Table 1.
Age- and sex-specific effects significantly contribute to DNAm aging in schizophrenia
| Hannum Δage | Horvath Δage | Levine Δage | |||||
|---|---|---|---|---|---|---|---|
| Model variables | Comparison | R2 (%) | P-value | R2 (%) | P-value | R2 (%) | P-value |
|
Model 0: baseline |
– | 6.9 | – | 3.6 | – | 2.1 | – |
|
Model 1: + status |
Model 0 vs. 1 | 6.9 | 1.00 | 4.0 | 9.8E−03 | 3.2 | 6.7E−06 |
|
Model 2: + status*age.cont |
Model 1 vs. 2 | 7.1 | 0.34 | 4.3 | 0.13 | 3.7 | 5.3E−03 |
|
Model 3: + status*age.groups |
Model 1 vs. 3 | 7.4 | 0.24 | 5.5 | 2.0E−05 | 4.0 | 0.02 |
|
Model 4: + status*age.groups*sex |
Model 3 vs. 4 | 7.7 | 1.00 | 5.9 | 0.34 | 4.7 | 0.02 |
Shown are the contributions of interaction effects between disease status and age and sex on Δage. The baseline model corresponds to Δage ~ dataset + cohort + platform + age.continuous + sex. For other models, the variable(s) in addition to the baseline variables are shown with the corresponding variance explained (R2) in Δage. Interaction terms with chronological age are modeled as a continuous variable (age.cont) or a categorical variable (age.groups). The latter uses previously defined decades. Model comparison is performed to assess whether the contribution of an interaction term is significant compared to a model without that term. The Chi-square test is used to test two models with corresponding p-value presented. The results of these analysis are shown for both the Horvath and Levine clock. P-values are corrected for the number of tests performed (3 clocks × 4 comparisons = 12)