Use of serum Prostate Specific Antigen doubling time and change to determine need for secondary treatment for prostate cancer following radical prostatectomy
Importance: Biochemical recurrence (BCR) following radical prostatectomy (RP) is an unreliable predictor of distant metastatic progression or prostate cancer death, consequently resulting in overtreatment. Following BCR, guidelines recommend that patients are treated with radiation therapy. However, little has been published about observation without secondary treatment management recommendations.
Objectives: Establish that a cohort of patients can be managed with observation without secondary intervention, and establish a process to safely manage men with BCR using observation versus therapeutic secondary intervention following RP using kinetics of PSA doubling time (PSAdt).
Methods: In a retrospective cohort analysis of 1865 patients following RP from June 2002 and September 2019 at a tertiary referral center, 407 patients experienced BCR as defined as two PSA levels >0.2ng/ml. 137 were managed with observation compared to 270 treated with secondary intervention. Using PSAdt graphs, patient doubling times (DT) kinetics were (re)calculated with each new date and PSA level and categorized as Increasing or Decreasing. Kaplan-Meier analysis and multivariate logistic regression were used to model PCSM and no need for treatment, respectively.
Results: Table 1 describes the patient Demographics between the Observation and Treatment groups. Significant univariate differences include preoperative PSA, pathologic stage (p-stage), pathological Gleason Grade Group (pGGG), and DT change. The median follow-up was 7.3 years (IQR 3.9-10.7). 10-year Kaplan-Meier analysis demonstrated no PCSM in the observation group compared to 7.1% PCSM in the treated group (p=0.001).In adjusted logistic regression analysis, PSAdt > 12 months and increasing DT were significant predictors for continued observation without treatment (p<0.001), while pGGG was p=0.06. In ROC analysis, the model including these three variables were an excellent predictor of no need for treatment (AUC = 0.84), similar to the model including only PSAdt and pGGG (AUC = 0.82). In adjusted logistic regression analysis in patients with PSAdt < 12 months, PSAdt group 6-12 months and increasing DT were significant predictors of no treatment. Of interest, pGGG was not predictive of treatment.
Conclusions and Relevance: There exists a group of patients (137/407, 33%) who can be observed without secondary treatment following BCR after RP, with 0% PCSM. No patients with PSAdt > 12 months experienced PCSM. PSAdt > 12 months, increasing DT, and a low pGGG (1-2) were excellent predictors of no need for treatment. Further studies of DT kinetics in cohorts of patients with PSAdt < 12 months in subset analysis may help direct more tailored processes for timing and intensity of treatment.