Gender differences contribute to variability of serum lipid biomarkers for Alzheimer's disease.

Alzheimer's disease (AD) cannot currently be diagnosed by a blood test. One reason may be gender differences. Another may be the statistical methods used. The authors evaluate these possibilities. The authors applied serum lipidomics to find AD biomarkers in men and women. They hypothesized that AD biomarkers would differ between genders and that machine-learning algorithms would improve diagnostic performance. Serum lipids were analyzed by mass spectrometry for a training set of AD cases and controls and in a blinded test set. Statistical analyses considered gender differences. Lipids best classifying AD subjects differed significantly between men and women. Robust statistical algorithms did not improve diagnostic performance. Poor performance of AD biomarkers appears to be due primarily to inherent variability in AD patients.