Classification of delusions in Alzheimer's disease and their neural correlates

Previous findings on neural correlates of delusion in Alzheimer's disease (AD) have been inconsistent because of methodological issues, such as treating multiple delusions as a single entity. In this retrospective study, we classified AD delusions and investigated their neural correlates by using single-photon emission computed tomography data.

We selected AD patients with delusions from our consecutive outpatients from 2004 to 2010. In this study, eight types of delusions were evaluated with Neuropsychiatric Inventory and classified by factor analysis. Twenty-five of the patients also had single-photon emission computed tomography data, which we used to assess the relationships between cerebral regions of hypoperfusion and hyperperfusion and each classified delusion. The relations were assessed using Statistical Parametric Mapping with normalization to the white matter cerebral blood flow.

The delusions were classified into three factors. Factor 1 consisted of a belief that his/her house is not his/her home, phantom boarder symptom, delusion of abandonment, and belief that one's spouse or others are not who they claim to be. Factor 1 was related to hypoperfusion in the right temporal pole and hyperperfusion in the medial frontal and precentral regions. Factor 2 consisted of delusion relating to the television and delusion of persecution. Factor 2 was related to hypoperfusion in the precuneus and hyperperfusion in the insula and thalamus. Factor 3 consisted of delusion of abandonment and delusional jealousy. Factor 3 was related to hypoperfusion in the right inferior temporal and frontal regions and hyperperfusion in the middle frontal gyrus, insula and posterior cingulate gyrus. Delusion of theft was not included in any factors, and it was related to hypoperfusion in the bilateral thalami and left posterior cingulate gyrus and hyperperfusion in the left inferior frontal regions and anterior cingulate gyrus.

Delusions in AD were classifiable, and each classified delusion was related to different neural networks.