Long-term outcomes of salvage radical cystectomy for recurrent urothelial carcinoma of the bladder following partial cystectomy


Siamak Daneshmand, Department of Urology, 1441 Eastlake Avenue, Suite 7416, Los Angeles, CA 90089, USA. e-mail: daneshma@usc.edu


Study Type – Therapy (outcomes)

Level of Evidence 2b

What's known on the subject? and What does the study add?

Data on the oncological outcomes in patients undergoing salvage cystectomy for recurrent disease following bladder-sparing treatment is limited and mostly based on case reports.

We present the clinical outcomes and prognostic factors in patients undergoing radical cystectomy for recurrent disease following partial cystectomy with long-term follow-up.


  • • To report the clinical outcomes and prognostic factors in patients undergoing salvage radical cystectomy (sRC) for recurrent urothelial carcinoma (UC) of the bladder following partial cystectomy (PC).


  • • Between 1971 and 2011, a total of 2290 patients underwent radical cystectomy for UC of the bladder, including 72 patients (3.1%) who underwent sRC following PC.
  • • Clinical and pathological data at the time of both PC and sRC were collected.
  • • Median follow-up time after sRC was 10.9 years. Overall survival and recurrence-free survival were the primary outcomes of interest.
  • • Univariate and multivariate analyses were performed to identify prognostic factors after sRC.


  • • The median time from PC to sRC was 1.6 years. Median age at sRC was 64 years. Peri-operative mortality was 2.8%.
  • • After sRC, 44 patients (61.2%) had pathologically organ-confined disease, 14 patients (19.4%) extravesical disease and 14 patients (19.4%) lymph node positive disease.
  • • Five-year recurrence-free survival and overall survival following sRC were 56% and 41%, respectively.
  • • On multivariate analysis, the presence of pathological tumor stage ≥pT3a (hazard ratio 6.86, P < 0.001) and the presence of lymph node metastases (hazard ratio 8.78, P < 0.001) were associated with increased risk of recurrence after sRC.


  • • sRC can provide prolonged survival following failure of PC.
  • • Prognosis, however, is highly dependent on pathological tumour stage and nodal status at sRC.
  • • Only 15% of patients with locally advanced recurrent disease were salvaged by sRC.

radical cystectomy


urothelial carcinoma


partial cystectomy


muscle-invasive bladder cancer


salvage radical cystectomy






lymph node positive


lymphovascular invasion


overall survival


recurrence-free survival


Radical cystectomy (RC) and lymph node dissection is considered the gold standard treatment for muscle-invasive urothelial carcinoma (UC) of the bladder. RC, however, carries significant peri-operative morbidity. Bladder-sparing treatments, such as partial cystectomy (PC), are occasionally used in order to reduce peri-operative morbidity and improve functional outcomes. Unfortunately, initial series from the 1960s demonstrated that PC compromised oncological outcomes with recurrence rates up to 50–75% [1–3]. As a result, PC was abandoned as the standard treatment for muscle-invasive bladder cancer (MIBC). Recent series, however, have demonstrated that in highly selected patients similar survival outcomes to RC can be achieved [4,5]. Using strict selection criteria, in addition to improvement of radiotherapy and systemic chemotherapy regimens, interest in bladder-sparing surgery for MIBC has emerged again. However, a Canadian population-based study reported that almost 25% of patients required salvage radical cystectomy (sRC) for recurrent disease in the preserved bladder after partial cystectomy [6]. Similarly, in the largest reported series of definitive chemoradiation for MIBC, 29% of patients eventually required sRC [7]. Thus, the risk of cancer in the preserved bladder that ultimately requires sRC remains an important concern.

Whilst a higher postoperative complication rate in sRC has been reported compared with patients undergoing immediate RC [7], the prognosis following sRC has not been previously investigated except for a few case reports. Knowledge of the success rate of salvage therapy is important and may even alter the selection criteria for PC. In this study we report the clinical outcomes and prognostic factors of patients who underwent sRC with intent to cure for recurrent UC after PC.



A total of 2290 patients underwent RC for UC of the bladder between November 1971 and April 2011. Clinicopathological characteristics of all subjects were collected using standardized retrospective medical record review and were stored in our Institutional Review Board-approved bladder cancer database. For the current study, we included all patients who underwent sRC for recurrent UC after PC. A total of 93 patients were identified. Next, the following exclusion criteria were applied: (i) tumour originating from the upper urinary tract or urethra; (ii) gross disease remaining after PC; (iii) patients who underwent palliative sRC. A total of 72 patients (3.1%) met the inclusion criteria and were the focus of this study.


