What's known on the subject? and What does the study add?
Lymph node (LN) metastasis is a critical predictor for disease recurrence and cancer-specific survival in patients with urothelial carcinoma of the bladder (UCB) treated with radical cystectomy. Patients with a low LN disease burden (pN1) might be cured by surgery alone, while patients with a high LN disease burden (stage ≥ pN2) might benefit most from adjuvant chemotherapy.
We found that outcomes of patients with pN1 UCB are significantly affected by pathological stage and soft tissue surgical margin status. Our nomogram may help to improve outcomes prediction in patients with pN1 UCB. An accurate prediction of the individual risk of outcomes may help risk stratifying patients with pN1 UCB to help improve clinical decision-making.
To identify clinicopathological factors that predict outcomes in patients with a single lymph node (LN) metastasis (pN1) treated with radical cystectomy (RC) for urothelial carcinoma of the bladder (UCB).
LN metastasis is an established predictor of clinical outcomes in patients. While most patients with large LN burden experience disease recurrence, lymphadenectomy can be curative in patients with pN1 disease.
Patients and Methods
We analysed 381 patients with pN1 UCB from a multi-institutional cohort of 4335 patients with UCB treated with RC and lymphadenectomy without preoperative chemo- or radiotherapy.
Subgroup analyses were performed for patients with ≥9 LNs removed and according to adjuvant chemotherapy administration (n = 215).
The median (interquartile range, IQR) LN number was 15 (19) and the median (IQR) LN density was 6.7 (7.5)%.
Within a median follow-up of 41 months, the mean (+/− sd) 2- and 5-year cancer-specific survival (CSS) rates were 55 (3)% and 46 (3)%, respectively.
On multivariable analysis that adjusted for the effects of standard clinicopathological features, female gender (hazard ratio [HR] 1.48, P = 0.023), higher tumour stage (HR 1.68, P = 0.007), positive soft tissue surgical margin (STSM; HR 2.06, P = 0.004), higher LN density (HR 2.99, P = 0.025) and absence of adjuvant chemotherapy (HR 0.70, P = 0.026) were independently associated with CSS.
In subgroup analyses of patients with ≥9 LNs removed, tumour stage and STSM status remained independent predictors for CSS (P = 0.009 and P < 0.001, respectively).
About half of the patients with pN1 UCB died from UCB within 5 years of RC.
Pathological stage and STSM status are strong predictors for outcomes.
Accurate prediction of the individual risk of CSS may help risk stratifying pN1 UCB in order to help improve clinical-decision making. Patients with pN1 UCB presenting with additional unfavourable risk factors need a closer follow-up scheduling and might receive adjuvant therapy.
About a quarter of patients treated with radical cystectomy (RC) and lymphadenectomy (LND) for urothelial carcinoma of the bladder (UCB) have lymph node (LN) metastasis at time of surgery [1, 2]. LN metastasis is a critical predictor for disease recurrence and cancer-specific survival (CSS) and therefore an important determinant of the therapeutic course after surgery . While most patients with LN positive UCB have disease recurrence and eventually death, up to 30% of LN-positive patients are cured with RC and LND as monotherapy [4-6]. Different variables have been identified to provide risk stratification in patients with LN metastasis after RC, e.g. pathological tumour stage, the extent of LND, the number of positive LNs detected, and LN density [7-14]. Accurate identification of patients who are cured with surgery alone would help in patient counselling and more importantly clinical decision making about follow-up scheduling and adjuvant chemotherapy.
Adjuvant chemotherapy is generally offered to LN-positive patients who can tolerate it, with the goal of delaying disease recurrence and prolonging survival [15-17]. However, as clinical trials were underpowered, the efficacy of adjuvant chemotherapy remains greatly debated . It has been suggested that patients with a high LN disease burden (stage ≥ pN2) after RC benefit most from adjuvant chemotherapy . Indeed, the likelihood of being cured with pN1 UCB after RC and LND is higher than for pN2–3 patients . In addition, patients with low volume LN disease (1 or 2 positive LNs) have 10-year recurrence-free survival (RFS) rates of up to 50% [7, 18, 19].
