SEARCH

SEARCH BY CITATION

Keywords:

  • lymph node dissection;
  • nephrectomy;
  • renal cell carcinoma;
  • survival;
  • extent

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Patients and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Conflict of Interest
  9. References

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

  • A recent population-based analysis suggested a potential survival benefit with respect to performing lymph node dissection at nephrectomy in node-positive patients with RCC.
  • The findings of the present study failed to corroborate the association of a survival benefit with the performance of lymph node dissection at nephrectomy.

Objective

  • Previous studies showed no survival benefit with respect to performing lymph node dissection (LND) at nephrectomy, whereas a recent population-based analysis suggested otherwise, although the latter relied on imputation. To reconcile the findings of that study by critically evaluating the handling of missing data.

Patients and Methods

  • Study participants comprised patients diagnosed with non-metastatic renal cell carcinoma (RCC) of all stages who underwent LND at nephrectomy (n = 10 596).
  • Multivariable Cox regression models were performed to predict cancer-specific mortality (CSM), where the primary variable of interest was the extent of LND.
  • To examine differences in approaches with respect to handling missing data, separate analyses were performed: (i) imputed population; (ii) exclusion of patients with missing data; and (iii) inclusion of patients with missing data as a sub-category.

Results

  • Overall, 2916 (28%) patients had missing tumour grade.
  • In multivariable analyses, our findings showed that increasing the extent of LND was associated with a significant protective effect on CSM in patients with pN1 after imputation (hazard ratio [HR], 0.82; P = 0.04).
  • By contrast, the extent of LND was no longer significantly associated with a lower risk of CSM after excluding patients with a missing tumour grade (HR, 0.83; P = 0.1) or when including patients with missing tumour grade as a sub-category (HR, 0.82; P = 0.05).

Conclusions

  • The findings of the present study failed to corroborate the association of a survival benefit with increasing extent of LND at nephrectomy.
  • The different methodologies employed to account for missing data may introduce important biases.
  • Such considerations are non-negligible with respect to the interpretation of results for investigators who rely on administrative cohorts.

Abbreviations
CSM

cancer-specific mortality

LND

lymph node dissection

SEER

Surveillance, Epidemiology and End Results

Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Patients and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Conflict of Interest
  9. References

The effect of lymph node dissection (LND) on the natural history of treated RCC is controversial. To date, a randomized trial showed no benefit of LND on survival [1]. In the setting of low-risk disease, previous population-based analyses of the Surveillance, Epidemiology, and End Results (SEER) data suggest the same lack of benefit [2, 3]. In high-risk RCC, only one institutional study showed improved survival [4], reporting that the survival of patients with lymph node metastases treated without LND was significantly worse than for patients treated with LND at cytoreductive nephrectomy.

Recently, Whitson et al. [5] performed a contemporary analysis of the SEER data showing that increasing the extent of LND at nephrectomy confers a survival benefit in patients with lymph node metastases. Although this finding is in direct disagreement with previous SEER analyses [2, 3], the study by Whitson et al. [5] also relied on multiple imputation, which is defined as a means of statistically filling in missing information with plausible values [6]. Specifically, patients with missing tumour grade information had their grade information ‘guesstimated’. Although this method is appealing in that dealing with missing data can be problematic for scientific research, the shortcomings of such data manipulation should not be ignored [7]. Moreover, the use of imputation on tumour grade is more disconcerting given that a previous study suggested that the inference of tumour grade using patient and tumour characteristics was poor [8].

In the context of conflicting messages between the novel data presented by Whitson et al. [5] suggesting a survival benefit with an increasing extent of LND and previous population-based data indicating the absence of such benefit, we aimed to perform a detailed re-analysis of this association by employing the same data set as that used by Whitson et al. [5]. Our analyses fitted models on imputed and non-imputed cohorts. Our hypothesis stated that the handling of missing data may be associated with significant variations in the results, which can lead to important systematic biases.

Patients and Methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Patients and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Conflict of Interest
  9. References

Data Source

The study cohort consisted of individuals who were diagnosed with RCC (International Classification of Disease for Oncology C67.0–C67.9) from the SEER database reported by the National Cancer Institute statistics programme, between 1988 and 2008. The SEER routinely collects patient demographics and publishes cancer incidence and survival data from population-based cancer registries covering ≈26% of the US population.

