Comparison of plasma Ab42, Ab40, and p‐tau181 assessed on two different platforms with CSF AD biomarkers and amyloid (PiB) PET

Abstract

AbstractBackgroundAs blood‐based biomarkers of Alzheimer’s disease enter clinical diagnostic use, key considerations remain including the concordance with CSF or amyloid PET and the comparability of results between different assay platforms.MethodParticipants (n = 42) from the Clinical Core of the Wake Forest ADRC, were selected using CSF Aβ42/Aβ40 values which clearly indicated amyloid pathology status (‐ or +). A subset of 29 participants had amyloid (PiB) PET measures within one year of CSF. Participants were considered amyloid positive when CSF Aβ42/40 ≤0.058 and/or amyloid PET was adjudicated as “positive”. CSF and plasma Aβ42, Aβ40, and pTau181 were assessed using the Lumipulse platform at the Wake Forest ADRC; plasma Aβ42, Aβ40, and pTau181 were also assessed using the Quanterix SIMOA platform at the National Centralized Repository for Alzheimer’s disease. CSF and plasma from each participant were collected on the same day.ResultAmyloid positive and negative groups differed significantly in age and APOE‐ε4 status (Table 1). Although raw values between plasma platforms differed significantly, biomarkers were highly correlated (Table 2). The correlation between CSF and plasma was: Lumipulse for Aβ42/Aβ40 (ρS = 0.76, p<0.0001) and pTau181 (ρS = 0.64, p<0.0001) and SIMOA Aβ42/Aβ40 (ρS = 0.63, p<0.0001) and pTau181 (ρS = 0.50, p = 0.0009). When comparing groups positive or negative for either CSF or PET, the ROC AUC for plasma Aβ42/Aβ40 was 0.94 for Lumipulse and 0.92 for SIMOA. Ptau181 AUC was 0.80 for Lumipulse and 0.82 for SIMOA. Figure 1 shows a clear relationship between decreasing Aβ42/Aβ40 and increasing pTau181 in the CSF which is present but diminished in Lumipulse plasma and largely absent in SIMOA plasma data.ConclusionRaw values from both platforms differed significantly which may limit generalizability and translatability of plasma data across platforms, but results were correlated. CSF values correlated marginally with both Lumipulse and SIMOA plasma. Plasma indices performed better than previously reported in other AD cohorts, likely due to our stringent CSF‐based selection criteria, and AUC did not differ between Lumipulse and SIMOA for either Aβ42/Aβ40 or pTau181. Further work is underway to expand the dataset and define population‐based cutoffs which also consider comorbidities, such as chronic kidney disease.