. The purpose of the present study was to investigate the feasibility of deriving accurate CBF, OEF and CMRO2 data from a single dynamic PET scan using a short bolus inhalation of O2.
Methods
Previously, a simulation study was performed showing that accurate CBF, OEF and CMRO2 values could be obtained from a single (bolus inhalation) O2 scan, but only by fixing the volume of distribution of water (Vd) parameter 1 . Therefore, these simulations were extended to investigate the effects of using incorrect values of Vd on the accuracy and precision of CBF, OEF and CMRO2. Moreover, this approach was validated using ten clinical studies with dynamic H2O, CO and O2 scans and continuous arterial blood sampling. Clinical data were analysed in two ways. First, CBF, OEF and CMRO were calculated using data from all three scans, i.e. OEF and CMRO2 were derived from the oxygen scan by reusing CBV, CBF and Vd from the CO and H2O scans. Secondly, data from the O2 scan alone were used in combination with a co-registered and segmented MRI scan. The MRI scan was used to assign Vd values to grey and white matter voxels. To this end, a whole brain time activity curve of the O2 scan was fitted without any fixed parameter, thereby estimating global Vd per patient. This global value was subsequently used to scale the gray and white matter Vd values based on the MRI data.
Results
Simulations showed that errors of ∼10% in Vd value may lead to a bias of ∼20% in CBF and OEF (but only ∼3% in CMRO2), indicating that using an average (population) Vd may not be feasible in clinical practice and that individual patient values should be assigned. Clinical evaluation revealed reasonable correlations between CBF, OEF and CMRO2 using both methods. Linear regression showed average (±SD) slopes of 1.07±0.07, 1.04±0.10 and 1.02±0.05 and average correlation coefficients of 0.88, 0.64 and 0.99 for CBF, OEF and CMRO2, respectively, after discarding outliers. However, large bias (>50%) and/or outliers were obtained in case of poor oxygen gas delivery and/or small amounts of C-11 contamination of the oxygen gas (∼0.5% or more), which occurred in 50% of the studies.
Conclusions
Calculation of CBF, OEF and CMRO2 from a single dynamic O2 scan in combination with a co-registered and segmented MRI scan is feasible. This method is, however, very sensitive to poor gas delivery and to small amounts of C-11 contamination of the oxygen-15 tracers. Therefore, it should only be used when online assessment of the quality of the acquired data is available. Moreover, both simulations (data not shown) and clinical data suggest that CBF, OEF and CMRO2 derived from a single O2 scan show somewhat poorer reproducibility than those derived by reusing CBF and Vd from an H2O scan.