Preoperative favourable characteristics in bladder cancer patients cannot substitute the necessity of extended lymphadenectomy during radical cystectomy: A sensitivity curve and a survival analysis

INTRODUCTION &amp; OBJECTIVES: Pelvic Lymphadenectomy (PLND) in pat ients treated with Radical Cystectomy (RC) for Bladder Cancer (BCa) represents an accuratestaging procedure and it may have a therapeut ic role. On the other hand PLND is associated with increased perioperative morbidity and operating t ime. In communityset tings, it has been reported that up to 25% of patients submitted to RC due to BCa did not receive any PLND. Considering only patients submit ted to RC due to nonmuscleinvasive BCa, this rate ranges from 35% to 50%. No evidence exists about the ideal extension of PLND on the basis of pre- or int ra-operat ive parameters. Weinvestigated this hypothesis, in a large contemporary cohort t reated in a single center.MATERIAL &amp; METHODS: Between 1995 and 2012, 1,016 RC and extended PLND due to BCa were performed at a single tertiary care institut ion. The relat ionshipbetween the number of nodes removed (RLNs) and the probability to find nodal metastases (LNM) at final pathology was assessed using Receiver Operat ing Characterist ic(ROC) analyses, in the overall population and according to pathological stage and different clinical and pathological scenarios. Multi-variable Cox regression analyses tested therelationship between RLNs and CSM rate in overall population and according pathological T stage (pT0-p1 vs pT2 vs. pT3-pT4). Covariates consist of age, gender,pathological T and N stage, grading, surgical margin, CIS, lymph vascular invasion, peri-op chemotherapy and number of LNM. Survival curves were strat ified according to thenumber of RLNs, using the points of maximum separat ion, as described by Harrell.RESULTS: The median number of nodes removed were 18 and the LNM prevalence was 35.7% (363 of 1,016). According to pathological stage, LNM were recorded in7%, 20%, 56%, of pT0/pT1, pT2 and pT3/pT4 (p<0.001), respect ively. The ROC curve indicated that 25, 35 and 45 nodes need to be removed to achieve 75, 90% and95% probability respectively of detect ing one or more LNM. When the analyses were strat ified according to pre-op characteristics, no significant differences were recordedfor NMIBC vs. MIBC at last TUR, first episode or recurrent/progressive tumour, radiological N or T stage (all p>0.2). At multi-variable analyses, RLNs (HR 0.97, p<0.001)exerted a protective role on CSM, in the overall populat ion subjected to RC. The number of RLNs needed to improve CSM were 17 (HR: 0.71), 14 (HR: 0.63) and 10 (HR:0.56) in overall, pT2 and pT3-4 patients, respect ively (all p<0.001). No survival differences were recorded in pT0-pT1 pat ients (all p>0.4), based on the number of LRNs.CONCLUSIONS: Our results show that is necessary to perform an extended PLND in order to offer an accurate staging in patients t reated with RC for BCa, if a curat iveintent is intended. However no survival benefits were recorded in extending PLND in pat ients with pT0-pT1 at RC.