The impact of inspired oxygen levels on calibrated fMRI measurements of M, OEF and resting CMRO(2) using combined hypercapnia and hyperoxia

Recent calibrated fMRI techniques using combined hypercapnia and hyperoxia allow the mapping of resting cerebral metabolic rate of oxygen (CMRO(2)) in absolute units, oxygen extraction fraction (OEF) and calibration parameter M (maximum BOLD). The adoption of such technique necessitates knowledge about the precision and accuracy of the model-derived parameters. One of the factors that may impact the precision and accuracy is the level of oxygen provided during periods of hyperoxia (HO). A high level of oxygen may bring the BOLD responses closer to the maximum M value, and hence reduce the error associated with the M interpolation. However, an increased concentration of paramagnetic oxygen in the inhaled air may result in a larger susceptibility area around the frontal sinuses and nasal cavity. Additionally, a higher O(2) level may generate a larger arterial blood T(1) shortening, which require a bigger cerebral blood flow (CBF) T(1) correction. To evaluate the impact of inspired oxygen levels on M, OEF and CMRO(2) estimates, a cohort of six healthy adults underwent two different protocols: one where 60% of O(2) was administered during HO (low HO or LHO) and one where 100% O(2) was administered (high HO or HHO). The QUantitative O2 (QUO2) MRI approach was employed, where CBF and R2* are simultaneously acquired during periods of hypercapnia (HC) and hyperoxia, using a clinical 3 T scanner. Scan sessions were repeated to assess repeatability of results at the different O(2) levels. Our T(1) values during periods of hyperoxia were estimated based on an empirical ex-vivo relationship between T(1) and the arterial partial pressure of O(2). As expected, our T(1) estimates revealed a larger T(1) shortening in arterial blood when administering 100% O(2) relative to 60% O(2) (T(1LHO) = 1.56±0.01 sec vs. T(1HHO) = 1.47±0.01 sec, P < 4*10(−13)). In regard to the susceptibility artifacts, the patterns and number of affected voxels were comparable irrespective of the O(2) concentration. Finally, the model-derived estimates were consistent regardless of the HO levels, indicating that the different effects are adequately accounted for within the model.