This project supports research to improve methods for estimating absolute and attributable risk and collaborative studies to estimate such risks for various cancers. We found that mammographic density was a promising predictor of breast cancer risk using data from the Breast Cancer Detection and Demonstration Project (BCDDP), and we gathered additional information on mammographic density from BCDDP to construct a risk model that incorporates this factor. Studies to assess the reliability of these data and to incorprate mammographic data into a model to project breast cancer risk are in progress. We have begun to gather information from a large case-control study in African-American women to improve the model for projecting the risk of breast cancer in African-American women.We validated a previously developed model for projecting breast cancer risk using independent data from the Breast Cancer Prevention Trial. We incorporated a version of this model for projecting the risk of invasive breast cancer into a computer program that also offers other medical information useful in deciding whether or not to take tamoxifen to prevent breast cancer. This material has been distributed widely on diskette by NCI's Office of Cancer Communications and is also available at NCI's web site. We continue to update this program to increase flexibility and provide additional information to help a woman put her risk in perspective.We presented a detailed analysis of the risks and benefits of tamoxifen prophylaxis against breast cancer for women with various levels of projected breast cancer risk. The benefit-risk ratio is most favorable for young women with high risk of breast cancer, because young women are less likely to have serious side effects from tamoxifen, such as stroke or endometrial cancer.We began a collaboration to develop a model for the individualized absolute risk of colon cancer.We analyzed the strengths and weaknesses of the kin-cohort design for estimating the risk of disease from an autosomal dominant gene. We developed a test to detect sources of familial aggregation in kin-cohort studies that are not related to the genes being studied. We developed marginal methods of analysis of kin-cohort data that are more robust to such residual familial correlation than are likelihood-based methods that ignore such correlation. We developed maximum likelihood and pseudo-likelihood to estimate monotone genotype-specific survival functions from kin-cohort data. We developed a bivariate copula models to accommodate residual familial correlation and to project risk based not only on genotype but also on the phenotypes of other family members.We developed bivariate cure models for familial survival data from randomly selected families. We developed methods for estimating covariate-adjusted survival curves and population attributable risk from stratified case-cohort studies.
| Gail, MH; Costantino, JP; Bryant, J et al. (2000) RESPONSE: re: weighing the risks and benefits of tamoxifen treatment for preventing breast cancer J Natl Cancer Inst 92:758 |
| Gail, M H; Costantino, J P; Bryant, J et al. (2000) Re: Risk/benefit assessment of tamoxifen to prevent breast cancer-still a work in progress? J Natl Cancer Inst 92:574-5 |
| Gail, M H; Costantino, J P; Bryant, J et al. (1999) Weighing the risks and benefits of tamoxifen treatment for preventing breast cancer. J Natl Cancer Inst 91:1829-46 |
| Costantino, J P; Gail, M H; Pee, D et al. (1999) Validation studies for models projecting the risk of invasive and total breast cancer incidence. J Natl Cancer Inst 91:1541-8 |