We hypothesise that despite the heterogeneity of outcomes of patients with UCB with only one LN metastasis (pN1 stage), standard clinicopathological features could help predict the individual risk of disease recurrence. Therefore, we assessed clinical outcomes of patients with pN1 UCB at the time of RC and identified independent risk factors for disease recurrence and cancer-specific death in a large, international, multicentre cohort.
Patients and Methods
This was an Institutional Review Board-approved study, with all participating sites providing the necessary institutional data-sharing agreements before initiation of the study. In all, 12 academic centres worldwide provided data. A computerised databank was generated for data transfer. After combining the data sets, reports were generated for each variable to identify data inconsistencies and other data integrity problems. Through regular communication with all sites, resolution of all identified anomalies was achieved before analysis. Before final analysis, the database was frozen and the final data set was produced for the present analysis.
From 4335 patients who underwent RC with bilateral LND between 1980 and 2008, 381 (8.8%) had a single LN metastasis (pN1) at the time of RC and comprised the study cohort. No patient received preoperative radiotherapy or chemotherapy. No patient had distant metastatic disease at the time of RC. In all, 215 of the 381 patients (56.4%) received adjuvant systemic chemotherapy administered at the investigator's discretion based on patients' tumour stage and overall health status.
For pathological evaluation, all surgical specimens were processed according to standard pathological procedures as previously described . Genitourinary pathologists assigned tumour grade according to the 1973 WHO grading system. Pathological stage was reassigned according to the 2002 American Joint Committee on Cancer TNM-staging system. The presence of concomitant carcinoma in situ (CIS) was defined as the presence of CIS in conjunction with another tumour other than CIS alone. Pelvic LNDs were examined grossly, and all lymphoid tissue was submitted for histological examination. The extent of LND was at the surgeon's discretion. Positive soft tissue surgical margin (STSM) was defined as the presence of tumour at inked areas of soft tissue on the RC specimen . Urethral or ureteric margin were not considered as STSM. Lymphovascular invasion (LVI) was defined as the unequivocal presence of tumour cells within an endothelium-lined space without underlying muscular walls [21, 22].
Follow-up was performed according to institutional protocols. Patients generally were seen after RC at least every 3–4 months for the first year, semi-annually for the second year, and annually thereafter. Follow-up visits consisted of a physical examination and serum chemistry evaluation, including liver function tests and alkaline phosphatase. Diagnostic imaging of the upper tracts (e.g. ultrasonography and/or IVU, CT of the abdomen/pelvis with i.v. contrast) and chest radiography were performed at least annually or when clinically indicated. Additional radiographic evaluations, such as bone scan and/or CT, were performed at the discretion of the treating physician when indicated. Detection of cancer in the ureter, renal pelvi-calyceal system and/or urethra were coded as a second (metachronous) primary and not as local or distant recurrence. When patients died, the cause of death was determined by the treating physicians, by chart review corroborated by death certificates, or by death certificates alone . Perioperative mortality (i.e. death ≤30 days of RC) was censored at time of death for bladder CSS analyses.
The Kolmogorov–Smirnov test was used to assess the normal distribution of variables. Fisher's exact test and the chi-square test were used to evaluate the association between categorical variables. Differences in variables with a continuous distribution across categories were assessed using the Mann–Whitney U-test (two categories) and Kruskal–Wallis test (three and more categories). Actuarial method was used to estimate RFS and CSS probabilities and the differences were assessed with the log-rank statistic. The Kaplan–Meier method was used to graphically display survivor functions. Univariable and multivariable Cox regression models addressed time to recurrence and time to cancer-specific mortality after RC. In all models, proportional hazards assumptions were systematically verified using the Grambsch–Therneau residual-based test.
