Abstract
Arterial spin labelling (ASL) enables non-invasive quantification of regional brain perfusion using MRI. ASL was used in the Reducing Pathology in Alzheimer's Disease through Angiotensin TaRgeting (RADAR) multi-centre trial to pilot the assessment of the effects of the anti-hypertension drug losartan on cerebral blood flow (CBF). In the multi-centre setting, disparities in ASL implementation on scanners from different manufacturers lead to inherent differences in measured CBF and its associated parameters (e.g. spatial coefficient of variation (sCoV) of CBF, a proxy of arterial arrival times). In addition, differences in ASL acquisition parameter settings can also affect the measured quantitative perfusion values. In this study, we used data from the RADAR cohort as a case study to evaluate the site-dependent systematic differences of CBF and sCoV, and show that variations in the readout module (2D or 3D) and the post-labelling delay acquisition parameter introduced artifactual group differences. When accounting for this effect in data analysis, we show that it is still possible to combine ASL data across sites to observe the expected relationships between grey matter CBF and cognitive scores. In summary, ASL can provide useful information relating to CBF difference in multi-centre therapeutic trials, but care must be taken in data analysis to account for the inevitable inter-site differences in scanner type and acquisition protocol.