Pregnancy history and risk of premenopausal breast cancer-a nested case-control study.

BACKGROUND: Reproductive factors are well-known risk factors for premenopausal breast cancer (PBC). It is unknown whether these associations are modified by familial factors, including genetic and early environment factors.

METHODS: Using Swedish health registries, we performed a nested case-control study with two control groups: sister controls and population controls. The study population included women with live singleton births between 1973 and 2010, who also had a full sister who gave birth during this period. All women subsequently diagnosed with PBC were selected as cases (n = 8327). Sisters with the least age difference and without PBC at the time of her sister's diagnosis were selected as sister controls. For each incident case, one population control without previous PBC was selected.The population controls were individually matched with the sister controls on year of birth. Conditional logistic regression was used to estimate associations between reproductive factors and PBC.

RESULTS: Increasing parity was inversely associated with PBC using population controls, and multiparity was a risk factor using sister controls. Very preterm delivery (≤ 31 weeks) was associated with a slightly higher PBC risk using sister controls. Preeclampsia was associated with a slightly protective effect using population controls. With respect to other factors, there were no substantial differences in risks of PBC by choice of control group.

CONCLUSIONS: The divergent results with regard to parity and PBC risk when using sister and population controls suggest that the influence of childbearing may be modified by genotype. Selection bias when using different control groups must also be considered.