Voxel-wise correlation of binding potential of Pittsburgh compound B to CBF in older subjects as revealed by joint frequency histograms

Background and Aims: Although amyloid deposition is negatively correlated with regional glucose metabolism in patients with dementia, little is known about associations between amyloid deposition and functional activity in pre-morbid stages. We hypothesized that amyloid accumulation is proportional to functional demand in normal aging until exceeding a threshold. Accordingly, we investigated the correlation between binding potential (BP) of PIB, an estimate of amyloid deposition, and resting CBF, an estimate of functional demand, using joint frequency histogram analysis of gray matter voxels. Methods: Thirty-two participants of the Baltimore Longitudinal Study of Aging (age: 80.3 +/-6.2) had one [11C]PIB PET scan and one static [15O]water scan. Participants had normal cognition (n=26) or mild cognitive impairment by clinical dementia rating of 0.5 (n=6). BP maps of PIB scans were constructed by the bolus-plus-infusion transformation (BPIT; Kuwabara et al., 2002). CBF images were normalized (nCBF) to the cerebellum value (50 ml/mg/min). BP and CBF maps were coregistered by SPM2 and smoothed with a Gaussian kernel (8 mm FWHM). Gray voxels defined on MRI were transferred to PET space by SPM2. PIB-BP (cutoff: -0.7 and 1.5; 100 bins; x-axis) versus CBF (cutoff: 30 and 80 ml/mg/min; 100 bins; y-axis) joint frequency histograms were constructed (voxel value = number of BP-CBF voxels within the histogram voxel limits). Results: The figure shows typical joint histograms. On visual inspection, one group of subjects displayed one long cluster only (Type T for 'tight' correlation; n=20) while another showed a broader distribution (Type D for 'diffuse' correlation; n=12). Some Type T subjects displayed a small 'hub' of negative correlation at the highest end of PIB-BP (Type T1; n=12) while others did not (Type T0; n=8). Voxel-wise regression analysis confirmed positive slopes in all cases (range 9.5 ? 42.3; F>8120; p<0.001). Individuals with mild cognitive impairment showed a significant tendency toward the diffuse pattern of correlation (p < 0.05). Conclusion: We demonstrate a positive voxel-wise correlation of PIB-BP with resting CBF in normal and mildly impaired older subjects except for a small 'hub' of negative correlation in some individuals. The positive correlation suggests the possibility that amyloid accumulation is greater in functionally more active brain tissue, reflecting a compensatory increase in CBF proportional to amyloid deposition. The tendency for cognitively impaired individuals to show more diffuse associations suggests a failure of these compensatory processes. The clinical significance of these patterns of association will be determined through continued longitudinal follow-up.