A simple urine test to detect cancer at an early stage is every oncologist's dream, but for childhood neuroblastoma, for which just such a test is available, two groups report that it doesn't save lives.

These common childhood tumours produce catecholamines, the metabolites of which can be measured in urine. This forms the basis of a diagnostic test that is mandatory in Japan. But the validity of the study that first showed the test to improve survival has been questioned because data from the control group were collected before data from the screened group. Furthermore, childhood neuroblastoma can regress spontaneously, so detecting these tumours might actually increase mortality from unnecessary surgery.

Two groups have now compared large populations of screened and unscreened babies contemporaneously. William G. Woods and colleagues screened over 425,000 babies in the Quebec province, Canada, at three weeks and six months of age and followed them until the age of 8. Compared with four different populations of unsreened children, the cumulative mortality rate from neuroblastoma was no lower in the screened group. Moreover, the test failed to detect almost all the tumours with poor prognostic indicators, such as MYCN amplifications.

Freimut Schilling and colleagues screened nearly 1.5 million babies at one year of age in six states in Germany, and compared the incidence of late-stage neuroblastoma (stage 4) and mortality with that in babies born at the same time in the other ten states. The incidence of stage-4 neuroblastoma and mortality were the same in the screened and unscreened groups, despite the fact that more neuroblastomas were detected in the screened population.

Both studies therefore argue against population screening for neuroblastoma in infancy. They also highlight the need for more research into the two forms of the disease — a highly aggressive one that seems to evade the urine test, and a more insidious form that is readily detected but frequently regresses of its own accord.