A statistical analysis of murine stem cell suicide techniques.

The clinical application of soft agar cloning techniques for granulocyte-macrophage stem cells (CFU-C) has resulted in a number of contradictory reports that may in part be due to an inadequate data base. Growth of murine CFU-C is more reproducible and less variable than that of human CFU-C. We utilized in vivo hydroxyurea suicide of murine marrow CFU-C to address the question of how many experiments are needed to detect a specific difference with a p of less than 0.05. In 66 experiments the mean marrow CFU-C hydroxyurea kill was 23.3%; 6-9 separate experiments were necessary to detaect differences of 25%-30%. In order to be sure that a 25%-30% difference is not present, 15-21 experiments were required. Using a Dec-20 computer, 1000 samples of sample size 3, 4, or 10 were drawn from the 66 experiments; it was found that with 3 experiments and a true value of 23%, the actually observed value was below 10%, 17% of the time, and was over 40% in 10% of the samplings. In a smaller number of experiments similar results were obtained analyzing 3HTdR suicide of pluripotent stem cells and CFU-C. These data could provide a base from which to judge the validity of studies utilizing the CFU-C technique.