Impact of renal impairment on eligibility for adjuvant cisplatin-based chemotherapy in patients with urothelial carcinoma of the bladder




Perioperative cisplatin-based chemotherapy has shown benefit in patients with high-risk localized urothelial bladder cancer, but it is not widely used. Renal impairment may be a major factor limiting its use. The current study was designed to determine the proportion of patients ineligible to receive adjuvant cisplatin-based chemotherapy based on inadequate renal function alone.


Patients who underwent radical cystectomy for urothelial cancer of the bladder with evidence of extravesical disease (≥pT3 or any N+) were identified. Patients who received neoadjuvant chemotherapy were excluded. Serum creatinine immediately before and nadir serum creatinine after cystectomy were used to calculate creatinine clearance (CrCl) or glomerular filtration rate (GFR) using the Cockroft-Gault (CG), Jelliffe, and Modification of Diet in Renal Disease (MDRD) study formulas. A cutoff of CrCl <60 mL/min or GFR <60 mL/min/1.73 m2 was used to determine ineligibility for cisplatin-based chemotherapy. The proportion of patients ineligible by each formula was compared by univariate logistic regression. Univariate linear regression was performed to determine the effect of age on CrCl or GFR.


Most patients were pT3 or greater; 39% were lymph node-positive. The overall proportion of patients ineligible for cisplatin-based chemotherapy was 28% by the CG formula, 52% by Jelliffe, and 24% by MDRD. Concordance between formulas was low. With all formulas the probability of ineligibility increased with age: by the CG equation, >40% of patients age >70 years were ineligible.


The widespread use of cisplatin-based perioperative chemotherapy in patients with high-risk localized bladder cancer may be significantly limited by the high prevalence of baseline renal insufficiency in this population. This finding is most striking in the elderly. Better selection of patients who may safely receive cisplatin and more effective regimens devoid of cisplatin are required to optimize outcomes in this group of patients. Cancer 2006. © 2006 American Cancer Society.

Radical cystectomy (RC) is the primary treatment modality for patients with muscle-invasive urothelial cancer of the bladder. Patients with extravesical or lymph node-positive disease, however, are at high risk for disease recurrence after surgery alone.1 Therefore, the integration of perioperative chemotherapy has been explored in an attempt to improve disease control and patient survival. Two large randomized trials and a metaanalysis have now confirmed a survival advantage with the use of cisplatin-based combination neoadjuvant chemotherapy.2–4 The data supporting adjuvant chemotherapy are less compelling. However, this is likely a result of small, underpowered trials utilizing suboptimal chemotherapeutic regimens.5–9 Many oncologists, as well as national consensus guidelines,10 advocate the use of chemotherapy in high-risk patients postcystectomy. Despite the supporting evidence, perioperative chemotherapy in urothelial cancer of the bladder is still not widely practiced in the community.11 Enrollment in past and present trials exploring perioperative chemotherapy in urothelial carcinoma has required adequate renal function. However, it has long been appreciated that a large proportion of patients with urothelial cancer have impaired renal function13 due to multiple factors, including medical comorbidities, age-related decline in glomerular filtration rate (GFR), and ureteral obstruction. The degree to which impaired renal function limits the widespread generalizable applicability of the results of perioperative cisplatin-based chemotherapy trials to the “real world” setting has never been systematically explored.

To appreciate the extent of this problem at our institution, we evaluated the proportion of patients with high-risk disease (extravesical tumor) after RC who would be deemed ineligible to receive adjuvant cisplatin-based chemotherapy due to impaired renal function alone. Whereas past and present trials exploring perioperative chemotherapy in urothelial carcinoma have used slightly different definitions of acceptable renal function for enrollment, we utilized the criterion common to the majority of these studies, a creatinine clearance >60 mL/min as determined by the Cockroft-Gault equation, as our “cutoff” for cisplatin eligibility. Notably, these are the same eligibility criteria utilized by ongoing randomized studies of adjuvant cisplatin-based chemotherapy in urothelial carcinoma.15, 16 We also explored the impact of different methods of estimating renal function and the impact of surgery on patients' eligibility for cisplatin-based chemotherapy.


