Characterizing Rare Variants in Drug Metabolism Genes in Refractory Pediatric High-Risk Pre-B ALL

We are working to understand how rare genetic variation can influence incidence and outcomes of pediatric high-risk leukemia. Despite improvements in standard-risk outcomes, children with high-risk precursor-B acute lymphoblastic leukemia (HR-ALL) have not enjoyed similar success. Recent studies demonstrate that there are critical metabolic pathways involved in the transition from normal precursor-B (pre-B) cells to leukemic cells as well as involved in proper therapeutic response (Mullighan, Nature 2007; Kang, Blood 2011). The Rare Variant Hypothesis predicts that a population of affected individuals would harbor a diverse collection of functionally significant variants in the genes involved in these pathways. Thus, the manner in which these critical genes or pathways become disrupted may be quite variable, but the outcome is very similar. Therefore, we hypothesize that rare variants in critical metabolic pathways influence the incidence and outcome of pediatric HR-ALL. We established Children's Oncology Group (COG) protocol AALL10B2 to study this question from existing patient DNA samples.

Using our pooled sequencing strategy (Druley, Nat Methods 2009) and novel computational analysis software, SPLINTER (Vallania, Genome Res 2010), we have sequenced 54 candidate genes associated with pediatric leukemia in three pools of genomic DNA. One pool from 96 germline DNA samples from patients enrolled on COG HR-ALL protocol P9906, a second pool from 96 matched P9906 leukemia DNA samples, and a third from 93 ethnically matched unaffected pediatric controls. We called an average of 3,987 variants per pool (range 3822–4209) and validated variants by individual custom genotyping. The correlation between aggregated individual minor allele frequencies and pooled sequencing variant calling was excellent for all three pools (R² range = 0.90–0.93).

By directly mapping the variants from the three pools, we identified gene regions with variation in the patients, but not controls. We found two genes with regions of excess patient-specific germline genetic variation, cytochromes 1A1 and 3A5 (CYP1A1 and CYP3A5). The CYP3A isozymes are involved in the activation of epipodophyllotoxins, anthracyclines and cyclophosphamide along with the clearance of vincristine and glucocorticoids, and common variants in these genes have previously been correlated with an increased incidence of pediatric ALL (Joseph, Pediatr Blood Cancer 2004; Aydin-Sayitoglu, Am J Hematol 2006; Borst, Eur J Haematol 2011). A heat map of genome-wide expression array results for our patient population (our 96 pilot patients from the P9906 trial and an additional 250 enrolled in the AALL0232 trial) against the 54 genes in our survey identified a previously unappreciated subpopulation (of ∼10% of patients) that trended toward inferior relapse-free survival and demonstrated significant overexpression of 14 genes: (in alphabetical order): ATM, CDKN1A, CYP1A1, CYP3A5, IKZF1, MDM2, MLL, MTHFR, NAT2, NQO1, PAX5, PTPN11, TCF3, TPMT. These patients did not possess other high-risk chromosomal translocations. Validation in 250 matched germline and HR-ALL DNA samples from participants in the COG AALL0232 study is underway, along with functional characterization and validation of the identified variants.

We have identified a potential new genomic signature for a subset of HR-ALL patients who appear predisposed to poor therapeutic outcomes. Our preliminary results support the hypothesis that pediatric HR-ALL incidence and outcomes are influenced by a pre-existing profile of germline genetic variation, which is reasonable given the fact that it is exceptionally mathematically unlikely that these children would acquire the entire cadre of deleterious genetic variants required for malignant transformation solely through somatic mutation. Such results are “clinically-actionable” now. Children demonstrating functionally significant variants in these genes or similar patterns of gene expression could be immediately referred for hematopoietic stem cell transplantation in first remission rather than being exposed to years of highly toxic chemotherapy that is unlikely to be successful anyway. Meanwhile, in vitro functional studies will facilitate the discovery of alternate therapeutic strategies.

No relevant conflicts of interest to declare.