Childhood leukemia: The role of maternal pregnancy conditions and childhood infections, and medical record validation and reliability of maternally reported pregnancy conditions

Leukemia is the most common form of childhood cancer in the United States and other western countries and its etiology remains largely unknown. Using two completed case-control studies conducted by the Children's Cancer Group (CCG) we examined the role of maternal pregnancy conditions and risk of ALL (cases n = 1914 and controls n = 1842) and AML (cases n = 530 and controls n = 610), separately. We also tested Greaves' and Kinlen's hypotheses to study the role of childhood infections in ALL using medical records. Lastly, we investigated the validity and reliability of maternal report in comparison to maternal medical records.;Maternal use of oral contraceptives at least 5 days in the year before or during pregnancy was associated with an increased risk of ALL (odds ratio [OR] 1.48, 95% confidence interval [CI] 1.02--2.15). Borderline reductions in risk of AML were seen for children whose mothers were hypertensive (OR 0.69, 95% CI 0.45--1.06) and children whose mothers were diabetic (OR 0.54, 95% CI 0.28--1.04). These findings suggest a potential role of in utero estrogen exposure and risk of childhood leukemia since the use of oral contraceptives increases pregnancy estrogen concentrations, and high blood pressure and diabetes decrease pregnancy estrogen concentrations. No significant associations were found between ALL and surrogate markers of infection, types of infections, or number of infections regardless of the time at which they occurred. When validity and reliability of maternal reporting was examined only birthweight, preeclampsia or toxemia and diabetes appeared to be accurately reported by mothers. Differential misclassification was present for high blood pressure, threatened miscarriage, vaginal bleeding, and anemia, which may have led to overestimates or underestimates of the risk of ALL.;In summary, a possible role of estrogen exposure in utero and increased risk of ALL and AML was suggested. Our data neither supported nor refuted Greaves' or Kinlen's hypotheses. Future research is necessary to examine these results, and our finding of differential misclassification of some pregnancy conditions.