With Institutional Review Board approval, we retrospectively reviewed the Memorial Sloan-Kettering Department of Urology Radical Cystectomy database for patients who underwent surgery between January 1990 and March 2005. The database contains patient demographics, perioperative clinical details, operative details, pathologic information, laboratory values, and follow-up information for all patients. Important operative details included type of urinary diversion: ileal conduit versus reservoir-type, such as orthotopic neobladder or continent cutaneous catheterizable reservoir. We identified patients who underwent RC for bladder cancer with urothelial histology and any evidence of extravesical disease (at least stage pT3 and or lymph node-positive disease) documented in the surgical pathology report. Postoperative records were evaluated to identify and exclude any patients who had received prior neoadjuvant chemotherapy or radiotherapy. Only patients who had weight data were included in the final analysis. The serum creatinine (SCr) obtained nearest to but before the date of RC was considered the preoperative value. We designated the nadir postoperative SCr level as the lowest documented value between 8 days to 3 months after RC. A minimum of 8 days after RC was chosen to decrease the rate of spuriously low SCr measurements due to perioperative intravenous hydration. Patients were included in the analysis only if SCr data were available for this period.

We estimated the creatinine clearance (CrCl) from the SCr using the most widely used equation,18 the Cockroft-Gault (CG) formula.19 For comparison, we also calculated the estimated CrCl with the Jelliffe formula20 and estimated the glomerular filtration rate (GFR) using the 4-variable MDRD Study equation (Fig. 1).17

Figure 1.

Formulas used for creatinine clearance (CrCl) calculations. SCr indicates erum creatinine; MDRD, Modification of Diet in Renal Disease; GFR, glomerular filtration rate.

The Kidney Diseases Outcomes Quality Initiative of the National Kidney Foundation defined patients with GFR <60 mL/min/1.73 m2 as having chronic kidney disease.21 Furthermore, entry into recent cisplatin-based chemotherapy trials for urothelial cancers has required adequate renal function defined as serum CrCl ≥60 mL/min.15, 16, 22 We therefore used this level as our primary threshold.

Statistical Analysis

For each formula, we evaluated the effect of age on the absolute value of CrCl in univariate linear regression models and the probability of ineligibility in logistic regression models. We calculated overall concordance among the 3 formulas and concordance between each pair. All statistical analysis was performed using Stata 8.2 (StataCorp., College Station, TX).


From January 1990 until May 2005, 1284 patients underwent RC at our institution for urothelial cancer of the bladder. Eighty patients who received neoadjuvant chemotherapy and 11 patients who underwent radiation therapy before RC were excluded. Of the remaining 1199 patients, 620 patients were either pathologic stage pT3a or greater (N0 or N+), or were stage pT0 through pT2b with a positive lymph node. We excluded 112 of these patients who lacked weight data, leaving 508 patients who met criteria for inclusion in the study. Fourteen patients were missing data for urinary diversion, and 3 patients were missing data for preoperative SCr. Baseline characteristics and demographics of the patients are shown in Table 1.

Table 1. Patient Characteristics and Demographics
Characteristic No. patients (%) or value (Interquartile range)
  1. LN indicates lymph nodes.

No. of patients 508
Median weight, kg 77 (68–87)
Interval from cystectomy to postcreatinine measurement, d 15 (8–37)
 Male 392 (77)
 Female 116 (23)
 White 440 (87)
 Black 14 (3)
 Other 54 (11)
Urinary diversion
 Ileal conduit 380 (77)
 Reservoir type 114 (23)
Pathologic stage
 pT0 to pT2 with positive LN 50 (10)
 pT3 with negative LN 210 (41)
 pT3 with positive LN 106 (21)
 pT3 with unknown LN 42 (8)
 pT4 with negative LN 40 (8)
 pT4 with positive LN 39 (8)
 pT4 with unknown LN 21 (4)
Age, y
 ≤50 29 (6)
 51–60 75 (15)
 61–70 162 (32)
 71–80 180 (35)
 >80 62 (12)
Serum creatinine, mg/dLPreoperationPostoperation
 0.0–1.0178 (35)301 (59)
 1.1–2.0304 (60)194 (38)
 2.0+23 (5)13 (3)

Impact of Renal Impairment on Eligibility for Cisplatin

Using the CG formula to calculate CrCl, 28% (95% confidence interval [95% CI], 24%–31%) of patients were ineligible for cisplatin-based chemotherapy using an estimated CrCl <60 mL/min as the threshold for eligibility. As demonstrated in Figure 2, a greater proportion of patients were deemed ineligible by the CG formula with increasing age; >40% of patients age >70 years would have been considered ineligible based on CrCl alone.

