RESUMO
OBJECTIVES: The impact of seven hemoglobin variants (Hb Q-Thailand, Hb G-Honolulu, Hb Ube-2, Hb New York, Hb J-Bangkok, Hb G-Coushatta, and Hb E) on the outcome of HbA1c was investigated for six methods by comparing with liquid chromatography-tandem mass spectrometry (LC/MS/MS) reference method. METHODS: Twenty-nine normal and 112 variant samples were measured by LC/MS/MS, Sebia Capillarys 3 TERA, Intelligene Biosystems QuanTOF, Premier Hb9210, Arkray HA-8190V, Bio-Rad D-100, and Tosoh G11, then evaluated for correlation, consistency, and mean relative bias among six methods. The lowest biological variation bias of ±2.8â¯% was an acceptable standard. RESULTS: All methods showed poor correlation and consistency with LC/MS/MS for Hb E. The unacceptable biases were observed for Capillarys 3 TERA (-14.4 to -3.7â¯% for Hb Q-Thailand, Hb Ube-2, Hb New York, Hb J-Bangkok and Hb E), QuanTOF (-8.3 to -2.9â¯% for Hb Ube-2, Hb New York and Hb G-Coushatta), Premier Hb9210 (-18.3 to -3.6â¯% for Hb Q-Thailand, Hb Ube-2, Hb New York, Hb J-Bangkok and Hb E), HA-8190V variant mode (-17.3 to 6.6â¯% for Hb G-Honolulu, Hb Ube-2, Hb New York, Hb G-Coushatta and Hb E). All variant samples showed larger biases than ±2.8â¯% comparing HA-8190V fast mode, D-100, and G11 with LC/MS/MS. CONCLUSIONS: The accuracy of different HbA1c methods was influenced by some Hb variants, especially Hb Ube-2 and Hb New York. Thus, laboratories need to choose appropriate methods to measure HbA1c with different Hb variants.
RESUMO
Objective: This study aimed to assess the current status of carcinoembryonic antigen (CEA) detection. We evaluated the correlation, consistency, and comparability of CEA results among six automated immunoassays, and combined with the results of CEA trueness verification of the Beijing Center for Clinical Laboratories (BCCL) for further analysis. Methods: Abbott Architect i2000, Beckman DxI800, Roche Cobas E601, Diasorin Liaison XL, Maccura IS1200, and Autolumo A2000 were used to detect 40 individual serum CEA samples. Taking the optimal analytical quality specifications calculated from data on biological variation as the evaluation criterion. Passing-Bablok regression and Bland-Altman analysis were performed between each assay and all-assays median values to evaluate the correlation and relative difference. The concordance correlation coefficient (CCC) was used for consistency analysis. Additionally, the trueness verification program used samples at three concentration levels to assess the bias, coefficient of variation (CV), and total error (TE) between the average measured values and the target value. Results: The Spearman's rank correlation coefficient (rs) was ≥0.996 and the CCC ranged between 0.9448 and 0.9990 for each assay vs. all-assays median. Considering the all-assays median value of each sample as a reference, there were proportional and systematic differences according to the Passing-bablok regression analysis. The relative difference of the four assays (Abbott Architect i2000, Autolumo A2000, Diasorin Liaison XL, and Maccura IS1200) met the optimal analytical quality specifications. On the other hand, Beckman DxI800 (13.2 %) and Roche Cobas E601 (-9.0 %) were only able to fulfill the desirable analytical quality specifications. The average pass rates for bias, CV, and TE of the trueness verification program were 80 %, 98 %, and 96 %, respectively. Conclusions: The six automated immunoassays vs. all-assays median have a good correlation in CEA detection. However, there is a lack of comparability of CEA results. Further improvements are needed in harmonization among CEA detections.