Residual disease detection using fluorescent polymerase chain reaction at 20 weeks of therapy predicts clinical outcome in childhood acute lymphoblastic leukemia

Ninety-five percent of children with acute lymphoblastic leukemia (ALL) will achieve a remission, but approximately 25% will relapse. Identifying these patients is difficult, as patients with adverse prognostic features at presentation are rare and the majority are standard risk. Analysis of minimal residual disease (MRD) may be able to determine those at risk of relapse, but the best method by which this can be accomplished has yet to be defined. The object of this study was to determine the predictive value of residual disease detection in a group of standard-risk patients with precursor-B ALL at a fixed point in therapy (week 20) using a simple fluorescent consensus immunoglobulin H (IgH) heavy chain polymerase chain reaction (PCR). Forty-two patients who presented with precursor-B ALL with standard-risk clinical features and treated according to either the Medical Research Council (MRC) UKALL X or XI protocols were assessed using a combination of both fluorescent consensus framework I and framework III Ig heavy-chain PCR. The results of the PCR were analyzed on an ABI 373 gene sequencer with genescan software (Applied Biosystems, Foster City, CA). Clonal rearrangements detected at presentation were looked for at week 20. Of 42 patients, 35 had a clonal population detectable at presentation; of these, seven had more than two clonal rearrangements; this latter group showed a similar disease-free survival (DFS) to the group as a whole. Thirty of 35 patients were analyzed before their second course of intensification therapy at week 20. At this point, nine of 30 had a detectable clonal rearrangement, eight (89%) of whom have since relapsed with a median DFS of 27.5 months. Of the rest of the group (n=21), in whom no clonal rearrangement was detectable, only six (21%) have relapsed. Fluorescent IgH PCR at week 20 provides a sensitive and specific means to predict ultimate relapse (57% and 89%, respectively) and is a simple yet promising technique for the identification of patients at risk of poor outcome.