Management of a positive surgical margin after radical prostatectomy: decision analysis

We created and tested a decision analysis model to help determine the preferred management of a positive surgical margin(s) after radical prostatectomy.

We constructed a decision tree modeling surveillance versus immediate prophylactic adjuvant radiation in patients with a positive surgical margin(s) after radical prostatectomy. Literature and institution based estimates were determined for certain factors, including the probability of undetectable prostate specific antigen (PSA) in patients followed expectantly postoperatively and those treated with immediate adjuvant radiotherapy, complications of radiotherapy after prostatectomy and probability of undetectable PSA in those treated with therapeutic radiation for detectable PSA postoperatively. A panel of experts assigned utilities to the various outcomes. Sensitivity analysis was performed to determine threshold values required to change the model outcome.

Using average probability estimates from a literature review the decision model recommended initial surveillance. Sensitivity analysis demonstrated that the model depended on the probability of disease recurrence in men followed expectantly after surgery as well as the efficacy of therapeutic radiation. We tested the decision model again for patient groups based on tumor grade, pathological stage, preoperative PSA and number of positive margins. The model recommended initial radiation for patients with low to intermediate grade disease, no evidence of seminal vesicle invasion and multiple positive margins.

The results of our decision analysis imply that immediate radiation may be appropriate for patients with a positive surgical margin(s) and a high likelihood of recurrent local rather than distant disease. This model may be useful to physicians and patients who use individual probability estimates and utility values to determine the preferred course of management after surgery.