Racine et al,

Association of longitudinal white matter degeneration and cerebrospinal fluid biomarkers of

neurodegeneration, inflammation and Alzheimer’s disease in late-middle-aged adults

SUPPLEMENTARY MATERIAL

Table of Contents
Page
Supplementary Figure 1. JHU-ICBM regions of interest / 2
Supplementary Figure 2. Visualization of random samplings of whole brain white matter changes / 3
Supplementary Figure 3. Scatter plots of CSF biomarkers levels in batch 1 and batch 2 and their regression fit for the subset of samples from the same lumbar punctures analyzed in both batches / 4
Supplementary Table 1. Summary of all results from models examining baseline CSF levels as predictors of fractional anisotropy (FA) in select white matter ROIs. / 5
Supplementary Table 2. Summary of all results from models examining baseline CSF levels as predictors of mean diffusivity (MD) in select white matter ROIs. / 6
Supplementary Table 3. Summary of all results from models examining longitudinal change in CSF levels (CSF) as predictors of fractional anisotropy (FA) in select white matter ROIs. / 7
Supplementary Table 4. Summary of all results from models examining longitudinal change in CSF levels (CSF) as predictors of mean diffusivity (MD) in select white matter ROIs. / 8

Supplementary Figure 1. JHU-ICBM regions of interest thresholded at FA>0.2 displayed on a template brain. Top panel: global white matter (pink). Bottom panel: Left (blue) and right (red) cingulum-CC defined as the portion of the cingulum that runs within the cingulategyrus, traveling dorsally around the corpus callosum and then transitions to left (yellow) and right (green) cingulum-HC as it progresses from the splenium of the corpus callosum along the ventral surface of the hippocampus, terminating at entorhinal cortex.

Supplementary Figure 2.Visualization of random samplings of whole brain white matter changes. Spaghetti plots of random samples without replacement (n=30 per graph) of diffusion changes in a whole brain white matter mask.Top panel: fractional anisotropy; bottom panel: mean diffusivity. While there is considerable individual variability in the data, FA tends to decrease over time and MD tends to increase over time.

Supplementary Figure 3.Scatter plots of CSF biomarkers levels in batch 1 and batch 2 and their regression fit for the subset of samples from the same lumbar punctures analyzed in both batches. Sample size for each biomarker and a measure of reliability, R2, is displayed in the bottom right corner.

Supplementary Table 1.Summary of all results from models examining baseline CSF levels as predictors of fractional anisotropy (FA) in select white matter ROIs.

BASELINE RESULTS –FRACTIONAL ANISOTROPY
Intercept (CSF) / WM Mask / CC Cing R / CC Cing L / HC Cing R / HC Cing L
p-tau/Aβ42 / N.S. / N.S. / N.S. / N.S. / N.S.
Neurogranin / N.S. / N.S. / N.S. / N.S. / N.S.
YKL-40 / N.S. / N.S. / N.S. / N.S. / N.S.
NFL / N.S. / N.S. / N.S. / N.S. / β= -.000211(p= .047)
Slope (CSF×time) / WM Mask / CC Cing R / CC Cing L / HC Cing R / HC Cing L
p-tau/Aβ42×time / N.S. / N.S. / N.S. / N.S. / N.S.
Neurogranin×time / N.S. / N.S. / N.S. / N.S. / N.S.
YKL-40×time / N.S. / N.S. / N.S. / N.S. / N.S.
NFL×time / N.S. / N.S. / N.S. / N.S. / N.S.

WM = white matter; CC Cing = portion of the cingulum that runs within the cingulategyrus, traveling dorsally around the corpus callosum; HC Cing = portion of the cingulum that progresses from the splenium of the corpus callosum along the ventral surface of the hippocampus; R= Right; L = Left; ROI=region of interest; p-tau/Aβ42=phosphorylated-tau to beta-amyloid-42 ratio; YKL-40= chitinase-3-like protein; NFL=neurofilament light; N.S. = Not Significant, p>.05.

Supplementary Table 2.Summary of all results from models examining baseline CSF levels as predictors of mean diffusivity (MD) in select white matter ROIs.

