Skip to Main Content
An official website of the United States government

Predicting how the ROCA algorithm would have performed in PLCO

Principal Investigator

Name
Paul Pinsky

Institution
NCI, DCP, EDRG

Email
pinskyp@mail.nih.gov

About this CDAS Project

Study
PLCO (Learn more about this study)

Project ID
2011-0089

Initial CDAS Request Approval
May 19, 2011

Title
Predicting how the ROCA algorithm would have performed in PLCO

Summary
The PLCO trial utilized TVU and CA-125 at a fixed cutoff for ovarian cancer screening. The risk of ovarian cancer (ROCA) algorithm is being used currently in the UK ovarian cancer screening trial. It utilizes serial values of CA-125 as well as age. With the PLCO ovarian results being reported, it is of interest to try to estimate how screening with the ROCA algorithm could have possibly altered the observed PLCO result. The idea is to utilize a version of ROCA that can be implemented with the serial screening CA125 values in PLCO. At each PLCO screen, the ROCA algorithm would give a positive or negative result. This ROCA screening result can be compared to the PLCO screening result (positive or negative). The ROCA could be used alone or in conjunction with the (PLCO) TVU, where in the latter case either test being positive would trigger a positive result. Based on the PLCO result and the ROCA result, various analyses can be made of the screened arm mortality data, including the most favorable case for ROCA. In this, any woman dying of ovarian cancer whose data of diagnosis would have been earlier based on the ROCA result could be assumed to not have died of ovarian cancer. Other analyses will also be used.

Aims

1. To estimate the sensitivity of the ROCA algorithm at each screening round in PLCO and compare to CA-125 and TVU. 2. To examine those who died from ovarian cancer and see whether the time of diagnosis might have been altered had they been screened with ROCA instead of CA-125 and TVU. 3. To project possible mortality rate ratios under a ROCA screening algorithm, using various assumptions, including a "best case" assumption. In the "best case" scenario, if ROCA caused the diagnosis date of a woman who subsequently died of ovarian cancer to be moved up, which could be a PLCO screen diagnosed case moved up to an earlier screen diagnosis with ROCA or a PLCO interval case being moved into a ROCA screened diagnosed case, then that women would be designated as not having died of ovarian cancer. If, on the other hand, ROCA caused the woman's diagnosis to be moved into the future (say by giving a negative screen where PLCO gave a screen detected cancer) and that woman did not die of ovarian cancer, the woman would still be designated as not having died of ovarian cancer.

Collaborators

Amanda Black (NCI)
Ed Partridge (UAB)
Steve Skates (Partners)
Claire Zhu (Division of Cancer Prevention)

Related Publications