To examine the utility of Arterial Spin Labeling (ASL) Cerebral Blood Flow (CBF) in assessing and predicting clinical progression of Alzheimer’s disease.

Alzheimer’s disease (AD) is characterized by progressive cognitive impairment and development of biomarkers for diagnosis, prognosis and accurate tracking of the disease is of significant importance.¹ ASL² provides a non-invasive technique for measuring regional CBF, which is tightly coupled to regional neural activity and is an effective biomarker for several cerebral disorders.³ CBF measured by ASL has been used to differentiate controls and AD subjects in a number of studies.⁴ Recently ASL was included as a substudy of the Alzheimer’s Disease Neuroimaging Initiative (ADNI) (http://adni.loni.usc.edu), an ongoing, longitudinal, multicenter study directed towards development of enhanced biomarkers for AD. Here, we used ADNI ASL data to study the change in regional CBF with cognition and disease progression as measured by clinical dementia rating scale sum of boxes (CDR-SB), a numerical scale to quantify cognitive and functional dysfunction in individuals with mild cognitive impairment and dementia.⁵

ADNI data from baseline, Year 1, and Year 2 were considered. Subjects were classified as Controls, Early Mild Cognitive Impairment (EMCI), Late MCI (LMCI) and AD. Only subjects (i) who had baseline CDR and ASL data within 3 months of each other and (ii) who showed evidence of cerebral amyloid as measured by amyloid PET imaging were considered, as this cohort is most susceptible to develop Alzheimer’s disease.¹ The number of subjects for baseline, Year 1 and 2 for different diseased groups was as follows: Control (20, 19, 15), EMCI (41, 37, 29), LMCI (41, 40, 29) and AD (34, 34, 9). ADNI ASL data were acquired using the Siemens product PICORE pulsed ASL (PASL) sequence using TR/TE=3400/12 ms, TI1/TI=700/1900 ms (other details of acquisition parameters can be found in ⁶). Each EPI time series was first motion corrected, and then the CBF maps were estimated using pairwise subtraction, dividing by the M0 image and following the model in ⁷. Mean CBF maps were computed first by applying an outlier rejection of CBF time series,⁸ and then applying a Bayesian Robust Regression method to the remaining CBF volumes. The mean CBF within precuneus, posterior cingulate cortex (PCC) and hippocampus, which have previously been demonstrated to be sensitive to AD-related changes,⁴ were considered. Three different analyses were performed. First, the mean CBFs within each ROI were compared with CDR-SB of all the subjects for different time points. Second, for individual subjects, the difference of CDR-SB between baseline and Year 2 (or Year 1 if Year 2 was unavailable), denoted as $$$\Delta$$$CDR-SB, and the same for mean CBF ($$$\Delta$$$CBF) in ROIs were computed and correlated with each other. Finally, for ROI(s) showing significant correlations in the previous two analyses, subjects were divided into two groups; those whose cognition deteriorated in two years and those who didn't ($$$\Delta$$$CDR-SB$$$>0$$$ vs $$$\Delta$$$CDR-SB$$$\leq 0$$$). Mean CBF values in these ROI were compared at baseline to see if CBF at baseline predicted disease progression. This analysis was performed separately for each group and also for the combined incipient groups.Correlations between mean CBF in precuneus, PCC and hippocampus and CDR-SB were $$$-0.23$$$, $$$-0.29$$$ and $$$-0.27$$$ respectively (all statistically significant, $$$p<0.0001$$$) demonstrating that CBF correlates with CDR-SB in general. On the other hand, $$$\Delta$$$ CDR-SB was statistically significantly correlated with $$$\Delta$$$CBF only for PCC $$$(r=-0.24, p=0.006)$$$. Figure 1 shows scatter plots for CBF in PCC vs CDR and $$$\Delta$$$CBF in PCC vs $$$\Delta$$$CDR. Figure 2 shows mean CBF in PCC (with standard errors) at baseline for controls, EMCI, LMCI and AD divided into groups based on whether their cognitive performance deteriorated in the future. For all the incipient groups, mean CBF was lower in subjects whose cognition deteriorated in the future, although the difference was not statistically significant because of small number of subjects in each group. However when all the progressors were combined and compared against all clinically stable individuals, the difference became statistically significant. Hence, CBF may be a potential predictor of disease progression at different stages of incipient AD. A similar result was shown in ⁹ in Controls. Progression in mean CDR and CBF in PCC with time for the different groups are shown in Figure 3. Increases in CDR over time parallels decreases in CBF. In the EMCI group, both CDR and PCC CBF show a subtle reverse trend, potentially indicating a compensatory response to neurodegeneration in this phase as has previously been suggested.¹⁰