AZ Sint-Lucas Gent
The aim of this research is to evaluate the performance of the new body fluid module on the Sysmex-XN hematology analyzer (XN-BF) for blood cell count and differential in body fluids in the laboratory of AZ Sint-Lucas Ghent. Along with other validation parameters, a method comparison with manual microscopy was performed to evaluate the accuracy of the body fluid module.
Automated blood cell count and differential in body fluids add a great value to medical laboratories in comparison to the labor-intensive and time-consuming manual microscope method. The hematology analyzer needs to be validated according to the laboratories standards; therefore the XN-BF was evaluated according to these standards.
During a period of nine weeks, 110 samples (42 pleural fluids, 17 synovial fluids, 22 ascites, 2 continuous ambulatory peritoneal dialysis (CAPD) and 27 cerebrospinal fluids (CSF)) were used for method comparison between the XN-BF and manual microscopy for blood cell counting (Fuchs-Rosenthal counting chamber) and differential (cytospins). Further evaluation included inter-observer variation (manual method), precision, carry-over, linearity and Lower Limit of Quantitation (LLoQ).
Inter-observer variation for the manual counting method showed a coefficient of variation of 8,17% for total nucleated cell count (TC) and 4,66% for red blood cell count (RBC). For the method comparison in pleural fluids, good correlations (R2=0,972; R2=0,994) were found for TC and RBC >1000/µl counts, despite the bias found with the Passing and Bablok regression analysis in TC (y=80,7980+1,1919x). Excellent correlation for TC and RBC (R2>0,980) was found in synovial fluids, yet a constant bias was found again for TC (y=154,4702+0,9890x). In ascites and CAPD, good correlations are found as well (TC: R2=0,964; RBC: R2=0,989). However, the XN-BF systematically counted more TC compared to the manual microscopy (y=-0,03084+1,2419x). In CSF <10 TC/µl (n=21) no significant or constant bias was found, but correlations were worse compared to the other fluids (R2=0,817). Regardless, good correlations were found for RBC (n=25) and TC >5/µl (n=7). For the differential count, the XN-BF systematically counted less monocytes compared to the manual method in all fluids. Results of comparing the manual reporting for malignant cells with the High-Fluorescent fraction on the XN-BF showed that these cells are located in this region, but that the XN-BF is not specific enough for quantitation of these cells. Furthermore carry-over was negligible, precision was excellent (CV%: TC=2,3%; RBC=0,0%; neutrophils=6,4%; lymphocytes=4,2%; monocytes=5,1%), linearity was good for both RBC and TC (R2=0,99) and the LLoQ for TC was defined at 4/µl.
The XN-BF can be used in the laboratory for fast counting of TC and RBC, if nucleated cell counts <10/µl are counted again manually (CSF). Differentiation through microscopy is still needed due to the low correlations in monocyte differentiation and the lack of precision for reporting of malignant cells.