Figure 2.

Proportion of patients deemed ineligible (i.e., creatinine clearance <60 mL/min) by the Cockroft-Gault formula by age group.

Comparison of Different Formulas to Estimate Renal Function

We chose to compare the Jelliffe and MDRD study equation to the popular CG formula. The Jelliffe is a frequently used equation that requires only age and SCr to estimate CrCl. The MDRD Study equation was recently developed as the most accurate estimation of GFR specifically in patients with chronic kidney disease,17 a population in whom accurate determination of kidney function is crucial. Because of its improved accuracy, the application of the MDRD Study equation in oncology has been encouraged.23 Table 2 details the proportion of patients determined ineligible by each formula.

Table 2. Proportion (95% Confidence Interval) of Patients Ineligible for Chemotherapy by Each Formula
FormulaPreoperative (n = 505)Postoperative (n = 508)
  1. MDRD indicates Modification of Diet in Renal Disease Study.

Cockroft-Gault43% (39–48%)28% (24–31%)
Jelliffe83% (80–86%)52% (48–56%)
MDRD47% (42–51%)24% (20–28%)

Of particular interest is how many patients receive different recommendations for eligibility or ineligibility according to the different formulas. All 3 tests were in concordance for only 324 patients (64%), that is, approximately one-third of patients would receive chemotherapy on the basis of 1 formula, but not on the basis of another. Pairwise concordance between tests is shown in Table 3.

Table 3. Pairwise Concordance between Equations to Estimate Postoperative Renal Function
  1. MDRD indicates Modification of Diet in Renal Disease Study.

 Ineligible236 (46%)8 (2%)  
 132 (26%)132 (26%)  
 Concordance: 72%  
MDRDEligible335 (66%)52 (10%)244 (48%)143 (28%)
Ineligible33 (6%)88 (17%)0 (0%)121 (24%)
 Concordance: 83%Concordance: 72%

Impact of Age on Eligibility for Cisplatin-Based Therapy

Table 4 gives the proportion of ineligible patients by each formula for ages <70 years and age ≥70 years. Figure 3 shows absolute values of CrCl for each formula by years of age. Not only do the formulas give different values for each age, but changes in CrCl with age vary between formulas. Figure 4 shows similar data in the form of the predicted probability of ineligibility by age. Starting at age 50 years, large differences (as much as 20%) are observed among the 3 formulas. The probability of ineligibility is initially lower under CG than MDRD until the 2 curves cross at around age 70 years. The Jelliffe formula produced the highest probability of ineligibility for all ages over 50 years. Although the 3 equations have variability in their predicted probability of ineligibility, it is important to recognize that all the equations demonstrate an increased probability of ineligibility with age.

Figure 3.

Change in absolute value of creatinine clearance with increasing age. Solid line indicates Cockroft-Gault; dotted line, Jelliffe; dashed line, Modification of Diet in Renal Disease Study. GFR indicates glomerular filtration rate.

Figure 4.

Change in predicted probability that a patient is ineligible for chemotherapy with increasing age. Solid line indicates Cockroft-Gault; dotted line, Jelliffe; dashed line, Modification of Diet in Renal Disease Study.

Table 4. Proportion (95% Confidence Interval) of Patients Ineligible for Chemotherapy by Each Formula by Age Group
  Age <70 yearsAge 70+ years
  1. MDRD indicates Modification of Diet in Renal Disease Study.

PreoperationNo. of patients263242
Cockroft-Gault25% (20%–31%)63% (57%–69%)
Jelliffe83% (78%–87%)83% (79%–88%)
MDRD44% (38%–50%)50% (44%–56%)
PostoperationNo. of patients265243
Cockroft-Gault12% (8%–16%)44% (38%–50%)
Jelliffe38% (32%–44%)67% (62%–73%)
MDRD18% (14%–23%)30% (24%–35%)

Impact of Surgery on Eligibility for Cisplatin-Based Therapy

Given that neoadjuvant chemotherapy before RC has been gaining acceptance4, 24 and that some patients have improvement in renal function with relief of tumor-related urinary obstruction after RC, we sought to determine whether more patients would become eligible for chemotherapy after RC. As seen in Tables 2 and 4, the proportion of eligible patients was higher after RC regardless of method to estimate renal function. Table 5 shows the median and interquartile range for the CrCl values of the 3 formulas, pre- and postoperation. For all of the formulas, the absolute values increased after surgery. To appreciate the proportion of patients whose eligibility status changed with surgery, we tabulated status before and after RC by the CG estimation alone for the 505 patients with complete data before and after RC (Table 6). Eligibility status changed for 18% of patients; 17% changed from ineligible to eligible after RC, and 1% changed from eligible to ineligible.

