Time-dependent associations between presentation-based infections and subsequent risk of childhood immune cell malignancies.

BACKGROUND: Infection is hypothesised as a contributory cause of childhood immune cell malignancies. Although some have reported associations between individual infections and immune cell malignancies, some could be spurious due to infections caused by malignancies that were already active prior to diagnosis. METHODS: Identified from Taiwan Cancer Registry, ∼3000 children with four commonest immune cell malignancies diagnosed during 2001-2015 at age 1-20 years were identified and matched with 1:10 controls. Using logistic regression, we estimated the time-specific case-versus-control odds ratios of seven common infection presentations in their health records. We also compared recorded unexplained lymph nodes between cases and controls to explore for how long malignancy may be active prior to diagnosis. RESULTS: Unexplained lymph nodes were increasingly recorded months before the diagnosis of childhood leukaemias and years before the diagnosis of childhood lymphomas. When using p < 0.01 as a guide, large case-control differences in infection records were found mostly within 0-2 months prior to the diagnosis (15 out of 28 comparisons). Changes in odds ratios within 3-35 months (2 out of 28 comparisons) and case-control differences beyond 36+ months prior to diagnosis (7 out of 28 comparisons) was relatively small (∼10 % difference in leukaemias). Statistical power varied according to incidence of malignancy, incidence of infection records, and the age distribution. CONCLUSION: Immune cell malignancies were likely to be active some time before the diagnosis. Previous studies using conventional population-based methods may not be able to distinguish any small causal link between infection and immune cell malignancies from spurious associations.