Abstract: PD09-06
Sources of Funding: None

Introduction

Physicians need practical methods to accurately estimate life expectancy when counseling older men with comorbidities regarding treatment of prostate cancer. Although numerous nomograms exist for prediction of life expectancy (LE), few are used in practice due to the difficulty of integration into busy clinical workflows. We sought to determine if survival could be accurately predicted if reduced to a count of several common comorbidities that pose a high risk to mortality. In selecting these comorbidities, we aimed to balance frequency and risk in order to maximize identification of men at risk for overtreatment based on <10-year LE.

Methods

We sampled 1,598 men with newly diagnosed prostate cancer at two Southern California Veterans Affairs Medical Centers from 1998 to 2004. We created rank-ordered lists of comorbidities organized by frequency and highest risk of mortality. Separate ranked lists were then created by differentially weighting comorbidities by frequency to risk ratios: 1:6, 1:4, 1:2, 1:1, 2:1, 4:1, and 6:1. By successively adding comorbidities from highest- to tenth highest-ranked, a set of 10 candidate comorbidity indices was constructed for each list. Using competing risks regression analysis, we determined c-index, the number of men with <10-year LE, and the number of men with <10-year LE treated with surgery or radiation for each index.

Results

Candidate comorbidity indices heavily weighted by frequency were poor at identifying men with <10-year LE, while indices heavily weighted by risk of mortality failed to identify men who were overtreated. Six candidate indices each found more than 300 men with <10-year LE (range 303-392); all six were weighted either 2:1, 1:1, or 1:2 by frequency to risk ratio and included highly similar comorbidities. Two of the six indices identified more than 200 men with <10-year LE overtreated with surgery or radiation (range 173-203). The candidate index with the highest number overtreated was weighted 1:1 by frequency to risk and included six comorbidities: 1) chronic obstructive pulmonary disease 2) congestive heart failure 3) peripheral vascular disease 4) stroke 5) myocardial infarction 6) exertional angina. C-index for this index was 0.66.

Conclusions

A simple count of six comorbidities predicts the risk of 10-year other-cause mortality and robustly identifies men who are overtreated for early stage prostate cancer. Simplifying estimation of life expectancy may be key to operationalizing this critical variable for prostate cancer decision-making.

Funding

None