Multivariate regression coefficients were then used to construct two sets of nomograms for disease recurrence and cancer-specific mortality prediction at 2 years after RC and bilateral LND. The first nomogram set included all patients and the second only patients with ≥9 LN removed. Using the threshold of 9 LNs based on previous reports indicating that 9 LNs was the minimum number required during LND . Predictions were quantified using the area under the curve of the receiver operator characteristic. Internal validation with 200-bootstrapping samples was performed to reduce overfit bias . The extent of overestimation or underestimation of the observed cancer-specific death rate was explored graphically using nonparametric loss-calibration plots.
All reported P values are two-sided, and statistical significance was set at P < 0.05. Statistical tests were performed with SPSS® 17 (SPSS Inc., IBM Corp., Somers, NY, USA). Nomogram calculation was conducted using the statistical package for R (the R-foundation for Statistical Computing, version 2.1.13).
Clinicopathological Features and Prognostic Factors in All PN1 Patients
The median (interquartile range [IQR]; range) number of removed LNs was 15 (19; 1–98). In all, 294 (77.2%) patients had ≥9 LNs removed and 149 (39.1%) had ≥20 LNs removed. The median (IQR; range) LN density was 6.7 (7.5; 1–100)%. There were significant differences in the number of LN removed across the participating centres (P = 0.001). Table 1 shows the descriptive characteristics of the study cohort. Most of the patients had features of aggressive UCB, i.e. non-organ confined (71.4%), grade 3 (54.1%) disease and LVI (58.3%).
Table 1. Clinical and pathological characteristics of 381 patients with pN1 UCB after RC and bilateral pelvic LND.
Median (IQR) age, years
Age (years; categorical):
Ta, Tis, T1
No grading (pT0 disease)
The median (IQR; range) follow-up of patients alive at censor was 41 (53.9; 1–299) months. During follow-up, disease recurred in 205 patients (53.8%), 221 (58.0%) died and 170 (44.7%) died of UCB. Overall, actuarial RFS and CSS estimates at 2, 5 and 10 years are given in Fig. 1.
On univariable analyses, female gender, higher tumour stage, positive STSM, higher LN density and not receiving adjuvant chemotherapy were associated with disease recurrence (P ≤ 0.039) and cancer-specific mortality (P ≤ 0.009; Table 2A and Fig. 2A–F). In addition, on univariable analyses, older age and lower LN count were associated with cancer-specific mortality (P ≤ 0.039; Table 2A).
Table 2. Univariable (A) and multivariable (B) Cox regression analyses predicting disease recurrence and cancer-specific mortality in 381 patients with a single LN- positive UCB treated with RC and bilateral LND.
HR (95% CI)
HR (95% CI)
Pathological T-stage (trend):
pTa/is/1 vs pT0
pT2 vs pT0
pT3 vs pT0
pT4 vs pT0
Organ-confined vs non-organ-confined
Number LNs removed (continuous)
Organ-confined vs non-organ-confined
Number LN removed (continuous)
In a multivariable Cox regression analysis that adjusted for the effects of standard clinicopathological characteristics, non-organ confined tumour stage (Hazard ratio [HR] 1.56, P = 0.009) and positive STSM (HR 1.89, P = 0.007) were independently associated with disease recurrence (Table 2B). Moreover, in multivariable Cox regression analysis that adjusted for the effects of standard clinicopathological characteristics, female gender (HR 1.48, P = 0.023), non-organ confined tumour stage (HR 1.68, P = 0.007), positive STSM (HR 2.06, P = 0.004), higher LN density (HR 2.99, P = 0.025) and absence of adjuvant chemotherapy (HR 0.70, P = 0.026) were independently associated with cancer-specific mortality (Table 2B).
When the models were restricted to patients with ≥9 LNs removed (n = 294), non-organ confined disease (HR 1.78, 95% CI 1.217–2.615, P = 0.003) and positive STSM (HR 2.92, 95% CI 1.537–5.562, P = 0.001) were independently associated with disease recurrence. Non-organ confined disease (HR 1.78, 95% CI 1.154–2.749, P = 0.009) and positive STSM (HR 3.49, 95 CI% 1.537–5.562, P < 0.001) were also independently associated with cancer-specific mortality, while female gender (P = 0.070) and adjuvant chemotherapy (P = 0.065) showed only a trend towards significance.
