Automated flow cytometric cell count and differentiation of canine cerebrospinal fluid cells using the ADVIA 2120

Background: Microscopic cell counts in cerebrospinal fluid (CSF) are time-consuming and prone to imprecision. The recently introduced automated hematology analyzer ADVIA 2120 offers an automated cell count and differential for CSF in the veterinary software mode based on laser light scatter and absorbance measurements. Objectives: The purpose of this study was to evaluate the precision, linearity, and accuracy of the ADVIA 2120 CSF assay. Methods: Sixty-seven CSF samples were analyzed on the ADVIA 2120 and total nucleated cell counts (TNCC) and RBC counts were compared with the hemocytometer results. In 21 samples with TNCC >5/muL, ADVIA 2120 results were compared with 100-300 cell manual differentials performed on cytocentrifuged preparations. Statistical analysis included Spearman's rank correlation, Passing-Bablok regression, and Bland-Altman analysis. Results: Repeatability (intra-assay) coefficients of variation (CVs) ranged from 4.19% to 25.94%. Interassay CVs ranged from 2.56% to 28.67%. Accurate results within 30% were achieved for TNCC up to 4000/muL. Except for low TNCC, deviation from the expected value was higher (TNCC of 8/muL instead of 4/muL). The following correlation coefficients (r) and biases were achieved compared with the reference method: r=.90 and bias 2.3/muL for TNCC; r=.88 and bias 32.0/muL for RBC counts; r=.86 and bias +/-13.4% for mononuclear and polymorphonuclear cell percentages; r=.88 and bias -6.1% for lymphocyte percentage; r=.56 and bias 19.4% for monocyte percentage; and r=.75 and bias -9.7% for neutrophil percentage. Conclusion: Our results demonstrated that the automated ADVIA 2120 CSF assay generally compares well with reference methods although there are some limitations for the automated monocyte count and for samples with only mild pleocytosis.