Race affects access to nephrectomy but not survival in renal cell carcinoma

Authors


Pierre I. Karakiewicz, Cancer Prognostics and Health Outcomes Unit, University of Montreal Health Center (CHUM), 1058, rue St-Denis, Montréal, Québec, Canada H2X 3J4.
e-mail: pierre.karakiewicz@umontreal.ca

Abstract

OBJECTIVES

To assess whether, in contemporary patients with renal cell carcinoma (RCC), access to nephrectomy is the same between the Blacks and Whites, and that there is no difference in mortality after stratification for treatment type.

PATIENTS AND METHODS

The effect of race has received little attention in RCC; only two reports have addressed and suggested the presence of racial disparities, including access to nephrectomy and survival after nephrectomy, where Black patients were disadvantaged relative to Whites. We used the Surveillance, Epidemiology and End Results data from 12 516 patients of all stages diagnosed and treated for RCC between 2000 and 2004. The effect of race (Black vs White) on nephrectomy rate was addressed in logistic regression and binomial regression models, and Cox regression models tested the effect of race on overall survival.

RESULTS

Black patients were 50% less likely to have a nephrectomy than their White counterparts. However, race had no effect on overall survival when the entire cohort was assessed, as well as in subgroups of patients with or without nephrectomy.

CONCLUSIONS

Although race is a determinant of access to nephrectomy, it should not be interpreted as a barrier to care, as survival was unaffected by race in patients having a nephrectomy or not. Instead, race might represent a proxy of comorbidity and life-expectancy, which represent surgical selection criteria for nephrectomy.

Abbreviations
SEER

Surveillance, Epidemiology and End Results

ICD-O

International Classification of Disease for Oncology

OR

odds ratio

HR

hazard ratio

RR

relative risk.

INTRODUCTION

Kidney cancer ranks seventh among solid tumours when mortality figures are analysed in the USA population [1–5]. Recently, important disparities were reported in the rates of access to care between Black and White populations with RCC in patients aged ≥65 years from the USA [6].

We hypothesized that in contemporary patients access to nephrectomy is the same between the races, and that there is no difference in mortality before and after stratification for treatment type. We used the most contemporary Surveillance, Epidemiology and End Results (SEER) cohort (2000–2004) and included all adult patients. This allowed the general applicability of our findings across all adult age strata.

PATIENTS AND METHODS

Patients diagnosed with primary invasive kidney cancer between 2000 and 2004 were identified within nine SEER cancer registries [7] (Atlanta, Detroit, San Francisco-Oakland, Seattle-Puget Sound metropolitan areas, and the states of Connecticut, Hawaii, Iowa, New Mexico, and Utah). The characteristics of the SEER population are comparable to the general USA population [7]. Two kidney cancer diagnostic codes (International Classification of Disease for Oncology, Second edition, ICD-O-2, C64.9 code and the 9th revision, ICD-O-9, 189.0 code) were used as inclusion criteria. The presence of both diagnostic codes resulted in the identification of 15 776 patients with RCC. This did not include upper tract TCC or ureteric, non-cortical renal tumours (i.e. melanomas, sarcomas and lymphomas). Exclusions were cases with unknown tumour size (1718, 10.9%) or of unknown stage (742, 4.7%). The analysis was limited to adult participants (≥16 years old) and classified as non-Hispanic white or black according to the SEER race variable (Race Re-code). This variable is based on race and ethnic information from registration data and medical records, and is thought to have a high specificity [8]. This resulted in the exclusion of 1065 patients (6.7%) of other race or ethnic origin, and 1490 patients aged <16 years (9.4%). Additional exclusions consisted of patients treated with local tumour destruction, e.g. thermal ablation, electrocautery or cryosurgery (433, 2.7%). This resulted in 12 516 assessable patients treated with either surveillance (1810, 14.5%), radical nephrectomy (8957, 71.5%), or partial nephrectomy (1749, 14.0%). Subsequently, for further comparisons with previous reports, we repeated the analyses with an age restriction (≥65 years), which resulted in 5853 assessable patients. For all surgery patients, malignant histology was confirmed with the ICD-O-3 SEER histological codes. Vital status and survival-time records were used for survival analyses.

