highly significantp < 0.001
For the PUT, a highly significant and robust GLM model emerged (adjusted R 2 = 0.623, F(12,167) = 25.62, p < 0.001), explaining over 60 % of the variance in susceptibility.
For the PUT, a highly significant and robust GLM model emerged (adjusted R 2 = 0.623, F(12,167) = 25.62, p < 0.001), explaining over 60 % of the variance in susceptibility.
In contrast, while the CN model showed a nominally significant interaction (p = 0.044), the non-significant slope in the LL/LD subgroup (β = 0.191, p = 0.153) fails to support robust moderation.
The direct effect of putamen volume on SDMT remained significant (β = -0.41, p = 0.02), while the total effect approached significance (β = -0.20, p = 0.06).
Critically, the alternative model failed to reach significance in the HL/HD subgroup (SDMT: β = 0.207, p = 0.095; MoCA: β = 0.348, p = 0.053), the subgroup most relevant to the clinico-radiological paradox, where the original model showed full mediation for SDMT (β = -0.28, p = 0.04) accounting for 45.8 % of the total effect.