Evaluation of the Sysmex DI-60 digital morphology analyzer on Wright-stained samples with a focus on prevalence-dependent quality indicators.

BACKGROUND: This study aims to evaluate the trueness of the DI-60 Digital Cell Imaging Analyzer on Wright-stained samples with a focus on prevalence-dependent quality indicators for differential blood counts requested from non-hematology wards. METHODS: Two hundred and ninety-nine samples were included into this performance evaluation study at the Department of Laboratory Medicine, Medical University of Vienna. The following aspects were verified: (a) the reliability of automatedly pre-classified differential counts, (b) the concordance of DI-60 counts with manual-microscopic differential counts and (c) the agreement of DI-60 and manual-microscopic results regarding clinically relevant findings. RESULTS: 82.3% of all leukocytes were correctly pre-classified. Cell categories with a low prevalence (eosinophils, basophils, progenitors/precursors) in non-hematological patients presented with a low positive predictive value (PPV), indicating a high frequency of false positives. Comparisons between visually adjusted results of the DI-60 and manual-microscopic differential counts revealed a good concordance for neutrophil and lymphocyte counts. Besides the detection of precursors/progenitors and normoblasts, no relevant systemic errors were detected. However, due to their low prevalence and technical aspects, the detection of basophilia, monocytosis or the presence of precursors/progenitors showed comparably low accuracies (error rates of 7.4%-24.1%). CONCLUSION: The DI-60 system works well for Wright-stained samples collected in the non-hematology ward. Due to the varying prevalence of cell categories found in peripheral blood, a low PPV can be expected with automatic assignment for those cells with low prevalence (e.g., basophils, eosinophils, precursor and progenitor cells, plasma cells). If the pre-test probability of these conditions is increased, manual microscopic processing may be recommended.