Journal of Urology 138(1): 92-98, 1987.
Oesterling JE, Brendler CB, Epstein JI, et al.
The usefulness of clinical stage, serum prostatic acid phosphatase and preoperative Gleason grade in predicting final pathological stage in patients with adenocarcinoma of the prostate remains controversial. To determine the predictive value of these 3 preoperative variables we reviewed 275 patients with clinically localized disease who were treated between April 1982 and February 1986. All patients were examined preoperatively and subsequently were operated upon by 1 urologist. Serum prostatic acid phosphatase was determined in all patients by the Roy method using thymolphthalein monophosphate as the substrate. The Gleason grade of each prostatic biopsy specimen was determined preoperatively by 1 pathologist, who also examined the final pathological specimen with respect to capsular penetration, and seminal vesicle and pelvic lymph node involvement. Using logistic regression analysis with the likelihood ratio chi-square test, clinical stage and Gleason grade had a direct correlation with capsular penetration (p less than 0.0001 and less than 0.0001, respectively), seminal vesicle involvement (p less than 0.0001 and less than 0.0001, respectively) and positive lymph nodes (p less than 0.0001 and less than 0.0002, respectively). Within the normal range of values (0.0 to 0.8 IU/l.) serum prostatic acid phosphatase correlated directly with capsular penetration (p less than 0.003) and seminal vesicle involvement (p less than 0.01) but not with lymph node involvement (p equals 0.08). Again with logistic regression analysis we determined that the best predictors of final pathological stage are not individual variables but models that use combinations of preoperative variables. The models generated are as follows: capsular penetration--serum prostatic acid phosphatase and Gleason grade (p less than 0.00001), seminal vesicle involvement--clinical stage and Gleason grade (p less than 0.00001), and lymph node involvement--clinical stage and Gleason grade (p less than 0.00001). With these models probability plots have been constructed so that the final pathological stage in patients with clinically localized prostatic cancer can be predicted preoperatively.
Rheinische Friedrich- Wilhelms- Universität Bonn