Prognostic Factors in PN1 Patients Who Received Adjuvant Chemotherapy
In total, 215 patients (56.4%) received adjuvant chemotherapy. The median (IQR; range) follow-up of these patients alive at censor was 49 (56.3; 1–299) months. In patients who received adjuvant chemotherapy, disease recurred in 116 patients (54.0%), 112 (52.1%) died and 92 (42.8%) died from UCB. Actuarial RFS and CSS estimates at 2, 5 and 10 years according to adjuvant chemotherapy administration are shown in Fig. 3. On univariable analyses, higher tumour stage (HR1.74, P = 0.011) and positive STSM (HR 1.99, P = 0.05) were associated with disease recurrence. In addition, on univariable analyses higher tumour stage (HR 1.95, P = 0.008) was associated with cancer-specific mortality. On multivariable analyses, higher tumour stage was independently associated with disease recurrence (P = 0.026) and cancer-specific mortality (P = 0.027; Table 3A).
Table 3. Multivariable Cox regression analyses predicting disease recurrence and cancer-specific mortality in 215 patients with adjuvant chemotherapy (A) and 166 patents without adjuvant chemotherapy (B) in single LN-positive UCB treated with RC and bilateral LND.
HR (95% CI)
HR (95% CI)
A. Adjuvant chemotherapy (n = 215)
Organ-confined vs Non organ-confined
Soft tissue surgical margin
Number LN removed (continuous)
B. No adjuvant chemotherapy (n = 166)
Organ-confined vs Non organ-confined
Soft tissue surgical margin
Number LN removed (continuous)
Prognostic Factors in PN1 Patients Who Did Not Receive Adjuvant Chemotherapy
In total, 166 patients (43.6%) did not receive adjuvant chemotherapy. The median (IQR; range) follow-up of these patients alive at censor was 35 (48.9; 1–232) months. In patients who did not receive adjuvant chemotherapy, disease recurred in 89 patients (53.6%), 109 (65.7%) died and 78 (47.0%) died from UCB. On univariable analyses, female gender, higher tumour stage, positive STSM and higher LN density were associated with disease recurrence (HRs 1.68, 1.81, 2.40 and 3.81, respectively; P ≤ 0.027) and cancer-specific mortality (HRs 1.92, 2.17, 3.46 and 5.58, respectively; P ≤ 0.007). On multivariable analyses, only positive STSM was independently associated with disease recurrence (P = 0.029) and cancer-specific mortality (P = 0.001; Table 3B).
Nomograms for Prediction of Disease Recurrence and Cancer-Specific Mortality
The multivariable nomograms for prediction of disease recurrence and cancer-specific mortality had bias-corrected predictive accuracy of 0.63 and 0.66, respectively. Both nomograms were well calibrated (Fig. 4). When restricted to patients with ≥9 LNs removed, the nomograms' bias-corrected predictive accuracies were 0.65 and 0.69 for disease recurrence and cancer-specific mortality (not shown).
In accordance with previous studies, we found that even a single LN metastasis in patients treated with RC is associated with high risk of disease recurrence and cancer-specific mortality [7, 14, 18, 19]. Nevertheless, comparable with previous studies we found that RC and LND resulted in adequate long-term disease control in >40% of pN1 patients [18, 19, 25]. Therefore, it is of great importance to identify the pN1 patients who are at increased risk of disease recurrence after RC and LND, and therefore are likely to benefit from multimodal therapy, such as adjuvant chemotherapy.