The independent-sample t-test and chi-square tests were, respectively, used for comparing means and proportions, between nephrectomy and no-nephrectomy patients. Categorical and multivariable logistic regression analyses were used to test the association between race and treatment type (nephrectomy vs no nephrectomy). Covariates consisted of age at diagnosis, gender, tumour size, stage at diagnosis or nephrectomy (localized, regional or metastatic), as well as year of diagnosis or of nephrectomy [9]. As logistic regression models can provide a biased estimate (away from the null) of the relative risk (RR) in cohort studies [10], we conducted a second analysis of the relationship between race and nephrectomy using log-binomial modelling. Finally, univariable and multivariable Cox regression analyses tested the effect of race on overall survival. Three separate analyses were used; the first addressed the effect of race on survival in the entire cohort and included patients treated with nephrectomy or not; the second exclusively addressed patients treated with nephrectomy; and the third exclusively addressed patients not treated with nephrectomy. The date of diagnosis was considered as ‘time zero’ for surveillance cases and the date of surgery represented the start of follow-up for nephrectomy cases. To ensure comparability with the previous analysis [6], we repeated all logistic regression, binomial regression and Cox regression models in patients aged ≥65 years.

Kaplan-Meier analyses were used to graphically depict survival rates according to the race. Actuarial survival rates were calculated with life-tables. All statistical tests were performed using the S-PLUS Professional, version 1 (Mathsoft, Seattle, WA) or the Statistical Package for Social Science, version 15.0 (SPSS, Chicago, Illinois). Moreover, all tests were two-sided with a significance level set at 0.05.

RESULTS

The patient characteristics are shown in Table 1; data are tabulated for the entire cohort and for patients aged ≥65 years. In the overall and age-restricted cohorts, relative to Whites, Black patients were more likely to have localized RCC (72.6% vs 66.9%, P < 0.001) and had smaller primary tumours (5.5 vs 5.8 cm, P = 0.003).

Table 1.  The characteristics of the 12 516 patients diagnosed with RCC between 2000 and 2004
VariableAll patients (12 516)Patients aged ≥65 years (5853)
BlacksWhitesPBlacksWhitesP
Total count1386 11 130 5405313 
Mean age, years  59.8    63.3<0.001 73.6  74.7<0.001
Males 845 (61.0)  7 011 (63.0)0.1301 (55.7)3198 (60.2)0.05
Mean tumour size, cm  5.5     5.80.003  5.4   5.70.01
n (%)
Stage  <0.001  0.005
 Localized1006 (72.6) 7 450 (66.9) 372 (68.9)3357 (63.2) 
 Regional 177 (12.8) 1 972 (17.7)  78 (14.4)1058 (19.9) 
 Metastatic 203 (14.6) 1 708 (15.3)  90 (16.7) 898 (16.9) 
Treatment type  <0.001  0.001
 Nephrectomy 1139 (82.2) 9 567 (86.0) 393 (72.8)4215 (79.3) 
 No nephrectomy 247 (17.8) 1 563 (14.0) 147 (27.2)1098 (20.7) 
Year of diagnosis or of nephrectomy  0.7  0.8
 2000 242 (17.5) 2 046 (18.4) 103 (19.1)1030 (19.4) 
 2001 279 (20.1) 2 126 (19.1) 125 (23.1)1057 (19.9) 
 2002 286 (20.6) 2 204 (19.8)  95 (17.6)1058 (19.9) 
 2003 279 (20.1) 2 371 (21.3) 106 (19.6)1052 (19.8) 
 2004 300 (21.6) 2 383 (21.4) 111 (20.6)1116 (21.0) 
Mean (range) follow-up, months  21.7 (0.1–59)    22.4 (0.1–59)0.1 21.2 (0.1–59)  25.5 (0.1–59)0.3

Regarding access to nephrectomy, in univariable logistic regression models that addressed the entire cohort, Black patients were 25% less likely to have a nephrectomy (odds ratio, OR, 0.75, P < 0.001, Table 2). After adjusting for all the covariates, Black patients were half as likely to have a nephrectomy (OR 0.52, P < 0.001). The same univariable and multivariable OR and P values were obtained when the cohort was restricted to older patients. Moreover, the RR calculated in univariable and multivariable binomial regression models, and the P values exactly replicated the results obtained in logistic regression models.

Table 2.  Univariable and multivariable logistic regression models predicting the probability of nephrectomy after a diagnosis of RCC and univariable and multivariable Cox regression models testing the effect of race (Black vs White) on all-cause mortality after a diagnosis or nephrectomy for RCC
ModelAll patients, OR or HR (95% CI); P≥65 years, OR or HR (95% CI); P
UnivariableMultivariable*UnivariableMultivariable*
  • *

    Covariates in all multivariable analyses consisted of: age categorized into decades, gender, tumour size continuously coded, stage and year of diagnosis or of nephrectomy.

