Analytical comparison between two hematological analyzer systems: CAL‐8000 vs. XN‐9000. (4th November 2016)

• Record Type: Journal Article
• Authors: Buoro, S.; Mecca, T.; Seghezzi, M.; Manenti, B.; Azzarà, G.; Dominoni, P.; Crippa, A.; Ottomano, C.; Lippi, G.
• Is Part Of: International journal of laboratory hematology. Volume 39:Number 2(2017:Apr.)
• Page Start: 147
• Page End: 162
• Publication Date: 2016-11-04
• Subjects: XN‐9000 -- CAL‐8000 -- automated count -- optical microscopy -- mindray -- sysmex; Hematology -- Periodicals; Blood -- Diseases -- Periodicals; Hematology -- Periodicals; 616.15005
• DOI: 10.1111/ijlh.12597
• Languages: English
• ISSNs: 1751-5521

Abstract

Summary: Introduction: This study was aimed to compare the analytical performance of traditional and new parameters and morphological flags of CAL‐8000 and XN‐9000. The automated differential leukocyte count (DIFF profile) and morphological flags were compared with optical microscopy (OM). Methods: A total of 1025 peripheral blood samples, collected in K3 EDTA tubes, were analyzed by CAL‐8000, by XN‐9000, and by OM. Within‐run imprecision was performed in low cellularity samples. The comparison was made using Spearman's correlation, Passing–Bablok regression, Bland–Altman bias, and Cohen's K test. Results: Within‐run imprecision in low cellularity samples yielded reproducible data between the instruments (imprecision was higher than 10% on samples with platelet count <21 × 10 9 /L using impedance technology). Passing–Bablok regression (CAL‐8000 vs . XN‐9000) yielded slopes ranging between 0.2 to 1.16 and intercepts from −6.54 to 21.63. The bias for leukocytes parameters ranged from −1.8% to −82.2%, the red blood cell parameters from −2.9% to 3.1%, platelets parameters from −27.8% to 26%, and reticulocyte parameters from −115.3% to 4.5%. The comparison of morphological flags yielded a K value always <0.55. The DIFF profile vs . OM had a Passing–Bablok regression with slopes ranging between 0.34 to 1.00 and intercepts from −0.01% to 0.11 and bias ranging from −42.9% to 2.6% for XN‐9000 parameters and from −2.7% to 35.0% for CAL‐8000 parameters. The comparison of morphological flags showed a K value ranging from 0.35 to 0.77 for XN‐9000 and from 0.17 to 0.54 for CAL‐8000. Conclusion: Differences exist between the two analyzers, especially in the generation of morphology flags, thus emphasizing the need of pursuing a major degree of harmonization and/or adopting instrument‐specific reference ranges.