Rearrangement of CRLF2 is associated with mutation of JAK kinases, alteration of IKZF1, Hispanic/Latino ethnicity, and a poor outcome in pediatric B-progenitor acute lymphoblastic leukemia

Abstract

Gene expression profiling of 207 uniformly treated children with high-risk B-progenitor acute lymphoblastic leukemia revealed 29 of 207 cases (14%) with markedly elevated expression of CRLF2 (cytokine receptor-like factor 2). Each of the 29 cases harbored a genomic rearrangement of CRLF2: 18 of 29 (62%) had a translocation of the immunoglobulin heavy chain gene IGH@ on 14q32 to CRLF2 in the pseudoautosomal region 1 of Xp22.3/Yp11.3, whereas 10 (34%) cases had a 320-kb interstitial deletion centromeric of CRLF2, resulting in a P2RY8-CRLF2 fusion. One case had both IGH@-CRLF2 and P2RY8-CRLF2, and another had a novel CRLF2 rearrangement. Only 2 of 29 cases were Down syndrome. CRLF2 rearrangements were significantly associated with activating mutations of JAK1 or JAK2, deletion or mutation of IKZF1, and Hispanic/Latino ethnicity (Fisher exact test, P < .001 for each). Within this cohort, patients with CRLF2 rearrangements had extremely poor treatment outcomes compared with those without CRLF2 rearrangements (35.3% vs 71.3% relapse-free survival at 4 years; P < .001). Together, these observations suggest that activation of CRLF2 expression, mutation of JAK kinases, and alterations of IKZF1 cooperate to promote B-cell leukemogenesis and identify these pathways as important therapeutic targets in this disease.

Trial registration: ClinicalTrials.gov NCT00005603.

Figure 1

Association of elevated gene expression of CRLF2 with JAK mutations and IKZF1 alteration in high-risk B-progenitor ALL. (A) CRLF2 expression levels generated by Affymetrix gene expression profiling (circles) in 207 high-risk B-precursor ALL patients. y-axis represents CRLF2 intensity; x-axis, 207 ALL patients ordered from lowest to highest CRLF2 expression levels. Red circles represent samples with CRLF2 rearrangement; blue circles, patients with no detectable CRLF2 rearrangements; and yellow triangles, patients with IKZF1 alterations. Dashed box highlights samples with CRLF2 expression levels more than 10-fold higher than the median level. (B) Expanded view of the 29 ALL cases with CRLF2 rearrangement showing the association of CRLF2 rearrangement with IKZF1 alterations (yellow triangles), and mutations in JAK1 (orange diamonds) or JAK2 (blue diamonds). Detailed information on each of the 29 cases may be found in Table 2.

Figure 2

Design of FISH assays to detect genomic CRLF2 rearrangements. (A-B) The 3 BAC clones flanking the CRLF2 locus and 1 centromeric to IGH@ on chromosome 14, respectively. The red arrow in panel A highlights the PAR1 region involved in the deletions that join P2RY8 to CRLF2. The colors and names of the BAC probes used to perform FISH in panels C to H are shown in the lower left of each panel. (C-D) Results from the IGH and the CRLF2 break-apart assays, respectively. Arrows indicate the split signals. (E-F) The IGH-CRLF2 fusion probes on interphase and metaphase cells, respectively. Arrows in panel E highlight the fusion signal, whereas arrows in panel F indicate normal signals from X, Y, and 14 as well as the fusion signal on der14. (G) The loss of 261P4 in 1 sample resulting from PAR1 deletion. (H) The same sample shows that the signal is regained when a more centromeric BAC (74L17) is used, and 2 normal fusion signals are seen. Areas of cellular debris and nuclei lacking red/green signals in the same focal plane as the other cells were masked during image capture.

Figure 3

Characterization of PAR1 deletion centromeric to CRLF2 and PCR detection of the P2RY8-CRLF2 fusion. (A) Representative log₂ ratio SNP 6.0 microarray DNA copy number data for 5 samples with PAR1 deletions. SNP array data derived from matched normal DNA is indicated by “N” and the leukemic sample with “T.” The black box highlights the region of PAR1 deletions. (B) RT-PCR demonstrating the fusion transcript of P2RY8 and CRLF2. Full-length cDNA transcripts were subjected to PCR with primers from P2RY8 and CRLF2 to generate the fusion products. (C) Genomic PCR of the chromosomal breakpoints joining CRLF2 and P2RY8. (D) Sequence of the RT-PCR product showing the junction of exon1 of P2RY8 with exon 1 of CRLF2. P2RY8 exon 1 is noncoding. (E) Germline sequence of the breakpoint. The exact junction of 5′ flanking sequence of CRLF2 to intron 1 of P2RY8 is shown. Several nonconsensus nucleotides are present at the junction.

Figure 4

Flow cytometric detection of CRLF2 protein expression. Flow cytometric measurements of CRLF2 protein expression are shown for 3 ALL samples and a normal control: 1 indicates normal buffy coat; 2, 9906_019; 3, 9906_055; and 4, 9906_258. These are ordered by their increasing levels of CRLF2 expression (determined by microarray). The leftmost panels (A) show scatter plots of the control antibodies (IgG1 and IgG2a). The middle panels (B) represent scatter plots for CRLF2 (y-axis) and CD34 (x-axis). The right panels (C) chart the relative intensity shift by histogram overlay, with the dashed lines corresponding to CRLF2. Sample 9906_258 (A-C) has the highest expression of CRLF2 in the cohort and contains the CRLF2-IGH@ translocation. The other 2 samples (9906_019 and 9906_055) have CRLF2 levels less than median.

Figure 5

Kaplan-Meier survival curves showing interaction of CRLF2 with IKZF1 alterations and JAK mutations. (A) Survival for patients with or without CRLF2 rearrangements with or without JAK mutations is shown. x-axis is time in years; y-axis, probability of RFS. The CRLF2-rearranged samples are shown in red, whereas the nonrearranged samples are in black. Solid lines indicate JAK mutations; and dashed lines, no detected JAK mutations. (B) Survival for patients with or without CRLF2 rearrangements with or without IKZF1 mutations/deletions is shown. x-axis represents time in years; y-axis, probability of RFS. The CRLF2-rearranged samples are shown in red, whereas the nonrearranged samples are in black. Solid lines represent IZKF1 alterations; dashed lines, none.