Computational analysis of flow-cytometry antigen expression profiles in childhood acute lymphoblastic leukemia: an MLL/AF4 identification

Precursor B-acute lymphoblastic leukemia (pB-ALL) is a heterogeneous disease and multiparameter flow cytometry, molecular genetics, and cytogenetic studies have all contributed to classification of subgroups with prognostic significance. Recently, gene expression microarray technology has been used to investigate lymphoblastic leukemias, demonstrating that known and novel pB-ALL subclasses can be separated on the basis of gene expression profiles. The strength of microarray technique lays in part in the multivariate nature of the expression data. We propose a parallel multiparametric approach based on immunophenotypic flow-cytometry expression data for the analysis of leukemia patients. Specifically, we tested the potential of this approach on a data set of 145 samples of pediatric pB-ALL that included 46 samples positive for mixed lineage leukemia (MLL) translocations (MLL+) and 99 control pB-ALLs, negative for this translocation ( MLL-). The expression levels of 16 marker proteins have been monitored by four-color flow cytometry using a standardized diagnostic panel of antibodies. The protein expression database has been then analyzed using those univariate and multivariate computational techniques normally applied to mine and model large microarray data sets. Marker protein expression profiling not only allowed separating pB-ALL cases with an MLL rearrangement from other ALLs, but also demonstrates that MLL+ leukemias constitute a heterogeneous group in which MLL/ AF4 leukemias represent a homogenous subclass described by a specific expression fingerprint.