Gene Expression and Ex Vivo Drug Sensitivity Profiles in Children with Acute Leukemia: An Insight for High-Dose Therapy and Role of Etoposide and Fludarabine.

A distinct phenotype of combined ex vivo drug resistance and gene expression associated with this phenotype discriminates treatment outcome and identifies a subset of patients with a markedly inferior outcome in childhood acute lymphoblastic leukemia (ALL). No similar relationship was found so far in acute myeloblastic leukemia (AML). We hypothesized that drug sensitivity profile combined with gene expression profile can provide a new insight into selection of drugs used in high-dose therapy before hematopoietic stem cell transplantation and determine the role of specific drugs.

The aim of the study was to analyze the gene expression profile in correlation to the ex vivo obtained chemosensitivity profile of drugs used in high-dose therapy before hematopoietic stem cell transplantation in children with ALL and AML.

We tested leukemic cells from 56 children (43 ALL de novo, 8 relapsed ALL, 5 AML de novo) for ex vivo sensitivity to etoposide, fludarabine, 4-HOO-cyclophosphamide, busulfan, treosulfan and melphalan. The cells were then subjected to an assessment of global profile of gene expression to identify differentially expressed genes in drug-sensitive and drug-resistant ALL. Cells were isolated from bone marrow at the initial diagnosis of leukemia or its relapse. In vitro cytotoxicity was analyzed by means of the MTT assay. The drug resistance was expressed as the IC50, the inhibitory concentration to 50% of the cells. RNA was extracted with the use of Trizol, purified and assessed for integrity. Samples were hybridized to the Human Genome U133A Chip oligonucleotide microarrays (Affymetrix) according to manufacturer's protocol. Dataset was pre-processed by RMA method and small variability genes were filtered out. The final analysis was carried out in 13 835 probe sets. Non-parametric Spearman's correlations of IC50 and gene expression values were analyzed.

From all analyzed drugs, the largest number of genes significantly associated with the chemosensitivity profile across samples was observed for etoposide. We found 386 probests significantly correlated with the IC50 values (Spearman's correlation coefficient ranging from R==0.7 to R=0.61, p<0.001). NUDT21 gene (Nudix, nucleoside diphosphate linked moiety X-type, motif 21) was found to be highly significantly (False Discovery Rate, FDR<0.0001%) associated with chemosensitivity (larger expression associated with decreased IC50). Global test of association confirmed (p<0.001) that overall gene expression profile is associated with etoposide resistance, among selected genes there was over-representation of transcripts from chromosomes 1, 16 and 19. Two other drugs were found to exhibit noticeable association: fludarabine chemosensitivity was correlated with the expression of 197 probesets, global test of significance p=0.002; for cyclophosphamide 111 probesets and p=0.006 in global test was obtained. Much less significant association was observed for melphalan (72 genes, global test p=0.02) and almost no association was found for treosulfan (28 probesets, global test p=0.12) and busulfan (22 probesets, global test p=0.18). Fludarabine resistance was associated with an increased expression of FRY gene (Drosophila Furry homolog), Spearman's correlation coefficient R=0.61 (FDR=0.27%). Melphalan sensitivity was associated with biologically interesting differences in gene expression of PIP4K2A (Phosphatidyloinositol-5-phosphate 4-kinase, type II, alpha; over-expressed in patients with high IC50 for this drug) and POLR2C (RNA Polymerase II, DNA directed, polypeptide C, 33 kDa; down-regulated in this subset of cases).

Expression of a single genes is strongly associated with drug resistance in childhood acute leukemia, and it may aid to determine drug sensitivity to high-dose chemotherapy. Etoposide and fludarabine sensitivity seem to be related to the significant number of biologically important genes.

This study was supported by grants MNiSW N407 078 32/2964 and UMK BW 10/09. JSz and MJ contributed equally to the study.

No relevant conflicts of interest to declare.