Cisplatin is the only chemotherapeutic agent shown in randomized trials to improve survival in any urothelial cancer disease state; however it carries significant risk for toxicity particularly in the setting of renal impairment. Adjuvant cisplatin-based chemotherapy, in contrast to neoadjuvant, overcomes the inaccuracies of clinical staging [36], and allows for surgical pathology-driven selection of patients at highest risk of recurrence and therefore highest probability of deriving benefit from treatment. However, study of this agent in patients with upper tract urothelial cancer is fraught with challenges in large part due to the high rate of impaired renal function both before and in particular after nephroureterectomy [37]. Combined with the rarity of the disease, it is not surprising that there are no prospective data evaluating cisplatin-based chemotherapy in the adjuvant setting. Limited retrospective data and extrapolation from adjuvant chemotherapy trials in bladder cancer provide some justification for its use in otherwise fit patients at high risk for recurrence.

Two retrospective experiences with cisplatin-based chemotherapy in the adjuvant setting provide conflicting data. Hellenthal et al. reported on 542 patients with pT3-4 N0/+ upper tract urothelial cancer of whom 121 received adjuvant chemotherapy (89 % receiving cisplatin-based chemotherapy) [38]. Cancer-specific survival (CSS) and OS were 45 % and 38 % at 5 years, respectively, for both groups, suggesting no benefit to adjuvant therapy; however baseline factors for both groups were not well balanced. A similar, yet smaller, retrospective experience reported by Kwak et al. included 39 patients with pT2-3 N0/+of whom 32 received adjuvant cisplatin-based chemotherapy [39]. Disease-free survival (DFS) and OS were substantially higher at 5 years for the group receiving chemotherapy (63 % vs. 36 % DFS and 78 % vs. 36 % OS); however the interpretation of the results is limited by the small sample size.

In bladder cancer, similar confusion also exists in the study of adjuvant cisplatin-based chemotherapy. Several studies have attempted to evaluate adjuvant therapy in bladder cancer; however most have suffered from poor accrual and noncompliance with study treatment and have ultimately closed early [40]. Further, a meta-analysis of six trials provides limited evidence for its benefit [41]. However, Spanish Oncology Group Trial 99/01, a randomized trial of adjuvant chemotherapy in patients with invasive urothelial cancer at high risk for recurrence, did provide some evidence that adjuvant chemotherapy can perhaps improve survival in selected patients. With a target enrollment of 340 patients, this trial randomized high-risk patients after radical cystectomy with pT3-T4 or lymph node positive urothelial cancer of the bladder to observation versus adjuvant paclitaxel, gemcitabine, and cisplatin (PGC) for four cycles. Although it was closed early due to poor accrual in 2007, a preliminary analysis at a median follow-up of 51 months for the 142 patients enrolled demonstrated a significant improvement in overall survival at 5 years with adjuvant PGC (60 % vs. 30 %, HR 0.44, P < 0.0009) [42]. This trial demonstrated a survival benefit with adjuvant chemotherapy despite accruing less than half the target enrollment likely because the trial selected for patients with the highest risk for recurrence, and therefore the highest likelihood of deriving benefit from adjuvant chemotherapy. Small sample size and incomplete accrual limit interpretation of this trial; however it does provide a basis for considering adjuvant chemotherapy in selected high-risk patients. In the absence of prospective data in upper tract urothelial cancer to suggest the contrary, there is no evidence to suggest similar patients with upper tract disease should be treated differently. Therefore, patients with pT3 or greater or lymph node positive upper tract urothelial cancer who are cisplatin eligible should be considered for cisplatin-based adjuvant chemotherapy.

Alternatively, carboplatin-based therapy is a less toxic approach and safer in the setting of renal impairment and has been investigated prospectively. Bamias et al. reported a prospective trial of adjuvant paclitaxel and carboplatin in 36 patients with pT2-4 N0/+upper tract urothelial cancer that demonstrated a 40 % DFS and 52 % OS at 5 years [43]. Thirty-four of these patients had ≥pT3 disease and outcomes reported in this study compare favorably to the poor long-term survival (40 % 5-year survival for pT3 and 6 months median survival for pT4 [44]) typically observed for these patients with surgery alone.

Regardless of the primary site of origin, metastatic urothelial cancer is a chemotherapy-sensitive disease. Early clinical trials of single-agent chemotherapy identified doxorubicin, 5-fluorouracil, vinblastine, vincristine, and mitomycin C as active agents with response rates of approximately 15 %. Cisplatin and methotrexate were associated with the highest response rates of approximately 30 %, and thus served as the backbone for the development of older combination strategies such as MVAC, CMV, and CISCA (cisplatin, cyclophosphamide, and adriamycin) and modern regimens that have included ifosfamide [45, 46]. Outcomes reported with cisplatin-based chemotherapy, the only regimens associated with a survival benefit in randomized trials, include response rates of approximately 50–60 % and median survival between 13 and 15 months [10, 46–48]. MVAC represented the standard of care in metastatic urothelial cancer until a seminal phase III trial compared MVAC to the novel regimen of gemcitabine and cisplatin (GC) in 405 patients with metastatic urothelial cancer and found similar efficacy, but improved tolerability with GC, establishing this regimen as a new standard of care for this disease [10].

However, an increased proportion of patients with metastatic urothelial cancer who have previously undergone nephroureterectomy for upper tract disease, as compared to radical cystectomy for bladder cancer, will be ineligible for cisplatin-based therapy due to impaired renal function [37, 49, 50]. Typically, these patients have been treated with carboplatin-based regimens. The EORTC 30986 phase II/III trial compared gemcitabine and carboplatin (GCa) versus methotrexate, carboplatin, and vinblastine (MCAVI) in 238 patients with metastatic urothelial cancer who were ineligible for cisplatin (poor performance status, impaired renal function, or both) and demonstrated a 41.2 % response rate and 9.3 months median survival for GCa. Similar response and survival outcomes were observed for MCAVI; however it was less well tolerated, and thus this trial provided the first level one evidence for a non-cisplatin-based regimen in this patient population. More recently the combination of gemcitabine and paclitaxel with doxorubicin has shown promise in this patient population. In a trial in patients with poor renal function, objective responses rates were approximately 50 %, with a median survival of 15 months [51].

Prospective randomized trials of chemotherapy in metastatic urothelial cancer are not typically stratified by primary site of disease and there are no definitive data to suggest outcomes are different between upper and lower tract urothelial cancer. A retrospective report of 72 patients with upper tract urothelial cancer who underwent nephroureterectomy suggested tumor location (renal pelvis versus ureter) independently impacted on 5-year survival outcomes [8]. More recent evidence includes a post hoc subset analysis from a randomized phase III trial of GC versus paclitaxel, gemcitabine, and cisplatin (PCG) in 626 patients with metastatic urothelial cancer and provides some rationale for separate investigation of tumors that originate from the upper versus lower urinary tract. For the entire study population, median survival for GC and PCG was 12.7 and 15.8 months, respectively, which did not reach statistical significance (HR 0.85; 95 % CI, 0.72–1.02; P = 0.075); however when the analysis was limited to the 81 % of patients who had bladder cancer as their primary, median survival with PCG was significantly longer than with GC (15.9 versus 11.9 months, respectively; HR, 0.80; 95 % CI, 0.66–0.97; P = 0.025). While not definitive, this observation in an unplanned subset analysis could serve as a basis for stratification by primary tumor site in future randomized trials of chemotherapy in metastatic urothelial cancer. Until such evidence is available, metastatic urothelial cancer should be treated similarly irrespective of primary tumor site.