The effect of race/ethnicity on the accuracy of the 2001 Partin Tables for predicting pathologic stage of localized prostate cancer.

OBJECTIVES

To test the accuracy of the 2001 Partin Tables in African American men who underwent radical prostatectomy at multiple centers throughout the United States.

METHODS

We compiled a large multiethnic cohort of men (n = 3748) treated with radical prostatectomy at multiple sites, including all of the sites of the Department of Veterans Affairs-based Shared Equal Access Regional Cancer Hospital (SEARCH) database (n = 1524), Wayne State University (n = 1305), the University of Texas Health Science Center (n = 522), and the Henry Ford Hospital (n = 397). We evaluated the accuracy of the 2001 Partin Tables using area under the receiver operator characteristic curve (AUC) separately among African American and white men.

RESULTS

African American men (n = 1188, 32%), despite being more likely to have clinical Stage T1c disease (56% versus 47%, chi-square P <0.001), had higher preoperative PSA values (9.1 versus 7.7 ng/mL, rank-sum P <0.001) and were more likely to have higher-grade disease on diagnostic biopsy (chi-square P = 0.005). Despite these differences in baseline clinical characteristics, the 2001 Partin Tables performed equally well in both racial groups. Specifically, there were no differences in the AUC for African American and white men for predicting organ-confined disease (AUC 0.73 versus 0.72; P = 0.56), extraprostatic extension (AUC 0.62 versus 0.62; P = 0.99), or seminal vesicle invasion (AUC 0.77 versus 0.79; P = 0.53).

CONCLUSIONS

These data lend further support to the idea that although baseline differences between the races existed that may underlie an overall more aggressive disease among African American men, for the individual patient, race is not valuable for prognostication.