Relation between subcortical grey matter atrophy and conversion from mild cognitive impairment to Alzheimer's disease

Objective To investigate whether subcortical grey matter atrophy predicts progression from mild cognitive impairment (MCI) to Alzheimer's disease (AD), and to compare subcortical volumes between AD, MCI and controls. To assess the correlation between subcortical grey matter volumes and severity of cognitive impairment. Methods We included 773 participants with three-dimensional T1-weighted MRI at 3 T, made up of 181 controls, who had subjective memory symptoms with normal cognition, 201 MCIs and 391 AD. During follow-up (2.0 +/- 0.9 years), 35 MCIs converted to AD (progressive MCI) and 160 MCIs remained stable (stable MCI). We segmented volumes of six subcortical structures of the amygdala, thalamus, caudate nucleus, putamen, globus pallidus and nucleus accumbens, and of the hippocampus, using FMRIBs integrated registration and segmentation tool. Results Analysis of variances, adjusted for sex and age, showed that all structures, except the globus pallidus, were smaller in AD than in controls. In addition, the amygdala, thalamus, putamen, nucleus accumbens and hippocampus were smaller in MCIs than in controls. Across groups, all subcortical greymatter volumes, except the globus pallidus, showed a positive correlation with cognitive function, as measured by Mini Mental State Examination (MMSE) (0.16< r<0.28, all p<0.05). Cox proportional hazards analyses adjusted for age, sex, education, Cambridge Cognitive Examination-Revised (CAMCOG-R) and MMSE showed that smaller volumes of the hippocampus and nucleus accumbens were associated with increased risk of progression from MCI to AD (HR (95% CI) 1.60 (1.15 to 2.21); 1.60 (1.09 to 2.35), p<0.05). Conclusions In addition to the hippocampus, the nucleus accumbens volume loss was also associated with increased risk of progression from MCI to AD. Furthermore, volume loss of subcortical grey matter structures was associated with severity of cognitive impairment.