Affiliations: [a] Department of Internal Medicine and Medical Specialties, University of Genova, Genova, Italy | [b] Department of Neurology, Galliera Hospital, Genova, Italy | [c] Department of Clinical and Behavioral Neurology IRCCS Santa Lucia Foundation, Roma, Italy | [d] Semeion Research Centre, Roma, Italy | [e] Bracco Medical Department, Milano, Italy
Correspondence:
[*]
Correspondence to: Prof. Massimo Tabaton, Department of Internal Medicine and Medical Specialties, University of Genova, 16132 Genova, Italy. Tel.: +39 010 3537064; Fax: +39 010 506938; E-mail: mtabaton@neurologia.unige.it.
Abstract: The search for markers that are able to predict the conversion of amnestic mild cognitive impairment (aMCI) to Alzheimer's disease (AD) is crucial for early mechanistic therapies. Using artificial neural networks (ANNs), 22 variables that are known risk factors of AD were analyzed in 80 patients with aMCI, for a period spanning at least 2 years. The cases were chosen from 195 aMCI subjects recruited by four Italian Alzheimer's disease units. The parameters of glucose metabolism disorder, female gender, and apolipoprotein E ε3/ε4 genotype were found to be the biological variables with high relevance for predicting the conversion of aMCI. The scores of attention and short term memory tests also were predictors. Surprisingly, the plasma concentration of amyloid-β42 had a low predictive value. The results support the utility of ANN analysis as a new tool in the interpretation of data from heterogeneous and distinct sources.