Pathological stage and STSM status were strong predictors of disease recurrence in pN1 patients. In addition, female gender, higher LN density and absence of adjuvant chemotherapy were associated with cancer-specific mortality. Some of these factors were previously shown to discriminate patients with either low positive-LN burden or general LN-positive patients into different risk groups [7, 10, 12-14]. In contrast to a recent report of 181 patients with one or two LN metastases , we identified female gender and STSM positivity as independent predictors of CSS. Both, female gender and STSM involvement are established predictors of poor outcomes in patients treated with RC and LND [20, 26, 27]. However, in the subgroup of patients with ≥9 LNs removed and those stratified by adjuvant chemotherapy, these parameters did not retain their independent predictive value. A possible explanation may be the loss of statistical power due to decreased sample sizes in subgroup analyses. Patients in the present study presented with high rates of LVI (58%), but LVI was not a prognostic factor for outcomes as reported in LN-negative patients before [21, 22, 28]. Nevertheless, while identification of risk factors is interesting, integration of these into decision tools is necessary to provide patients and physicians with individualised probabilities of disease recurrence and mortality.
We developed an adequately accurate (68% predictive accuracy for cancer-specific mortality) and well-calibrated nomogram based on four established clinicopathological parameters for pN1 patients treated with RC and LND without neoadjuvant chemotherapy. This tool may help in patient counselling and decision-making about adjuvant chemotherapy. However, before using in daily clinical practice it needs to be externally validated in a robust cohort. To date nomograms represent one of the most accurate tools for predicting outcomes in patients with cancer [6, 29]. However, evaluation of the clinical consequences of a new prediction tools is necessary. In the absence of prospective randomised trials to answer the value of decision tools, one simple way to test the clinical consequences of using prediction tools is to use decision curve analysis . For example, Vickers et al.  showed in a decision analytic approach that referring patients to chemotherapy on the basis of a multivariate model is likely to lead to better patient outcomes than the use of pathological groups.
We found that patients who received adjuvant chemotherapy had better outcomes than those who did not receive chemotherapy. Indeed, adjuvant chemotherapy was an independent protective factor for cancer-specific mortality. This finding is in agreement with a previous study that reported improved outcomes in patients with low LN burden who received adjuvant chemotherapy . Moreover, Svatek et al.  reported in 3947 patients treated with RC that off-protocol adjuvant chemotherapy is associated with a significant improvement in survival in patients with LN involvement. However, conclusions about the efficacy of adjuvant chemotherapy cannot be made on these data. Patients in the present study were not randomised to adjuvant chemotherapy, different chemotherapy regimens were used, and administration of adjuvant chemotherapy was influenced by physician preferences and general health issue parameters that were not controlled for.
The present study is not devoid of limitations. First and foremost are limitations inherent to its retrospective design and multicentre nature including multiple surgeons, oncologists and pathologists. However, all surgeons operated at high-volume tertiary care centres with significant experience in RC and LND and all pathologists were experienced genitourinary specialists. We did not perform a central pathological review, which might have an undefined impact, given differences in the rigor used by different pathologists to identify LNs and metastasis within LNs . Conversely, the present data reflect a real-world experience. Furthermore, the number of LNs removed is not only a factor of the extent of LND but is also dependent on the pathological evaluation and inherent differences between patients . In addition, the location of LNs may be important. Removing LNs from an area with a high likelihood of malignancy may be more valuable than removing LNs in areas less likely to be involved with cancer . We did not adjust for the template, which might have influenced the present results and outcomes. Finally, the nomogram is only internally tested requiring external validation. Therefore, further studies are warranted to replicate the present results and to externally validate the performance of our predictive model.
In conclusion, over half of all patients with a single LN metastasis die within 5 years after RC. Pathological stage and STSM status are strong predictors for outcomes in these patients. We integrated these risk factors into a decision tool to predict individualised probabilities of cancer-specific mortality in pN1 patients. Together with further research, such information may help risk stratifying patients with pN1 UCB regarding prognosis and the need for multimodal therapy such as adjuvant chemotherapy.
Dr. Michael Rink and Dr. David A. Green are supported by The Frederick J. and Theresa Dow Wallace Fund of the New York Community Trust.
Conflict of Interest
Shahrokh F. Shariat is an Advisory Board Member of Ferring Pharma.