Statistical Analysis of Monocyte Monolayer Assay Validation in Brazilian Blood Donors

Abstract

Aims: To validate the monocyte monolayer assay (MMA) technique concerning its suitability for blood donor screening and its cost per test performed. The MMA is an in vitro simulation of the behavior of the antibodies, demonstrating the reactions that would occur in the endothelial reticulum system after a transfusion of incompatible red blood cells, indicating the risk of a hemolytic transfusion reaction and therefore of the clinical significance of the antibodies. Study Design: Blood samples of alloimmunized patients, selected at random from a blood donation bank, were submitted to validation tests recommended by the Brazilian National Health Surveillance Agency for the approval of new testing procedures. Place and Duration of Study: The following Brazilian institutions were involved between June 2009 and July 2010: Immunohematology laboratory of the Hematology and Hemotherapy Center of Santa Catarina state in Florianópolis, Department of Medicine (Medical Unit IV) and Department of Radiology of the Institute of Medical Sciences, Hospital Lahore in São Paulo. Methodology: Ninety samples of alloimmunized patients treated by the Santa Catarina blood donors were used. The validation tests evaluated the selectivity, linearity, precision, and accuracy of the MMA method and determined the limits of detection and quantification. External validation of the method was performed by comparing these results with those of an independent laboratory in São Paulo, while making sure that the latter was blind to the results of the former. The coefficient of variation was used to express the MMA testing precision of 5 replicates across 5 different concentration levels. Type I error for evaluating statistical significance was set at 5%. Results: Selectivity assessment of the impact of multiple alloantibodies on the MMA test result showed no statistically significant difference (P>0.05) across the titers of 64, 256, and 2048, each with three replications, thus confirming the test specificity. Homoscedasticity of the monocyte index (MI) data was not refuted by Levine's test with the F-value of 0.746, much below the value of 3.056 needed to achieve a statistical significance level of P<0.05. MI linearity against the logarithm of the alloantibody concentration was shown in a simple linear regression where the latter predicted 83% of the variation in the former, and the regression slope of 0.4 (95% confidence interval 0.32, 0.48). The limits of detection and quantification on the logarithm scale were 0.28 and 0.84, respectively. External validation found no statistically significant difference between the MMA test results from the two independent laboratories. The coefficient of variation of <15% indicated good MMA testing precision under routine laboratory conditions. Conclusion: The assay met all validation criteria and was therefore effective in assessing the clinical significance of alloantibodies.