Table 4.
Baseline DTI measures predicting dementia conversion while accounting for the clinical markers. Baseline PSMD and PC1 measures predicted dementia conversion in SCANS, RUN DMC and HARMONISATION. In RUN DMC MD median and DSEG was not associated with dementia. In HARMONISATION baseline Geff did not predict dementia conversion.
| Dementia conversion | ||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| SCANS | RUN DMC | HARMONISATION | ||||||||||
|
Change Markers |
HR (95 % CI) |
P-Value | AIC | AUC |
HR (95 % CI) |
P-Value | AIC | AUC |
HR (95 % CI) |
P-Value | AIC | AUC |
| MD Median | 2.19 (1.51, 3.16) |
<0.001 | 138.37 | 0.832 | 1.33 (1.00, 1.76) |
0.05 | 515.69 | 0.828 | 1.78 (1.08, 2.93) |
0.02 | 173.51 | 0.761 |
| PC1 | 2.28 (1.51, 3.44) |
<0.001 | 139.60 | 0.825 | 1.57 (1.15, 2.14) |
<0.001 | 511.60 | 0.837 | 1.74 (1.03, 2.92) |
0.04 | 174.17 | 0.765 |
| PSMD | 1.74 (1.29, 2.34) |
0.001 | 143.70 | 0.804 | 1.45 (1.14, 1.83) |
<0.001 | 511.51 | 0.831 | 1.73 (1.13, 2.65) |
0.01 | 172.72 | 0.765 |
| DSEG θ | 3.52 (2.09, 5.92) |
<0.001 | 128.21 | 0.908 | 1.33 (0.91, 1.93) |
0.14 | 517.17 | 0.821 | 1.94 (1.06, 3.53) |
0.03 | 173.67 | 0.753 |
| Geff | 0.37 (0.23, 0.61) |
<0.001 | 138.39 | 0.842 | 0.64 (0.46, 0.89) |
0.01 | 512.59 | 0.832 | 0.79 (0.49, 1.30) |
0.36 | 177.58 | 0.702 |
MD Median = mean diffusivity median of the WM histogram, PC1 = scores of the first principal component, PSMD = peak width of skeletonized mean diffusivity, DSEG θ = diffusion tensor image segmentation θ, Geff = global efficiency network measure, AIC = Akaike information criterion, HR = hazard ratio, AUC = area under the curve, AIC = Akaike information criterion, 95 % CI = 95 % confidence interval, P-Value = statistical value of significance with p < 0.05.