Clonal Origins of 'late' Relapses in ETV6-RUNX1 Acute Lymphoblastic Leukemia..

Abstract 1583

Poster Board I-609

Approximately a quarter of B cell precursor childhood acute lymphoblastic leukemia (ALL) is characterized by an ETV6-RUNX1 (TEL-AML1) fusion gene and has an overall good prognosis. The majority of these children will be treated on the standard risk arm of the United Kingdom ALL treatment protocols. Relapse usually occurs after cessation of treatment but remarkably can present many years later. The incidence of ETV6-RUNX1 at relapse has been reported to be less than or similar to de novo ALL. Molecular studies on neonatal bloodspots and on twins with concordant ALL have demonstrated the prenatal origin of major subtypes of childhood ALL, including most ETV6-RUNX1 fusion gene positive cases. In addition these investigations have suggested the existence of a preleukaemic stem cell requiring additional mutations or ‘hits’ in order to develop frank leukemia. To understand the genetic basis and clonal origin of late relapses we have compared the profiles of genome-wide copy number alterations (CNA) at relapse versus presentation in samples matched with remission DNA from 24 patients. The selected samples had tumor cell purity >75% before DNA extraction. DNA copy number alteration data was generated using the Affymetrix 500K SNP arrays. LOH analysis was performed using CNAG 3.0 and dCHIP 2008. Overall we identified 168 CNA at presentation and 252 at relapse (excluding deletions at IgH and TCR loci), equating to 6.96 and 10.3 CNA at presentation and relapse respectively. Although the number of CNA increased at relapse, no single gene or pathway was uniquely targeted in relapse. The most frequent alterations involved loss of 12p3.2 (ETV6), 9p21.3 (CDKN2A/B), 6q16.2-3 and gain of 21q22.1-22.12. A novel observation was gain of part or whole of chromosome 16 (2 patients at presentation, 5 at relapse) and deletion of the oncogene Plasmocytoma Variant Translocation 1 (PVT1) in 3 patients. Pathway analysis demonstrated frequent involvement at presentation and relapse of genes implicated in both B cell development (44 versus 46%) and cell cycle control (46 versus 71%). In order to study the clonal origin of relapse, we devised a classification describing the change in CNA between presentation and relapse in each individual patient. The clonal relationship between the presentation and relapse clone was established by the persistence of both the ETV6-RUNX1 fusion and at least 1 Ig and/or TCR rearrangement. We used a classification focussed on ‘driver’ CNA, defined as CNA that target genes functionally involved in leukemogenesis or CNA that are recurrently targeted as described in the literature. The four categories of relapse were type 1 (the dominant clone at presentation presented unchanged at relapse), type 2 (the relapse clone was derived from the major subclone at presentation with additional CNA), type 3 (the relapse clone was derived from a minor clone at presentation with gains and losses of CNA) and type 4 (the relapse clone is derived from an ancestral or preleukemic clone at initial presentation with all CNA gained). Twenty-one of the 24 patients were classifiable in this way (Figure 1). Although comparative relapse / presentation CNA profiles cannot identify precise clonal origins of relapse, the data indicate that irrespective of time to relapse (<2 to 9.9 years), the relapse clone appeared to be derived from either a major or minor clone at diagnosis with none (0/6) of the very late relapses (>5 years) derived from pre-leukemic cells lacking CNA. This data indicate diverse clonal origins of relapse and extended periods of dormancy, possibly via quiescence, for stem cells in ETV6-RUNX1⁺ ALL.