Affiliations: [a] Cognitive Impairment Center, General Hospital of Treviso, Treviso, Italy | [b] Department of Clinical Pathology, General Hospital of Treviso, Treviso, Italy | [c] Neuroradiology Unit, General Hospital of Treviso, Treviso, Italy | [d] Department of Rehabilitative Medicine, General Hospital of Treviso, Treviso, Italy | [e] Bracco Foundation, Milan, Italy
Abstract: Background:The role of folate and homocysteine in brain atrophy associated with Alzheimer’s disease is not completely understood. Objective:The aim of this study was to investigate the relationships between serum folate and homocysteine levels and the degree of cortical-subcortical and hippocampal atrophy in a first relatively preliminary sample of the Treviso Dementia (TREDEM) study using a potent data mining method. Methods:Physiological data, biochemical parameters, clinical assessment data, brain atrophy severity assessed with CT scans, and neuropsycological and disability data were assessed in a group of 232 outpatients (93 men and 139 women, aged 40.2–100 years) enrolled in the TREDEM study carried out in Treviso (Italy). A semantic connectivity map obtained through the Auto-CM system, a fourth generation artificial neural network (ANN), was used to offer some insight regarding the complex biological connections between the studied variables and the degree of brain atrophy. Results:Close associations between low serum folate levels and severe cortical-subcortical atrophy along with severe hippocampal atrophy measured by the width of the temporal horns of lateral ventricles were found. We also showed an association between high homocysteine levels and severe cortical-subcortical and hippocampal atrophy. Conclusion:The role of folate, which is inversely associated with the severity of brain atrophy, was confirmed. Our results also confirm the association between high homocysteine levels and severe cortical-subcortical and hippocampal atrophy. Auto-CM ANN is able to highlight associations sometimes visible only in longitudinal studies through intelligent data mining of a cross-sectional study.
