TABLE 2.
Sample characterization
| Aβ‐negative (n = 334) | Aβ‐positive (n = 74) | Aβ‐negative vs Aβ‐positive | Unadjusted association with WMHV(n = 408) | |
|---|---|---|---|---|
| Age, years, median (IQR) | 70.7 (1.2) | 70.7 (1.1) | P = 0.66a | Rho = 0.12; P = 0.013*,b |
| Male sex, n (%) | 166 (49.7%) | 40 (54.1%) | OR 0.84; P = 0.5c | P = 0.17a |
| Earliest childhood cognitive ability, z‐score | 0.41 (0.74) | 0.44 (0.75) | P = 0.74d | Rho = −0.05; P = 0.27b |
| Advanced level of education, n (%) | 181 (54.2%) | 37 (50%) | OR 0.85; P = 0.51c | P = 0.86a |
| MMSE, median (IQR), maximum score = 30 | 30 (1) | 29 (1) | P = 0.063a | Rho = −0.03; P = 0.52b |
| Logical memory score, mean (SD), maximum score = 25 | 11.7 (3.6) | 11.3 (3.7) | P = 0.33d | Rho = −0.05; P = 0.35b |
| Digit‐symbol substitution score, mean (SD), maximum score = 93 | 48.8 (10.1) | 46.9 (9.7) | P = 0.14d | Rho = −0.11; P = 0.03*,b |
| Matrix reasoning, median (IQR), maximum score = 32 | 26 (4) | 25 (4) | P = 0.037* , a | Rho = −0.04; P = 0.47b |
| 12‐Item Face‐Name test, median (IQR), maximum score = 96 | 66 (28) | 68 (27) | P = 0.29a | Rho = 0.05; P = 0.34b |
| APOE ε4 carrier, n (%) (missing data: n = 2) | 76 (22.9%) | 45 (60.8%) | OR 5.22; P < 0.0001c | P = 0.49a |
| Hypertension, n (%) (missing data: n = 1) | 217 (64.9%) | 42 (57.5%) | OR 0.73; P = 0.23c | P = 0.004* , a |
| Hypercholesterolemia, n (%) | 116 (34.7%) | 28 (37.8%) | OR 1.14; P = 0.61c | P = 0.22a |
| Diabetes, n (%) (missing data: n = 4) | 35 (10.6%) | 9 (12.1%) | OR 1.17; P = 0.7c | P = 0.14a |
| Ever‐smoked, n (%) | 216 (64.7%) | 46 (62.1%) | OR 0.89; P = 0.68c | P = 0.32a |
| WMHV, milliliters, median (IQR) | 2.85 (4.84) | 3.30 (4.97) | P = 0.48a | ‐ |
Wilcoxon rank‐sum tests, or two‐sample t tests if the outcome was normally distributed, were used to compare continuous clinical and cognitive characteristics between Aβ‐positive and Aβ‐negative groups. Logistic regression models were used to compare categorical variables between Aβ‐positive and Aβ‐negative groups. Spearman correlation coefficients were used to assess unadjusted relationships between WMHV and continuous variables. Wilcoxon rank‐sum tests were used to test associations between WMHV and categorical variables. IQR, interquartile range; MMSE, Mini‐Mental State Examination; OR, odds ratio; WMHV, white matter hyperintensity volume (milliliters).
Wilcoxon rank‐sum test.
Spearman rank correlation.
Logistic regression.
t test.
* = P < 0.05.