Estimating prostate cancer screening, diagnostic follow-up, and detection processes for a natural-history microsimulation model

Principal Investigator

Name
ALEX BRYANT

Degrees
M.D.

Institution
Regents of the University of Michigan

Position Title
Principal Investigator

Email
bralex@med.umich.edu

About this CDAS Project

Study
PLCO (Learn more about this study)

Project ID
PLCO-2077

Initial CDAS Request Approval
Sep 8, 2026

Title
Estimating prostate cancer screening, diagnostic follow-up, and detection processes for a natural-history microsimulation model

Summary
Individual-level microsimulation models of prostate cancer screening depend critically on the observation process linking underlying disease to clinical detection, including screening participation, screen positivity, diagnostic follow-up, biopsy, diagnosis, and downstream consequences of diagnostic evaluation. These processes are often represented using assumptions or sparse external evidence because few data sources directly characterize what occurs after a positive screening test. The PLCO intervention arm provides protocol PSA and DRE screening together with ascertainment of subsequent diagnostic evaluation, including biopsy and other diagnostic procedures, as well as prostate cancer characteristics, diagnostic complications, and treatment information. We propose to use PLCO prostate data to estimate key components of this observation process: the distribution of PSA and DRE results by age and screening round; the probability and timing of diagnostic follow-up and prostate biopsy after a positive screen, conditional on PSA level, age, round, DRE findings, and prior screening history; cancer yield, Gleason grade, and stage at diagnosis conditional on the pathway to detection; and selected downstream consequences of diagnostic evaluation and detection. These estimates will parameterize the observation layer of an individual-level prostate cancer natural-history microsimulation designed for screening-policy evaluation. We will assess internal consistency by comparing model-predicted screening and detection outcomes with observed PLCO intervention-arm outcomes.

Aims

Aim 1. Characterize protocol screening results in the PLCO intervention arm.
Estimate distributions of PSA level and DRE outcome by screening round and age, characterize screen positivity under the PLCO protocol criterion, and evaluate alternative PSA thresholds relevant to screening-policy analyses. Where informative, characterize the features of abnormal DRE findings using the detailed screening-abnormalities data.

Aim 2. Estimate diagnostic follow-up after positive screening tests.
Estimate the probability and timing of diagnostic evaluation and prostate biopsy following a positive screen as a function of PSA level, DRE result, age, screening round, prior screening history, and prior diagnostic work-up. Distinguish participants with documented absence of diagnostic procedures from those with incomplete or unknown follow-up ascertainment, and characterize common follow-up pathways, including repeat PSA testing, imaging or other diagnostic evaluation, and biopsy.

Aim 3. Estimate cancer detection outcomes conditional on the diagnostic pathway.
Estimate the probability of prostate cancer diagnosis following biopsy and characterize Gleason grade and clinical stage at diagnosis conditional on PSA level, age, and detection pathway, including screen-prompted and interval or non-protocol detection.

Aim 4. Characterize selected downstream consequences of diagnostic evaluation and cancer detection.
Describe complications associated with prostate diagnostic evaluation and characterize initial treatment patterns following prostate cancer diagnosis. These estimates will support representation of downstream harms and treatment consequences in screening-policy simulations and will provide additional checks on the internal coherence of modeled detection pathways.

Aim 5. Parameterize and evaluate the prostate cancer screening observation submodel.
Embed estimates from Aims 1–4 in the observation layer of an individual-level prostate cancer natural-history microsimulation and assess internal consistency by comparing model-predicted, round-specific screening positivity, diagnostic follow-up, biopsy use, cancer detection, and other observable outcomes with corresponding PLCO intervention-arm results.

Collaborators

ALEX BRYANT Regents of the University of Michigan