December 12, 2018
Breast cancer is the most commonly diagnosed cancer in women. Researchers have long since confirmed a number of the risk factors that contribute to its development, such as body mass index and alcohol consumption, and have developed well-validated risk prediction models used to estimate an individual woman’s breast cancer risk. Such models, which include the Gail and Rosner–Colditz prediction models, have been used to set clinical trial entry criteria and provide tailored recommendations to women for screening, chemoprevention, and other risk-reducing strategies. However, the performance of these models is known to be less than desired.
Until recently, these breast cancer prediction models have not included any of the more recently validated biological markers of risk, such as a multiple gene (polygenic) risk score (PRS), breast density as measured on a mammogram (mammographic density; MD), and postmenopausal endogenous hormone levels. Further, the combined influence of these biomarkers to risk prediction models had not previously been assessed in detail.
Xuehong Zhang of Harvard Medical School, with UMass Amherst Professor of Epidemiology and senior author Susan Hankinson and others, sought to incorporate these biomarker risk factors into the existing Gail and Rosner–Colditz models to see if their joint contribution could increase the models’ accuracy. Their initial findings are promising, significantly improving predictive outcomes for invasive breast cancer, particularly among postmenopausal women not using hormone therapy (HT).
Using a nested case–control study within the prospective Nurses’ Health Study (NHS) and Nurses’ Health Study II (NHSII) among women ages 34-70, the researchers examined data from 4,006 cases of invasive breast cancers and an additional 7,874 women included as controls. They added the biologic markers including PRS, MD, and circulating testosterone, estrone sulfate, and prolactin hormone levels to the data used in the existing risk models. Each of the factors was significantly associated with invasive breast cancer risk. Further, including PRS, MD, and hormones also better identified low- versus high-risk women. In particular, the new model improved the most among postmenopausal women not using HT, where each of the biomarkers could be included and significantly predicted risk. Model improvement appeared even greater for hormone receptor positive breast cancer.
The authors note that although promising, further studies will be needed to confirm these findings in other populations and to determine whether improved risk prediction models have practical value in identifying women at higher risk who would most benefit from chemoprevention, screening, and other risk-reducing strategies. This additional work also is ongoing through a $4.5 million NIH grant awarded to Hankinson.
Additional contributors to the study include Associate Professor of Biostatistics Jing Qian of the Department of Biostatistics and Epidemiology in the School of Public Health and Health Sciences, and colleagues at the Harvard Medical School, Massachusetts General Hospital, Harvard T.H. Chan School of Public Health, H. Lee Moffitt Cancer Center and Research Institute, University of Washington, and the Washington University School of Medicine.
Their findings appear in PLOS Medicine.