Biostatistics and Computational Biology
Using Bayesian methods to study the time-dependent behavior of the ovarian cancer marker CA-125, Professor Stephen Skates has developed an algorithm for predicting the risk of ovarian cancer. This algorithm now forms the basis of two prospective screening trials among women at risk for ovarian cancer:
The U.K. trial will randomize 200,000 post-menopausal women above the age of 50 into three groups:
The endpoint of the trial is ovarian cancer mortality, and the main goal is to determine if either of the experimental approaches reduces mortality compared to the control group. Professor Skates is PI of the U.S.-based study and co-PI of the U.K. study.
This program is a collaboration of seven research groups modeling the impact of screening and adjuvant treatment on breast cancer mortality in the United States over the period of 1975-2000.
In 1975, breast cancer mortality among women 30-79 years of age was 48.3 deaths per 100,000 women. This rate increased slightly to 49.7 per 100,000 in 1990, then fell to 38.0 per 100,000 in 2000 (a 24% drop since 1990). Dana-Farber investigators and six other research groups developed independent statistical models to assess the relative and absolute contributions of screening mammography and treatment to the reduction in breast cancer mortality.
The proportion of the total reduction in the breast cancer mortality rate attributed to screening varied from 28% to 65% in seven groups, with a median of 46%. For adjuvant treatment, this proportion ranged 35% to 72% with a median of 54%. Although there was some variation in actual estimates, all seven models concluded that both screening mammography and treatment have contributed in reducing the breast cancer morality rate in the United States.
Professor Marvin Zelen and Dr. Sandra Lee lead a joint research program investigating the effectiveness of screening in breast cancer, and they currently lead one of the sites in CISNET.