152 Chapter 6 Annual change in PET as determinant for cognitive decline When annual change in tau PET BPND and R1 were incorporated in the model simultaneously, additional associations were found between higher accumulation rates in Braak I tau PET BPND and steeper decline on MMSE, as well as higher accumulation rates in Braak III/IV tau PET BPND with steeper decline on RAVLT immediate recall in CI individuals (Figure 4). For R1, an association was found between larger decrease in R1 with steeper decline on MMSE (Supplementary Figure 4). In CU individuals, the association between Braak I tau PET BPND with RAVLT immediate recall lost significance when annual change in R1 was included in the model. By contrast, associations with rate of decline were unmasked, as higher tau PET BPND accumulation rates in all three ROIs were associated with steeper decline on the three cognitive tests, except for accumulation rates in Braak I and decline on MMSE and TMT-B (Figure 4). For R1, we now found associations between larger decreases in R1with steeper decline on MMSE and RAVLT immediate recall (Supplementary Figure 4). Sensitivity analyses Baseline PET as determinant for prospective cognitive decline When models only included cognitive data that was obtained from the moment of the baseline PET scan onward, results remained generally unchanged (Supplementary Figure 5). Overall, estimates from models with baseline tau PET BPND as determinant became somewhat larger (i.e. more negative in case of negative coefficients). Most pronounced increases in estimates were found for the association between baseline tau PET BPND in Braak I and decline on MMSE (CI; β= -0.30 SE=0.17 when all cognitive data was included versus β= -0.80 SE=0.33 in sensitivity analyses), and baseline tau PET BPND in Braak V/VI with decline on RAVLT immediate recall (CU; β= -0.45 SE=0.17 when all cognitive data was included versus β= -0.92 SE=0.30 in sensitivity analyses). Results for baseline R1 also remained similar, but the estimate from the association with decline in RAVLT immediate recall was attenuated (CU; β= 0.85 SE=0.35 when all cognitive data was included versus β= 0.34 SE=0.61 in sensitivity analyses) (Supplementary Figure 6). Annual change in PET as determinant for cognitive decline Results from sensitivity analyses remained generally the same. Overall, estimates from models with annual change in tau PET BPND as determinant became larger, especially in models including change in Braak I and V/VI tau PET BPND (Supplementary Figure 7). Most pronounced increase in estimate was found for the association between change in tau PET BPND in Braak I and improvement on RAVLT immediate recall (CU; β= 0.06 SE=0.08 when all cognitive data was included versus β= 0.34 SE=0.14 in
RkJQdWJsaXNoZXIy MjY0ODMw