A statistical analysis of human lymphocyte transformation data

Abstract

The lymphocytes from 107 maternal-foetal pairs were examined for their in vitro responsiveness, as determined by the incorporation of tritiated thymidine following stimulation with phytohaemagglutinin (PHA), candida, varicella, mumps, streptokinase-streptodornase (SKSD) and tetanus toxoid. The data were collected and analysed in two sequential groups (forty-seven and sixty) in order to determine whether the results were reproducible. The variable chosen for analysis was the difference (d) between the square roots of the isotope incorporation in the stimulated and control cultures because it gave the most symmetrical distribution of the data. The experimental error in the determination of maternal lymphocyte stimulation was 1.4--8.6% and of the foetal lymphocytes, 1.0--16.6%, depending upon the antigen or mitogen and its concentration. The data in the two sets of patients were statistically the same in forty-eight of the fifty-six analyses (fourteen antigen or mitogen concentrations in autologous and AB plasma for maternal and foetal lymphocytes). The statistical limits of the distribution of responses for stimulation or suppression were set by an analysis of variance taking two standard deviations from the mean as the limits. When these limits were translated into stimulation indices, they varied for each antigen or mitogen and for different concentrations of the same antigen. Thus, a detailed statistical analysis of a large volume of lymphocyte transformation data indicates that the technique is reproducible and offers a reliable method for determing when significant differences from control values are present.