Validation of genes for H-ARS severity prediction in leukemia patients - interspecies comparison, challenges, and promises.

In a previous baboon-study, a total of 29 genes were identified for clinical outcome prediction of the hematologic, acute, radiation, syndrome (H-ARS) severity. Among them, four genes ( Peripheral blood was drawn from 10 leukemia patients before and up to 3 days during a fractionated (2 Gy/day) total-body irradiation (TBI) with 2-12Gy. After RNA-isolation, gene expression (GE) was evaluated on 31 genes widely used in biodosimetry and H-ARS prediction employing qRT-PCR. A customized low-density-array (LDA) allowed simultanously analyzing all genes, the 96-well format further examined the four most promising genes. Fold-changes (FC) in GE relative to pre-irradiation were calculated. Five patients suffering from acute-lymphoblastic-leukemia (ALL) respectively non-Hodgkin-lymphoma (NHL) revealed sufficient RNA-amounts and corresponding lymphocyte and neutrophile counts for running qRT-PCR, while acute-myeloid-leukemia (AML) and one myelofibrosis patient could not supply enough RNA. Generally, 1-2µg total RNA was isolated, whereas up to 10-fold differences in RNA-quantities (associated suppressed GE-changes) were identified among pre-exposure and exposure samples. From 31 genes, 23 were expressed in at least one of the pre-exposure samples. Relative to pre-exposure, the number of expressed genes could halve at 48 and 72h after irradiation. Using the LDA, 13 genes were validated in human samples. The four most promising genes (vid. sup.) were either undetermined or too close to pre-exposure. However, they were measured using the more sensitive 96-well format, except Identified genes for H-ARS severity prediction, previously detected in baboons, were validated in ALL but not in AML patients. Limitations related to leukemia type, associated reduced RNA amounts, suppressed GE changes, and methodological challenges must be considered as factors negatively affecting the total number of validated genes. Based on that, we propose additional controls including blood cell counts and preferably fluorescence-based RNA quantity measurements for selecting promising samples and using a more sensitive 96-well format for candidate genes with low baseline copy numbers.