Abstract
Characterization of the genetic landscape of relapsed pediatric acute lymphoblastic leukemia (ALL) and changes that occur with disease progression provides insight into the molecular basis of relapse and may identify new therapeutic targets. We analyzed 20 diagnosis-remission (germline)-relapse trio samples of pediatric B-ALL by high-coverage (>200x) whole-exome sequencing. Included patients were originally NCI high risk (HR) by either age (≥10 years) or white blood count (≥50,000/microliter), enrolled on a Children’s Oncology Group B-ALL trial, and experienced bone marrow relapse. Samples were selected based upon availability of sufficient high quality material from all three time points. We identified recurrent relapse-specific somatic mutations in 5 genes with significant frequency found in genes encoding the purine 5’ nucleotidase NT5C2 (n=7, 35%) and the histone acetyltransferase CREBBP (n=2, 10%). Furthermore, we discovered novel recurrent somatic mutations that were highly enriched in relapsed ALL (20%) compared with diagnosis (5%) in WHSC1, USH2A and NT5C1B, another enzyme involved in purine metabolism. Three of the four WHSC1 mutations cause the same amino acid change E1099K in the highly conserved SET domain in which structural modeling predicts perturbed WHSC1-substrate interactions resulting in increased WHSC1 activity. The WHSC1 and NT5C2 mutations are mutually exclusive with a combined prevalence of 55% in relapsed tumors. Analysis of a validation cohort of 63 independent trios from both NCI standard risk and HR cohorts replicated the findings in NT5C2 (n=8, 13%) and WHSC1 E1099K (n=6, 10%). Five pathways were significantly mutated at relapse with high-frequency somatic mutations present at diagnosis and/or relapse of the 20 ALL trio samples: the Ras signaling pathway (NRAS, KRAS, PTPN11, FLT3; 65%), genes involved in histone modification (MLL2, WHSC1, SETD2, CREBBP; 50%), purine metabolism (NT5C2 and NT5C1B; 45%), tyrosine kinase signaling (JAK2, CRLF2; 25%) and genes regulating B-cell development (PAX5, IKZF1; 15%). The median number of non-silent coding region sequence mutations in diagnostic samples was 10 (range 4-25) while that of the relapse samples was 25 (range 7-506) including 3 hypermutated samples with >100 non-silent mutations accompanied by a dominance of C(G)>T(A) substitution in relapse-specific mutations, suggesting a possibility that these mutations may be induced by a specific mutagen. Most of the diagnostic and relapse tumors were polyclonal based on diagnosis-relapse comparison of mutant allele fraction (MAF). Inter-tumor MAF of a recurrently mutated gene was highly heterogeneous despite an estimated >70% leukemia involvement for most specimens, suggesting presence of subclonal mutations. For example, NT5C2 MAF in the relapsed specimens ranges from a low of 0.08 to a high of 0.93. Ten relapsed specimens demonstrated evolution from a minor subclone (<10%) in the diagnostic specimen; 8 of the subclones have oncogenic mutations in NRAS, KRAS, JAK2, WHSC1 or CRLF2. A notable finding was the lack of preservation of specific clonal RAS pathway mutations from diagnosis to relapse as subclonal mutations in KRAS, PTPN11 and FLT3 present in diagnosis were replaced by a dominant NRAS mutation in relapse. We were also able to identify structural alterations present in both relapse and diagnosis tumors from exome sequencing data, including IGH@-CRLF2 fusion (n=2), BTG1 deletion (n=2), ETV6-RUNX1 fusion (n=1), intragenic deletion of RUNX1 (n=1) and MAP3K2 (n=1), focal amplification in the last exon of MYC (n=1) and a t(1;14)(p36;q32) translocation resulting in truncation of SLC2A5 and BTBD7. These results provide new insights into the genetic events contributing to the relapse of pediatric B-ALL, and suggest new potential therapeutic targets.
No relevant conflicts of interest to declare.
Author notes
Asterisk with author names denotes non-ASH members.