Logistic regression predicting the probability of nephrectomy
No. of patients12 516 5853 
0.75 (0.65–0.87); <0.0010.52 (0.43–0.63); <0.0010.69 (0.57–0.85); <0.0010.53 (0.43–0.63); <0.001
Cox regression predicting testing the effect of race on all-cause mortality
Testing the effect of race in all patients (nephrectomy and no nephrectomy)
No. of patients12 516 1.16 (1.05–1.29); <0.011.09 (0.94–1.27); 0.245853 1.03 (0.89–1.20); 0.68 0.81 (0.64–1.03); 0.09
Testing the effect of race in patients treated with nephrectomy
No. of patients10 706 4608 
1.10 (0.95–1.27); 0.180.97 (0.77–1.21); 0.770.86 (0.68–1.09); 0.210.97 (0.77–1.22) 0.78
Testing the effect of race in patients treated without nephrectomy
No. of patients1 810 1.08 (0.92–1.26); 0.331.13 (0.97–1.32); 0.121245 0.99 (0.80–1.21); 0.90 1.06 (0.86–1.30); 0.57

For overall survival, in univariable Cox regression models that addressed the entire cohort, there was a 1.16-fold higher rate of death in Black patients than in Whites (hazard ratio, HR, 1.16, P < 0.01, Table 2; Fig. 1A). This effect dissipated after adjusting for all the covariates (HR 1.09, P = 0.24). When the analyses were repeated in the cohort of older patients (≥65 years old), the effect of race was not statistically significant (HR 1.03, P = 0.68; Fig. 1B) and independent predictor status (0.81; P = 0.09) in, respectively, univariable and multivariable models.

Figure 1.

Kaplan-Meier survival curves depicting the overall survival for all 12 516 patients and for those aged ≥65 years (5853 patients) stratified according to race (Black vs White, A and B, respectively) and stratified according to race and treatment type (nephrectomy vs no-nephrectomy, C and D, respectively).

In the overall survival analyses in patients treated with nephrectomy, univariable Cox regression models that addressed patients of all ages showed no statistical significance of race (HR 1.10, P = 0.18, Table 2; Fig. 1C,D). Multivariable Cox regression models also showed no independent predictor status for the variable defining race (HR 0.97, P = 0.77). There was also no statistical significance (HR 0.86, P = 0.21) and no independent predictor status (HR 0.97, P = 0.78) in the same analyses on older nephrectomy patients (≥65 years old).

In the overall survival analyses of the patients not treated with nephrectomy, univariable Cox regression models for all ages showed no statistical significance of race (HR 1.08, P = 0.33, Table 2; Fig. 1C,D). Multivariable Cox regression models also showed no independent predictor status for the variable defining race (HR 1.13, P = 0.12). There was also no statistical significance (HR 0.99, P = 0.90) or independent predictor status (HR 1.06, P = 0.57) in the same analyses on older patients not treated with nephrectomy.

DISCUSSION

A recent report based on the 1986–99 SEER database population showed that access to nephrectomy varies according to race [6]. Specifically, Black patients were 10% less likely to be treated with nephrectomy than their White counterparts. Moreover, the same analysis showed that the survival of Black patients diagnosed with RCC was inferior to that of Whites, even after adjusting for socio-economic status, comorbidity and cancer stage. In patients treated with nephrectomy, excess mortality was also associated with Black race and persisted after adjusting for potential confounders [6].

Racial differences in access to care and subsequent survival are worrisome and prompted us to examine access to nephrectomy and the mortality trends in the 2000–2004 version of the SEER database. We hypothesized that there were no racial differences in the rate of nephrectomy in the contemporary (2000–2004) SEER database, and that the survival after nephrectomy or non-surgical management of RCC is also unaffected by race.

To maximize the sample size, the present analysis was not restricted to patients aged ≥65 years, as was done in the previous analysis. Instead, we examined the entire adult SEER cohort, but to allow comparisons with the previous analyses, we also analysed a subset with age restricted to ≥65 years.

There were two main findings: First, access to nephrectomy was not improved but actually decreased for Black patients, relative to the previous findings [6]; second, there were no survival differences according to race in all the analyses, where diagnosis or nephrectomy vs no nephrectomy were used as starting points.

For access to nephrectomy, contemporary Black patients were 30% less likely to have a nephrectomy than their White counterparts. The racial disparity in access to nephrectomy increased even further (disfavouring Black patients by 50%), when age, gender, stage, tumour size and year of diagnosis or of nephrectomy were considered. These results indicate an even lower access to nephrectomy for Black patients than reported in the more historic SEER cohort. The same results were obtained in logistic and in binomial regression models.