All patients underwent a standard surgical procedure of RC and were examined using the same pathological protocol, as previously described [8]. Pathological tumour staging and grading were standardized to the 2002 TNM classification of the American Joint Committee on Cancer and 1973 WHO recommendations, respectively [9,10]. Pathological subgroups at RC were divided into organ-confined (OC; pTa–pT2b, N0), extravesical (EV; pT3a–pT4b, N0) and lymph node positive (LN+; any pT, N+) disease. Lymph node density was defined as the ratio of positive lymph nodes to total removed lymph nodes [11]. Lymphovascular invasion (LVI) in the RC specimen was defined as vascular and/or lymphatic invasion by tumour.


Neoadjuvant treatment to PC was defined as non-surgical treatment prior to PC. Adjuvant therapy was defined as non-surgical treatment following surgery, not related to recurrent disease, and was reported for both PC and sRC. Neoadjuvant treatment to sRC was defined as non-surgical treatment prior to sRC but following PC. The indication for sRC was based on the surgeon's discretion. Whilst sRC was not systematically offered to all patients with non-muscle-invasive recurrence, patients with a high risk profile or failed conservative treatment were considered for sRC. This is a similar approach to that for patients who primarily present with non-muscle-invasive bladder cancer at our institution. Patients were routinely followed postoperatively at 4-month intervals in the first year, 6-month intervals in the second year and annually thereafter unless otherwise clinically indicated. The median follow-up time following sRC was 10.9 years (range 5 days to 18.8 years). Overall survival (OS) and recurrence-free survival (RFS) were the primary outcomes of interest. Time to OS was calculated from the date of sRC to the date of death (from all causes). RFS was calculated from the date of sRC to the date of first documented clinical recurrence. If lost to follow-up, subjects were censored from the analysis at the date of last follow-up. Early recurrence was defined as recurrence within 12 months following PC. Time from PC to sRC was calculated using both dates of surgery. Peri-operative mortality was defined as any death that occurred within 30 days after sRC.


Kaplan–Meier plots were used to estimate the probabilities of OS and RFS over time since sRC. Log-rank tests were used to compare the differences in survival or recurrence in subgroups. The Cox proportional hazard model was used to test the association of clinical and pathological variables with survival or recurrence. All P values reported in the analyses are two-sided. Statistical analysis was performed using SAS version 9.2.


At the time of PC, 48 patients (66.6%) had pathological OC disease, 11 patients (15.3%) pathological EV disease, two patients (2.8%) pathological node positive disease and for 11 patients (15.3%) no data were available. Four patients (5.6%) had positive surgical margins at PC. All clinical and histopathological features at the time of PC are presented in Table 1.

Table 1. Clinical and pathological characteristics at the time of partial cystectomy
VariableNumber of patients (percentage) (total N= 72)
Pathological tumour stage 
 T06 (8.3%)
 Tis/Ta14 (19.5%)
 T18 (11.1%)
 T2a14 (19.5%)
 T2b7 (9.7%)
 T3a6 (8.3%)
 T3b5 (6.9%)
 T4a1 (1.4%)
 Unknown11 (15.3%)
Tumour grade 
 High50 (69.5%)
 Low8 (11.1%)
 Not applicable – no tumour6 (8.3%)
 Unknown8 (11.1%)
Pathological subgroup 
 Organ-confined disease48 (66.6%)
 Extravesical disease11 (15.3%)
 Lymph node positive disease2 (2.8%)
 Unknown11 (15.3%)
Surgical margin status 
 Positive4 (5.6%)
 Negative4 (5.6%)
 Unknown64 (88.8%)
Neoadjuvant therapy (N= 11) 
 Radiotherapy9 (12.5%)
 Chemotherapy2 (2.8%)
Adjuvant therapy (N= 12) 
 Radiotherapy7 (9.7%)
 Chemotherapy5 (6.9%)

The median time from PC to sRC was 1.6 years (range 24 days to 18.8 years). One patient underwent sRC 24 days after PC due to positive surgical margins at PC. The median age at the time of sRC was 64 years (range 27–83 years). Peri-operative mortality was 2.8% (N= 2). After sRC, 44 patients (61.2%) had pathological OC disease, 14 patients (19.4%) pathological EV disease and 14 patients (19.4%) pathological LN+ disease. Lymphadenectomy at the time of sRC was performed in 58 patients (80.6%). The median number of removed lymph nodes was 21. Of the four patients (5.6%) who received neoadjuvant chemotherapy prior to sRC, two patients had abdominal wall involvement, one patient positive surgical margins and one patient LN+ disease. Adjuvant chemotherapy was offered to 12 patients (16.7%) of whom six patients had LN+ disease, four patients EV disease, one patient positive surgical margins and one patient was treated as part of a trial. As expected, patients with EV and/or LN+ disease were more often exposed to adjuvant therapy compared with patients with OC disease (42.8% vs 2.2%). All clinical and histopathological features at the time of sRC are summarized in Table 2.

