Model Integration for Prostate Cancer Mortality Absolute Risk Prediction

• Activity Number: 197 - SPAAC Poster Competition
• Type: Topic Contributed
• Date/Time: Monday, August 8, 2022 : 2:00 PM to 3:50 PM
• Sponsor: Biometrics Section
• Author(s): Elizabeth C Chase* and Elise C Covert and Matthew Schipper and Robert Dess and William Jackson
• Companies: University of Michigan and University of Michigan and University of Michigan and University of Michigan and University of Michigan
• Keywords: Absolute risk; Competing risks; Cause-specific hazards; Prostate cancer

Abstract

Other-cause mortality (OCM) is the primary cause of mortality in early-stage prostate cancer patients. However, most research has focused on prostate-cancer specific mortality (PCSM), with less attention paid to the competing risk of OCM. The emphasis on PCSM may disguise the true drivers of absolute risk of mortality in prostate cancer patients. We explore the relative impact of predictors of OCM and PCSM on absolute risk of death in prostate cancer patients. To do so, we integrate two cause-specific hazard models––the STAR-CAP prediction model for PCSM and the OCCAM model for OCM––to provide predictions of the cumulative incidence of PCSM, OCM, and overall survival. We validate the integrated model’s performance for predicting overall survival in the Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial and implement it as a Shiny app. Predictions from the integrated model suggest that for early-stage prostate cancer patients, the impact of traits like smoking status, diabetes, and hypertension may play a comparable or larger role in absolute risk of death as prostate cancer features such as Gleason score, prostate specific antigen (PSA), or T stage.