Neuroimaging of tissue microstructure as a marker of neurodegeneration in the AT(N) framework: defining abnormal neurodegeneration and improving prediction of clinical status

Table 4 Comparison of mean robust NODDI z-scores between CU A − T − and A + T + CSF A/T status participants

GM ROIs (bilateral)	NDI z-score p-value (Kruskal–Wallis)	Cliff’s delta (95% CI)	ODI z-score p-value (Kruskal–Wallis)	Cliff’s delta (95% CI)
Superior frontal gyrus	0.30	0.17 (− 0.03–0.35)	0.06	0.22 (− 0.037–0.45)
Hippocampus	0.30	0.12 (− 0.13–0.35)	0.2	0.06 (− 0.19–0.31)
Posterior cingulate gyrus	0.30	0.16 (− 0.096–0.40)	0.01^a	0.30 (0.09–0.49)^d
Inf. parietal + precuneus	0.30	0.18 (− 0.04–0.39)	0.01^a	0.23 (0.027–0.46)
Parahippocampus	0.40	0.12 (− 0.13–0.36)	0.06	0.24 (− 0.010–0.47)
Temporal inferior	0.40	0.11 (− 0.11–0.32)	0.06	0.16 (− 0.10–0.40)
WM ROIs (bilateral)
Uncinate fasciculus	0.18	0.16 (− 0.08–0.38)	0.42	0.03 (− 0.24 to − 0.30)
Superior longitudinal fasciculus	0.61	0.04 (− 0.19–0.26)	0.68	− 0.11 (− 0.30–0.08)
Inferior longitudinal fasciculus	0.16	0.19 (− 0.05–0.41)	0.11	0.21 (− 0.05–0.44)
Cingulum (hippocampus)	0.09	0.24 (0.014–0.45)^d	0.19	0.12 (− 0.12 to − 0.35)
Cingulum (cingulate gyrus)	0.16	0.14 (− 0.07–0.35)	0.29	0.21 (0.01–0.40)^d
Forceps major	0.62	0.06 (− 0.18–0.29	0.055	0.38 (0.16–0.60)^d

Z-scores for NODDI-ODI and NODDI-NDI were calculated with robust norms analysis as described. CSF A − T − status group included participants with a clinical diagnosis of cognitively unimpaired (CU) and A − T − CSF (n = 231). A + T + CSF (n = 28) status group included CU-diagnosed participants with A + T + CSF. CSF status cutoffs were determined by cutoffs through ROC analysis as described. Omnibus comparison (KW test, FDR correction) evaluated mean difference in ROIs z-score across all CSF groups. For ROIs that had a significant difference in Z-score across groups with omnibus comparison, an additional follow-up pairwise comparison (KW) was conducted across CSF status groups (A − /T − , A + /T +). All ROIs that showed significance in the omnibus test showed significance difference in pairwise comparison test. Thereby in this table, ROIs with non-significant P-values were those that only underwent initial omnibus comparison, whereas ROIs with significant p-values underwent both omnibus and pair-wise comparison testing. Effect size was calculated by Cliff’s delta with 95% confidence intervals
^a = P < 0.05
^b = P < 0.01
^c = P < 0.001, Kruskal–Wallis
^d = Sig. Cliff’s delta)

ISSN: 1758-9193