Table 1.
Parameters | Baseline | PRS P < 1 × 10−5 | PRS.no.APOE P < 1 × 10−5 | PRS.no.APOE + APOE alleles | ||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|
B | SE | P(>|z|) | B | SE | P(>|z|) | B | SE | P(>|z|) | B | SE | P(>|z|) | |
Intercept | −1.56 | 0.20 | 4.89E−15 | −1.67 | 0.21 | 8.88E−16 | −1.58 | 0.20 | 2.89E−15 | −1.82 | 0.22 | <2.20E−16 |
MZ | 0.04 | 0.16 | 7.92E−1 | 0.06 | 0.15 | 6.79E−1 | 0.08 | 0.16 | 6.15E−1 | 0.07 | 0.15 | 6.08E−1 |
Sex (0 = M, 1 = F) | 0.44 | 0.09 | 3.83E−6 | 0.44 | 0.10 | 4.15E−6 | 0.44 | 0.09 | 3.02E−6 | 0.41 | 0.09 | 1.40E−5 |
LastAge | 0.97 | 0.12 | 8.88E−15 | 0.97 | 0.13 | 2.75E−14 | 0.95 | 0.13 | 4.31E−14 | 0.96 | 0.13 | 6.51E−14 |
LastAge2 | −0.55 | 0.09 | 5.12E−10 | −0.50 | 0.09 | 1.29E−8 | −0.53 | 0.09 | 1.78E−9 | −0.47 | 0.09 | 4.27E−8 |
Array | 0.40 | 0.16 | 1.55E−2 | 0.44 | 0.17 | 9.83E−3 | 0.40 | 0.17 | 1.69E−2 | 0.38 | 0.17 | 2.88E−2 |
PRS | — | — | — | 0.38 | 0.06 | 7.53E−12 | 0.16 | 0.05 | 1.41E−3 | 0.16 | 0.05 | 1.06E−3 |
APOE ɛ2 alleles | — | — | — | — | — | — | — | — | — | −0.18 | 0.12 | 1.38E−1 |
APOE ɛ4 alleles | — | — | — | — | — | — | — | — | — | 0.75 | 0.11 | 2.05E−12 |
Random·MZ | 2.23 | 0.76 | 3.37E−3 | 1.77 | 0.62 | 4.49E−3 | 2.06 | 0.73 | 4.53E−3 | 1.61 | 0.58 | 5.60E−3 |
Random·DZ | 0.37 | 0.23 | 1.04E−1 | 0.39 | 0.24 | 1.02E−1 | 0.39 | 0.24 | 1.02E−1 | 0.36 | 0.24 | 1.40E−1 |
Fit statistics | ||||||||||||
Deviance | 1703.37 | 1637.55 | 1692.10 | 1610.81 | ||||||||
AIC | −859.69 | −827.78 | −855.05 | −816.40 | ||||||||
SBC | −879.83 | −850.44 | −877.71 | −844.10 | ||||||||
ICC·MZ | 0.691 | 0.638 | 0.673 | 0.616 | ||||||||
ICC·DZ | 0.269 | 0.282 | 0.280 | 0.263 | ||||||||
AUC | 0.976 | 0.972 | 0.977 | 0.970 | ||||||||
R 2 Nagelkerke | 0.084 | 0.062 | 0.011 | 0.076 |
Regression analyses estimating varying intraclass correlations (ICCs) with clustered twin data were adapted from code in Archer et al.32 using mixor. MZ, monozygotic twin; DZ, dizygotic twin; LastAge, age at last follow-up, death or dementia onset, centred on age 80 years and divided by 10; Array, Human OmniExpress = 0, Illumina PsychArray = 1; PRS, polygenic risk score at P < 1 × 10−5 residualized for four PCs and standardized within array type; PRS.no.APOE, PRS without APOE region; Random, random effect; Deviance, –2ln(Likelihood); AIC, Akaike Information Criteria; SBC, Schwarz Bayesian Criterion; ICCs measured as Random·MZ/(1 + Random·MZ) and Random·DZ/(1 + Random·DZ)32; AUC, area under the curve.