Study Population

Patients diagnosed with non-metastatic RCC of all stages treated with a LND in addition to nephrectomy between 1988 and 2008 were abstracted. Patients aged <18 years were removed from the analyses. Only patients with clear cell, papillary, chromophobe, collecting duct, medullary, granular and sarcomatoid histological subtypes were included. Patients with missing tumour size, pathological tumour stage and/or nodal stage were excluded from the analyses. Death certificate only or autopsy cases were removed from the analysis. This resulted in 10 596 patients being available for assessment.

Description of Covariates

The primary variable of interest, as reported by Whitson et al. [5], was the number of lymph nodes removed/examined at LND. This information was derived from the SEER variable ‘eod10lnexam’ and was examined both as a discrete variable and categorized into quartiles: 1–5, 6–10, 11–15 and >15. Nodal involvement was defined using the SEER-derived number of positive lymph nodes variable (‘eod10posnodes’) and categorized as pN0 (all examined nodes were negative) and pN1 (at least one positive lymph node).

Baseline characteristics included patient age, sex (male, female), race (white, black, other), marital status at diagnosis (married, single, unmarried, unknown) and year of diagnosis. Histological subtype was classified into clear cell (clear cell and granular), papillary, chromophobe and other (collecting duct, medullary, sarcomatoid). Tumour size was coded as a continuous variable (cm). Pathological tumour stage was defined as pT1a, pT1b, pT2, pT3a, pT3b/c and pT4. Tumour grade was categorized as G1, G2, G3, G4 and GX.

Statistical Analysis

Frequencies and proportions, as well as means, medians and interquartile ranges, were reported for categorical and continuously coded variables, respectively. Chi-square and Mann–Whitney U-tests were used to compare proportions and medians, respectively. For the purpose of the present analyses, three separate cohorts were examined and defined in accordance with approaches for handling missing tumour grade information. The first sub-cohort was identified by re-applying the methodology used by Whitson et al. [5]. Specifically, we relied on ordered logistic regression to impute tumour grade information in patients with missing grade (GX), using sex, tumour size, tumour and nodal stages (function ‘polr’; R statistical package ‘MASS’; R Foundation for Statistical Computing, Vienna, Austria). The second sub-cohort was identified by excluding all patients with unknown tumour grade (GX). The third sub-cohort was identified by including patients with missing tumour grade, as a sub-category, which resulted in groups: G1, G2, G3, G4 and GX.

Our primary endpoint of interest was cancer-specific mortality (CSM). Multivariable Cox regression analyses for the prediction of CSM were performed in the three separate sub-cohorts noted above. Similar to the study by Whitson et al. [5], all analyses were fitted in individuals with (pN1) and without (pN0) lymph nodes metastases at nephrectomy. P < 0.05 (two-tailed) was considered statistically significant. Analyses were conducted using the R statistical package, version 2.15.0 (R Foundation for Statistical Computing).

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Patients and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Conflict of Interest
  9. References

Baseline Descriptives

Between 1988 and 2008, 10 596 patients who underwent a LND at nephrectomy were identified (Table 1). Baseline descriptives of patients were stratified according to nodal status (pN0 vs pN1). The rate of pN1 was 13%. Out of all the patients, 2916 (28%) patients had a missing tumour grade. Patients with missing tumour grade information had a greater probability of harbouring larger tumours (mean 7.6 vs 6.9 cm; P < 0.001), a more advanced tumour extent (≥T3 RCC: 37% vs 27%; P < 0.001), a non-clear cell histological subtype (11% vs 9%; P < 0.001) and positive lymph node metastases at nephrectomy (12% vs 7%; P < 0.001) relative to their lower grade (G1–2) counterparts. In univariable Cox regression analyses, patients with missing tumour grade information had a significantly greater probability of CSM than their low-grade counterparts (hazard ratio [HR], 1.7; P < 0.001, data not shown).

Table 1. Descriptive characteristics of patients who underwent lymphadenectomy at nephrectomy, stratified according to nodal status (n = 10 596).
CharacteristicpN0pN1P
  1. a

    Includes separated, widowed and divorced. GX, grade unknown; IQR, interquartile range; LN, lymph node.

