|
Cognitive test
|
Predictors
|
β-coefficients
|
R
2
|
AIC
|
|---|
|
APOE ε4+
|
18F-AV-1451
|
|---|
|
MMSE
|
APOE ε4+
|
−0.581 (P = 0.67)
|
NA
|
0.056
|
379.1
|
|
18F-AV-1451
|
NA
|
−3.838 (P = 0.054)
|
0.112
|
375.2
|
|
APOE ε4+ and 18F-AV-1451
|
−0.946 (P = 0.48)
|
−4.024 (P = 0.046)
|
0.12
|
376.6
|
|
ADAS immediate recall
|
APOE ε4+
|
0.352 (P = 0.45)
|
NA
|
0.091
|
211.7
|
|
18F-AV-1451
|
NA
|
0.968 (P = 0.14)
|
0.118
|
209.9
|
|
APOE ε4+ and 18F-AV-1451
|
0.426 (P = 0.36)
|
1.03 (P = 0.12)
|
0.132
|
211
|
|
ADAS delayed recall
|
APOE ε4+
|
2.124 (P = 0.01)
|
NA
|
0.168
|
260.9
|
|
18F-AV-1451
|
NA
|
1.724 (P = 0.13)
|
0.086
|
266.3
|
|
APOE ε4+ and 18F-AV-1451
|
2.17 (P = 0.01)
|
1.832 (P = 0.09)
|
0.214
|
259.7
|
|
TMT-A
|
APOE ε4+
|
−47.961 (P = 0.03)
|
NA
|
0.158
|
634.5
|
|
18F-AV-1451
|
NA
|
57.62 (P = 0.07)
|
0.131
|
636.2
|
|
APOE ε4+ and 18F-AV-1451
|
−44.371 (P = 0.04)
|
51.109 (P = 0.09)
|
0.204
|
633.3
|
|
Letter fluency (S)
|
APOE ε4+
|
2.537 (P = 0.11)
|
NA
|
0.167
|
327.1
|
|
18F-AV-1451
|
NA
|
−0.659 (P = 0.79)
|
0.123
|
329.9
|
|
APOE ε4+ and 18F-AV-1451
|
2.54 (P = 0.12)
|
0.023 (P = 0.99)
|
0.167
|
329.1
|
|
Animal fluency
|
APOE ε4+
|
1.969 (P = 0.27)
|
NA
|
0.108
|
340.1
|
|
18F-AV-1451
|
NA
|
−6.637 (P = 0.01)
|
0.198
|
334.4
|
|
APOE ε4+ and 18F-AV-1451
|
1.238 (P = 0.47)
|
−6.304 (P = 0.02)
|
0.207
|
335.8
|
|
AQT-CF
|
APOE ε4+
|
−9.364 (P = 0.47)
|
NA
|
0.078
|
518.4
|
|
18F-AV-1451
|
NA
|
21.59 (P = 0.23)
|
0.097
|
517.3
|
|
APOE ε4+ and 18F-AV-1451
|
−7.717 (P = 0.55)
|
20.403 (P = 0.26)
|
0.104
|
518.9
|
- Linear regression models predicting different cognitive tests. β-coefficients are on the original scales. For each test, models were evaluated with different predictors: APOE ε4, 18F-AV-1451, or both APOE ε4 and 18F-AV-1451 (no interactions). In all cases when APOE ε4 or 18F-AV-1451were significant predictors alone, they remained significant with only minor changes in β-coefficients when adjusted for each other. All models were adjusted for age, sex and education. Comparisons of AIC never favored the combined models, since ΔAIC ranged from −1.2 to 2 for the combined model minus the best (lowest AIC) individual model (a ΔAIC ≤ 2 is generally not considered a meaningful model difference) Significant results (P < 0.05) are indicated in bold ADAS Alzheimer’s disease Assessment Scale, AIC Akaike information criterion, AQT-CF A Quick Test of cognitive speed—color and form, MMSE Mini-Mental State Examination, NA not applicable, TMT-A Trail Making Test-A
