Background: In Japan, a nationwide programme between 1984 and 2003 screened all infants for urinary catecholamine metabolites as a marker for neuroblastoma. Before 1989, this was done by qualitative spot tests for vanillylmandelic acid in urine, and subsequently by quantitative assay with high-performance liquid chromatography (HPLC). However, the Japanese government stopped the mass-screening programme in 2003, after reports that it did not reduce mortality due to neuroblastoma. We aimed to assess the effectiveness of the programme, by comparing the rates of incidence and mortality from neuroblastomas diagnosed before 6 years of age in three cohorts. Methods: We did a retrospective population-based cohort study on all children born between 1980 and 1998, except for a 2-year period from 1984. We divided these 22 289 695 children into three cohorts: children born before screening in 1980-83 (n=6 130 423); those born during qualitative screening in 1986-89 (n=5 290 412); and those born during quantitative screening 1990-98 (n=10 868 860). We used databases from hospitals, screening centres, and national cancer registries. Cases of neuroblastoma were followed up for a mean of 78·7 months. Findings: 21·56 cases of neuroblastoma per 100 000 births over 72 months were identified in the qualitatively screened group (relative risk [RR] 1·87, 95% CI 1·66-2·10), and 29·80 cases per 100 000 births over 72 months in the quantitatively screened group (RR 2·58, 2·33-2·86). The cumulative incidence of neuroblastoma in the prescreening cohort (11·56 cases per 100 000 births over 72 months) was lower than that in other cohorts (p<0·0001 for all comparisons), but more neuroblastomas were diagnosed after 24 months of age in this cohort (p=0·0002 for qualitative screening vs prescreening, p<0·0001 for quantitative screening vs prescreening). Cumulative mortality was lower in the qualititative screening (3·90 cases per 100 000 livebirths over 72 months) and quantitative screening cohorts (2·83 cases) than in the prescreening cohort (5·38 cases). Compared with the prescreening cohort, the relative risk of mortality was 0·73 (95% CI 0·58-0·90) for qualitative screening, and 0·53 (0·42-0·63) for quantitative screening. Mortality rates for both the qualitative and quantitative screening groups were lower than were those for the prescreening cohort (p=0·0041 for prescreening vs qualitative screening, p<0·0001 for prescreening vs quantitative screening). Interpretation: More infantile neuroblastomas were recorded in children who were screened for neuroblastoma at 6 months of age than in those who were not. The mortality rate from neuroblastoma in children who were screened at 6 months was lower than that in the prescreening cohort, especially in children screened by quantitative HPLC. Any new screening programme should aim to decrease mortality, but also to minimise overdiagnosis of tumours with favourable prognoses (eg, by screening children at 18 months). Funding: Japanese Ministry of Health, Labour, and Welfare.
All Science Journal Classification (ASJC) codes