PubMed2004
BACKGROUND: Due to the high cost, subject availability and ethical constraints, it is often critically important in pre-clinical and clinical studies to carry out an adequate statistical analysis of longitudinal multivariate data over several time points in trials in several small groups. OBJECTIVES: We aim to accurately assess and develop an appropriate distribution-free longitudinal model for an estimate of the comparative treatment effects of several biological factors in several small groups even if data sets should contain outlier measurements and censored values. This approach is used to evaluate the relative efficacy of mucosal and subcutaneous polypeptide vaccines in rhesus macaques exposed to SHIV. STUDY DESIGN: The algorithms of the non-parametric repeated measures ANOVA models [Mack GA. A quick and easy distribution-free test for main effects in a two-factor ANOVA. Communic Stat Part B: Simp Comp 1981;10:571-91; Akritas MG, Brunner, E. A unified approach to rank tests for mixed models. J Stat Plan Inference 1997;61:249-77; Brunner E, Puri ML. Nonparametric methods in factorial designs. Stat Pap 2001;42:1-52.] are described, programmed and assessed. The viral loads, CD4(+) and CD8(+) cell counts were analyzed at several time points in peripheral blood of the 11 MamuA(*)01 positive macaques intrarectally challenged with pathogenic SHIV-Ku2 and immunized intrarectally with synthetic HIV/SIV peptide vaccine. RESULTS: Using nonparametric ANOVA tests, we demonstrated with statistical significance that after intrarectal challenge with pathogenic SHIV-Ku2, intrarectally immunized monkeys expressed viral titers that fell below the level of detection in blood and intestine (which is a major reservoir of viral replication), whereas the subcutaneously immunized or control macaques had residual viraemia. Moreover, the proliferative response of T cells and both CD4(+) and CD8(+) cells were better preserved in intrarectally immunized animals. Robustness of the comparisons was confirmed by gradual removal of up to 50% of data points. CONCLUSION: Despite limited data, our analysis shows better preservation of both CD4(+) and CD8(+) cells in intrarectally immunized animals. The result is consistent with our hypothesis that mucosal immunization is more effective than systemic immunization and that an induction of specific CTLs in the intestinal mucosa, a key site of virus replication, with a mucosal AIDS vaccine ameliorates SHIV infection in non-human primates. Our analytical methodology can be applicable in comparative estimates of the different treatment-associated effects and their synergy for a variety of longitudinal data sets in small treatment groups.