Adding interaction terms between age and APOE to our model significantly improved its fit (P = 1.1 × 10−5). The negative and non‐linear association between total hippocampal volume and age remained with similar effect size estimates compared to the model without the interaction terms. However, we no longer observe a main effect of APOE status on hippocampal volume. Instead, we found significant interactions between age and APOE genotype (Figure 2). Among participants < 60 years old, APOE did not affect hippocampal volume (Table 2). The age‐dependent effect of APOE became apparent in older participants. Specifically, among those aged 60 to 69, homozygosity for the ε4 allele was significantly associated with hippocampal volume loss (β = −97.6 mm3, P = 0.022), and the largest age‐dependent effect of APOE was observed in individuals aged ≥ 70 carrying two copies of the ε4 allele (β = −246.2 mm3, P = 1.3 × 10−6). We found no significant interactive effect of ε3/ε4 among those aged 60 to 69, but in the ≥ 70 age group, we did observe significant associations of APOE ε3/ε4 with
