Background Breastfeeding has been linked to positive developmental and childhood health outcomes. It is unclear whether the beneficial effects of being breastfed, such as the small reduction in risk of childhood-onset cancers, persist into adulthood. In the Million Women Study, a cohort of UK women in middle age, we examined whether having been breastfed was associated with total cancer risk or risks of common specific cancers.

Methods Approximately 1.3 million women were recruited through the national breast screening programmes of England and Scotland in 1996–2001. Three years on average after recruitment, 560,879 women (mean age 60 years) reported whether they had been breastfed as infants. These women were linked to a second cohort with breastfeeding duration recorded when they were aged 2 years; among matched participants, self-report of having been breastfed was consistent with breastfeeding having stopped at around 6 months, on average, although breastfeeding may have been non-exclusive before that time. All women were linked to deaths and cancer registrations via NHS Central Registers. Cox regression was used to estimate relative risks and 99% confidence intervals for invasive cancers of 16 common sites or types with at least 500 cases (mouth and pharynx, oesophagus, stomach, colorectal, pancreas, lung, malignant melanoma, breast, endometrium, ovary, kidney, bladder, central nervous system, non-Hodgkin lymphoma, multiple myeloma, and leukaemia), and for all invasive cancers combined. Multivariable models were adjusted for age and 14 other known cancer risk factors.

Results Baseline characteristics did not differ greatly according to whether or not the women were breastfed, except that parous women who had been breastfed tended to breastfeed their own children for slightly longer. During an average of 9.3 years of follow-up per woman, there were 48,610 incident invasive cancers. Having been breastfed was not associated with overall cancer risk (adjusted RR=1.02, 99% CI 0.99–1.05). Having been breastfed was associated with a slightly increased risk of colorectal cancer, but not with risk of the 15 other specific cancer sites or types (all P > 0.05).

Conclusion Our results suggest that of having been breastfed as an infant is not an important modifier of overall cancer risk.