To find out the impact of lymph node dissection (LND) on the prognosis of patients undergoing cystectomy for non-muscle-invasive bladder cancer (NMIBC) and determine the minimum number of lymph nodes (LNs) for resection.

Patients who had NMIBC and underwent cystectomy between 2004 and 2015 were retrospectively included from the Surveillance, Epidemiology, and End Results (SEER) database. LND was confirmed by the existence of LNs under postoperative pathological examination. Restricted cubic spline (RCS) analysis was utilized to determine the ideal threshold for minimal number of lymph node examined (nLNE). Survival analysis was conducted using Kaplan–Meier (KM) curves, log-rank tests and forest plots. Propensity score matching (PSM) was utilized to balance baseline characteristics.

Of 1950 patients with NMIBC, 1,353 underwent LND, of which 58(4.2%) exhibited positive LNs. LND were associated with better prognosis in the whole cohort (P<0.001). Patients had positive lymph nodes were found to have the worst prognosis (P<0.001). RCS found patients with at least 8 LNs examined show longer survival than those with nLNE<8, which was further validated in multivariable cox regression [HR=0.66(0.59, 0.76)]. Subgroup analysis indicated that nLNE≥8 was associated with significantly improved overall survival in T1 and Tis patients both before and after PSM. While for Ta patients, the impact of LND or nLNE≥8 on prognosis was insignificant. Additionally, for patients with negative lymph node, nLNE≥8 still resulted in better prognosis [HR=0.73(0.61, 0.87)]. Furthermore, sensitivity analyses replicated the same results in patients more than 80 years old and found patients with nLNE≥8 demonstrated superior prognosis than those with less than 8 lymph node retrieved.

Among patients with NMIBC, it is crucial to remove 8 or more lymph node effectively during cystectomy, except in Ta subgroup. Fewer than 8 LNs retrieved should be regarded as an independent risk factor.