Prognostic Impact of Ikaros (IKZF1) Gene Alterations In Childhood ALL Treated with ALL IC-BFM 2002 Protocol: A Comparison of Gene Expression and Genomic-Based Methods.

Recently, Ikaros (IKZF1) gene alterations were found to predict poor prognosis in childhood acute lymphoblastic leukemia (ALL). Thus, the implementation of IKZF1 status into the risk group stratification is discussed. So far, limited data are available concerning both IKZF1 importance in different treatment protocols for Ph-negative ALL and the choice of the best diagnostic method. In this study, we compared two methods based on either genomic DNA examination or gene expression analysis, and their prognostic impact within a treatment protocol for childhood ALL.

Gene expression of functional (IK1, IK2) and non-DNA binding (IK4, IK6, IK8) IKZF1 isoforms was semi-quantitatively evaluated using Lab-on-a-chip (Agilent) electrophoresis and reported either as an absolute level or relatively to the total isoform signal. The thresholds for abnormal gene expression were set based on the analysis of peripheral blood (PB) of healthy donors, remission bone marrow (BM) samples of children with ALL, and sorted B- and T-cell precursor subpopulations. MLPA (multiplex ligation-dependent probe amplification) reaction including probes for Ikaros exons 1 to 8 was performed on BM DNA samples and the products were analyzed with the Coffalyser v9.4 software.

Results of both gene expression and MLPA analysis in the BM were available for 120 of 193 children diagnosed with ALL between 2002 and 2005. MLPA analysis revealed a deletion of at least one exon of IKZF1 gene in 17/120 (14%) patients. Of note, two patients with this deletion underwent a lineage switch (LS) from ALL to AML during the induction treatment. The entire IKZF1 gene was mononoallelically deleted in 3 patients. The ratio between non-DNA binding (IK4, IK4del, IK4A, IK6, Ik6del, IK8) and functional (IK1 and IK2) isoform expression was significantly elevated (non-DNA binding isoforms>70%) in 21 of 120 (18%) patients. The expression of a dominant-negative IK6 isoform was significantly elevated (>20% of total) in 7 of 120 (6%) patients. Surprisingly, gene deletion on one allele was not accompanied by decrease of total IKZF1 gene expression. On the contrary, patients with IKZF1 deletion had higher total IKZF1 transcript level than those without deletion (p=0.008, Mann Whitney), but it was not possible to set any reliable expression threshold for the prediction of gene deletion. The change on a DNA level was not always reflected in relative gene expression: Of 17 patients with gene deletion, only 7 had significantly altered short/long isoform ratio and 5 patients had an increased expression of IK6. Conversely, of 7 patients with IK6 overexpression, two patients had no DNA alteration, suggesting a different mechanism of altered gene expression. We next evaluated the prognostic impact of IKZF1 alterations in 113 patients treated with ALL IC-BFM 2002 protocol (5 patients were excluded due to Ph-positive ALL with imatinib-based treatment and 2 patients due to treatment change after LS). Patients with IKZF1 gene deletion had significantly worse relapse-free survival (RFS) than other patients (5-year RFS 50.0±14.4% vs. 90.8±2.9%, p=0.0002). The presence of IKZF1 deletion did not correlate with minimal residual disease (MRD) during induction treatment (days 8, 15, 33), neither in BM nor in PB. Patients with IK6 overexpression had 5-year RFS 50.0±20.4% compared to 88.4± 3.2% in those with low IK6 expression (p=0.004). The elevated short/long isoform ratio (>70%) had no prognostic impact. The best prediction of relapse was achieved via combining two factors: the presence of IKZF1 deletion detected by MLPA or the relative IK6 overexpression (>50% of total isoform signal). The 5-year RFS was 50.0±13.4% for this group (14 pts, 7 relapses) compared to 91.6±2.8% for other patients (99 pts, 8 relapses, p<0.0001).

This study confirmed that the presence of Ikaros gene alterations was connected with a high risk of relapse also in a BFM-based protocol for Ph-negative childhood ALL treatment. The deletion within IKZF1 locus did not necessarily correlate with an altered Ikaros gene expression. Ideally, both genomic and gene expression-based approach should be applied together for the evaluation of prognosis. However, if this is not possible, the examination of DNA changes by MLPA identifies more patients who subsequently relapse than the gene expression-based approach.

VZ MSM 0021620813, P301/10/1877, IGA NS/10472-3

No relevant conflicts of interest to declare.