MEASURING CHANGE IN BETA-AMYLOID BURDEN OVER TIME USING FLORBETAPIR-PET AND A SUBCORTICAL WHITE MATTER REFERENCE REGION

Abstract

Background: Estimation of longitudinal change in beta amyloid (Ab) burden may be important to assess disease progression or response to therapy. Using ADNI florbetapir data we evaluated whether measurement of longitudinal amyloid accumulation in Ab positive subjects can be improved with an alternative reference region. Methods: 187 subjects (61 CN, 83 early MCI, 42 late MCI and 1 AD) from ADNI who underwent an initial florbetapir imaging session followed by a second scan at approximately 2 years were analyzed. PET images were spatially normalized to MNI space using a florbetapir PET template. Mean cortical SUVR (mcSUVr) values were calculated as the ratio of the average of 6 cortical regions (medial orbital frontal, parietal, temporal, anterior cingulate, posterior cingulate and precuneus) with respect to whole cerebellum. A pre-specified threshold of mcSUVr>1.10 was used to identify Ab+ subjects. An alternative cortical SUVR was determined using an atlas-based sub-cortical white matter area (centrum semiovale) as a reference region (mcSUVr wm). Change as well as correlation in SUVr (from baseline to 2 years) was examined in Ab+ cases for both mcSUVr and mcSUVr wm. Results: Relative to mcSUVr, the mean increase from baseline SUVr in all Ab+ cases(N1⁄4 84) was slightly reduced with the white matter reference region, and the variability (as measured by the standard deviation, or SD, of the change) was markedly reduced by comparison to the cerebellum reference region (DmcSUVr 1⁄4 2.7 6 7.4%, D mcSUVr wm 2.3 6 4.2%,). As a result, the signal to noise ratio (Dmean/SD) for detecting an increase in SUVr increased from 0.34 to 0.53. The correlation between baseline and endpoint SUVr also increased with the use of a white matter reference region (r 21⁄4 0.73 for mcSUVr; r 21⁄4 0.90 for mcSUVr wm). Conclusions:While limited by a lack of truth standard, the analyses show the white matter reference region was associated with less unexplained variance (27% for mcSUVr ; 10% for mcSUVr wm) and increased signal-to-noise ratio (0.34 vs 0.53), potentially providing increased power to detect slowing of amyloid accumulation in therapeutic trials.