BASELINE RESULTS –MEAN DIFFUSIVITY
Intercept (CSF) / WM Mask / CC Cing R / CC Cing L / HC Cing R / HC Cing L
p-tau/Aβ42 / N.S. / β= .247
(p= 0.005) / β= 0.282
(p= 0.006) / N.S. / N.S.
Neurogranin / N.S. / N.S. / N.S. / N.S. / N.S.
YKL-40 / N.S. / N.S. / N.S. / N.S. / N.S.
NFL / N.S. / N.S. / N.S. / N.S. / N.S.
Slope (CSF×time) / WM Mask / CC Cing R / CC Cing L / HC Cing R / HC Cing L
p-tau/Aβ42×time / N.S. / N.S. / N.S. / β= 0.033
(p= .049) / β= 0.046
(p= .012)
Neurogranin×time / N.S. / N.S. / N.S. / N.S. / β= 0.000729
(p= .023)
YKL-40×time / N.S. / N.S. / N.S. / N.S. / β = 0.000310
(p= .002)
NFL×time / N.S. / N.S. / N.S. / N.S. / N.S.

WM = white matter; CC Cing = portion of the cingulum that runs within the cingulategyrus, traveling dorsally around the corpus callosum; HC Cing = portion of the cingulum that progresses from the splenium of the corpus callosum along the ventral surface of the hippocampus; R= Right; L = Left; p-tau/Aβ42=phosphorylated-tau to beta-amyloid-42 ratio; YKL-40= chitinase-3-like protein; NFL=neurofilament light; N.S. = Not Significant p>.05.

Supplementary Table 3.Summary of all results from models examining longitudinal change in CSF levels (CSF) as predictors of fractional anisotropy (FA) in select white matter ROIs.

LONGITUDINAL RESULTS –FRACTIONAL ANISOTROPY
Intercept (CSF) / WM Mask / CC Cing R / CC Cing L / HC Cing R / HC Cing L
p-tau/AB42 / N.S. / N.S. / N.S. / N.S. / N.S.
Neurogranin / β = -0.000026
(p= .010) / β = -0.00059
(p= .045) / β = -0.000069
(p= .030) / N.S. / N.S.
YKL-40 / N.S. / N.S. / N.S. / N.S. / N.S.
NFL / N.S. / N.S. / N.S. / β = 0.000011
(p= .049) / β = 0.00069
(p=.049)
Slope (CSF×time) / WM Mask / Cing CC R / Cing CC L / Cing HC R / Cing HC L
p-tau/AB42×time / N.S. / N.S. / N.S. / N.S. / N.S.
Neurogranin×time / N.S. / N.S. / N.S. / N.S. / N.S.
Total Tau×time / N.S. / N.S. / N.S. / N.S. / N.S.
YKL-40×time / N.S. / N.S. / N.S. / N.S. / N.S.
NFL×time / N.S. / N.S. / N.S. / N.S. / N.S.

WM = white matter; CC Cing = portion of the cingulum that runs within the cingulategyrus, traveling dorsally around the corpus callosum; HC Cing = portion of the cingulum that progresses from the splenium of the corpus callosum along the ventral surface of the hippocampus; R= Right; L = Left; p-tau/Aβ42=phosphorylated-tau to beta-amyloid-42 ratio; YKL-40= chitinase-3-like protein; NFL=neurofilament light; N.S. = Not Significant p>.05.

Supplementary Table 4.Summary of all results from models examining longitudinal change in CSF levels (CSF) as predictors of mean diffusivity (MD) in select white matter ROIs.

LONGITUDINAL RESULTS –MEAN DIFFUSIVITY
Intercept (CSF) / WM Mask / CC Cing R / CC Cing L / HC Cing R / HC Cing L
p-tau/AB42 / N.S. / N.S. / N.S. / N.S. / N.S.
Neurogranin / β =0.0000771
(p.001) / β =0.000112
(p< .001) / β =0.000112
(p=.001) / β =0.000589
(p=.015) / β =0.000575
(p=.035)
YKL-40 / N.S. / N.S. / N.S. / N.S. / N.S.
NFL / N.S. / N.S. / N.S. / N.S. / N.S.
Slope (CSF×time) / WM Mask / CC Cing R / CC Cing L / HC Cing R / HC Cing L
p-tau/AB42×time / N.S. / N.S. / β=0.128
(p=.044) / N.S. / N.S.
Neurogranin×time / N.S. / N.S. / N.S. / N.S. / N.S.
YKL-40×time / N.S. / N.S. / N.S. / N.S. / N.S.
NFL×time / N.S. / N.S. / N.S. / N.S. / N.S.

WM = white matter; CC Cing = portion of the cingulum that runs within the cingulategyrus, traveling dorsally around the corpus callosum; HC Cing = portion of the cingulum that progresses from the splenium of the corpus callosum along the ventral surface of the hippocampus; R= Right; L = Left; p-tau/Aβ42=phosphorylated-tau to beta-amyloid-42 ratio; YKL-40= chitinase-3-like protein; NFL=neurofilament light; N.S. = Not Significant p>.05.

1