Specificity of Markers of Leukemia Initiating Cells with a New Multiparameter Flow Cytometry Based Appraoch; Impact for Prognostic and Therapeutic Applications

Abstract

AbstractAbstract 1834Leukemia stem cells (LSC) are proposed to underly relapse of AML. In order to develop methods to specifically detect LSCs and to design LSC specific therapies, discrimination between LSC and normal hematopoietic stem cells (HSC) is a prerequisite. HSC and LSC may be either CD34-positive (CD34+CD38-) or CD34-negative. The latter may be contained in stem cell compartments defined by functional parameters: Hoechst 33342 efflux (side population, SP) and ALDH activity. To discriminate between LSCs and HSCs aberrant cell surface markers have been described. Additional flowcytometric parameters (scatter, and level of expression of CD34/CD45) allow further discrimination (Terwijn Blood 2009; 114:165). We applied this new approach to compare the specificity of expression of the LSC markers CD123, CLL-1, CD96, CD44, CD47, and CD33. We studied LSCs in diagnosis AML, HSCs in normal bone marrow (NBM) and regenerating BM (RBM) and HSCs present at diagnosis in the AML BM itself. The Table shows the results for the CD34+CD38-cells. Large differences were found for LSCs: CD44>CD33>CD47>CD123≥CLL-1 and CD96 (column 2; median and ranges shown in all columns). Expression on HSCs in NBM (column 4) shows the lack of LSC specificity of several of these markers: CD44>CD47>CD33>CD123>CD96 and CLL-1. A similar result was found for RBM for the three markers studied: CD33>CD123>CLL-1 (column 5). Expression in HSCs present in the AML BM itself was: CD33>CD123>CLL-1 and CD96 (column 3). CD44 and CD47 could not be properly analyzed due to extensive overlap of LSC and HSC (column 3). In order to determine whether expression on HSC would negatively impact diagnostic and targeted therapy, we defined an in vitro “therapeutic window” (TW=expr. in LSC/expr. in HSC) in columns 6–8: each individual value for LSC for a particular marker was divided by the median value of HSC in NBM (column 7), RBM (column 8) or divided by each corresponding HSC value in the AML BM itself (column 6). Overall, TW was slightly higher for CLL-1 than for CD96, CD123 and CD33, irrespective HSC source (CD33 had a high TW in diagnosis BM). The TW for CD44 and CD47 was low (column 7) and not analyzable in the AML BM itself. Experiments were repeated for SP cells: TW for CLL-1 (for the 3 HSC sources, resp) were 2.1, 3.8, and 3.8; for CD123: 2.0, 1.5, and 2.5 and for CD33: 1.9, 1.3, and 1.6. CLL-1 was thus slightly superior for SPs. The overall conclusions from this project are that, due to a high specificity, CD123, CLL-1, CD96 and CD33, but not CD44 and CD47 are suitable for diagnostic purposes. Whether or not a similar conclusion can be drawn for therapeutic applications remains to be investigated. Large differences exist in expression of the different antigens for most individual AML cases; for therapeutic applications, for each AML case, the optimal antigen or, most likely, combinations of antigens will be needed. The most important conclusion is that, by using our approach, it is now possible to discriminate between concomitantly present LSCs and HSCs. This approach may have high impact for both prognostic purposes (e.g. prediction of relapse) and prospective isolation of LSC and HSC from the same AML BM to design new therapeutic modalities. It may further serve as a surrogate marker to monitor effectiveness of a treatment and to design antibody/ligand combination therapies in a personalized setting. Disclosures:No relevant conflicts of interest to declare.