AS97 - Estimating Risk of Ovarian Cancer

This page provides study documentation for AS97. For description of the specimen results, see Specimen Results Description (open to public). Data sets of the specimen results are included in the existing WHI datasets located on the WHI Data on this site (sign in and a completed Data Distribution Agreement are required; see details on the Data site).

Investigator Names and Contact Information

Garnet Anderson, PhD, Director, Public Health Sciences, FHCRC

Introduction/Intent

The primary objective of this research is to develop a risk model to estimate a woman’s probability of developing ovarian cancer during a defined interval.  

Ovarian cancer afflicts approximately 26,000 women in the US each year.1  More than 75% of these cases will be diagnosed in late stage and their life expectancy is only 3 to 4 years after treatment.  Though treatment strategies have evolved over time, prevention and early detection hold the greatest potential for reducing the morbidity and mortality of this dread disease.   To succeed, research in prevention and screening requires that we be able to identify populations that are expected to experience large numbers of cases.

The more accurately we can target these cases in advance, the more cost-efficient and productive these efforts.  Risk-based eligibility has been used in other randomized trials to target individuals most likely to benefit from the intervention.  The Breast Cancer Prevention Trial, for example, used the Gail model2 to determine eligibility based on a woman’s estimated risk, and the estimated risks in the recruited population justified a reduction in  sample size from 16,000 to 13,388[1].  Risk-based screening, except as defined by age, has not been standard for the relatively common diseases of breast and prostate cancers. Recently, however, there have been some attempts to develop risk-based screening for younger women[2] where lower incidence and more false-positive screens make general population screening controversial.  Such tailoring to risk levels offers the potential of additional cost-effectiveness.  For ovarian cancer, with an incidence rate roughly 1/7th that of breast cancer, a cost-effective screening approach may very well depend on our ability to appropriately target interventions to individuals.

We are particularly interested in developing a multimodal screening strategy using TVS only in women at sufficiently high risk.  Prior work suggests that using a tumor marker such as CA-125 to define elevated risk is promising in that it may lower error rates and improve cost-effectiveness.[3]   The use of a risk model based on known risk factors but extended to include a small set of the most promising tumor markers may be even more effective in identifying women who would benefit from routine TVS screening.

Specific Aims:

1.     To develop and validate a statistical model for estimating ovarian cancer risk in post-menopausal women using demographic, reproductive, medical and family history data from the Women’s Health Initiative.

2.     To extend the risk model in specific aim 1 through a nested case-control design to include serum levels of CA-125 and other promising tumor markers and to determine whether these marker levels improve our ability to identify women at high risk of ovarian cancer. 

Specific aim 1 will produce a risk model that would support protocol design in ovarian cancer prevention trials both to determine eligibility and estimate incidence rates and power.  Such a model may also be used downstream to determine at what point biomarker screening is advised or to provide objective risk estimates as a basis of counseling women.  In specific aim 2 we will determine whether the best set of markers identified through Project 4 or other ongoing research can be used to improve upon the risk prediction developed in specific aim 1. 

Subsidiary aims include:

(a)   describing selected tumor marker behavior in the pre-diagnostic period for postmenopausal women diagnosed with ovarian cancer and over the same time interval in similar but healthy women;

(b)   correlating marker behavior with disease risk factors;

(c)   correlating marker behavior with disease characteristics in cases, in particular invasive versus borderline disease;

(d)   validating the test performance of the selected markers and screening rules produced in Project 1 in an independent population; and

(e)   evaluating the cost-effectiveness of risk-based screening using this panel of tumor markers through micro-simulation modeling.

References