Purpose: We investigated whether clinical information routinely available in community practice could predict extracapsular extension of clinically localized prostate cancer in men undergoing radical prostatectomy.
Materials and methods: We examined prostate cancer outcomes in a population based sample of 3,826 patients with primary prostate cancer in 6 regions of the United States covered by the Surveillance, Epidemiology, and End Results program. Stratified and weighted logistic regression was used to identify predictors of and probabilities for extracapsular extension of clinically localized tumors treated with radical prostatectomy.
Results: Nearly 47% of men undergoing radical prostatectomy had extraprostatic extension. The strongest predictors were elevated prostate specific antigen (PSA) greater than 20 versus less than 4 ng./ml. (odds ratio 5.88, 95% confidence interval 2.90 to 11.15), Gleason score greater than 8 versus less than 6 (1.73, 1.04 to 2.87) and age greater than 70 versus less than 50 years (1.91, 0.98 to 3.70). Ethnicity and region were not associated with increased risk of extraprostatic extension. A nomogram developed from our model predicts extracapsular extension ranging from 24% in men younger than 50 years with PSA less than 4 ng./ml. and a Gleason score of less than 7 to 85% in those 70 years old or older with PSA greater than 20 ng./ml. and a Gleason score of 8 or more. If prostatectomy were limited to patients with less than 60% probability of extraprostatic extension based on the nomogram, 95% of those with organ confined cancers would undergo definitive surgery and 18% of those with extracapsular extension would be spared the morbidity of surgery.
Conclusions: In a population based analysis of prostate cancer practice patterns PSA, Gleason score and age are clinically useful predictors of extracapsular extension. Although extracapsular extension may be an imperfect predictor of cancer outcomes, our nomogram provides more realistic probabilities for extracapsular extension than those based on institutional series.