PURPOSE:

To identify children with T-cell acute lymphoblastic leukemia (T-ALL) at high risk of induction chemotherapy failure by using DNA copy number analysis of leukemic cells collected at diagnosis.

PATIENTS AND METHODS:

Array comparative genomic hybridization (CGH) was performed on genomic DNA extracted from diagnostic lymphoblasts from 47 children with T-ALL treated on Children's Oncology Group Study P9404 or Dana-Farber Cancer Institute Protocol 00-01. These samples represented nine patients who did not achieve an initial complete remission, 13 who relapsed, and 25 who became long-term, event-free survivors. The findings were confirmed in an independent cohort of patients by quantitative DNA polymerase chain reaction (DNA-PCR), an assay that is well suited for clinical application.

RESULTS:

Analysis of the CGH findings in patients in whom induction chemotherapy failed compared with those in whom induction chemotherapy was successful identified the absence of biallelic TCRgamma locus deletion (ABD), a characteristic of early thymocyte precursors before V(D)J recombination, as the most robust predictor of induction failure (P < .001). This feature was also associated with markedly inferior event-free (P = .002) and overall survival (P < .001) rates: 25% versus 58% and 25% versus 72%, respectively. Using a rapid and inexpensive quantitative DNA-PCR assay, we validated ABD as a predictor of a poor response to induction chemotherapy in an independent series of patients.

CONCLUSION:

Lymphoblasts from children with T-ALL should be evaluated at diagnosis for deletion within the TCRgamma locus. Patients lacking biallelic deletion, which confers a high probability of induction failure with contemporary therapy, should be assigned to alternative therapy in the context of a prospective clinical trial.