Back to 2014 Annual Meeting Abstracts

INTRODUCTION: Guidelines recommend that clinicians consider life expectancy when counseling patients regarding treatment for localized prostate cancer. Most patients with localized prostate cancer die of other causes. Life expectancy is strongly associated with age, functional status, comorbidity, smoking status and socioeconomic factors. To explore whether life expectancy related factors are considered in treatment allocation, we examine their associations with types of definitive therapy in a cohort of older men, controlling for tumor characteristics and county-level prostatectomy rates.
METHODS: Using the linked Surveillance Epidemiology and End Results (SEER) and Medicare Health Outcomes Survey (MHOS) databases we identified men with newly diagnosed, clinically localized, prostate cancer from 2004-2009. Tumors were classified according to D’Amico risk group, utilizing PSA level, Gleason score and clinical tumor staging. Stepwise backward selection (with p<0.2 as a stopping rule) was performed for a multinomial logistic regression model with treatment allocation as the outcome (conservative management vs. radical prostatectomy vs. radiation therapy) and the candidate predictors: age at diagnosis, comorbidity count, smoking status, patient-reported physical health (SF-12 Health Survey PCS), race, household income, county-level prostatectomy rates from the Dartmouth Atlas of Health Care, and D’Amico risk group.
RESULTS: We identified 1,316 patients with cT1a-cT3a prostate adenocarcinoma and baseline data from the MHOS database, with a mean age of 75 at diagnosis. Within 1 year of diagnosis, 13% (n=172) underwent radical prostatectomy, 56% (n=733) received either external beam radiation or brachytherapy or a combination, and 31% (n=411) were conservatively managed. Significant predictors of treatment allocation in the final multivariate adjusted model were age, comorbidity count, smoking status, patient-reported physical health, D’Amico score, and county-level prostatectomy rates. The variables significantly associated (p<0.05) with radical prostatectomy (RP) vs. conservative management included age at diagnosis (31% less likely to undergo RP for each additional year at diagnosis), patient-reported physical function (35% less likely to undergo RP for one standard deviation decreased function), comorbidity count (18% less likely to undergo RP for each additional comorbidity), and D’Amico risk group (20.5 times more likely to undergo RP if high risk vs. low risk). The same predictors were less consistently associated with radiation therapy (RT) vs. conservative management. Patients were 9% less likely to undergo RT for each additional year of age at diagnosis (p<0.001) and were 71% more likely to undergo RT vs. conservative management with intermediate vs. low D’Amico risk tumors (p=0.001). County-level RP rates were not significantly associated with surgical treatment allocation, although, in regions where RP rates were above the national average, patients were less likely to receive RT (p=0.027).
CONCLUSIONS: Patient-level health factors traditionally predictive of mortality risk (age, comorbidity count, physical function, and smoking) are also significantly associated with surgical treatment allocation, after adjusting for clinical tumor characteristics and regional practice patterns. However, many of these same factors (comorbidity count, physical function, and smoking) may not be associated with the likelihood of receiving radiation therapy. Future work should explore the appropriateness of patient allocation to radiation therapy, particularly among older men.