Table 2. Clinical and pathological characteristics at the time of salvage radical cystectomy
VariableNumber of patients (percentage) (total N= 72)
 >65 years31 (43.1%)
 ≤65 years41 (56.9%)
 Male56 (77.8%)
 Female16 (22.2%)
Type of urinary diversion 
 Orthotopic18 (25.0%)
 Non-orthotopic54 (75.0%)
Clinical tumour stage 
 Missing4 (5.6%)
 ≤cT133 (45.8%)
 cT226 (36.1%)
 cT33 (4.2%)
 cT46 (8.3%)
Pathological tumour stage 
 pT08 (11.1%)
 pTa/Tis14 (19.4%)
 pT115 (20.8%)
 pT2a6 (8.3%)
 pT2b4 (5.6%)
 pT3a3 (4.2%)
 pT3b8 (11.1%)
 pT4a10 (13.9%)
 pT4b4 (5.6%)
Pathological subgroup 
 Organ-confined disease44 (61.2%)
 Extravesical disease14 (19.4%)
 Lymph node positive disease14 (19.4%)
Tumour grade 
 Low grade18 (25.0%)
 High grade54 (75.0%)
Lymphovascular invasion 
 Present15 (20.8%)
 Absent57 (79.2%)
Concomitant carcinoma in situ 
 Yes36 (50.0%)
 No36 (50.0%)
Multifocal disease 
 Yes19 (26.4%)
 No53 (73.6%)
Soft tissue positive margins 
 Present3 (4.2%)
 Absent69 (95.8%)
Lymph node density 
 ≤20%67 (93.1%)
 >20%5 (6.9%)
Neoadjuvant therapy (N= 17) 
 Radiotherapy13 (18.1%)
 Chemotherapy4 (5.6%)
Adjuvant therapy (N= 13) 
 Radiotherapy1 (1.4%)
 Chemotherapy12 (16.7%)
Time to recurrence 
 Early (≤12 months after PC)24 (33.3%)
 Late (>12 months after PC48 (66.7%)
Type of lymphadenectomy 
 Not performed14 (19.4%)
 Standard23 (32.0%)
 Extended (≥aortic bifurcation)35 (48.6%)

Five- and 10-year OS after sRC was 41% and 35% respectively. RFS after 5 and 10 years was 56% and 51%, respectively. All but one of the recurrences following sRC occurred within 5 years. When stratified by pathological subgroups, RFS and OS were significantly higher in patients with OC disease compared with patients with EV disease and patients with LN+ disease (both, P < 0.001). Between the subgroups EV and LN+ disease, however, no significant difference in RFS and OS was found (P= 0.52 and P= 0.62, respectively). As the numbers of patients in the separate subgroups of EV and LN+ were low (both N= 14) and survival did not significantly differ, these subgroups were merged in the Kaplan–Meier analysis (Figs 1 and 2).

Figure 1.

Recurrence-free survival for organ-confined disease (≤pT2, LN−) vs extravesical/lymph node positive disease (>pT2 and/or pLN+).

Figure 2.

Overall survival for organ-confined disease (≤pT2, LN−) vs extravesical/lymph node positive disease (>pT2 and/or pLN+).

To compare the outcomes of sRC to the baseline group of patients undergoing immediate RC for MIBC, we calculated survival data for the remaining 2218 patients in our cohort. Overall, 5-year RFS in patients undergoing sRC was 56% compared with 66% in patients undergoing immediate RC (log-rank; P= 0.50). In patients with OC disease, the 5-year RFS was 76% in sRC patients compared with 82% in immediate RC patients (log-rank; P= 0.33). Lastly, in the subgroup of pathological EV or LN+ patients (N= 28), 5-year RFS was 22% in the sRC group and 36% in the immediate RC group (log-rank; P= 0.027). After 10 years, 15% of patients in this EV/LN+ group were cured by sRC.