Patients, n (%)9196 (87)1400 (13)
Age (years)  0.008
Mean (median)60 (60)61 (62)
IQR51–6952–70
Sex, n (%)  <0.001
Male5762 (63)959 (69)
Female3434 (37)441 (32)
Race, n (%)  <0.001
White7784 (85)1136 (81)
Black762 (8)176 (13)
Other650 (7)88 (6)
Marital status, n (%)  0.788
Married6065 (66)930 (66)
Never married1237 (14)188 (13)
Unmarrieda1633 (18)249 (18)
Unknown261 (3)33 (2)
Tumour size (cm)  <0.001
Mean (median)7.4 (7.0)9.4 (9.0)
IQR4.5–9.56.5–12.0
Tumour grade, n (%)  <0.001
G1873 (10)35 (3)
G23177 (35)256 (18)
G32020 (22)477 (34)
G4556 (6)286 (20)
GX2570 (28)346 (25)
Histological subtype, n (%)  <0.001
Clear cell8288 (90)1112 (79)
Papillary446 (5)154 (11)
Chromophobe331 (4)26 (2)
Other131 (1)108 (8)
Tumour stage, n (%)  <0.001
T1a1768 (19)55 (4)
T1b2177 (24)146 (10)
T22112 (23)218 (16)
T3a1359 (15)381 (27)
T3b/c1601 (17)476 (34)
T4179 (2)124 (9)
Number of LNs removed  0.001
Mean (median)5 (2)5 (3)
IQR1–61–7
Number of LNs, n (%)  0.602
1–56776 (74)990 (71)
6–101276 (14)209 (15)
11–15526 (6)101 (7)
>15618 (7)100 (7)
Number of positive LNs 
Mean (median) 3 (1)
IQR 1–3

Tumour Grade Distribution and Handling of Missing Data

In the first sub-cohort (n = 10 596), patients with missing tumour grade information (n = 2916) had their grade imputed. This resulted in a G1, G2, G3 and G4 distributionof 9%, 52%, 31% and 8%, respectively. In the second sub-cohort (n = 7680), patients with missing tumour grade information were removed, which resulted in a G1, G2, G3 and G4 distribution of 12%, 45%, 33% and 11%. In the third sub-cohort (n = 10 596), patients with missing tumour grade were included as a sub-category, which resulted in a G1, G2, G3, G4 and GX distribution of 9%, 32%, 24%, 8% and 28%.

Results of Multivariable Analyses Using Imputation for Missing Tumour Grade

In multivariable analyses that focused on grade-imputed patients without lymph node metastases (n = 10 596), the extent of LND had no statistically significant effect on survival (HR, 0.96, 95% CI, 0.84–1.10; P = 0.550) (Table 2). In comparison, in patients with positive lymph nodes, increasing the extent of LND (per 10 node increase) exerted a protective statistically significant effect on CSM (HR, 0.82, 95% CI, 0.68–0.99; P = 0.04).

Table 2. Multivariable Cox regression analyses for the prediction of cancer-specific mortality amongst all non-metastatic RCC patients who underwent lymph node dissection at nephrectomy, where a missing tumour grade was imputed.
VariableNode-negativeNode-positive
HR (95% CI)*PHR (95% CI)*P
  1. HR, hazard ratio; ref, referent; CI: confidence interval; cm: centimeter.

Number of nodes removed (per 10-node increase)0.96 (0.84–1.10)0.60.82 (0.68–0.99)0.04
Age at diagnosis (years)1.01 (1.01–1.02)<0.0011.01 (1.00–1.02)<0.001
Sex    
Male1.0 (ref) 1.0 (ref) 
Female1.01 (0.91–1.12)0.91.05 (0.90–1.22)0.5
Race    
White1.0 (ref) 1.0 (ref) 
Black1.01 (0.82–1.23)0.91.23 (0.98–1.53)0.1
Other0.76 (0.61–0.94)0.010.96 (0.72–1.28)0.8
Marital status    
Married1.0 (ref) 1.0 (ref) 
Single0.95 (0.81–1.11)0.51.16 (0.92–1.45)0.2
Previously married1.12 (0.98–1.28)0.11.02 (0.84–1.23)0.9
Unknown0.78 (0.55–1.11)0.20.54 (0.33–0.88)0.01
Tumour size (cm)1.002 (1.001–1.003)<0.0011.001 (0.999–1.003)0.2
Tumour stage    
pT1a1.0 (ref) 1.0 (ref) 
pT1b2.17 (1.65–2.85)<0.0011.26 (0.76–2.10)0.4
pT23.66 (2.81–4.77)<0.0011.21 (0.73–2.01)0.5
pT3a4.79 (3.67–6.26)<0.0012.07 (1.29–3.34)0.003
pT3b/c8.12 (6.25–10.56)<0.0012.44 (1.51–3.94)<0.001
pT416.43 (11.90–22.67)<0.0014.13 (2.47–6.92)<0.001
Tumour grade    
G11.0 (ref) 1.0 (ref) 
G21.55 (1.22–1.98)<0.0011.63 (0.94–2.84)0.9
G32.28 (1.78–2.93)<0.0012.31 (1.35–3.96)0.002
G43.68 (2.77–4.89)<0.0013.47 (2.00–6.02)<0.001
Year of diagnosis0.97 (0.96–0.98)<0.0010.98 (0.97–0.99)0.002

