Abstract
Purpose :
Automated hematology is used to count and differentiate cells in blood samples, but there are no validated instruments for use in ocular surface tear samples. This study was done to evaluate an automated hematology analyzer (DxH 500, Beckman Coulter) for use in human tear samples.
Methods :
Nine tear samples were collected by tear wash (TW), using 1-2 mL non-preserved saline (0.9% NaCl, Addipak, Teleflex Medical) to wash and collect ocular surface tears. Samples were centrifuged and the cell pellet resuspended in 200 uL saline and run 3 consecutive times on the DxH 500 (12 uL per run). Two methods were compared to the DxH outputs: standard hemocytometry (done 3x) to compare total cells/uL, and fixation (4% PFA), staining (modified H&E protocol), imaging (40X, EVOS M5000), and manual counting of 100 cells per subject (done 3x) to compare cell differentiations. Proportions of neutrophils (NE), eosinophils (EO), lymphocytes (LY), basophils (BA), and monocytes (MO) were determined. The methods were compared using Bland-Altman (BA) analysis and paired t-tests.
Results :
Total cell concentration (median; IQR) was significantly greater with the DxH (603 cells/uL; 327, 2365) compared to manual hemocytometry (441 cells/uL; 265, 2067), P = .01, with a BA bias of 213 ± 213 cells/uL (LOA: -204 to 629 cells/uL) between methods. However, a smaller coefficient of variation was found using the DxH (7 ± 8%) versus manual hemocytometry (14 ± 9%). Cell differentiation comparisons show significant differences between NE (BA bias of 30 ± 15%; LOA: 1 to 59; P < .01), EO (-8 ± 6%; LOA: -20 to 4; P < .01), and BA proportions (-2 ± 2%; LOA: -6 to 2; P = .03), although 19 ± 12% of granulocytes were unidentifiable by manual counting. There were no significant differences in the proportion of LY (-3 ± 4%; LOA: -10 to 5; P = .05) or MO (-8 ± 23%; LOA: -52 to 36; P = .31) detected between the two methods.
Conclusions :
Although the DxH consistently detected greater cell concentrations than manual hemocytometry, there was general agreement between DxH and manual hemocytometry and the repeatability of the DxH was better. Cell differentiation outcomes generally agreed for non-granulocytes, but did not agree for NE and EO, which are phenotypically similar, which may be due to the difficulty in manual differentiation, warranting further comparisons of the DxH to reliable differentiation methods.
This abstract was presented at the 2023 ARVO Annual Meeting, held in New Orleans, LA, April 23-27, 2023.