ASPO Abstracts
Predicting breast cancer risk in a community-based sample of high-risk survey respondents
Category: Early Detection & Risk Prediction
Conference Year: 2021
Abstract Body:
Introduction: Identifying women with high risk of breast cancer is necessary for clinicians to
deliver guideline-recommended cancer risk management care. Risk prediction models estimate
individuals' lifetime risk of breast cancer. However, risk models have rarely been applied in
community-based settings among women not yet receiving specialized care.
Purpose: The aims of this study were twofold: (1) to apply three breast cancer risk models (i.e.
Gail, Claus, IBISv7) to a community-based sample of racially diverse women and (2) to assess
the feasibility and results of risk model estimates from self-reported information.
Methods: A total of 4,502 women who self-identified as "high risk†for breast cancer were
recruited from mainly non-clinical settings and screened for eligibility; 1,053 women (23%) were
initially eligible & completed an online survey of information needed for the risk models. Risk
models were used to estimate lifetime risk of breast cancer for each participant as applicable
(e.g., Gail only applies to women ≥35 years old without BRCA mutations). Final eligibility
required meeting a threshold of ≥20% lifetime risk per ≥1 model. Descriptive statistics were
used to assess the feasibility of running each model, proportion of high-risk women identified by
each model, and the subsequent lifetime risk estimates.
Results: A total of 717 women (68% of those initially eligible) met final eligibility criteria of ≥20%
lifetime risk. Participants were 18-74 years of age, 65% White, and 35% African American. All
women self-reported the information necessary to run at least one model; >90% had sufficient
information to run >1 model. Most participants (76%) were identified as high risk by one model
only; 73% of these were identified by IBIS, 2% by Claus, 0.85% by Gail. Twenty percent were
identified by 2 models; 3.2% were identified by all 3 models.
Conclusions: Risk reduction modeling is feasible using self-reported data from a community-
based sample. Gail, Claus, and IBIS models have low levels of agreement in identifying racially
diverse women at high risk of breast cancer. The IBIS model identifies high-risk women most
often. Researchers and clinicians should consider the use of multiple models to avoid
misidentifying potentially high-risk women.
Keywords: Breast cancer risk, risk prediction, community- based sample, breast cancer prevention,