Table 5. Creatinine Clearance Values by Equation
 Preoperation (n = 505)Postoperation (n = 508)
  1. MDRD indicates Modification of Diet in Renal Disease Study.

  2. Data are shown as the median (interquartile range).

Cockroft-Gault in mL/min63.3 (47.4, 79.4)76.2 (57.0, 97.0)
Jelliffe in mL/min49.5 (38.8, 60.7)59.0 (45.5, 73.9)
MDRD in mL/min/1.73 m264.6 (50.9, 77.5)77.5 (61.6, 98.8)
Table 6. Relationship between Preoperative and Postoperative Eligibility Status, as Determined by the Cockroft-Gault Formula
Eligible279 (55%)7 (1%)
Ineligible87 (17%)132 (26%)

Patients who received a continent urinary diversion (cutaneous reservoir or neobladder) had a lower probability of ineligibility as compared with patients with an ileal conduit urinary diversion: for CG, Jelliffe, and MDRD, the relative risk was 27%, 63%, and 61%, respectively (P < 0.0005, P < 0.0005, and P = 0.02 by the chi-square test). Moreover, patients with continent urinary diversions were on average 12 years younger than their ileal conduit counterparts (P < 0.0005 by linear regression). This suggests that type of urinary diversion may have influenced our findings based on differences in age and the effects of age between formulas. Figure 5 shows the predicted probability of ineligibility separately by specific type of urinary diversion. Although rates of ineligibility were higher in the ileal conduit group, there were important differences in eligibility rates between formulas irrespective of urinary diversion type.

Figure 5.

Change in predicted probability that a patient is ineligible for chemotherapy with increasing age, separately by urinary diversion type. Solid line indicates Cockroft-Gault; dotted line, Jelliffe; dashed line, Modification of Diet in Renal Disease Study.


Our study documents that a large proportion of patients with high-risk bladder cancer were ineligible for cisplatin-based adjuvant chemotherapy based on impaired renal function. This finding was most striking in the elderly population. For patients age >70 years, over 40% would be considered ineligible based on a CrCl <60 mL/min estimated by the Cockroft-Gault equation. This finding is of particular relevance given that the majority of patients diagnosed in the U.S. with bladder cancer are age >60 years.25 Several recent studies have demonstrated that aggressive surgical treatment of invasive bladder cancer in elderly patients is beneficial and associated with acceptable morbidity.26, 27 However, a recent series noted that elderly patients were less likely than younger patients to receive adjuvant chemotherapy, despite the observations that they were likely to have higher stage disease at RC and decreased disease-free survival.28 Given our current findings, impaired renal function likely significantly limits the widespread applicability of the results of perioperative chemotherapy trials in urothelial carcinoma completed to date.

Tactics to improve therapeutic options among this patient population include improving the selection of patients who may safely receive cisplatin-based therapy, optimizing the timing of administration of perioperative chemotherapy, and developing more effective chemotherapy regimens devoid of cisplatin. Various methods of estimating renal function have been explored in the past. Urine collection methods have proven cumbersome and unreliable, whereas isotope methods for calculation of GFR are accurate but labor-intensive and not universally available. Therefore, in the clinical setting, estimation methods are desirable because of their simplicity using readily available information and low cost. Whereas the CG equation is utilized in most clinical trials, and recognized by the Food and Drug Administration (FDA) in studies of patients with renal impairment,29 other formulas have been proposed as more applicable for use in cancer patients.30 For example, in a study of 116 ovarian cancer patients, Calvert et al.31 reported that GFR determined by chromium-51-EDTA clearance had correlation coefficients of 0.75 with the Jelliffe and 0.65 with the CG formula estimates of CrCl. More recently, investigators have called for exploration of the MDRD study equation in oncology trials.23 When we compared the concordance of eligibility for cisplatin (by an estimated CrCl <60 mL/min or GFR <60 mL/min/1.73 m2) between the 3 equations, we noted that the overall concordance was approximately two-thirds. These findings highlight the fact that current means of assessing eligibility for cisplatin-based therapy may be suboptimal. Prospective studies are needed to determine which of these equations best indicates renal function in cancer patients and, more importantly, predicts cisplatin toxicity.

