HAI Book 2025 - Flipbook - Page 533
Vrillon, Agathe
117
Concordance between visual and quantitative amyloid PET and CSF
biomarkers in the ADNI-DOD cohort
Agathe Vrillon1,6, Ganna Blazhenets1, Julien Lagarde1, Konstantinos Chiotis1, Maison Abu
Raya1,6, David Soleimani-Meigooni1, Gil Rabinovici1, Michael Weiner1, Lisa Silbert2, Nathan
Hantke2, Daniel Schwartz2, Abigail Livny Ezer3, Orit Lesman Segev3, Kristine Yaffe1, Leslie
Shaw4, Susan Landau5, Renaud La Joie1,6, Raquel Gardner1,3
1
University of California San Francisco, San Francisco, CA, US
Oregon Health & Science University, Portland, OR, US
3
Sheba Medical Center, Tel-Hashomer, Ramat Gan, IL
4
University of Pennsylvania, Philadelphia, PA, US
5
University of California Berkeley, Berkeley, CA, US
6
Global Brain Health Institute, San Francisco, CA, US
2
Background: Clinical factors, including comorbidities, may affect visual interpretation of amyloid PET. We
evaluated the agreement between PET visual read, centiloid (CL) quantification and CSF AD biomarkers in the
ADNI-DOD cohort enrolling US Veterans with a high prevalence of traumatic brain injury (TBI) and post-traumatic
stress disorder (PTSD).
Methods: ADNI-DOD participants who underwent [18F]Florbetapir amyloid PET were included (Figure 1). Visual
reads (VR) were performed by one of five trained neurologists blinded to quantification. CL quantification used an
MRI-free pipeline (Landau et al., positivity threshold: CL g27). Visually and quantitatively discordant and/or
challenging scans were reviewed in consensus meetings to achieve a final read. CSF AD biomarkers measured
using Elecsys (Roche®) were examined for a subsample of n=140 individuals. Agreement was evaluated using
Cohen's kappa (κ).
Results: We included 274 Veterans (mean age: 70.1 years; 99% male; 84% cognitively unimpaired [CU]; 16% MCI;
60% PTSD, 46% TBI). Overall, 36% (n=100/274) of scans were initially read as positive and 24% (n=67/274) were
quantitatively positive (Figure 2). Initial visual read and quantification were concordant for 86% of scans (κ=0.67).
Agreement tended to be lower in CU participants (85% agreement, κ=0.63, P=0.053) than in MCI (91% agreement,
κ=0.82) and lower in the PTSD group than in the TBI group (P=0.036). Discordance (14%) was principally
constituted of VR-positive/quantitatively-negative scans (92% [n=36/39 of overall disagreement]). VRpositive/quantitatively-negative scans were characterized mostly by focal tracer uptake or posterior patterns not
captured by CL quantification (Figure 3a). Among CSF biomarkers, the CSF Ab42/40 ratio showed the highest
concordance with PET, with both final VR (84% agreement, κ=0.56) and quantification (85% agreement, κ=0.52)
(Figure 3).
Conclusion: We observed 14% of visually/quantitatively discordant scans, mostly visually positive with low CL,
more common in CU than in MCI participants. PET visual read and quantification had equivalent concordance with
CSF-based A status.
HAI2025 - 533