On univariate analysis, EV and LN+ disease (P < 0.001), LVI (P= 0.03), lymph node density > 20% (P < 0.001) and positive surgical margins (P < 0.001) were significantly associated with decreased RFS. Age (P= 0.91), gender (P= 0.37), focality of disease (P= 0.10), time to recurrence (P= 0.52), concomitant carcinoma in situ (P= 0.20) and use of adjuvant chemotherapy (P= 0.06) were not significantly associated with RFS on univariate analysis (Table 3). Due to the limited size of the cohort, a limited number of variables were included in the multivariate analysis (Table 4). Presence of EV disease (hazard ratio 6.86 [2.69–17.49], P < 0.001) and presence of LN+ disease (hazard ratio 8.78 [2.81–27.39], P < 0.001) were independently associated with recurrence after sRC.

Table 3. Recurrence-free and overall survival after salvage radical cystectomy
 No. of patientsProbability of not recurring ±se (%)Probability of overall survival ±se (%)
5 years10 years P 5 years10 years P
Age   0.91  0.029
 ≤65 years4154.4 ± 8.351.4 ± 8.348.7 ± 7.840.8 ± 7.8
 >65 years3158.6 ± 10.252.1 ± 10.931.1 ± 8.627.3 ± 8.4
Gender   0.37  0.07
 Male5658.3 ± 7.253.4 ± 7.446.3 ± 6.738.6 ± 6.7
 Female1646.9 ± 14.146.9 ± 14.125.0 ± 10.825.0 ± 10.8
Focality of disease   0.10  0.54
 Unifocal5360.7 ± 7.360.7 ± 7.343.1 ± 6.936.5 ± 6.9
 Multifocal1942.6 ± 12.828.4 ± 11.836.8 ± 11.131.6 ± 10.7
Time to recurrence   0.52  0.08
 Early (≤12 months)2463.4 ± 10.353.7 ± 10.858.3 ± 10.049.7 ± 10.3
 Late (>12 months)4851.0 ± 10.351.0 ± 8.332.4 ± 6.927.6 ± 6.7
Pathological subgroup   <0.001  <0.001
 Organ-confined disease4475.7 ± 7.172.4 ± 7.559.6 ± 7.649.2 ± 7.9
 Extravesical disease1418.9 ± 11.97.1 ± 6.8
21.4 ± 11.021.4 ± 11.0
 Lymph node positive disease1427.7 ± 13.127.7 ± 13.1
Lymphovascular invasion   0.03  0.023
 Absent5763.7 ± 7.058.6 ± 7.248.2 ± 6.740.4 ± 6.6
 Present1522.8 ± 13.122.8 ± 13.114.3 ± 9.414.3 ± 9.4
Lymph node density   <0.001  <0.001
 ≤20%6759.3 ± 6.754.9 ± 6.844.6 ± 6.237.9 ± 6.1
Carcinoma in situ   0.20  0.50
 Absent3647.2 ± 9.247.2 ± 9.234.3 ± 8.031.4 ± 7.8
 Present3663.0 ± 9.154.6 ± 9.648.3 ± 8.538.7 ± 8.4
Surgical margin status   <0.001  0.006
 Negative6958.3 ± 6.653.9 ± 6.843.2 ± 6.136.8 ± 6.0
Adjuvant chemotherapy   0.06  0.44
 No6060.1 ± 7.055.1 ± 7.344.8 ± 6.537.4 ± 6.4
 Yes1231.1 ± 14.831.1 ± 14.825.0 ± 12.525.0 ± 12.5
Table 4. Multivariable Cox regression analysis for recurrence-free and overall survival in patients undergoing salvage radical cystectomy
VariableNumber of patientsRecurrence-free survivalOverall survival
HR (95% CI) P HR (95% CI) P
  1. HR, hazard ratio.

Patient age  0.59 0.07
 ≤65 years411.00 (Reference)1.00 (Reference)
 >65 years310.81 (0.38–1.75)1.70 (0.97–2.97)
Tumour stage  <0.001 0.004
 ≤pT2b471.00 (Reference)1.00 (Reference)
 ≥pT3a256.86 (2.69–17.49)2.76 (1.38–5.50)
Lymph node status  <0.001 0.10
 pN0581.00 (Reference)1.00 (Reference)
 pN1–3148.78 (2.81–27.39)1.86 (0.89–3.89)


