Table 2. Linear model that regresses DNAm age on HD status and other covariates.
Model 1 | Model 2 | Model 3 | |||||
---|---|---|---|---|---|---|---|
Covariate | Contrast | Coef (SE) | P-value | Coef (SE) | P-value | Coef (SE) | P-value |
Huntington | 2.06 (0.517) | 6.7×10−5 | 1.704 (0.503) | 0.00070 | 0.9 (0.486) | 0.065 | |
Age | 0.646 (0.012) | <2×10−16 | 0.64 (0.011) | <2×10−16 | 0.632 (0.011) | <2×10−16 | |
Sex | Female vs Male | −0.981 (0.49) | 0.046 | −0.84 (0.474) | 0.077 | 0.611 (2.637) | 0.817 |
Brain Bank | UCLA vs NewZealand | −0.093 (1.224) | 0.94 | 1.049 (1.198) | 0.382 | 1.32 (1.139) | 0.247 |
Tissue | Caudate Nucleus vs Frontal | −1.237 (1.266) | 0.33 | −3.412 (1.278) | 0.008 | −3.239 (1.201) | 0.007 |
Cingulate Gyrus vs Frontal | −1.631 (1.224) | 0.18 | −1.961 (1.183) | 0.098 | −0.729 (1.119) | 0.52 | |
CRBM vs Frontal | −5.353 (1.121) | 1.8×10−6 | −3.854 (1.113) | 0.001 | 16.194 (10.299) | 0.12 | |
Hippocampus vs Frontal | 1.327 (1.191) | 0.27 | −0.077 (1.175) | 0.95 | 1.08 (1.131) | 0.34 | |
Midbrain vs Frontal | −1.115 (1.274) | 0.38 | −4.12 (1.334) | 0.002 | −1.327 (1.331) | 0.32 | |
Motor Cortex vs Frontal | 1.539 (1.224) | 0.21 | 1.699 (1.182) | 0.151 | 1.64 (1.112) | 0.14 | |
Occipital vs Frontal | −2.886 (1.115) | 0.01 | −1.704 (1.096) | 0.121 | −2.218 (1.037) | 0.033 | |
Parietal vs Frontal | 0.835 (1.065) | 0.43 | 1.781 (1.042) | 0.088 | 1.382 (0.982) | 0.16 | |
Sensory Cortex vs Frontal | −0.173 (1.224) | 0.89 | 0.079 (1.183) | 0.95 | 0.179 (1.116) | 0.87 | |
Temporal vs Frontal | 0.191 (1.156) | 0.87 | 0.228 (1.116) | 0.84 | 0.621 (1.053) | 0.56 | |
Visual Cortex vs Frontal | 0.4 (1.233) | 0.75 | 2.178 (1.23) | 0.077 | 0.408 (1.192) | 0.73 | |
Prop. Neurons | −13.966 (2.395) | 5.5×10−9 | |||||
PC1 | 862.505 (313.408) | 0.006 | |||||
PC2 | 147.452 (64.232) | 0.022 | |||||
PC3 | −36.821 (13.872) | 0.008 | |||||
PC4 | 59.519 (11.864) | 5.3×10−7 | |||||
PC5 | 8.251 (26.496) | 0.76 |
Coefficients, standard error, and corresponding p-values for three multivariate models.
According to model 1, the age acceleration due to HD status amounts to 3.3 years (=2.159/0.646).
Model 2 is similar to model 1 but includes the (estimated) proportion of neurons as covariate. Model 3 is similar to model 1 but includes principal components. Since the analysis ignores the dependency of observations (due to multiple brain regions coming from the same individual), the p-values should only be interested as descriptive measures (as opposed to inferential measures). The multivariate models allow us to estimate the increase in biological age due to HD status. HD status is associated with an increase of 3.2 years (=2.06/0.646) according to model 1, an increase of 2.7 years (=1.704/0.64) according to model 2, and am increase of 1.4 years according to model 3.