AUA 2019: Understanding of Prognosis in Non-Metastatic Prostate Cancer: A Randomized Comparative Study of Clinician Estimates Against the PREDICT Prostate Tool

Chicago, IL (Urotoday.com) Decision-making around treatment for non-metastatic prostate cancer is notoriously complex. Treatment counseling is performed by clinicians with different perceptions on disease lethality and hence biased opinions on the value of radical treatment. At today’s Epidemiology session at the AUA 2019 meeting, Vincent Gnanapragasam and colleagues discussed findings of their multivariable prognostic model PREDICT Prostate tool.   They assessed clinician understanding of prognosis in prostate cancer, comparing responses to the PREDICT Prostate, which is derived from high-quality survival data and validated in three prostate cancer cohorts (including >80,000 men).

For this study, the participation of prostate cancer specialists was requested predominantly through professional mailing lists. Respondents were randomized into group A or B and presented with opposing hypothetical vignettes: (i) 6 with clinical diagnostic information only and (ii) 6 with these details plus PREDICT Prostate estimates.

AUA 2019 common opening questions


Comparisons were made between groups for clinician-estimated and model-predicted 15-year outcomes.

There were 190 responses received, of which 63% were urologists and 17% were oncologists. Furthermore, 60% of respondents work in specialist cancer centers and 82% counsel men with prostate cancer at least weekly. Interestingly, only 19.3% reported using any survival prediction tool in their current routine practice. Clinician estimates of 15-year prostate cancer mortality exceeded PREDICT Prostate estimates in 92% of the case vignettes; clinicians estimated 1.9-fold greater disease lethality than PREDICT. Clinician perceptions of overall survival benefit from radical treatment at 15 years were over-optimistic in every vignette, with mean clinician estimates 5.4-fold greater than PREDICT. Concomitantly viewing data from PREDICT Prostate led to reductions in the likelihood of recommending radical treatment in 75% of vignettes, with reductions most evident in intermediate-risk cases. Dr. Gnanapragasam notes for example, in a fit 75-year old man with PSA 5.1, Gleason 3+4 prostate cancer in 2/12 biopsy cores: likelihood of recommending treatment dropped from 32.5% to 19.1% when PREDICT estimates were also shown (p=0.009).

This is an important study and results may help with appropriately managing patients considering active surveillance for appropriate risk of prostate cancer. Dr. Gnanapragasam concluded with several summary points:
  • Clinicians tend to overestimate prostate cancer lethality and the benefits of radical treatment
  • Significant variation in baseline estimation of lethality and the benefits gained from treatment
  • A standardized individual prognostic model may help inform true relative survival benefits and reduce treatment recommendation variation.
This tool is available free online at: prostate.predict.nhs.uk

Presented by: Vincent Gnanapragasam, University of Cambridge, Cambridge, United Kingdom

Written by: Zachary Klaassen, MD, MSc – Assistant Professor of Urology, Georgia Cancer Center, Augusta University - Medical College of Georgia, @zklaassen_md at American Urological Association's 2019 Annual Meeting (AUA 2019), May 3 – 6, 2019 in Chicago, Illinois