Amyloid and tau imaging biomarkers explain cognitive decline from late middle-age

Abstract

This study investigated differences in retrospective cognitive trajectories between amyloid and tau PET biomarker stratified groups in initially cognitively unimpaired participants sampled from the Wisconsin Registry for Alzheimer's Prevention. One hundred and sixty-seven initially unimpaired individuals (baseline age 59 ± 6 years; 115 females) were stratified by elevated amyloid-β and tau status based on 11C-Pittsburgh compound B (PiB) and 18F-MK-6240 PET imaging. Mixed effects models were used to determine if longitudinal cognitive trajectories based on a composite of cognitive tests including memory and executive function differed between biomarker groups. Secondary analyses investigated group differences for a variety of cross-sectional health and cognitive tests, and associations between 18F-MK-6240, 11C-PiB, and age. A significant group × age interaction was observed with post hoc comparisons indicating that the group with both elevated amyloid and tau pathophysiology were declining approximately three times faster in retrospective cognition compared to those with just one or no elevated biomarkers. This result was robust against various thresholds and medial temporal lobe regions defining elevated tau. Participants were relatively healthy and mostly did not differ between biomarker groups in health factors at the beginning or end of study, or most cognitive measures at study entry. Analyses investigating association between age, MK-6240 and PiB indicated weak associations between age and 18F-MK-6240 in tangle-associated regions, which were negligible after adjusting for 11C-PiB. Strong associations, particularly in entorhinal cortex, hippocampus and amygdala, were observed between 18F-MK-6240 and global 11C-PiB in regions associated with Braak neurofibrillary tangle stages I-VI. These results suggest that the combination of pathological amyloid and tau is detrimental to cognitive decline in preclinical Alzheimer's disease during late middle-age. Within the Alzheimer's disease continuum, middle-age health factors likely do not greatly influence preclinical cognitive decline. Future studies in a larger preclinical sample are needed to determine if and to what extent individual contributions of amyloid and tau affect cognitive decline. 18F-MK-6240 shows promise as a sensitive biomarker for detecting neurofibrillary tangles in preclinical Alzheimer's disease.

Keywords: Alzheimer’s disease; amyloid imaging; dementia: biomarkers; neurofibrillary tangles; tau imaging.

Figure 1

Parametric MK-6240 and PiB images, and biomarker group stratification. Mean parametric MK-6240 SUVR (A, left) and PiB DVR (A, right) images for each biomarker group. Individuals that were A−T+ only had elevated MK-6240 SUVR in the entorhinal cortex, whereas the A+T+ group had elevated MK-6240 SUVR extending into the greater neocortex including regions associated with Braak neurofibrillary tangle stages I–IV. Individuals that were A+T− had lower PiB DVR throughout the cortex and subcortex compared to the A+T+ group. Quadrant plots indicating the observed relationship between global PiB DVR and MK-6240 SUVR in the entorhinal cortex are shown in B with positivity thresholds for each tracer indicated with dashed lines. Individuals generally required sufficient levels of global PiB to show elevated MK-6240 in the entorhinal cortex, suggesting detectable changes in PiB precede detectable changes in entorhinal MK-6240 in most cases. Individuals with mild cognitive impairment at PET (triangles) were more likely to be in the A+T+ group than any other group.

Figure 2

Observed and group-modelled cognitive performance by biomarker group. Observed longitudinal PACC-3 performance organized by biomarker groups (A). Triangles indicate individuals with mild cognitive impairment at their cognitive assessment most proximal to PET imaging. Individuals who had elevated global PiB DVR and entorhinal MK-6240 SUVR (top right) had more precipitous decline in scores over time several years prior to mild cognitive impairment diagnosis and prior to imaging. The difference in rates of cognitive decline between biomarker groups was characterized using linear mixed effects analysis (model outcomes given in Table 3). Panel B shows the group-level modelled PACC-3 simple slopes and confidence levels over the range of ages present in each group, with the individual observed PACC-3 performance displayed in the background (points). Results of the linear mixed effects model indicated a significant group × age interaction. Tukey-adjusted pairwise comparisons indicated that the A+T+ group (orange in plots) declined approximately three times faster on average during the retrospective observation period compared to all other groups.

Figure 3

Associations between MK-6240, age and PiB. Surface rendered T-statistics of the positive association between MK-6240 and Global PiB DVR covaried for age for: (A) the entire study sample, and (B) in the subset without mild cognitive impairment at their PET scan. Images are thresholded such that t-statistics for voxels below the FWE-adjusted significance threshold are not shown. No negative associations with global PiB DVR or associations with age survived FWE correction (P_FWE= 0.05) in the voxelwise analysis. Panel C shows the post hoc region of interest analyses in composite regions spanning Braak I–VI neurofibrillary tangle stages. The percentage of variance in MK-6240 explained by both global PiB DVR and age (${\mathrm{R}}_{\text{Age},\text{PiB}}^2$) is shown under the composite region name with the partial correlation between MK-6240 and age (_Age, global PiB DVR partialled out) and the partial correlation between MK-6240 and global PiB DVR (_PiB, age partialled out) shown in the bottom of the title for each plot. The simple correlation between MK-6240 and age is shown in the top left corner of each plot. Region of interest analyses indicated small parametric associations with age, but these associations were mostly explained by global PiB DVR when partial correlations between MK-6240 and age and global PiB DVR were examined. The amount of variance in MK-6240 SUVR explained by PiB and age was highest in regions associated with early Braak neurofibrillary tangle staging, and decreased progressively in regions associated with later Braak stages. (C) *P < 0.05, unadjusted for multiple comparisons.