Results from Multivariable Analyses on Individuals with Available Tumour Grade

In multivariable analyses that excluded patients with missing tumour grade (n = 7680), increasing extent of LND failed to predict CSM in pN0 patients (HR, 0.89, 95% CI, 0.75–1.04; P = 0.1) (Table 3). Similarly, among patients with the presence of positive lymph nodes, increasing the extent of LND (per 10 node increase) did not achieve independent predictor status (HR, 0.83, 95% CI, 0.66–1.05; P = 0.1).

Table 3. Multivariable Cox regression analyses for the prediction of cancer-specific mortality amongst all non-metastatic renal cell carcinoma patients who underwent lymph node dissection at nephrectomy, where patients with missing tumour grade information were removed from the analyses.
VariableNode-negativeNode-positive
HR (95% CI)*PHR (95% CI)*P
  1. HR, hazard ratio; ref, referent; CI: confidence interval; cm: centimeter.

Number of nodes removed (per 10-node increase)0.86 (0.75–1.04)0.10.83 (0.66–1.05)0.1
Age at diagnosis (years)1.011 (1.006–1.017)<0.0011.009 (1.003–1.016)0.006
Sex    
Male1.0 (ref) 1.0 (ref) 
Female0.96 (0.84–1.09)<0.0011.10 (0.92–1.32)0.3
Race    
White1.0 (ref) 1.0 (ref) 
Black0.77 (0.59–1.01)0.11.19 (0.92–1.54)0.2
Other0.89 (0.70–1.13)0.30.98 (0.72–1.34)0.9
Marital status    
Married1.0 (ref) 1.0 (ref) 
Single0.96 (0.79–1.16)0.71.08 (0.83–1.41)0.6
Previously married1.16 (0.99–1.37)0.11.13 (0.91–1.42)0.3
Unknown0.80 (0.53–1.21)0.30.43 (0.23–0.81)0.009
Tumour size (cm)1.003 (1.002–1.004)<0.0011.001 (0.999–1.004)0.3
Tumour stage    
pT1a1.0 (ref) 1.0 (ref) 
pT1b2.31 (1.63–3.28)<0.0011.21 (0.67–2.17)0.5
pT23.59 (2.54–5.06)<0.0011.35 (0.76–2.40)0.3
pT3a4.47 (3.17–6.32)<0.0012.23 (1.31–3.81)0.003
pT3b/c8.31 (5.95–11.61)<0.0012.47 (1.44–4.23)0.001
pT416.51 (11.09–24.56)<0.0014.43 (2.47–7.95)<0.001
Tumour grade    
G11.0 (ref) 1.0 (ref) 
G21.54 (1.19–1.98)0.0011.64 (0.94–2.88)0.1
G32.43 (1.88–3.14)<0.0012.36 (1.36–4.08)0.002
G43.91 (2.94–5.20)<0.0013.59 (2.05–6.27)<0.001
Year of diagnosis0.96 (0.95–0.97)<0.0010.98 (0.96–0.99)0.005

Results from Multivariable Analyses on Individuals with Missing Tumour Grade Included as a Sub-category

Finally, in multivariable analyses that included patients with GX (n = 10 596), increasing the extent of LND remained unassociated with CSM in pN0 patients (HR, 0.95, 95% CI, 0.83–1.09; P = 0.4) (Table 4) and pN1 patients (HR, 0.82, 95% CI, 0.68–1.01; P = 0.05).

