Genetic characteristics and treatment outcomes of pediatric acute lymphoblastic leukemia patients with CRLF2 rearrangements Free

Background. Acute lymphoblastic leukemia (ALL) with CRLF2 rearrangements (CRLF2r) is one of the most common subtype of BCR::ABL1-like ALL.

Objective. To characterize clinical, immunological and genetic parameters of pediatric ALL patients carrying CRLF2r.

Patients and methods. Nine hundred and fifty-eight children with B-lineage ALL, enrolled into the national multicenter ALL-MB 2015 trial, were included in the current study. The median age was 4.2 years (range 1.02-17.9). All initial patients' samples were screened by flow cytometry (FC) using TSLP receptor monoclonal antibody (clone 1A6) and split-signal FISH for CRLF2 rearrangement. Positive by each method cases were further investigated by multiplex ligation-dependent probe amplification (MLPA) with P335 and P327 kits (both MRC-Holland, Netherlands) and customized next-generation sequencing panels. Hybridization-based capture targeted RNA panel (Nanodigmbio, Singapore), that covered all coding regions and sites of splicing in genes, previously described in BCR::ABL1-like ALL and hybridization-based capture DNA panel, covered coding and non-coding regions of IGH gene were applied. Informed consents were obtained in all cases.

Results. CRLF2r were detected in 58 patients (6.1%). Overall correct prediction of CRLF2r by FC TSLP receptor detection was 97.7% with only 2 false-negative cases. All cases where CRLF2 gains were revealed (n=256) by MLPA or FISH (mainly high hyperdiploid patients) were correctly interpreted as CRLF2-negative by FC. Among CRLF2r, 42 P2RY8::CRLF2-positive (72.4%) and 16 IGH::CRLF2-positivecases (27.6%) were verified. There were 10 (17.2%) Down syndrome ALL (DS-ALL) cases, all of these patients carried P2RY8::CRLF2 fusion gene. Patients with IGH::CRLF2 were significantly older (p=0.001), more often had initial WBC ≥50*10⁹/L (p<0.001) and allocated to the NCI high-risk group (p<0.001). JAK1/2 mutations (n=6, 14.0%) were exclusively found in P2RY8::CRLF2-positive group. The proportion of JAK1/2-mutated cases in our cohort was significantly lower, than previously reported, regardless of NCI risk group allocation (14.3% in NCI standard risk group and 13.6% in NCI high-risk group). RAS mutations (4 in KRAS, 5 in NRAS, 1 in PTPN1, 23.4% in total) were equally distributed in IGH::CRLF2-positive (26.7%) and P2RY8::CRLF2-positive (21.4%) groups (p=0.268). PAX5alt (4 point mutations and 4 fusion genes, including PAX5::ZCCHC7, PAX5::CBFA2T3, PAX5::NOL4L) were detected in 18.6% of cases. Concomitant PAX5alt and hotspot NRAS/KRAS mutations were found in 3 cases, PAX5alt and JAK2 in 1. IKZF1 deletions were found in 17 out of studied 45 (37.7%) cases, while IKZF1^plus profile was verified in 10 (22.2%). In IGH::CRLF2-positivecases genomic DNA breakpoint positions in IGH were identified within the following genomic coordinates of chr14:105862758-105916831 (HG38) and intergenic region between CRLF2 и CSF2RA (chrX:1213579-1239697). Event-free survival (EFS), overall survival (OS) and cumulative incidence of relapse (CIR) in the whole cohort were as follows - 0.49 (SE 0.22), 0.81 (SE 0.12) and 0.46 (SE 0.12), respectively with median follow-up time of 5.3 years. Median time to relapse was 39.4 months (range 11.8-125.2), Isolated BM relapses occurred in 9 cases (60.0%), extramedullary and combined relapses in 3 cases each (20.0%). HSCT were done in 9 cases (15.5%), 8 of them in CR2, 1 in CR1. Two patients after HSCT died of relapsed, the rest 7 are alive in CCR. Of note, end of induction MRD did not allow predicting outcome (EFS 0.73 (SE 0.18) in fast MRD-responders vs 0.61 (SE 0.12) in slow MRD-responders, p=0.16). The only significant co-variate within total cohort was presence of IKZF1^plus profile, that led to dismal outcome (EFS 0 vs 0.50 (SE 0.31), p=0.003 and CIR 1.0 vs 0.47 (SE 0.17), p<0.001). Interestingly, treatment outcome did not differ significantly between DS-ALL and non-DS-ALL with CRLF2r (EFS 0.48 (SE 0.39) vs 0.47 (SE 0.27) p=0.54 and CIR 0.41 (SE 0.21) vs 0.48 (SE 0.14) p=0.45).

Conclusions. The incidence of CRLF2r was 6.1% in our cohort of pediatric B-lineage ALL. FC with TSLP receptor monoclonal antibody (clone 1A6) can efficiently predict the presence of CRLF2r. High genetic heterogeneity with relatively high rate of RAS mutation and PAX5alt was observed. All, except one, relapses occurred after the end of maintenance therapy and the majority of them were salvaged by HSCT in CR2. MRD did not distinguish patients with various treatment outcomes.