Screening for neuroblastoma: a review of the evidence

Screening infants for the early detection of neuroblastoma is advocated by many paediatric oncologists and is practised in a limited number of places in the developed world, most notably in Japan where a national screening programme has been in operation since 1985. The screening test consists of measurements of the levels of vanillylmandelic acid and homovanillic acid in the urine; these metabolites of catecholamine are excreted in the urine of 92% of patients with clinically presenting neuroblastoma. The prognosis for children with symptomatic neuroblastoma is dependent both on age and stage, with children aged under 1 year and those with tumours of stages I, II, and IVS having a much better prognosis. Screening aims at detecting and treating during the first year those neuroblastomas which would otherwise present at an advanced stage in older children. Evidence from Japan shows that screening achieves the interim outcomes of a shift in the age distribution and stage distribution of neuroblastomas in populations for whom screening has been provided, and that survival of subjects detected by screening is over 90%, compared with around 50% for symptomatic subjects. However, there is not yet any clear evidence that screening results in a reduction in the incidence of advanced neuroblastoma in children over the age of 1, nor a reduction in mortality. Recent cross sectional analyses of age specific incidence and mortality suggest that screening may be having a limited effect, but as yet no analysis of these outcomes in cohorts for whom screening has been provided has been published. Other factors, such as improved chemotherapy, may also be contributing to lower mortality. A number of missed (interval) cancers have been diagnosed in children who screened negative both in the Japanese programme and in Canadian and English studies, indicating that there is a problem with the sensitivity of screening. But the screening test is highly specific with less than 0.1% of infants having false positive results requiring investigation. The natural history of neuroblastoma ranges from highly malignant tumours to biologically benign variants that regress without active treatment, the prevalence of the latter being inversely related to age. Serial measurements of biological markers, including ploidy, chromosome 1p deletion, and N-myc amplification, performed within the same patient at different times indicate that malignant potential does not progress over time. The distribution of these markers in cases detected by screening shows that they are inherently tumours with a good prognosis, whereas the reverse is true of interval cases. Thus screening is differentially picking up the tumours that are least likely to progress and failing to detect at least some tumours of those destined to die from the disease. Comparison of the yield of cancers detected by screening and the expected cumulative incidence of neuroblastoma throughout childhood suggest that screening "overdiagnoses" many non-progressive cases, with consequent physical and psychological morbidity. On balance present evidence suggests that the number of deaths that could be prevented by screening is small and the potential for overdiagnosis is great. Unless further evidence from Japan or the results of a current North American trial conclude otherwise, screening cannot be recommended.