COMPLEX ANALYSIS OF PROGNOSTIC FACTORS IN PATIENTS WITH ADVANCED AND LOCALLY ADVANCED PROSTATE CANCER

Objective: to study the likelihood of tumor extracapsular invasion (EI) and lymphogenous metastasis (N+) in patients with prostate cancer (PC) after radical prostatectomy (RPE) on the basis of prognostic factors. Subjects and methods. Fifty hundred patients who had undergone RPE in 1999-2008 were enrolled in the study. The patients' age was 63.2 +/- 6.2 (range 46-78) years; median prostate-specific antigen (PSA) was 11.0 ng/ml (interquartile range 7.4-19.5 ng/ml). According to preoperative findings, 418 (78.9%) and 112 (21.1%) patients were diagnosed as having advanced and locally advanced PC, respectively. According to Gleason tumor grades, the patients were distributed as follows: 2-4 scores in 94 (17.7%) patients; 5-6 scores in 266 (50.2%); 7 scores in 103 (19.4%); 8-10 scores in 26 (4.9%). Planned histological studies revealed no tumor in 1.5% of cases; advanced and locally advanced PC was diagnosed in 54.5 and 45.5% of cases, respectively; of them N+ was in 32.9%. Results. A correlation of prognostic factors and expected outcomes was analyzed, the results of which were used to select the most important predictors of the extension of a tumor process (percent of positive biopsy specimens, sum of Gleason scores, PSA level, clinical process stage, presence of EI from preoperative data). Regression models were created considering the results of the logistic regression analysis. To evaluate efficiency, the area under the AUC-ROC curve was calculated for each model. For evaluation of the comparative effectiveness, the AUCROC curve was also estimated for each outcome, by applying the Partin tables. Conclusions. The developed models of predicting the presence of tumor EI and metastatic spread to the lymph nodes make it possible to have a good agreement of the predicted likelihood of an outcome and the results of a histological study. These models show a higher prognostic effectiveness than do the Partin tables.