Abstract
Introduction: Devices such as the CellaVision® DM96 (CellaVision AB, Lund, Sweden) locate and image nucleated cells on blood smears. Using image recognition software, the DM96 also pre-classifies those cells into differential categories similar to the most complex hematology analyzers. We compared the cell counts, differential counts and flagging information gained from a complex hematology analyzer, the XE5000 (Sysmex, Kobe, Japan), with information from a minimal hematology analyzer (Sysmex PocHi) plus the DM96. We found that the cell counts, differential and flagging capabilities are similar, but the PocHi plus DM96 advantages include allowing remote review of the blood smear.
Methods: 210 blood samples, selected for various abnormalities, had complete blood counts with automated differentials produced by a Sysmex XE5000 hematology analyzer. These results were compared with cell counts from the Sysmex PocHi hematology analyzer, their 100-cell DM96 post reclassification differentials, and with DM96 pre-classification differentials using standard regression analyses and Rumke 95% confidence intervals (CI) as calculated using the Clopper-Pearson method. Flagging by the XE5000 for immature granulocytes (IG's) and for blasts/abnormal cells was compared to the DM96 pre-classification using truth tables with the DM96 post reclassification as the gold standard. The following translations were used to compare flagging: IG's > 2 for either post reclassification DM96 differential, XE5000 or the DM96 pre-classification differential; Any blast cells on the manual differential were compared to XE5000 flags WBC Abn Scg; NRBC Abn Scg; Blasts?; Atyp LY?; Abn Ly/ L_Bl? and to DM96 pre-classification % Blasts > 0%; unidentified cells >3%. .
Results: Non-differential blood count parameters including white blood count, red blood count, hemoglobin, mean corpuscular volume (MCV), and platelet count showed excellent correlation between the PocHi and the XE5000 with R2>0.95. Differential-dependent blood count parameters including neutrophils, lymphocytes, monocytes, eosinophils, basophils and immature granulocytes showed excellent correlation between the XE5000 and pre-classification DM96 with R2>0.95. Nucleated red cells also showed excellent correlation between the XE5000 and the DM96 with R2>0.85.
For blasts/abnormal cells, the DM96 showed 100% sensitivity and 40% specificity with 0% false negatives. The XE5000 showed 93% sensitivity and 19% specificity with 3% false negative. Two of the false negatives were shared by both instruments and were 1% blasts. Of the three false negatives with the XE5000 that were true positives with the DM96, two had 1% blasts while one had 2% blasts. For immature granulocytes (IG's), the XE5000 showed 94% sensitivity and 79% specificity with 2% false negatives. The DM96 showed 85% sensitivity and 95% specificity with 4% false negatives. All of the false negatives were for IG's < 5%
Conclusions: Pairing the DM96 or a similar imaging instrument with a relatively inexpensive hematology analyzer, such as those commonly used in physician office laboratories, would provide all of the information available from expensive, complex hematology analyzers in high throughput laboratories AND allow remote review of the blood smear findings by experts.
Disclosures
Broome: CellaVision: Consultancy, Research Funding. Bengtsson:CellaVision AB: Employment, Equity Ownership. Palm:CellaVision: Employment, Equity Ownership.