Demonstration of Microbial Subgroups among Normal Vaginal Microbiota Data

Abstract

In this study we identified subgroups of observations relating to the healthy vaginal microbiota. This microbiota resides in a dynamic environment that undergoes cyclic change during the menstrual cycle. Cluster analysis procedures were applied to divide a set of 226 normal microbiota observations into groups. Three subgroups containing 100, 65, and 61 observations were identified. Plots of principal components determined by canonical analysis were obtained to demonstrate graphically the clustering of normal vaginal microbiota observations into subgroups. Results from the cluster analyses were verified using a predictive logit regression model. Analysis of mean logit values for the three clusters demonstrated that they differed from each other. Further verification was obtained using data from women with abnormal vaginal microbiota, either due to a yeast infection or the use of an iodine-based medicated douche product, and applying them to the clustering results developed using the normal microbiota observations. Observations from both abnormal groups clustered outside the boundaries of the three normal microbiota subgroups. These results indicate that it may be possible to identify normal microbiota subgroups that are prone to the development of an abnormal microbiota. Keywords: cluster analysis, logit regression model, vaginal microbiota, normal ecosystem