Allergy-associated symptoms in relation to childhood non-Hodgkin's as contrasted to Hodgkin's lymphomas: A case-control study in Greece and meta-analysis

An increase of the prevalence of childhood allergic diseases and the incidence of childhood Hodgkin's (HL) and non-Hodgkin's lymphoma (NHL) were reported in the late 20th century. Among adults, several studies point to an inverse association with lymphoma; it remains to be confirmed whether allergy is also related to childhood lymphomas and whether the association, if any, is of an aetiologic nature. Between 1996 and 2008, 277 children (aged 0-14 years) with HL (N = 111) or NHL (N = 166) were enrolled in Nationwide Registry for Childhood Hematological Malignancies (NARECHEM), a Greek hospital-based-registry of childhood hematological malignancies. Hospital controls were individually matched to cases on age and sex. Multivariate conditional logistic regression was used to estimate odds ratios (ORs) with 95% confidence intervals (CIs) for associations of allergic diseases and other covariates with childhood HL or NHL risk. Subsequently, we combined our results with those of a French case-control study in a meta-analysis amounting to a total of 330 NHL cases/1478 controls and 239 HL cases/959 controls. After controlling for sociodemographic, perinatal and environmental factors, childhood NHL was less prevalent among children with allergy-associated symptoms overall (OR:0.50, 95% CI:0.27-0.92) or a history of asthma (OR:0.43, 95% CI:0.21-0.88). By contrast, allergy did not seem to be associated with childhood HL risk, although statistical power was limited. Fewer seaside holidays and higher birth weight were also associated with increased childhood NHL risk. The combined OR of the two studies for the association of asthma with NHL risk was: 0.52, 95% CI:0.32-0.84, whereas for HL: 0.86, 95% CI:0.51-1.45. Allergy seems to be strongly and inversely associated with childhood NHL. It remains to be elucidated in future investigations comprising larger populations, focusing on specific disease subtypes and employing more pertinent study-designs, whether this association is genuinely protective.