Neuromarkers predicting neurogranin in preclinical Alzheimer’s using multilayer perceptron

Alberto Guevara-Tirado, School of Human Medicine, Universidad Científica del Sur, Lima, Peru

Background: Neurogranin is a marker of synaptic dysfunction. Objective: To predict neurogranin concentrations based on neurodegenerative biomarkers in preclinical stages of Alzheimer’s disease (AD). Method: Analytical and cross-sectional study of 367 adults in preclinical AD. The variables were: neurogranin, synaptotagmin-1 (SYT-1), synaptosome-associated protein-25 (SNAP-25), growth-associated protein-43 (GAP-43), and amyloid beta protein. Multilayer perceptron-type neural networks were used. Results: The correlation of neurogranin with GAP-43, SNAP-25 and SYT-1 was high. The relative error of the perceptron was 0.10. In the scatter plot, there was a homoscedastic distribution, with R2 coefficient of 0.879, implying that 88% of the cerebrospinal fluid neurogranin values can be explained by the neurogranin values indirectly detected by GAP-43, SNAP-25 and SYT-1. Conclusions: SNAP-25, GAP-43 and SYT-1 are strong predictors of neurogranin concentrations in preclinical AD, suggesting a high pathophysiological interrelation in the progression towards this disease.

Keywords: Synaptotagmin-1. Neurogranin. Synaptosomal-associated protein-25. Nerve growth factors. Neural networks. Computer.