Impact of a secondary biomarker on time-to-event risk prediction of prostate cancer mortality in a screening cohort

Principal Investigator

Name
Kristian Stensland

Degrees
MD MPH MS

Institution
The University of Michigan

Position Title
Assistant Professor of Urology and Learning Health Sciences

Email
kstens@med.umich.edu

About this CDAS Project

Study
PLCO (Learn more about this study)

Project ID
PLCO-2012

Initial CDAS Request Approval
Jan 26, 2026

Title
Impact of a secondary biomarker on time-to-event risk prediction of prostate cancer mortality in a screening cohort

Summary
Time-to-event prediction is important in prostate cancer screening given the long interval between an abnormal screening test and eventual prostate cancer mortality. Our recent work has developed and externally validated a prediction model for time-to-prostate cancer mortality for use in the screening setting, with the potential advantage of identifying patients who are at a greater risk of losing more life-years to prostate cancer. Many screening PSA tests are followed by a secondary biomarker, but it is unclear what impact these tests have on time-to-event prediction and how they can be used to optimize prostate cancer screening strategies. We hope to explore this question using the Free PSA dataset.

Aims

Most prostate cancer screening relies on use of prostate specific antigen (PSA), with frequent use of a secondary biomarker such as free PSA. However, the application of these biomarkers in existing models has historically not included key results to inform decision making, like time-to-event (e.g., time to prostate cancer mortality) predictions, limiting the utility of these biomarkers in shared decision making. Our recent work addressed this gap by developing and externally validating a prediction model for time to prostate cancer mortality in the screening setting, relying on PSA and incorporating competing risks based on patient comorbidities and other factors. In addition to a 'standard' screening PSA, other biomarkers such as free PSA are often used to complement screening PSA alone. We seek to expand on our prior analyses by analyzing the additive benefit of free PSA in predicting prostate cancer specific mortality and informing prostate cancer shared decision making. For these reasons, we propose these specific aims:

1. To model lifetime risk of prostate cancer mortality amongst a screening population including Free PSA .

2. To model lifetime risk of non-prostate cancer mortality amongst patients screened for prostate cancer in light of their Free PSA results.

3. To create a multivariable prediction model integrating these two calculations.

Impact: Through these analyses, we hope to further improve the ability to prognosticate prostate cancer relevant outcomes using common biomarkers. With these data, we hope to improve shared decision making surrounding prostate cancer to better tailor decisions surrounding prostate cancer screening, workup, and treatment.

Collaborators

Kristian Stensland The University of Michigan
Patrick Lewicki The University of Michigan