Subject classification obtained by cluster analysis and principal component analysis applied to flow cytometric data

Background: Polychromatic flow cytometry (PFC) allows the simultaneous determination of multiple antigens in the same cell, resulting in the generation of a high number of subsets. As a consequence, data analysis is the main difficulty with this technology. Here we show the use of cluster analysis (CA) and principal component analyses (PCA) to simplify multicolor data visualization and to allow subjects' classification. Methods: By eight-colour cytofluorimetric analysis, we investigated the T cell compartment in donors of different age (young, middle-aged, and centenarians). T cell subsets were identified by combining positive and negative expression of antigens. The resulting data set was organized into a matrix and subjected to CA and PCA. Results: CA clustered people of different ages on the basis of cytofluorimetric profile. PCA of the cellular subsets identified centenarians within a different cluster from young donors, while middle-aged donors were scattered between these groups. These approaches identified T cell phenotypes that changed with increasing age. In young donors, memory T cell subsets tended to be CD127+ and CD95- whereas CD127-, CD95+ phenotypes were found at higher frequencies in people with advanced age. Conclusions: Our data suggest the use of bioinformatic approaches to analyze large data-sets generated by PFC and to obtain the rapid identification of key populations that best characterize a group of subjects.