Reproducibility of pharmacological ASL using sequences from different vendors: implications for multicenter drug studies

The current study assesses the multicenter feasibility of pharmacological arterial spin labeling (ASL) by comparing a caffeine-induced relative cerebral blood flow decrease (%CBFa dagger") measured with two pseudo-continuous ASL sequences as provided by two major vendors. Twenty-two healthy volunteers were scanned twice with both a 3D spiral (GE) and a 2D EPI (Philips) sequence. The inter-session reproducibility was evaluated by comparisons of the mean and within-subject coefficient of variability (wsCV) of the %CBFa dagger", both for the total cerebral gray matter and on a voxel level. The %CBFa dagger" was larger when measured with the 3D spiral sequence (23.9 +/- A 5.9 %) than when measured with the 2D EPI sequence (19.2 +/- A 5.6 %) on a total gray matter level (p = 0.02), and on a voxel level in the posterior watershed area (p < 0.001). There was no difference between the gray matter wsCV of the 3D spiral (57.3 %) and 2D EPI sequence (66.7 %, p = 0.3), whereas on a voxel level, the wsCV was visibly different between the sequences. The observed differences between ASL sequences of both vendors can be explained by differences in the employed readout modules. These differences may seriously hamper multicenter pharmacological ASL, which strongly encourages standardization of ASL implementations.