Prostate Modeling to Identify Surveillance Strategies

Graph showing data from four active surveillance cohorts (GS6) before integratice analysis.
Graph showing data from four active surveillance cohorts (GS6) before integratice analysis.


The widespread adoption of PSA screening in the US has created an epidemic of low-risk prostate cancer. The vast majority of low-risk cases will not die of their disease yet they continue to seek curative treatment. Active surveillance (AS) has become a preferred option for managing low-risk prostate cancer, but no evidence-based standard exists for how to implement it. In AS, cases are closely monitored with intent to treat if there is any evidence of disease progression. Several AS studies are ongoing, but they use different approaches and their results cannot be readily compared or integrated.

This primary objective of this research is to determine an optimal approach to AS given patient characteristics and preferences. We are collaborating with investigators conducting four of the largest and most well-followed AS cohorts in North America to (a) integrate the data to produce a representative estimate of the risk of disease progression on AS and (b) develop models projecting the long-term outcomes of different AS policies for men with low-risk disease.

We have received and processed date from the four cohorts listed below (a total of over 3,000 cases) and have implemented two integrative analyses – Bayesian joint modeling and Hidden Markov modeling – designed to produce comparable estimates of the risk of biopsy upgrading from Gleason Score (GS) 6 to GS 7 and above. The derived estimates account for differing intervals between biopsies and different risks of competing treatment. The resulting estimates are interpretable as natural history estimates in that they represent the cumulative upgrading risk on an annual basis and in the absence of competing treatment. The results reveal persistent differences among cohorts likely reflecting differences in underlying risk. Thus, not all AS cohorts are comparable and we recommended against developing recommendations for AS strategies based on any single cohort.


Ruth Etzioni
Daniel Lin
David Penson
Jane Lange
Roman Gulati
Lurdes Inoue
John Gore
Bruce Trock
Matthew Cooperberg
Janet Cowan
Peter Carroll
Lawrence Klotz

Active Surveillance Cohorts

Johns Hopkins University
University of Toronto

Overview of 5 active surveillance cohort studies participating in PROMISS

Overview of 5 active surveillance cohort studies participating in PROMISS

Overview of aims in PROMISS

Overview of aims in PROMISS

PROMISS Publications

Lange JM, Gulati R, Leonardson AS, Lin DW, Newcomb LF, Trock BJ, Carter HB, Cooperberg MR, Cowan JE, Klotz LH, Etzioni R. Estimating and comparing cancer progression risks under varying surveillance protocols. Ann Appl Stat. 2018 Sep;12(3):1773-1795.

Inoue LYT, Lin DW, Newcomb LF, Leonardson AS, Ankerst D, Gulati R, Carter HB, Trock BJ, Carroll PR, Cooperberg MR, Cowan JE, Klotz LH, Mamedov A, Penson DF, Etzioni R. Comparative analysis of biopsy upgrading in four prostate cancer active surveillance cohorts. Ann Intern Med. 2017 Nov 28.

Ankerst DP, Xia J, Thompson IM Jr, Hoefler J, Newcomb LF, Brooks JD, Carroll PR, Ellis WJ, Gleave ME, Lance RS, Nelson PS, Wagner AA, Wei JT, Etzioni R, Lin DW. Precision medicine in active surveillance for prostate cancer: development of the Canary-Early Detection Research Network active surveillance biopsy risk calculator. Eur Urol. 2015 Dec;68(6):1083-8.