Development and Validation of an Improved Pathological Nodal Staging System in Men With Prostate Cancer.

Background Prostate cancer pathological nodal staging uses a single category for all node-positive patients. We sought to improve risk stratification by creating and validating a novel pathological nodal staging system incorporating number of metastatic lymph nodes (+LNs). Methods 118,450 men who underwent radical prostatectomy (RP) for nonmetastatic prostate cancer in the National Cancer Database comprised our development cohort. Multivariable Cox proportional hazards analysis with restricted cubic splines was used to assess the non-linear association between number of +LNs and overall mortality (OM). A novel staging system based on number of +LNs was derived by recursive partitioning analysis (RPA). The staging system was validated for prediction of overall and prostate-specific mortality (PCSM) in 105,568 men with nonmetastatic prostate cancer undergoing RP from the Surveillance, Epidemiology, and End Results database. Discrimination was assessed via Harrell's c-index. Results In multivariable Cox analysis, OM risk increased with higher number of +LNs up to four (HR1.30 per each LN+, 95%CI 1.23-1.38), with a non-statistically significant increase in risk (HR 1.05, 95%CI 0.99-1.11) beyond four +LN. In the development cohort, RPA identified optimal cutoffs at 0 (N0:Ref), 1 (N1:HR1.40, 95%CI 1.25-1.58), 2 (N2:HR1.67, 95%CI 1.40-1.99), 3-5 (N3a: HR2.18, 95%CI 0.84-2.60), and ≥6 (N3b:HR3.00, 95%CI 2.37-3.79) +LNs. In the validation cohort, these groups had markedly different 10-year OM (0+LNs (N0:15%), 1+LN (N1:35%), 2+LNs (N2:43%), 3-5+LNs (N3a:52%), and ≥6+LNs(N3b:59%) (p Conclusions Pathological nodal staging in prostate cancer is improved with stratification by number of +LNs.