Multi-site reproducibility of a human immunophenotyping assay in whole blood and peripheral blood mononuclear cells preparations using CyTOF technology coupled with Maxpar Pathsetter, an automated data analysis system

Abstract

High-dimensional mass cytometry data potentially enable a comprehensive characterization of immune cells. In order to positively affect clinical trials and translational clinical research, this advanced technology needs to demonstrate a high reproducibility of results across multiple sites for both peripheral blood mononuclear cells (PBMC) and whole blood preparations. A dry 30-marker broad immunophenotyping panel and customized automated analysis software were recently engineered and are commercially available as the Fluidigm® Maxpar® Direct™ Immune Profiling Assay™. In this study, seven sites received whole blood and six sites received PBMC samples from single donors over a 2-week interval. Each site labeled replicate samples and acquired data on Helios™ instruments using an assay-specific acquisition template. All acquired sample files were then automatically analyzed by Maxpar Pathsetter™ software. A cleanup step eliminated debris, dead cells, aggregates, and normalization beads. The second step automatically enumerated 37 immune cell populations and performed label intensity assessments on all 30 markers. The inter-site reproducibility of the 37 quantified cell populations had consistent population frequencies, with an average %CV of 14.4% for whole blood and 17.7% for PBMC. The dry reagent coupled with automated data analysis is not only convenient but also provides a high degree of reproducibility within and among multiple test sites resulting in a comprehensive yet practical solution for deep immune phenotyping.

Keywords: cytometry automation; cytometry standardization; kits; percentage precision.

FIGURE 1

Assay workflow. Based on the broad immune cell phenotyping flow panels for the Human Immune Project (Maecker et al., 2012), the Maxpar Direct Immune Profiling Assay was designed as an optimized panel of 30 dry antibodies plus DNA intercalators in a single tube for staining whole blood and PBMC. Data were acquired on a Fluidigm Helios and analyzed using Maxpar Pathsetter, a customized automated analysis system powered by GemStone 2.0. Pathsetter software automatically cleans the data file by eliminating dead cells, debris, aggregates, and normalization beads. Modeling software then identifies and enumerates a broad spectrum of immune populations and presents the results in summary reports [Color figure can be viewed at wileyonlinelibrary.com]

FIGURE 2

Cleanup and analysis Cen-se′ maps: The top two panels are Cen-se′ maps created from the QC measurements: DNA1, DNA2, Live/ Dead, Beads, Event Length, Residual, Center, Width, and Offset. The topleft panel represents the raw normalized data from one file and the top-right the associated cleaned exported data. In the top-left panel, A (dark gray) are the live intact events, B (blue) are the low-DNA1 or debris events, C (yellow) are the normalization beads, D (blue) are events with zero pulse-processing parameters (Residual, Center, Width, and Offset), E (red) are “not cleaned events” with high Residual and Event Lengths, F (red) are true aggregates with high DNA1 and DNA2 intensities, G (yellow) are bead/cell aggregates, and H (red) are coincident ion clouds with low and high center values. The top-right panel is the Cen-se′ map of only the “cleaned” events. The bottom panel shows the same data with all markers selected after cleanup and modeling [Color figure can be viewed at wileyonlinelibrary.com]

FIGURE 3

Whole blood reproducibility. The top panel shows the mean and ± SD percentage of live intact cells for all 37 evaluated populations across all seven sites. The bottom panel shows the associated %CVs for each population where the average was 14.4% [Color figure can be viewed at wileyonlinelibrary.com]

FIGURE 4

Peripheral blood mononuclear cells (PBMC) reproducibility. The top panel shows the mean and ±SD percentage of live intact cells for all 37 evaluated populations. Absent from this plot are the granulocyte, neutrophils, basophils, eosinophils, and CD66b− granulocytes. The bottom panel shows the associated %CVs for each population, where the average was 17.7%. The percentages, SDs, and %CVs were an average of Cohort 1 (Week 1) and 2 statistics [Color figure can be viewed at wileyonlinelibrary.com]