Correlation of pretherapy prostate cancer characteristics with seminal vesicle invasion in radical prostatectomy specimens

Purpose: This study was conducted to identify pretherapy factors associated with seminal vesicle invasion (SVI) in patients with localized carcinoma of the prostate (CAP), and to develop a model that would allow estimation of the likelihood for SVI at the time of initial diagnosis. Methods and Materials: Between January 1988 and December 1993, 2959 patients underwent radical retropubic prostatectomy, with or without pelvic lymph node dissection, as initial therapy for clinical Stage T1a-3bN0-XM0 CaP. Preoperative patient and tumor-related characteristics were evaluated for an association with SVI in univariate and multivariate logistic regression analyses. A model was developed and probability plots were constructed to display the estimated likelihood for SVI in the patient with a new diagnosis of localized CaP. Results: Within clinical tumor stage, three groups (T1a- 2a, T2b-c, and T3a-b) were observed to have a distinctly different rate of SVI. Gleason primary grades were combined (1-2, 3, and 4-5) because of a similar observation. Univariate analysis identified clinical tumor stage (p < 0.0001), Gleason primary grade (p < 0.0001), and serum prostate-specific antigen level (p < 0.0001) as factors associated with the likelihood for SVI. Multivariate analysis confirmed the independent significance (p = 0.0001) of each of these factors. Patient age (p = 0.16) and history of prior transurethral resection of the prostate (p = 0.82) were not associated with this end point. Probability plots were constructed to display the likelihood of SVI as a function of pretherapy clinical tumor stage, Gleason primary grade, and serum prostate-specific antigen value. Conclusion: In the patient with a new diagnosis of localized CaP, clinical tumor stage as determined by digital rectal examination, diagnostic biopsy tumor (Gleason primary) grade, and pretherapy serum prostate-specific antigen value were significant factors for development of a model that estimated the likelihood of SVI. Estimates from this type of model may be of value in the pretherapy diagnostic evaluation of such patients, and may aid in the administration of radiation therapy.