Population-based relative risks for specific family history constellations of breast cancer

Purpose: Using a large resource linking genealogy with decades of cancer data, RRs were estimated for breast cancer (BC) based on specific family history extending to first cousins. Methods: RRs for BC were estimated in 640,366 females with breast cancer family histories that included number of first- (FDR), second-(SDR), and third-degree relatives (TDR), maternal and paternal relatives, and age at earliest diagnosis. Results: RRs for first-degree relatives of BC cases ranged from 1.61 (≥1 FDR affected, CI: 1.56, 1.67) to 5.00 (≥4 FDRs affected, CI: 3.35, 7.18). RRs for second degree relatives of probands with 0 affected FDRs ranged from 1.08 (≥1 SDR affected, CI: 1.04, 1.12) to 1.71 (≥4 SDRs affected, CI: 1.26, 2.27) and for second degree relatives of probands with exactly 1 FDR from 1.54 (0 SDRs affected, CI:1.47, 1.61) to 4.78 (≥ 5 SDRs; CI 2.47, 8.35). RRs for third-degree relatives with no closer relatives affected were significantly elevated for probands with >=5 affected TDRs RR=1.32, CI: 1.11, 1.57). Conclusions: The majority of females analyzed had a family history of BC. Any number of affected FDRs or SDRs significantly increased risk for BC, and more than 4 TDRs, even with no affected FDRs or SDRs significantly increased risk. Risk prediction derived from specific and extended family history allows identification of females at highest risk even when they do not have a conventionally defined high risk family; these risks could be a powerful, efficient tool to individualize cancer prevention and screening.