Prognostic Stratification of pN1 Prostate Cancer After Radical Prostatectomy: A Competing Risk Analysis from a Multi-institutional Cohort.

BACKGROUND AND

OBJECTIVE

Lymph node-positive (pN1) prostate cancer (PCa) is a heterogeneous disease, and a clear definition of prognostic groups is urgently needed. We aimed to assess cancer-related mortality (CRM) in different prognostic groups of pN1 patients, created based on the pathological PCa characteristics and number of positive lymph nodes (LN+).

METHODS

We conducted a retrospective, multicentre cohort study including 894 patients with pN1 disease treated at 15 European high-volume centres. Independent predictors for CRM were identified and pooled. A prognostic model was constructed for the prediction of CRM, accounting for death from other causes as a competing risk. The 10-yr cumulative risk of mortality was assessed.

KEY FINDINGS AND LIMITATIONS: Our model was based on pT stage (pT2-3a vs pT3b-4), surgical margin (SM) status (positive vs negative), and number of LN+ (1-4 vs >4), and included three prognostic groups. The favourable-prognosis group includes patients with pT2-3a and one to four LN+, or pT3b-4 with negative SM status. The intermediate-prognosis group included patients with pT2-3a disease and more than four LN+, or pT3b-4 disease, one to four LN+, and positive SM status. Patients in the poor-prognosis group had all three high-risk factors present. The C-index of this model was 0.73. The 10-yr cumulative CRM rates were 12% (95% confidence interval: 7.3-16%), 32% (24-40%), and 58% (40-76%), respectively, with significant differences between groups (hazard ratio 2.2-6.4, p < 0.005).

CONCLUSIONS AND CLINICAL IMPLICATIONS: The pN1 patient population is extremely heterogeneous, with an increased risk of death from PCa rather than death from other causes. In this group of patients, primary cancer characteristics (pT stage, number of LN+, and SM status) still represent the driving factors of CRM.

PATIENT SUMMARY: Men with positive lymph nodes on pathology have an increased risk of dying from prostate cancer, rather than from other causes. Our proposed model stratifies patients into groups with different cancer-related prognosis and may aid in personalised clinical decision-making in a postoperative setting.