Edward Atkinson

PhD

Rice University

Senior Lecturer

PLCO (Learn more about this study)

PLCO-25

Apr 2, 2013

Assesing the risk of ovarian cancer using serial ultrasounds and transvaginal ultrasound

The Risk of Ovarian Cancer Algorithm (ROCA) uses serial measurements of CA-125 to evaluate the risk a woman has developed undetected ovarian cancer. Based on predicted risk, a woman may be seen again I one year, seen again in three months, or referred for transvaginal ultrasound (TVU). This strategy is currently being assessed in both UKCTOCS and NROSS, large trials of women at normal risk for ovarian cancer. We are implementing ROCA in the RJAGS package of R and comparing the predictions from our implementation with those reported by the NROSS trial. Implementation of ROCA requires prior information on the distributions of CA125 over time in the cohort in question, both for healthy controls and for women who developed ovarian cancer. We will use data on serial measurements of CA 125 obtained from PLCO to derive the necessary priors.

Aim 1. We will use data on (a) serial CA125 measurements, (b) age, (c) TVU results, and (d) cancer status, and (e) menopausal status provided by the PLCO to derive prior distributions for the parameters of the ROCA model.

Aim 2. We will use our implementation of the ROCA method, along with the prior distributions derived from the PLCO data, to obtain predictions of risk for patients from the NROSS trial. We will compare these predictions against the predictions produced by the software used in the NROSS trial.