Given that ureteral obstruction before RC may have contributed to impaired renal function in some patients with bladder cancer, we compared estimates of renal function before and after RC. Interestingly, we found that the median CrCl or GFR estimate was higher after RC regardless of the equation used. Therefore, based on CrCl, more patients may be eligible for cisplatin-based adjuvant chemotherapy than for neoadjuvant chemotherapy. However, these findings should be interpreted with caution, as many patients' improvement in renal function after surgery may be due to perioperative hydration, rather than relief of obstruction. Although we found a lower probability of cisplatin-ineligibility for patients who received a continent urinary diversion regardless of estimating equation, this finding should also be interpreted with caution, given that patients who received a continent urinary diversion were younger and met criteria for minimal renal function.32 Furthermore, estimations of renal function based on SCr after RC may be confounded by urinary resorption in patients with a continent urinary diversion.33

Regardless of the method used for defining ineligibility for cisplatin-based chemotherapy, clearly a large proportion of patients with urothelial carcinoma have impaired renal function. New therapeutic approaches are needed for this subset of patients. The sequential doublet regimen of doxorubicin and gemcitabine followed by paclitaxel and carboplatin has been explored in patients with impaired renal function, both in the metastatic and adjuvant setting.34 This regimen has shown intriguing results in patients with locally advanced urothelial cancer compared with contemporary controls not receiving adjuvant treatment. Similarly, the 3-drug regimen of carboplatin, gemcitabine, and paclitaxel has shown impressive results in the metastatic setting,35 and a pilot trial has been performed using this regimen in the neoadjuvant setting.36 Ultimately, randomized trials are needed to determine if these carboplatin-based regimens confer a survival benefit in the perioperative setting.

There are several limitations to our study including the following: 1) Our institution is a referral center for bladder cancer treatment. As a result, our treatment population may be older and more likely to have comorbidities, including renal impairment. We chose the nadir SCr in the period of 1 week to 3 months after RC. For the majority of the patients the nadir was within 1 to 2 weeks after RC. This SCr value may not have reflected a steady state but the well-hydrated postoperative state of a patient in the hospital or recently discharged. This may partly explain the higher median CrCl after RC than before RC and may also underestimate the percentage of ineligible patients. Despite the risk that a single SCr measurement may not reflect patients' steady-state levels, we chose the nadir level so that such an artifact would have only underestimated the problem of ineligibility after RC. 2) We excluded 80 (6%) patients who received neoadjuvant chemotherapy to avoid the potential contribution of cisplatin-induced nephrotoxicity to the estimates of eligibility. Notably, only 27 of these 80 patients were age >70 years. Therefore, even if all of these patients had postoperative creatinine clearance values adequate for cisplatin, the effect of this exclusion was minimal among the elderly group in whom eligibility was lowest. 3) We chose to focus on the postoperative adjuvant setting in this analysis, as this population is well defined in our database. However, the results of this analysis can be readily extrapolated to the neoadjuvant setting where the benefit of cisplatin-based chemotherapy is more universally accepted. 4) Finally, we acknowledge that there are multiple factors beside renal impairment, particularly in the RC and elderly patient populations, that may limit the use of cisplatin. Ultimately, this only leads to an underestimation of the proportion of patients ineligible for cisplatin-based therapy.


Our study demonstrates that the widespread application of cisplatin-based perioperative chemotherapy in urothelial cancer in the “real world” may be significantly limited by the relatively high frequency of baseline renal impairment in this population. Because the aged population is increasing, we will be confronted with more elderly patients with advanced bladder cancer who may benefit from perioperative chemotherapy. Several important issues relevant to this population need to be explored. Which estimate of GFR or CrCl is ideal in this population of patients? Which cutoff most safely defines cisplatin eligibility? Finally, are there alternative regimens devoid of cisplatin that are beneficial in the perioperative setting? Prospective trials are required to answer these questions.


We thank Kirjal Vora for assistance in querying the cystectomy database.