Predicting the need for adjuvant systemic therapy in patients receiving postprostatectomy irradiation

Objectives. This study was initiated to determine if clinical, pathologic, or biochemical variables available in patients receiving postprostatectomy irradiation could be predictive of treatment outcomes, including the need for subsequent systemic therapy. Methods. Between January 1992 and January 1995, 50 patients received external beam irradiation for a nonzero or rising postprostatectomy prostate-specific antigen (PSA) level. The median PSA values preoperatively and preradiation were 24 and 2.7 ng/mL, respectively. At the time of treatment, no patient had evidence of disseminated disease. The patients received a combination of axial and noncoplanar irradiation to an average total dose of 65 Gy at 1.8 Gy fractions. Univariate analysis and multivariate logistic regression were performed to identify factors predictive of outcome. Results. Treatment was well tolerated with no grade 3 or higher gastrointestinal or genitourinary complications. A decline in the serum PSA during irradiation occurred in 82% of the patients. During radiation, 13 patients (26%) had a complete response (PSA less than 0.05 ng/mL), 56% had a partial response, and 18% had a rise in their PSA level. With a median follow-up of 16 months, 50% of all patients have an undetectable level of PSA. Twenty-six percent of patients have had evidence of biochemical progression. Seventeen percent of responding patients and 67% of nonresponders have evidence of progression (P= 0.002). On univariate analysis, other significant prognostic factors included clinical stage (P <0.01), the preradiation PSA level (P <0.05), and N stage (P <0.01). On stepwise multivariate logistic analysis, a predictive model was generated from which treatment outcome could be predicted with 80% accuracy (percent correct classification) and an 85% sensitivity. Conclusions. The outcome of patients with an elevated postprostatectomy PSA level who receive irradiation can be accurately predicted using a combination of clinical stage, preradiation PSA value, Gleason score, and the response to treatment. In this way, patients who do not require additional treatment (that is, hormonal) will avoid the unnecessary cost and toxicity.