INFERRING PROSTATE CANCER NATURAL HISTORY IN AFRICAN AMERICAN MEN IMPLICATIONS FOR SCREENING RUTH ETZIONI FRED HUTCHINSON CANCER RESEARCH CENTER SEATTLE, WASHINGTON RETZIONI@FREDHUTCH.ORG
EVIDENCE-BASED MEDICINE Evidence-based medicine is an approach to medical practice intended to optimize decision-making by emphasizing the use of evidence from well designed and conducted research The problem with evidence-based medicine is that often well-designed research either does not exist or does not provide the answer The question then is can we learn from non-well-designed research?
Observed OUR WORKdata Prostate cancer incidence Deeper knowledge about disease Overdiagnosis, natural history, screening
PROSTATE CANCER TRENDS IN THE US POPULATION Cases Deaths 52% decline 51% decline Screening begins Screening begins
USPSTF RECOMMENDATIONS 2017 DRAFT
USPSTF ON SCREENING IN AFRICAN AMERICAN MEN African American men are more likely to develop prostate cancer than white men AA men are also more than twice as likely as white men to die of prostate cancer Screening trials do not have enough AA participants to determine whether overall trial results differ for these men Evidence is insufficient to compare the risk of false positives, potential for overdiagnosis, and magnitude of harms from treatment in AA versus other men Based on the available evidence, the USPSTF is not able to make a separate, specific recommendation on PSA-based screening for prostate cancer in AA men
TODAY S PRESENTATION Evidence-based medicine Population trends in prostate cancer by race Absence of screening recommendations for African American men Modeling as a tool for learning about disease from population trends Translating model inferences into policy guidance A model for natural history in African American men Translating model inferences into screening recommendations
MODELING AS A TOOL Use data on disease incidence with and without screening to infer underlying disease natural history Healthy Onset Symptoms Death MM RR MM = duration of healthy state RR = latent preclinical duration Superimpose different screening schedules in a virtual trial, compare benefits and harms Do this for the general population and for African American men
DATA FOR INFERRING UNDERLYING NATURAL HISTORY 1. Incidence without and with screening PSA screening uptake 2. Screening patterns
A MODEL FOR INFERRING UNDERLYING NATURAL HISTORY For given values of MM and RR the model predicts incidence patterns and tries to match with observed Incidence without screening informs about MM + RR Incidence with screening informs about RR If RR is overestimated then the model will predict an inflated incidence with screening Healthy Onset Symptoms Death MM RR
FRED HUTCH MODEL OF PROSTATE CANCER GENERAL POPULATION Fred Hutch model Natural history PSA growth Risk of onset increases with age Risk of progression increases with PSA Survival depends on stage at diagnosis Use the fitted model to simulate a population and conduct a virtual trial
Reduced QOL Preferred strategies JAMA Oncology 2016
PROSTATE CANCER, RACE AND RISK OBSERVED AND MODELLED INCIDENCE BY RACE AND STAGE ALL MEN BLACK MEN Onset in lifetime Mets at dx given onset Mean lead time Fraction overdiagnosed ALL MEN 29% 45% 6% 10% BLACK MEN 7.3 years 7.4 years 43% 40% Allow risks of onset and progression to late stage and clinical state to depend on race Tsodikov, Gulati,. Etzioni Cancer 2017
IMPLICATIONS FOR SCREENING POLICIES I USPSTF 2017: US men age 55 should discuss prostate screening with their MD Among blacks incidence of potentially lethal disease reaches a level at age 45 that matches all races at age 55 Implications for black men Discuss with their MD s at age 45 Potentially screen more frequently subject to careful harm-benefit analysis
IMPLICATIONS FOR SCREENING POLICIES II BENEFIT VS HARM Results from Fred Hutch model of prostate cancer progression and survival Treatment distributions for blacks and all races based on SEER 2012 data Similar efficacy for RP, RT Similar treatment efficacy for blacks and all races
CONCLUSIONS Compared to the general population, African American men have Earlier age at onset Higher risk of developing disease Higher risk of progressing to metastasis This means that the same screening strategies will have Different benefits Different harms Different harm-benefit tradeoffs In African American men Present modeling as a tool to estimate these outcomes in absence of screening trial data Results to date point towards earlier age and potentially more frequent screening in AA men
ACKNOWLEDGMENTS Roman Gulati Lurdes Inoue Josh Roth Alex Tsodikov (Michigan) Harry de Koning (Erasmus) Eveline Heijnsdijk (Erasmus) Angela Mariotto (NCI) Eric Feuer (NCI) ERSPC and PLCO trial investigators CISNET support