Recurrent UC in the preserved bladder after PC is an important concern and an estimated 25–30% of patients ultimately require sRC [6,7]. The prognosis after sRC has not been extensively studied and current data are based on small series with conflicting results. Holzbeierlein et al. [5] reported on 58 patients undergoing PC of whom 11 patients (19%) had recurrence in the preserved bladder. A total of four patients (6.9%) had muscle-invasive cancer recurrence requiring sRC of whom only one patient survived [5]. Similar results were reported by Resnick et al. [1] in an older series including 86 PC patients with available follow-up data. A total of 65 patients (75.6%) developed recurrent disease of whom 47 had MIBC. A variety of salvage treatments were performed, but of the three patients treated with sRC, all died within 1 year [1]. On the other hand, other series have reported better outcomes of sRC. Kassouf et al. [4] reported on a PC series of 37 patients of whom five patients (13.5%) underwent sRC for MIBC recurrence. After a median follow-up of 37 months, four patients (80%) were free of disease. Similarly, Dandekar et al. reported a sRC success rate of 75% in seven patients with MIBC recurrence [12]. To our knowledge, this study is the first focusing on survival after sRC, demonstrating the 5-year RFS rate for all patients undergoing sRC to be 56%. To put these outcomes in perspective, we compared these results with data of 2218 patients undergoing immediate RC at our institution from the same era. Acknowledging the limitations of this comparison, we found the RFS to be fairly similar between sRC and immediate RC (56% and 66%, respectively). This suggests that, overall, a reasonable number of patients can be salvaged by RC.

Nonetheless, sRC will not cure all patients with recurrent disease. To identify which patients are likely to achieve long-term survival after sRC, we focused on what factors influence post-sRC survival. As expected, the prognosis after sRC was highly dependent on the pathological tumour stage and nodal status. While excellent RFS was achieved in patients with OC disease, patients with EV or LN+ disease had a 5-year RFS rate of only 22% which was worse than immediate RC (36%; P= 0.027). One explanation for the poor prognosis in EV and LN+ cases in the sRC group may be a more aggressive biological behaviour of the recurred tumour. It should be noted that a substantial number of patients were understaged at the time of sRC. While clinically 12.5% of patients were thought to have ≥cT3a disease before sRC, at final pathology 34.8% of patients had ≥pT3a disease. Moreover, 11 of 14 pN1-3 patients (78.6%) had ≤pT2b disease, demonstrating that a significant number of patients had nodal disease despite limited local disease. Based on these findings, we feel that muscle-invasive and possibly high grade non-muscle-invasive recurrence mandates aggressive management and that sRC should not be reserved as the ‘last resort’. While the use of neoadjuvant and adjuvant therapy at the time of sRC did not significantly affect outcomes in this study, it should be emphasized that this was a retrospective study reporting on a relatively small cohort. The role of neoadjuvant therapy in patients with locally advanced recurrence therefore remains unknown, but can probably be extrapolated from data documenting its benefit in de novo presentation of muscle-invasive disease.

Of all patients undergoing sRC in our cohort, most patients underwent sRC for recurrent disease within 2 years after PC. Nonetheless, late recurrences after PC did occur and one patient underwent sRC for recurrent disease nearly 19 years after PC. This finding underscores the need for long-term surveillance of the preserved bladder after PC, as also suggested by Kassouf et al. [4]. In their study, three patients (8.1%) had late MIBC recurrence up to 12 years after PC. As all three patients with late recurrence died, we investigated whether the time from PC to sRC was associated with survival after sRC. We found a trend towards decreased survival for a longer interval between PC and sRC; however, this effect did not reach statistical significance. A larger study would be needed to further investigate the effect of time to sRC and survival.

An inherent limitation of using tertiary referral centre data is the potential for selective referral bias. While this is true, performing this study outside a referral centre is likely to be underpowered with a limited number of patients undergoing PC and an even lower number of patients undergoing sRC. Additionally, as nearly all PCs were performed in other hospitals, we were unable to collect all pathological data at the time of PC. Lastly, we acknowledge that the comparison between outcomes in sRC and immediate RC may be biased by baseline differences. To reduce this effect, we used data from the remaining 2218 patients who underwent immediate RC for UC at the same institution during the same era.

Most patients with local recurrence following PC can be salvaged by RC. However, the success rate of sRC is highly dependent on pathological tumour stage and nodal status at sRC. In patients with recurrent disease confined to the bladder, aggressive surgical management provides a good chance of salvage. In the group of patients with EV and/or LN+ recurrent disease, however, only 15% of patients were salvaged by sRC.




None declared.