Variability in subjective review of umbilical cord blood colony forming unit assay

Abstract

Background: The goal of this study was to quantitatively evaluate the reproducibility of current manual counting methods of colony forming units (CFUs) from umbilical cord blood samples. Methods: Fresh and reconstituted frozen cells from 10 cord blood samples were cultured under standard conditions. The number of BFU-Es, CFU-GMs, and CFU-GEMMs were counted by three expert reviewers using the standard microscope method and manually traced CFUs on digital images of cell cultures. Results: The mean colony count based on the traced digital images was 82 (22% CV) and 52 (15% CV) for the fresh and frozen samples, respectively. This was significantly greater than that observed using the microscope, 61 (13% CV) for fresh and 43 (16% CV) for frozen. The difference was mainly due to the reviewers observing more CFU-GMs in the digital images than through the microscope review. All three reviewers agreed on the presence of a colony 72% of the time based on the digital review in both fresh and frozen samples. Reviewer agreement with respect to colony type in the fresh samples was 38% (22%CV), 25% (51%CV), and 6% (115%CV) for BFU-Es, CFU-GMs, and CFU-GEMMs, respectively. Reviewer agreement increased for BFU-Es and CFU-GMs in the frozen samples where fewer colonies were present. Conclusions: Although this study showed marked variability between reviewers, the analysis of manually traced digital images has the potential to improve inter-observer variation when compared to current methods by identifying features that lead to discrepancies in colony counting and providing cases with consensus results. © 2016 International Clinical Cytometry Society.