Although the 50% lower rate of nephrectomy recorded in Black patients relative to Whites might appear alarming, the second part of the present analysis showed that the racial differences in access to nephrectomy are not associated with suboptimal survival outcomes. We used three separate analyses to arrive at this conclusion. First, we examined the effect of race on overall survival using the date of RCC diagnosis or of nephrectomy as the starting point. This analysis showed that Black patients fared equally well to White patients. The second analysis was restricted to nephrectomy patients only and used the nephrectomy date as the starting point. Again, Black patients fared equally well to Whites. The third and final analysis was restricted to patients not treated with nephrectomy and used the diagnosis date as the starting point. As before, Black patients’ survival was no different to that of Whites. The three analysis steps were then repeated in older patients; again, race had no bearing on survival in all analyses. Taken together, our observations on access to nephrectomy and on survival after nephrectomy indicate that once eligibility (age, comorbidity, etc.) for nephrectomy is established, race has no effect on survival.

The racial disparities in access to nephrectomy might therefore be attributed to various pre-surgical considerations, possibly including chronological age, remaining life-expectancy and comorbidities. These considerations might bar access to nephrectomy to a greater extent in Black patients and represent confounders of the association between race and nephrectomy rate. The most prominent of these confounders, comorbidity, was unavailable in the current dataset and could not be examined. Changes in comorbidity patterns between the current cohort and the previous SEER cohort might have contributed to the more important racial disparity in access to nephrectomy in the current analysis. According to this argument, contemporary Black patients might be sicker than their historic counterparts. Alternatively, the contemporary clinicians might be better at recognizing important comorbidities in Black patients that could undermine their well-being and survival after nephrectomy. This attitude resulted in more restrictive access to nephrectomy in Black patients, as well as in better overall survival after nephrectomy, relative to previous analysis. Specifically, it is striking that at 5 years the survival rates of White and Black patients were, respectively, 35% and 30% in the study of Berndt et al.[6] study, and 56.6% and 56.4% in the present study. Similarly, in patients not treated with nephrectomy, the survival rates were, respectively, 60% and 54% in Berndt et al.[6] and 68% and 71% in the present study. Taken together, our findings indicate that clinicians’ selection criteria for or against nephrectomy are valid from a racial perspective, as the delivery of nephrectomy or of non-surgical therapy does not result in racial disparities in survival.

Our analysis represents the third formal assessment of racial disparities in the field of RCC [6,11]. Despite the paucity of RCC-specific analyses, several studies addressed racial disparities in other malignancies, e.g. breast and colon cancer [12,13]. These studies showed worse survival of Black patients than Whites [12,13]. However, survival differences for breast and colon cancers could not be solely explained by access to care [12,13]. Conversely, racial disparities in survival for early-stage lung cancer were attributed to a lower rate of surgical treatment [14]. Our analysis differs from the previous analyses, as we have shown that racial disparities in access to nephrectomy are unrelated to survival differences.

More detailed analyses of the relationship between race and nephrectomy rate and survival, as reported by Berndt et al.[6], where the SEER data were linked to Medicare data, to examine additional variables that might better explain the rationale for the observed racial disparity in nephrectomy access, are warranted. However, as was shown by Berndt et al.[6], access to comorbidity scores does not guarantee the ability to fully explain the nephrectomy selection patterns. The performance status, and laboratory values, jointly termed the Motzer criteria [15–18], represent established predictors of prognosis, especially in metastatic RCC, and might help in elucidating the link between race and nephrectomy rate. Finally, the date of diagnosis should be used in future studies for surgical cases. The use of date of surgery and date of diagnosis for, respectively, surgical and non-surgical cases, might introduce a lead-time bias that favours the non-surgical cases.

The implications of our findings are that Black race should not be regarded as a barrier to nephrectomy in the USA. Instead, Black race might be interpreted as a proxy of comorbidity and other characteristics, which are used by clinicians to discriminate between good and suboptimal candidates for nephrectomy. The use of these selection criteria results in balanced survival between Black and White patients if either nephrectomy or no nephrectomy is selected. However, this explanation, as well as others, is speculative, as we had no definitive data (comorbidity) to support them. Consequently, we can hypothesise that Black race and comorbidity might portend poorer surgical outcomes, but these variables can only be definitively tested within a study where these variables are available.

In conclusion, our report represents the third analyses of racial disparities in patients with RCC. Unlike the two previous analyses, that were based on more historic patients, our analysis focuses on the most contemporary patient cohort. It shows that race affects access to nephrectomy, but has no bearing on survival in patients treated with nephrectomy or not.

ACKNOWLEDGEMENTS

Pierre I. Karakiewicz is partially supported by the University of Montreal Health Center Urology Associates, Fonds de la Recherche en Santé du Quebec, the University of Montreal Department Of Surgery and the University of Montreal Health Center (CHUM) Foundation. Laurent Zini is partially supported by the Association pour la Recherche sur le Cancer, the Fondation de France- Fédération Nationale des Centres de Lutte Contre le Cancer, the Association Française d’Urologie and the Ministère Français des Affaires Etrangères et Européennes (Bourse Lavoisier).

CONFLICT OF INTEREST

None declared.

Ancillary