Surface expression of Cytokine Receptor-Like Factor 2 increases risk of relapse in pediatric acute lymphoblastic leukemia patients harboring IKZF1 deletions

We prospectively examined whether surface expression of Cytokine Receptor-Like Factor 2 (CRLF2) on leukemic blasts is associated with survival and induction treatment response in pediatric B-cell precursor acute lymphoblastic leukemia (BCP-ALL) patients. Flow cytometric analysis of bone marrow-derived leukemia cells revealed that 7.51% (29/286) of 386 pediatric BCP-ALL patients were CRLF2-positive (CRLF2pos) at diagnosis. The median minimal residual disease (MRD) was lower in CRLF2pos than CRLF2-negative (CRLF2neg) patients on day 15 (MRD15) after induction therapy [0.01% (0.001-0.42%) vs. 0.45% (0.05-3.50%); p=0.001]. By contrast, the MRD15 was higher in Ikaros family Zinc Finger Protein 1 (IKZF1)-deleted BCP-ALL patients than in BCP-ALL patients without IKZF1 deletions [1.18% (0.06-12.0%) vs 0.33% (0.03-2.6%); p=0.003]. Subgroup analysis showed that MRD15 levels were lower in IKZF1Δ/CRLF2pos patients than in IKZF1Δ/CRLF2neg patients [0.1% (0.02-5.06%) vs. 2.9% (0.25-12%); p=0.005]. Furthermore, MRD15 levels were higher in IKZF1WT/CRLF2neg patients than in IKZF1WT/CRLF2pos patients [0.40% (0.04-2.7%) vs. 0.001% (0.001-0.01%)]. Despite the low MRD15 levels, IKZF1Δ/CRLF2pos patients showed poorer relapse-free survival (RFS) than other patient groups (p=0.003). These findings demonstrate that surface CRLF2 expression is associated with increased risk of relapse in pediatric BCP-ALL patients harboring IKZF1 deletions.


Minimal residual disease monitoring
Bone marrow samples were processed according to lyse/wash protocol, with 1x FACS Lyse used for erythrocyte lysis (Becton Dickinson) and analysed with FACS Canto II flow cytometer (Becton Dickinson). To ensure the sensitivity level of (at least) 10 -4 , lowcellular samples, were first subjected to erythrocyte lysis in 1x ammonium chloride solution (Pharm Lyse, Becton Dickinson) in a so-called bulk lysis protocol. Subsequently, the suspension of leukocytes was stained in a single 8-color tube with antibodies adequate for the blasts' phenotype at diagnosis. CRLF2 expression was not assessed during follow-up studies.

DNA and RNA isolation
Genomic DNA from the leukemic blasts, was isolated from previously Ficoll-gradient separated bone marrow samples using the FlexiGene DNA Kit (Qiagen, Valencia, USA), according to the manufacturer's protocol. Total RNA from 1×10 7 of leukemic cells resuspended in TRIzol solution was extracted using chloroform extraction according to the Invitrogen protocol. Nucleic acid purity and concentration was assessed on NanoDrop Spectrophotometer-8000 (Thermo Fisher Scientific, Walthman, USA).

Multiplex Ligation-Dependent Probe Amplification (MLPA)
Targeted copy number screening of eight selected loci was performed in the cohort using multiplex ligationdependent probe amplification (MLPA) using both the P202-B1 and P335-B2 SALSA MLPA kits (MRC-Holland, Amsterdam, The Netherlands). Both assays include probes for each of the eight exons of the IKZF1 gene and are able to detect deletions of the whole gene as well as all types of focal intragenic deletions. Selected exons of the genes BTG1, CDKN2A/B, EBF1, ETV6, IKZF1, PAX5, RB1, IGH and the PAR1 region (located approximately 230 kbp downstream of SHOX, CRLF2, CSF2RA and IL3RA) are also covered. Probe mix and hybridization buffer (MRC-Holland) were added in equal amounts to 120 ng of genomic DNA followed by heat denaturation and overnight hybridization of gene-specific probes at 60°C. Subsequently, ligation was performed and the ligation products were amplified by PCR using a 6-FAM fluorophore-labelled primer set (MRC-Holland). The amplification products were quantified and identified by capillary electrophoresis on an ABI 3130 DNA analyser (Life Technologies, Carlsbad, USA). Data were analysed using Gene Marker v.5.4 software (Softgenetics, State College, USA). Normalization of the data was carried out by dividing the peak area of each probe by the average area corresponding peak of control probes. This normalized peak pattern was divided by the average peak pattern of all the samples in the same experiment. The resulting values were 1 for every wild-type peak, 0.5 for heterozygous deletions and 1.5 for heterozygous duplications.

