Concurrent prognostic utility of lymph node count and lymph node density for men with pathological node-positive prostate cancer.

Abstract

While both the number (+LN) and density (LND) of metastatic lymph nodes on radical prostatectomy lymphadenectomy predict mortality in prostate cancer, the independent impact of each on overall mortality (OM) is unknown. We sampled men who underwent radical prostatectomy and lymphadenectomy between 2004 and 2013 from the National Cancer Database. Multivariable Cox proportional hazards analysis with restricted cubic spline was used to assess the non-linear association of +LN count and LND with OM. Of 229,547 men in our sample, 3% (n = 7507) had +LNs, of which 89% had 1-3 +LN and 11% had ≥4 +LN. In multivariable Cox analysis across all patients, OM increased with each additional +LN up to four (HR 1.14, 95%CI 1.06-1.23 per node), with no increase beyond 4 +LN. LND was an independent predictor of OM (HR 1.09, 95%CI 1.06-1.12 per 10% increase). However, after excluding patients with inadequate nodal sampling (<5 LN examined), the variation in OM explained by LND was negligible for patients with ≤3 +LN. In men with 1, 2, and 3 +LN, there was a 0.28%, 0.02%, and 0.50% increase in OM for each 10% increase in LND, compared with 1.9% and 1.6% for men with 4 or 5+ LNs. While +LN count and LND independently predict OM, the impact of LND is negligible in men with ≤3 +LN, who comprise the vast majority of men with +LN. Pathological nodal staging should primarily rely on LN count rather than LND.