Sponsoring Section/Society: Caucus for Women in Statistics
Session Slot: 2:00- 3:50 Tuesday
Estimated Audience Size: 150
AudioVisual Request: Overhead Projector, Slide Projector
Session Title: Statistical Models for Individualized Counseling
about Breast Cancer
Theme Session: Yes
Applied Session: No
Session Organizer: Parmigiani, Giovanni Duke University
Address: Giovanni Parmigiani Box 90251, ISDS Duke University Durham, NC 27708-0251
Phone: 919-684-8064
Fax: 919-684-8594
Email: gp@stat.duke.edu
Session Timing: 110 minutes total (Sorry about format):
Opening Remarks by Chair - 5 minutes First Speaker - 30 minutes Second Speaker - 30 minutes Third Speaker - 30 minutes Discussant - 0 minutes Floor Discusion - 15 minutes
Session Chair: Stangl, Dalene Duke University
Address: Dalene Stangl Box 90251, ISDS Duke University Durham, NC 27708-0251
Phone: 919-684-4263
Fax: 919-684-8594
Email: dalene@stat.duke.edu
1. Predicting Risk of Breast Cancer
Benichou, Jacques, Centre Hospitalier Universitaire de Rouen, France
Address: Jacques Benichou, M.D., Ph.D. Division of Cancer Epidemiology and Genetics National Cancer Institute Executive Plaza North, Room 403 6130 Executive Blvd. MSC 7368 Bethesda, MD 20892-7368
Phone: 301-496-3356
Fax: 301-402-0081
Email: Jacques_Benichou@nih.gov
Abstract: When counseling women about their breast cancer risk, it is useful to provide a quantitative assessment of their individualized risk of developing breast cancer, or absolute risk. Absolute risk is defined as the probability that a woman with a given risk profile and given age will develop breast cancer over a defined time period.There are three general approaches to estimate absolute risk of breast cancer depending on how familial and genetic risk factors are taken into account. In the empirical approach, observable family history as well as other non familial risk factors are modeled and no genetic model is assumed nor any genotype observed. An intermediate or genetic-modeling approach consists in analyzing family history data relying on a genetic model, but genotype is again not observed. Finally, in the direct approach, genotype is directly measured and modeled as a risk factor.
In this paper, the principles, results, limitations and possible extensions of these three approaches are reviewed.
The empirical approach has been implemented in the Gail et al. model. Based on the data from the Breast Cancer Demonstration and Detection Project (BCDDP), absolute risk estimates of breast cancer were derived by combining data on the five-year follow-up of 243,221 white women and data from a nested case-control sample of 2,852 cases and 3,146 controls. Absolute risk estimates depend on current age as well as four risk factors, namely family history of breast cancer in first-degree relatives, age at menarche, age at first live birth and number of previous breast biopsies. A computer program and graphs allow counselors to easily implement this approach and provide point estimates as well as confidence intervals for the absolute risk of breast cancer. Besides individual counseling, these absolute risk predictions have been used to define inclusion criteria and calculate sample size in a trial to determine whether Tamoxifen can reduce the risk of developing breast cancer. Validation studies showed that this model provides accurate predictions for women in regular screening but tends to exaggerate the risk for unscreened or sporadically screened women.
2. Family History of Breast Cancer: Controlling for the Effect of BRCA1/BRCA2
Claus, Elizabeth B., Yale University
Address: Elizabeth Claus Department of Epidemiology & Public Health Yale University School of Medicine 60 College Street P.O. Box 20-8034 New Haven, Connecticut 06520-8034
Phone: 203-785-6050
Fax: 203-785-6912
Email: elizabeth.claus@yale.edu
Abstract: Approximately ten percent of women in the Unites States have a family history of breast cancer. Although genes for breast cancer has been identified, i.e. BRCA1 and BRCA2 among others, the attributable risk for these genes appears to be low. In fact it is unlikely that women with only one or even two family members affected with breast cancer are carriers of such genes. It is therefore of interest to clarify the role of a family history of breast cancer in risk prediction for non-carriers of the known susceptibility alleles as these women are likely to represent the majority of instances seen in clinic populations.
3. Towards a Comprehensive Model of Breast Cancer Risk
Parmigiani, Giovanni, Duke University
Address: Giovanni Parmigiani Box 90251, ISDS Duke University Durham, NC 27708-0251
Phone: 919-684-8064
Fax: 919-684-8594
Email: gp@stat.duke.edu
Abstract: Accurate breast cancer risk assessment is important for women who consider themselves at high risk for breast cancer and who seek counseling for decision-making purposes regarding prevention and for physicians interested in counseling women about their risks for breast cancer and their options. Also, it is important to improve accurate perceptions about risk among women are not currently seeking counseling, such as first degree relative of breast cancer patients, who may over- or under-estimate their risks. This talk describes our progress to date in developing a comprehensive model for risk assessment in breast cancer. The model has two main goals: including recent advances in genetic susceptibility to breast cancer, and incorporating evidence from several data sets, including a cohort study. The talk will cover some technical aspects of our modeling effort as well as broader issues regarding the current and future role of statistical risk models in cancer prevention and counseling.
List of speakers who are nonmembers: None