Gender differences contribute to variability of serum lipid biomarkers for Alzheimer’s disease - Supplemental_Figure_1
Background: 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. Objective: 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. Methods: 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. Results: Lipids best classifying AD subjects differed significantly between men and women. Robust statistical algorithms did not improve diagnostic performance. Conclusion: Poor performance of AD biomarkers appears to be due primarily to inherent variability in AD patients.
Plain language summary: Alzheimer’s disease (AD) cannot be diagnosed by a blood test or radiologic study. Newer laboratorymethods usingmass spectrometers have successfully identified molecules in blood that mark the presence of other diseases, but such approaches have failed to find diagnostic biomarkers for AD. Often, initial studies of serum from AD cases and controls have provided promising diagnostic biomarkers, but follow-up studies have not confirmed their usefulness. This study attempts to clarify why this is so by carrying out a serum lipid (fatty molecule) analysis using mass spectrometry (MS) in both an initial serum set of AD cases and matched controls and applying those results to AD diagnosis in a second, independent set of specimens. Sources of variability that could prevent the discovery of useful markers using MS analyses of serum include specimen integrity, variable MS results, problems with statistical methods that analyze MS data and inherent AD patient variability reflected in their sera.