Modelling a minimal residual disease-based treatment strategy in childhood acute lymphoblastic leukaemia.

The measurement of minimal residual disease (MRD) in childhood acute lymphoblastic leukaemia offers the promise of individualized, risk-stratified treatment, but an optimal protocol needs establishing. A model was developed to explore certain unanswered questions. The model assumes that all patients have MRD assessed after induction chemotherapy and children above a certain threshold are offered intensive chemotherapy. Using parameter estimates derived from published studies of MRD, the model predicted event-free survival (EFS) rates, relapse rates and treatment-related mortality for a cohort of children in the first presentation who were Philadelphia chromosome negative. Using the level of MRD after induction in order to decide on the use of intensive therapy resulted in an increase in EFS rates of up to 2.9 per 100 children, although if the error of the MRD measurement were too great, the benefit was almost nullified. Taking and analysing more than one marrow sample from the patient for the MRD measurement, in order to reduce sampling and measurement error, improved EFS by a further 1.1 patients per 100 treated, and decreased the number of patients offered intensive therapy by up to 2.6 per 100 treated. The optimal threshold for offering intensive therapy was in the range of 10-3.5- 10-4.5 cells if the intensive treatment-related mortality was 13-18% (allograft options), but 10-5- 10-6 cells if it was less than 8% (intensified chemotherapy). Using MRD to target patients at a high risk of relapse improved EFS rates, but the accuracy of measurements was of critical importance.