Table 4. Multivariable Cox regression analyses for the prediction of cancer-specific mortality amongst all non-metastatic renal cell carcinoma patients who underwent lymph node dissection at nephrectomy, where patients with missing tumour grade information were included as a sub-category.
VariableNode-negativeNode-positive
HR (95% CI)*PHR (95% CI)*P
  1. HR, hazard ratio; ref, referent; CI: confidence interval; cm: centimeter.

Number of nodes removed (per 10-node increase)0.95 (0.83–1.09)0.40.82 (0.68–1.01)0.05
Age at diagnosis (years)1.012 (1.008–1.017)<0.0011.010 (1.005–1.016)<0.001
Sex    
Male1.0 (ref) 1.0 (ref) 
Female0.99 (0.89–1.10)0.81.03 (0.88–1.20)0.7
Race    
White1.0 (ref) 1.0 (ref) 
Black1.01 (0.82–1.23)0.91.21 (0.97–1.52)0.1
Other0.77 (0.62–0.95)0.020.96 (0.72–1.28)0.8
Marital status    
Married1.0 (ref) 1.0 (ref) 
Single0.93 (0.80–1.10)0.41.17 (0.93–1.46)0.2
Previously married1.11 (0.97–1.27)0.11.02 (0.84–1.23)0.9
Unknown0.78 (0.55–1.11)0.20.54 (0.33–0.89)0.02
Tumour size (cm)1.002 (1.002–1.003)<0.0011.002 (1.000–1.003)0.1
Tumour stage    
pT1a1.0 (ref) 1.0 (ref) 
pT1b2.13 (1.62–2.80)<0.0011.28 (0.77–2.13)0.3
pT23.50 (2.68–4.56)<0.0011.27 (0.77–2.12)0.4
pT3a4.74 (3.63–6.19)<0.0012.18 (1.36–3.51)0.001
pT3b/c8.82 (6.81–11.41)<0.0012.58 (1.60–4.15)<0.001
pT417.17 (12.46–23.66)<0.0014.57 (2.74–7.62)<0.001
Tumour grade    
G11.0 (ref) 1.0 (ref) 
G21.51 (1.18–1.95)0.0011.59 (0.91–2.78)0.2
G32.34 (1.81–3.02)<0.0012.27 (1.31–3.91)0.003
G43.70 (2.80–4.90)<0.0013.42 (1.96–5.95)<0.001
GX1.69 (1.32–2.17)<0.0012.33 (1.35–4.01)0.002
Year of diagnosis0.97 (0.96–0.98)<0.0010.98 (0.97–0.99)0.003

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Patients and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Conflict of Interest
  9. References

In several solid tumours, LND and/or its extent are associated with a survival benefit. For example, in colon cancer, the extent of LND directly correlates with the magnitude of its protective effect [9, 10]. A similar concept has been described in bladder cancer [11]. The notion that a more extensive LND exerts a protective effect is conceptually very attractive. The approach related to the survival benefit of LND does not limit surgeons to improving cancer control only by removing the primary tumour. Instead, cancer control may be further improved by resecting the maximal number of lymph nodes that the primary tumour may drain into.

Numerous studies have examined the effect of LND in the context of RCC. To date, several studies found LND or its extent at nephrectomy to be of no therapeutic value [1, 2, 12-14]. Of these studies, a randomized trial [1] was heavily criticized for including patients at very low risk (T1–2; 71%) of lymph node metastases. Joslyn et al. [2] examined the value of LND in the SEER database (1983–1998) and found no association between LND and/or its extent and survival (n = 4453). Trinh et al. [3] examined a more recent version of the SEER database (1988–2006) and were unable to link increasing extent of LND with the survival in patients with nodal metastases at nephrectomy (n = 799). Taken together, these reports suggest a lack of benefit of LND and/or its extent when survival after nephrectomy represents the endpoint of interest.

