One of the reasons that treatment for paediatric acute lymphoblastic leukaemias (ALLs) has been so successful, achieving long-term event-free survival rates of almost 80%, is that the intensity of treatment is precisely tailored to the patient's risk of relapse. After diagnosis, patients are placed into a specific 'risk group' that is based on immunophenotype, cytogenetic and molecular diagnostic data. Therapy is then carefully selected to avoid undertreatment or overtreatment. Accurate assignment of patients to specific risk groups is a difficult and expensive process, however, requiring a large number of laboratory tests and health-care professionals. So could gene-expression profiling be an easier way to predict therapeutic response?

In the March issue of Cancer Cell, James Downing's group reports the use of oligonucleotide microarrays to analyse the expression patterns of 12,600 genes from leukaemic blasts of 360 paediatric ALL patients. The expression profiles were able to identify specific leukaemia subtypes, including E2A – PBX1 , BCR – ABL , TEL – AML1 and MLL gene rearrangement, hyperdiploidy and T-lineage leukaemias (T-ALL), with a diagnostic accuracy of 96%. The study also revealed some new leukaemia-associated genes, such as the MER receptor tyrosine kinase in E2A–PBX1, which might be developed as a therapeutic target.

But most importantly, the analysis was able to predict which patients were most likely to undergo relapse. For T-ALL and hyperdiploid >50 subgroups, expression profiling predicted which cases would relapse with an accuracy of 97% and 100%, respectively. There was no single common expression profile that predicted relapse, indicating that a unifying mechanism might not exist.

The authors suggest that this approach could be developed as a more straightforward means of identifying patients who are most likely to undergo relapse or therapy-induced acute myeloid leukaemia (AML) — the main causes of treatment failure in paediatric acute leukaemia. Gene-profiling approaches might also be developed to identify patients who are at risk of developing other therapy-induced complications, such as organ toxicity or infection.