Detection of the P2RY8-CRLF2 fusion transcript
For those samples showing PAR1 deletion in MLPA, we confirmed the presence P2RY8-CRLF2 fusion on mRNA level. First strand cDNA synthesis was carried out on 1 μg of RNA using High-Capacity cDNA Reverse Transcription Kit (Life Technologies). The presence of P2RY8-CRLF2 chimeric transcript caused by the fusion of the first noncoding exon of P2RY8 and the first coding exon of CRLF2 was identified by RT-PCR with primers derived from both genes. The sequences of designed primers are listed in Supplementary Table I. Subsequently the product was sequenced using Sanger's method with a BigDye Terminator Cycle Sequencing kit v. 3.1 (Life Technologies) on an ABI 3130 DNA analyser (Life Technologies) following the manufacturer's instructions.

Analysis of ERG deletions
The presence of ERG deletions was tested using multiplex-PCR based method. We followed conditions and primers sequences which were published by Zaliova M et al [18]. In each positive case, after the verification of the presence of PCR product by electrophoresis on a 2% agarose gel and visualization under UV light, amplicons were purified using QIAquick PCR Purification Kit (Qiagen, Valencia, USA). Direct DNA sequencing using Sanger's method was performed with a BigDye Terminator Cycle Sequencing kit v. 3.1 (Life Technologies) on an ABI 3130 DNA analyser (Life Technologies) following the manufacturer's instructions. Resulting sequences were analysed using Sequencher software (Gene Codes Corporation, Ann Arbor, USA).

Direct sequencing for detection of point mutations in CRLF2 and JAK2 genes
To search for CRLF2 and JAK2 point mutations, direct sequencing was performed on DNA from leukemic cells with CRLF2 overexpression detected by flow cytometry. Primers for PCR reactions were designed in all coding regions of CRLF2 and JAK genes. The sequences of designed primers are listed in Supplementary Table 4. Amplification was carried out with an initial denaturation step at 95 ºC for 5 min, followed by 30 cycles of denaturation at 95 ºC for 30 s, annealing at 60 ºC for 30 s, extension at 72 ºC for 60 s, and a final extension step at 72 ºC for 10 min. After the verification of PCR products by electrophoresis on a 2% agarose gel and visualization under UV light amplicons were purified using QIAquick PCR Purification Kit (Qiagen, Valencia, USA). Direct DNA sequencing using Sanger's method was performed with a BigDye Terminator Cycle Sequencing kit v. 3.1 (Life Technologies) on an ABI 3130 DNA analyser (Life Technologies) following the manufacturer's instructions. Resulting sequences were analysed using Sequencher software (Gene Codes Corporation, Ann Arbor, USA) and aligned against reference sequences obtained from UCSC Genome Bioinformatics (NM_022148.3 and NM_004972) (University of California, Santa Cruz, CA, USA).

Statistical analysis
Differences in categorical variables were evaluated using Chi 2 test or two-tailed Fisher's exact tests. For continuous variables the Mann-Whitney's U test or Kruskall-Wallis nonparametric analysis of variance were used depending on the number of compared groups. Post-hoc comparisons were performed using the Dunn's test if significance was ascertained in Kruskal-Wallis test. Overall survival (OS) was defined from diagnosis to last follow-up, with censoring of alive patients at last follow-up. Relapse-free survival (RFS) was defined as the time from diagnosis to relapse. Estimated RFS and OS analyses were compared using the log-rank test. Kaplan-Meier survival curves were used to represent probabilities of survival over time. Multivariate analysis of survival was performed using Cox' proportional hazard regression models. Multivariate analysis of MRD was performed using analysis of covariance (ANCOVA) after log 10 of MRD values. All analyses were performed in Statistica Software version 12.5 PL (StatSoft, Tulsa, USA). P values <0.05 were considered statistically significant. Figure 1: Probability of relapse-free survival (RFS) with respect to MRD level at day 15 in the study cohort.