Peak Location of Amyloid PET Tracer Uptake within Cortical Gray Matter: Topographic Patterns and Diagnostic Application in Alzheimer's Disease

Objectives This study is to investigate topographical analysis of peak location of cortical laminar deposition of amyloid tracer uptake within gray matter and its potential diagnostic application in Alzheimer's disease.

Methods Hypothesizing that progressive amyloid deposition would encompass a greater extent of cerebral cortical laminae, the distance between the peak gray matter uptake of amyloid tracer and the gray-white matter interface (determined by non-specific white matter uptake, normal atlas based) was estimated along the line perpendicular to the brain surface at each cortical pixel using the three-dimensional surface projections (3D-SSP) algorithm following spatial normalization of each amyloid PET scan. The estimated values were assigned to stereotactically pre-defined cortical surface pixels covering the entire cerebral cortex, constituting an absolute millimetric map of peak location of cortical laminar deposition (PCLD) of the amyloid tracer (3D-SSP-PCLD). Using [F-18]florbetapir PET data from the ADNI (165 normal controls, age 75±6.9 years old, 78 female and 148 AD patients, age 75±8.2 years old, 62 female), an image-based database of 3D-SSP-PCLD was constructed from the normal controls, and topographical distribution of PCLD was examined individually and for the entire AD group using 3D-SSP-PCLD Z-score maps in comparison to the database. We assessed the discrimination between normal controls vs AD patients by comparing 3D-SSP-PCLD and Standard Uptake Value Ratio (SUVR) analysis (3D-SSP-SUVR, pons reference).

Results In AD patients, the peak location of cortical laminar deposition of the amyloid tracer from the gray-white matter interface was estimated at 1.74±1.37mm (mean±sd) over the entire cerebral cortex, with regional values ranging from 1.58±2.29mm; 1.93±1.77mm; 2.18±2.31mm; and 4.20±2.81mm in the occipital, frontal, parietal, and temporal association cortices, respectively. Corresponding regional SUVR %increases in AD were 28%; 31%; 37%; and 36%, with the average increase of 30% in the entire cortex. Individual and group Z-score maps of 3D-SSP-PCLD and 3D-SSP-SUVR showed grossly concordant patterns of regional changes in AD patients, most conspicuously involving temporo-parieto-frontal association cortices and the posterior cingulate cortex, but also some regional discordance such as seen in the medial frontal cortex. The discrimination efficacy of normal controls vs AD patients was Z=12.2 (U=2462) using 3D-SSP-PCLD and Z=12.0 (U=2584) using 3D-SSP-SUVR (Mann-Whitney test using individual ranking). Receiver Operating Characteristics (ROC) showed the Area-under-the-ROC-Curve of 0.90±0.018 (mean±se) using 3D-SSP-PCLD and 0.91±0.018 3D-SSP-SUVR, indicating a nearly identical discriminatory power for normal controls vs AD patients.

Conclusions The regional pattern of topographical peak location of amyloid tracer uptake within the gray matter in AD was similar to the pattern shown with intensity-based SUVR assessment. Positive amyloid deposition places the peak location of gray matter uptake into cortical laminae distant from the underlying white matter. This analysis allows the construction of image-based normal database and the Z-score analysis of individual patients as well as groups similar to SUVR. 3D-SSP-PCLD, the absolute quantitative map of geometric location of peak cortical laminar deposition of the tracer, is an objective measure of amyloid deposition with potential diagnostic utility that does not require a reference region thus avoids a source of possible measurement variability inherent with SUVR.