Efficacy and Safety of Filgrastim in Alzheimer's Disease

The study was originally designed to have two arms (GCSF first followed by placebo= Arm 1; placebo followed by GCSF=Arm2). Total length of study for each subject was 14 weeks, with crossover on week 7. Linear regression models could not be used because of the small number of subjects; n = 5 in the first Arm who received G-CSF before crossover to the placebo phase and n = 3 in Arm 2 (who received placebo first before crossover to the G-CSF treatment phase). The general rule for the sample size required for regression analysis is n = 15 per covariate. Therefore, comparison of the final cognitive scores on visit 5 (week 14) were compared to scores at baseline (visit 1) for all subjects (n = 8) using the Wilcoxon signed rank test. Plasma cytokine changes were plotted to compare effects following G-CSF to that following placebo, regardless of the order of treatment. Wilcoxon sum rank test was used to test differences between G-CSF treatment and placebo treatment at specific intervals after treatment. Statistical Package for the Social Sciences (SPSS Version 19) was used for all data analysis.