Cancer Risk Prediction Models: Priorities for Future Research
More than 100 researchers met last May at an NCI-sponsored workshop on statistical models for predicting a person's risk of developing cancer. A detailed report on the meeting, including recommendations for future research on cancer risk prediction models, appears in a commentary in the May 18 Journal of the National Cancer Institute.
Participants identified research priorities and resources in the areas of: 1) revising existing breast cancer risk assessment models and developing new models; 2) encouraging the development of new risk models; 3) obtaining data to develop more accurate risk models; 4) supporting validation mechanisms and resources; 5) strengthening model development efforts and encouraging coordination; and 6) promoting effective cancer risk communication and decision-making.
The workshop included epidemiologists, statisticians, geneticists, clinicians, and genetic counselors, among others. They identified strengths and limitations of cancer and genetic susceptibility prediction models in use or under development, and they explored methodological issues related to their development, evaluation, and validation.
"This meeting brought together a diverse group of scientists to talk about how we can develop and improve our current statistical tools for predicting the development of cancer," says Dr. Andrew Freedman of NCI's Division of Cancer Control and Population Sciences, who co-chaired the meeting.
"There's been a lot of interest in cancer risk prediction models, and now is an important time to explore issues involved in developing, applying, and evaluating these models," says Dr. Freedman.
The workshop focused attention on risk prediction models and "actually has spurred quite a bit of research activity" since last spring, notes Dr. Ruth Pfeiffer of NCI's Division of Cancer Epidemiology and Genetics, (DCEG), the other co-chair. DCEG staff are working on models for melanoma and colorectal cancer, and criteria for evaluating risk models.
Among the findings to emerge from the workshop, Dr. Pfeiffer says, is the importance of communication between the people who develop the models and the people who use them.
The number of risk models has grown steadily since a model for predicting the risk of heart disease was published in 1976. In the late 1980s, researchers began to publish models that predicted a woman's risk of breast cancer based on such risk factors as age, age at menarche, age at first live birth, and family history of the disease.
Today, statistical models are widely used by physicians to make decisions about cancer prevention and treatment. Like clinicians and researchers, the public is interested in cancer risk prediction, as is clear from the number of related Web sites, handbooks, and information resources from professional societies.
A number of companies in the United States and the United Kingdom offer genetic risk profiling. "With the proliferation of new risk models, there's been a concern that the models are used appropriately, and this was one reason for the workshop," notes Dr. Freedman.