Study Highlights Need for Risk Model Calibration
Gail model's effectiveness altered by changes in invasive breast cancer rates
THURSDAY, April 8 (HealthDay News) -- The Gail model, used to assess risk of invasive breast cancer, may need to be recalibrated to reflect changes in the incidence of breast cancer, according to research published online April 5 in the Journal of Clinical Oncology.
Because of changes in breast cancer incidence in the 1990s, Sara J. Schonfeld, of the National Cancer Institute in Bethesda, Md., and colleagues evaluated the Gail model's calibration in two recent cohorts: white, postmenopausal women from the National Institutes of Health-AARP Diet and Health Study (NIH-AARP), and the Prostate, Lung, Colorectal and Ovarian Cancer Screening Trial (PLCO).
The researchers found that the Gail model significantly under-predicted the number of invasive breast cancers in NIH-AARP and PLCO, with expected-to-observed ratios of 0.87 and 0.86, respectively. An updated model that combined Gail model relative risks with 1995 to 2003 Surveillance, Epidemiology, and End Results cancer incidence rates fared better, with expected-to-observed ratios of 1.03 and 1.01, respectively. In addition, 13 to 14 percent of women aged 50 to 55 at baseline had a projected Gail model five-year risk lower than the recommended threshold of 1.66 percent for use of tamoxifen or raloxifene, but had a risk of at least 1.66 percent with the updated model.
"In conclusion, the calibration of risk prediction models is sensitive to trends in underlying population rates," the authors write. "This study highlights the importance of using appropriate baseline rates in absolute risk models and the importance of calibration of risk prediction models for clinical decision-making."