Profiling of polychromatic flow cytometry data on B-cells reveals patients' clusters in common variable...

The aim of this study was to find an objective computational approach for phenotype analysis of common variable immunodeficiency (CVID) patients that describes all differences in the six-color space and to form groups of patients using computational methods. CVID is a heterogeneous primary immunodeficiency disorder where molecular defect is recognized in <10% of the cases and is unknown in the majority of patients. The current CVID classification, EUROClass, is based on quantification of selected B-cell subsets. Using six-color polychromatic flow cytometry, we analyzed B-cell phenotypes in a cohort of 48 CVID patients and 49 healthy donors. We used a "probability binning" algorithm to create 1,024 bins (each bin is a six-color gate) that covered the cells' distribution within the entire B-cell compartment. A matrix file recording cellular content in all the bins was made. The hierarchical clustering of the individual samples was analyzed using a Pearson correlation of the bins' values. The Cut tree algorithm found 12 clusters. In six clusters, healthy individuals predominated; in one cluster, smB+CD21low (CVID patients by EUROClass) cells prevailed; in one cluster, smB-CD21norm cells prevailed; in one cluster, smB+CD21low cells prevailed; the remaining cluster was mixed. The overall reproducibility of probability binning clustering was confirmed by matching of replicates to the original cohort using the similarity matrix of the Pearson correlation, 15 replicates matched the same individual, three replicates matched a different individual within the same cluster, and three replicates matched to a different cluster. We were able to define B-cell subsets over- or under-represented in a particular cluster and display them back in the flow cytometry software. We describe a new analytical approach that enables a search in an objective computational environment for patient cohorts that are defined by similar B-cell profiles and thus contribute to the description of differences between CVID patient groups.