Survival outcomes of patients treated with local therapy for nonmetastatic prostate cancer with high prostate-specific antigen concentrations.

BACKGROUND

Patients with nonmetastatic prostate cancer (nmPCa) and high prostate-specific antigen (PSA) levels due to the high likelihood of metastasis pose a clinical dilemma regarding their optimal treatment and long-term outcomes after initial local therapy. We aimed to evaluate the oncologic outcomes of patients treated with radical prostatectomy (RP) or radiotherapy (RT) for nmPCa with high PSA levels.

METHODS

We queried the Surveillance, Epidemiology, and End Results (SEER) database to identify patients diagnosed with nmPCa who received RP or RT from 2004 through 2015. We included nmPCa patients with high PSA levels categorized as ≥50 and ≥98 ng/mL, the highest level recorded in SEER. We used the Kaplan-Meier method and Cox proportional hazards to analyze cancer-specific (CSS) and overall survival (OS).

RESULTS

We included 6177 patients with nmPCa and PSA ≥ 50 ng/mL at diagnosis; 1698 (27%) had PSA ≥ 98 ng/mL. Of these, 1658 (26.8%) underwent RP and 4519 (73.16%) patients received primary RT. Within a median of 113 months (interquartile range 74-150 months), the 5- and 10-year CSS estimates were 92.3% and 81.5% respectively; 10-year OS was 61%. In the PSA ≥ 98 ng/mL subgroup 5- and 10-year CSS estimates were 89.2% and 76%, respectively. In multivariable analyses for CSS, ISUP grade group (p < 0.001), N stage (p < 0.001), treatment with RP (hazard ratio [HR] = 0.60, 95% confidence interval [CI] 0.43-0.83, p < 0.001), and patient's age (p < 0.05) were associated with improved CSS. In the whole cohort of patients with PSA ≥ 50 ng/mL and RP subgroup, PSA failed to retain its independent prognostic value for CSS.

CONCLUSIONS

Patients treated with local therapy for nmPCa with very high PSA at diagnosis have relatively good long-term oncological outcomes. Therefore, among well-selected patients with nmPCa, high PSA levels alone should not preclude the use of radical local therapy. Potential selection bias limits inferences about the relative effectiveness of specific local therapies in this setting.