Discriminant function analysis as decision support system for the diagnosis of acute leukemia with a minimal four color screening panel and multiparameter flow cytometry immunophenotyping

Abstract

Despite several recommendations for standardization of multiparameter flow cytometry (MFC) the number, specificity and combinations of reagents used by diagnostic laboratories for the diagnosis and classification of acute leukemias (AL) are still very diverse. Furthermore, the current diagnostic interpretation of flow cytometry readouts is influenced arbitrarily by individual experience and knowledge. We determined the potential value of a minimal four-color combination panel of 13 monoclonal antibodies (mAbs) with a CD45/sideward light scatter-gating strategy for a standardized MFC immunophenotyping of the clinically most relevant subgroups of AL. Bone marrow samples from 155 patients with acute myeloid leukemia (AML, n=79), B-cell precursor acute lymphoblastic leukemia (BCP-ALL, n=29), T-cell precursor acute lymphoblastic leukemia (T-ALL, n=12) and normal bone marrow donors (NBMD, n=35) were analyzed. A knowledge-based learning algorithm was generated by comparing the results of the minimal panel with the actual diagnosis, using discriminative function analysis. Correct classification of the test sample according to lineage, that is, BCP-ALL, T-ALL, AML and differentiation of NBMD was achieved in 97.2% of all cases with only six of the originally applied 13 mAbs of the panel. This provides evidence that discriminant function analysis can be utilized as a decision support system for interpretation of flow cytometry readouts