Abstract
BACKGROUND: The aim of this study was to assess whether minimal residual disease (MRD) at the end of induction front-line treatment can serve as a surrogate endpoint for event-free survival (EFS) in childhood B-lineage acute lymphoblastic leukemia. METHODS: The analysis was based on individual data of 4830 patients from two large phase III trials that asked a randomized question on the effect of different corticosteroids (dexamethasone vs prednisone) during induction chemotherapy on EFS. The association between MRD classified in three ordered categories [negative = 0, low positive = (>0 and <5 × 10(-4)), and positive = (≥5 × 10(-4))] and EFS at the individual and trial levels was evaluated with the meta-analytic approach based on the Plackett copula model. Centers within trial were grouped according to geographical area, and a total of 28 units were identified for the analysis. RESULTS: MRD at the end of induction was a poor surrogate for treatment effect on EFS at the trial level, with Rtrial2 = 0.09 (95% confidence interval [CI] = 0.00 to 0.29), whereas at the individual level it was strongly associated with EFS, with an odds ratio of 3.90 (95% CI = 3.35 to 4.44) of failure for patients with higher compared with lower MRD levels. Additional sensitivity and relevant subgroup analyses confirmed these findings at both trial- and patient-level association. CONCLUSIONS: Although MRD is a robust biomarker highly predictive of outcome for individual patients, clinicians and regulatory bodies should be cautious in using early MRD response in the context of complex multiagent acute lymphoblastic leukemia therapy as an early surrogate endpoint to predict the effect of a randomized treatment intervention on long-term EFS.