Biofluid-based staging of Alzheimer's disease

Abstract

Recently, conceptual systems for the in vivo staging of Alzheimer's disease (AD) using fluid biomarkers have been suggested. Thus, it is important to assess whether available fluid biomarkers can successfully stage AD into clinically and biologically relevant categories. In the TRIAD cohort, we explored whether p-tau217, p-tau205 and NTA-tau (biomarkers of early, intermediate and late AD pathology, respectively) have potential for biofluid-based staging in cerebrospinal fluid (CSF; n = 219) and plasma (n = 150), and compared them in a paired CSF and plasma subset (n = 76). Our findings suggest a good concordance between biofluid staging and underlying pathology when classifying amyloid-positivity into three categories based on neurofibrillary pathology: minimal/non-existent (p-tau217 positive), early-to-intermediate (p-tau217 and p-tau205 positivity), and advanced tau tangle deposition (p-tau217, p-tau205 and NTA-tau positive), as indexed by tau-PET. Discordant cases accounted for 4.6% and 13.3% of all CSF and plasma measurements respectively (9.2% and 11.8% in paired samples). Notably, CSF- and plasma-based staging matched one another in 61.7% of the cases, while approximately 32% of the remaining participants were one to three biofluid stages higher in CSF as compared to plasma. Overall, these exploratory results suggest that biofluid staging of AD holds potential for offering valuable insights into underlying AD hallmarks and disease severity. However, its applicability beyond molecular characterization at research settings has yet to be demonstrated.

Keywords: Alzheimer; CSF; Plasma; Staging; Tau; p-Tau.

Fig. 1

CSF tau biomarker trajectories in the TRIAD cohort across (A) medial temporal and (B) neocortical tau-PET SUVRs using a local weighted regression method (Loess curve). Changes in CSF biomarker levels are represented as Z-scores using tau-PET SUVRs as a proxy of AD pathology progression. Abnormal biomarker levels are determined as two standard deviations (SD) above the mean. Biomarker thresholds for positivity are indicated by the dashed lines

Fig. 2

Biofluid-based staging across clinically relevant stratifications in non-paired CSF and plasma samples. Stacked bar charts represent the percentages of fluid stage-0, stage-1, stage-2 and stage-3 participants. In blue, CSF staging across (A) diagnostic groups, (B) AT groups, (C) tau-PET defined Braak stages, and (D) tau-PET status. In pink, plasma staging across (E) diagnostic groups, (F) AT groups, (G) tau-PET defined Braak stages, and (H) tau-PET status. Discordant cases with the biofluid staging criteria are shown in grey. Table shows the percentages and total number of individuals in each group

Fig. 3

Average tau-PET uptake across the brain for each of the (A) CSF-based stages and (B) plasma-based stages. The boxplots depict the median (horizontal bar), 25th to 75th percentiles (hinges) and whiskers indicate 10th and 90th percentiles. Linear regression models contrasted a higher stage to its preceding stage, adjusting for age and sex. The * indicates significantly different tau-PET uptake in medial temporal tau-PET SUVR in (C) CSF- and (D) plasma-based stages, followed by similar comparisons in (E–F) neocortical tau-PET SUVR

Fig. 4

Agreement between CSF- and plasma-based stages in the matched dataset. The mosaic plot shows in (A) the proportions of individuals classified at each biofluid-based stage. The stacked bar chart (B) shows the proportion of individuals with some level of agreement between CSF and plasma stages in green shades whilst the grey shades indicate disagreement between these biofluid-based stages in which the plasma is more “advanced” as compared to the CSF classification