Impact of the primary tumor location on secondary sites and overall mortality in patients with metastatic upper tract urothelial carcinoma

Background: To date it is unknown whether renal vs. ureteral urothelial carcinoma affects the type and the distribution of metastatic sites, and whether survival differs according to renal vs. ureteral location in metastatic patients. Methods: Two datasets were used, namely Surveillance, Epidemiology and End Results (SEER) and National Inpatients Sample (NIS). Multivariable logistic regression models tested whether renal pelvis vs. ureteral location predicts site-specific metastases. Kaplan-Meier plots and multivariable Cox regression models (CRMs) tested overall mortality (OM) according to renal pelvis vs. ureteral location. Results: In SEER (2010-2016), 623 (71.1%) metastatic renal pelvis urothelial carcinoma (RPUC) vs. 253 (28.9%) ureteral urothelial carcinoma (UUC) patients were identified. Patients with RPUC more frequently harbored lung (46.1% vs. 35.2%, P < 0.01; Odds ratio [OR]: 1.57, P < 0.01), but less frequently liver metastases (27.9% vs. 36.4%, P = 0.02; OR:0.66, P = 0.01). In RPUC, lung, liver, bone, and brain metastases independently predicted higher OM. Only liver metastases independently predicted higher OM in UUC. In NIS (2005 -2015), 818 (61.0%) RPUC vs. 522 (39.0%) UUC patients were identified. Patients with RPUC more frequently harbored lung (34.0% vs. 17.2%, P < 0.001; OR:2.36, P < 0.001), as well as brain (4.4% vs. 1.9%, P = 0.02; OR:2.00, P = 0.049) metastases, but less frequently har -bored retroperitoneal and/or peritoneal (12.3% vs. 21.8%, P < 0.001; OR:0.51, P < 0.001), urinary tract (9.3% vs. 14.0%, P = 0.01; OR:0.65, P = 0.01) and multiple metastatic sites (62.6% vs. 70.7%, P < 0.01; OR:0.69, P < 0.01). Conclusions: In both databases lung metastases were more frequent in RPUC and abdominal metastases were more frequent in UUC. Moreover, liver metastases independently predicted worse survival, regardless of primary site.