Simplified Quantification of Pittsburgh Compound B Amyloid Imaging PET Studies: A Comparative Analysis

Simplified Quantification of Pittsburgh Compound B Amyloid Imaging PET Studies: A Comparative Analysis

Brian J. Lopresti, BS¹, William E. Klunk, MD, PhD², Chester A. Mathis, PhD¹, Jessica A. Hoge, BS¹, Scott K. Ziolko, BS¹, Xueling Lu, MS¹, Carolyn C. Meltzer, MD¹^,2^,3, Kurt Schimmel, BS¹, Nicholas D. Tsopelas, MD², Steven T. DeKosky, MD³ and Julie C. Price, PhD¹

¹ Department of Radiology, University of Pittsburgh School of Medicine, Pittsburgh, Pennsylvania
² Department of Psychiatry, University of Pittsburgh School of Medicine, Pittsburgh, Pennsylvania
³ Department of Neurology, University of Pittsburgh School of Medicine, Pittsburgh, Pennsylvania

View larger version (88K):

[in a new window]

FIGURE 1. Representative VOIs defined on an SPGR MR image of an AD subject (A-3). Shown in white are enhanced examples of cortical ROIs examined in this study: ANC = anterior cingulate; FRT = frontal; LTC = lateral temporal; MTC = mesial temporal; OCC occipital; PAR = parietal; PCG = posterior cingulate. Also shown is CER (cerebellum reference region) and a carotid ROI (CAR) defined bilaterally over internal carotid artery on an early summed PIB PET scan (initial 45 s of acquisition).

View larger version (15K):

[in a new window]

FIGURE 2. (A) Population average PIB unchanged fraction in plasma determined in 16 subjects with complete metabolite data. (B) Average (n = 24) PIB input functions normalized for ID and body mass (%ID x kg/g) derived from both external arterial sampling and a carotid VOI. Population average PIB unchanged fraction was used to correct the carotid time–activity curve for metabolism, whereas individual data were applied to total plasma radioactivity measurements to perform metabolite correction of arterial input function.

View larger version (21K):

[in a new window]

FIGURE 3. Average (±1 SD) brain radioactivity concentrations normalized for ID and body mass (%ID x kg/g) after injection of PIB. Shown are PCG (A) and CER (B) regions for AD (n = 6) and control (n = 8) subjects as well as PCG-to-CER radioactivity concentration ratio (C).

View larger version (32K):

[in a new window]

FIGURE 4. Shown are outcome measures for individual AD (n = 6; red), MCI (n = 10; green), and control (n = 8; blue) subjects across all methods of analysis for PCG (A) and FRC (B). Outcome measure represented for all methods is the DVR, with the exception of SUVR90 and SUVR60 methods, for which tissue-to-CER ratio over 40–60 min or 40–90 min is shown. Numbered circles represent individual subjects (Table 1), whereas colored bars denote range of values within the group. Subjects with overlapping values are placed adjacent to one another.

View larger version (80K):

[in a new window]

FIGURE 5. Parametric images of Logan DVR using 90 min of emission data and either arterial data (ART90; top) or CER tissue (CER90; bottom) as input. Shown are a young control (C-4), a control with detectable amyloid deposition in FRC (C-2), an amyloid-negative MCI subject (M-2), an MCI subject with intermediate levels of PIB retention (M-10), an amyloid-positive MCI subject with levels of PIB retention characteristic of AD (M-4), and a typical AD subject (A-2).

View larger version (21K):

[in a new window]

FIGURE 6. Bias and correlation measures of various simplified methods with ART90. (A) Box plot shows percentage bias in simplified outcome measures relative to ART90 in PCG. Subjects were divided into high-binding (ART90 PCG DVR > 1.8) and low-binding (ART90 PCG DVR < 1.8) groups to determine whether methodologic bias was consistent across the spectrum of PIB retention for all simplified analysis methods. Boxes denote the interquartile range (50% of subjects) and median value (solid line), whereas box whiskers indicate 10th and 90th percentile. ${circ}$ , Individual subject values. (B) Slopes of linear correlations between ART90 and simplified methods. (C) Coefficient of determination (r²) for correlations between ART90 and simplified methods.

View larger version (22K):

[in a new window]

FIGURE 7. (A) Graph shows correlation of ART90 and CAR90 ( ${circ}$ , solid line) and SUVR90 and CER90 (•, solid line) outcome measures. (B) Graph shows correlation between ART90 and CER60 ( ${blacksquare}$ , solid line) and ART90 and CER60 ( ${square}$ , solid line) outcome measures. The equation describing the linear regression is shown for each comparison in the form y = mx + b as well as the coefficient of determination (r²). The thin dashed line in both graphs represents the line of unity.