No abstract

Gross cytogenetic anomalies are traditionally being used as diagnostic, prognostic and therapeutic markers in the clinical management of cancer, including childhood acute lymphoblastic leukemia (ALL). Recently, it has become increasingly clear that genetic lesions driving tumorigenesis frequently occur at the submicroscopic level and, consequently, escape standard cytogenetic observations. Therefore, we profiled the genomes of 40 childhood ALLs at high resolution. We detected multiple de novo genetic lesions, including gross aneuploidies and segmental gains and losses, some of which were subtle and affected single genes. Many of these lesions involved recurrent (partially) overlapping deletions and duplications, containing various established leukemia-associated genes, such as ETV6, RUNX1 and MLL. Importantly, the most frequently affected genes were those controlling G1/S cell cycle progression (e.g. CDKN2A, CDKN1B and RB1), followed by genes associated with B-cell development. The latter group includes microdeletions of the B-lineage transcription factors PAX5, EBF, E2-2 and IKZF1 (Ikaros), as well as genes with other established roles in B-cell development, that is RAG1 and RAG2, FYN, PBEF1 or CBP/PAG. The fact that we frequently encountered multiple lesions affecting genes involved in cell cycle regulation and Bcell differentiation strongly suggests that both these processes need to be targeted independently and simultaneously to trigger ALL development.

Purpose Outcome of childhood acute lymphoblastic leukemia (ALL) improved greatly by intensifying chemotherapy for all patients. Minimal residual disease (MRD) levels during the first months predict outcome and may select patients for therapy reduction or intensification. Methods Patients 1 to 18 years old with ALL were stratified on the basis of MRD levels after the first and second course of chemotherapy. Thereafter, therapy was substantially reduced in patients with undetectable MRD (standard risk) and intensified in patients with intermediate (medium risk) and high (high risk) levels of MRD. Seven hundred seventy-eight consecutive patients were enrolled. The method of analysis was intention-to-treat. Outcome was compared with historical controls. Results In MRD-based standard-risk patients, the 5-year event-free survival (EFS) rate was 93% (SE 2%), the 5-year survival rate was 99% (SE 1%), and the 5-year cumulative incidence of relapse rate was 6% (SE 2%). The safety upper limit of number of observation years was reached and therapy reduction was declared safe. MRD-based medium-risk patients had a significantly higher 5-year EFS rate (88%, SE 2%) with therapy intensification (including 30 weeks of asparaginase exposure and dexamethasone/vincristine pulses) compared with historical controls (76%, SE 6%). Intensive chemotherapy and stem cell transplantation in MRD-based high-risk patients resulted in a significantly better 5-year EFS rate (78%, SE 8% v 16%, SE 8% in controls). Overall outcome improved significantly (5-year EFS rate 87%, 5-year survival rate 92%, and 5-year cumulative incidence of relapse rate 8%) compared with preceding Dutch Childhood Oncology Group protocols. Conclusion Chemotherapy was substantially reduced safely in one-quarter of children with ALL who were selected on the basis of undetectable MRD levels, without jeopardizing the survival rate. Outcomes of patients with intermediate and high levels of MRD improved with therapy intensification.

Relapse is the most common cause of treatment failure in pediatric acute lymphoblastic leukemia (ALL) and is often difficult to predict. To explore the prognostic impact of recurrent DNA copy number abnormalities on relapse, we performed high-resolution genomic profiling of 34 paired diagnosis and relapse ALL samples. Recurrent lesions detected at diagnosis, including PAX5, CDKN2A and EBF1, were frequently absent at relapse, indicating that they represent secondary events that may be absent in the relapse-prone therapy-resistant progenitor cell. In contrast, deletions and nonsense mutations in IKZF1 (IKAROS) were highly enriched and consistently preserved at the time of relapse. A targeted copy number screen in an unselected cohort of 131 precursor B-ALL cases, enrolled in the dexamethasone-based Dutch Childhood Oncology Group treatment protocol ALL9, revealed that IKZF1 deletions are significantly associated with poor relapse-free and overall survival rates. Separate analysis of ALL9-treatment subgroups revealed that non-high-risk (NHR) patients with IKZF1 deletions exhibited a B12-fold higher relative relapse rate than those without IKZF1 deletions. Consequently, IKZF1 deletion status allowed the prospective identification of 53% of the relapse-prone NHR-classified patients within this subgroup and, therefore, serves as one of the strongest predictors of relapse at the time of diagnosis with high potential for future risk stratification.