Higher aluminum concentration in Alzheimer's disease after Box-Cox data transformation

Abstract

Evidence regarding the role of mercury and aluminum in the pathogenesis of Alzheimer's disease (AD) remains controversial. The aims of our project were to investigate the content of the selected metals in brain tissue samples and the use of a specific mathematical transform to eliminate the disadvantage of a strong positive skew in the original data distribution. In this study, we used atomic absorption spectrophotometry to determine mercury and aluminum concentrations in the hippocampus and associative visual cortex of 29 neuropathologically confirmed AD and 27 age-matched controls. The Box-Cox data transformation was used for statistical evaluation. AD brains had higher mean aluminum concentrations in the hippocampus than controls (0.357 vs. 0.090 μg/g; P = 0.039) after data transformation. Results for mercury were not significant. Original data regarding microelement concentrations are heavily skewed and do not pass the normality test in general. A Box-Cox transformation can eliminate this disadvantage and allow parametric testing.