Despite the relatively convincing negative data, some studies have suggested that LND may improve cancer control outcomes [4, 15, 16]. Schafhauser et al. [15] retrospectively studied 1035 patients with RCC. Of these, 531 patients underwent LND at nephrectomy, where most had pN0 disease after surgery. In patients with positive lymph node metastases, 4% survived >5 years. Based on these observations, Schafhauser et al. [15] recommended a systematic LND at nephrectomy. However, given the relatively few patients who actually benefitted from LND, a large number of patients will need to undergo LND before a single patient may benefit from the procedure. Similarly, Pantuck et al. [4] found a survival benefit in 112 pN1M1 patients who underwent LND compared to 17 other patients who did not (P = 0.002). However, the findings reported by Pantuck et al. [4] originate from a relatively small patient cohort treated at a single institution.

Most recently, Whitson et al. [5] reported the association between the extent of LND and survival in patients with established lymph node metastases. In their analyses, which relied on the same study cohort as that of Trinh et al. [3], increasing the extent of LND exerted a protective effect on survival. Their observation is based on a very large and highly generalizable patient cohort. However, the conflicting results relative to those of Trinh et al. [3], who used an almost identical database, raise concern about potential methodological issues that could reverse the direction of the association between LND and survival relative to previous studies. Specifically, Whitson et al. [5] relied on multiple imputation [6], which represents a valid approach for handling missing data. Nonetheless, this approach is not immune to pitfalls, which should be evaluated carefully when interpreting the results of such analyses. Furthermore, the rationale for claiming that the use of imputation is even less suitable in this setting is justified for two reasons. First, previous work by Trinh et al. [3], where imputation was not used, relied on the same data and failed to show any association between the extent of LND and survival. Second, previous attempts at predicting tumour grade using available parameters such as tumour size, sex, age and symptom classification resulted in inaccurate predictions [8, 17]. Because imputation relies on a similar methodology, its ability to correctly infer tumour grade will be equally limited.

Given the concerns noted above, we aimed to perform a detailed assessment of the hypothesis of Whitson et al. [5] with respect to the survival benefit of increasing extent of LND at nephrectomy. The analyses performed in the present study showed several important findings. First, more than one-quarter of patients had a missing tumour grade in the SEER database (28%). Moreover, patients with missing tumour grade information had a significantly greater probability of harbouring more aggressive disease at nephrectomy (i.e. a higher rate of ≥T3 disease, positive lymph node metastases and a higher risk of CSM). In the context of patients with an unknown tumour grade being predisposed to a higher risk of CSM, the handling of such individuals becomes primordial during the the evaluation of the final results. Indeed, in the subsequent part of the present analyses, the results obtained showed that, by replicating the approach of Whitson et al. [5] for handling missing data (i.e. imputation), increasing the extent of LND at nephrectomy exerted a protective effect on survival in pN1 patients, even after adjustment for all other covariates. By contrast, after the removal of patients with an unknown grade, the statistical significance of the extent of LND on survival in pN1 patients was dissipated. Similarly, no effect of the extent of LND and survival was recorded when patients with a missing tumour grade were included (GX) as a sub-category.

In essence, the re-analysis performed in the present study suggests that different approaches for handling missing data may be associated with significant variations in the results obtained, which may lead to important biases. Such biases could either obliterate existing relationships or create statistically significant associations that have not existed previously. Based on these considerations, the use of statistical tools aimed at handling missing data should be employed with great caution. The application of such tools to large proportions of observations (28% in the present study) may result in unanticipated results. The corroboration of findings with other studies that have relied on the same population becomes paramount in such a setting.

The results obtained in the present study are not meant to discredit the use of imputation. Rather, they are meant to highlight potential shortcomings in the handling of missing data. Undermining the necessary cautions that should be taken may result in conflicting data and misinterpretation. From a practical perspective, the present study did not intend to refute the value of LND at nephrectomy in high-risk patients. The current consensus of the urological community appears to be that universal LND is not recommended, especially in the context of low-risk disease. On the other hand, as recommended by national guidelines [18, 19], patients with adenopathy on preoperative imaging or palpable/visible-adenopathy at the time or surgery should undergo LND at nephrectomy. Future studies focusing on high-risk patients, especially those including a randomized design, may seek to refute or corroborate the presence of a survival benefit for LND at nephrectomy.

The present study is not devoid of limitations. As is applicable to all retrospective observational studies, the SEER database includes patients who were treated at different institutions across the USA, accompanied by the absence of a standardized LND template, the lack of a central pathological review and interobserver bias, as well as a lack of information on adjuvant and/or salvage treatment. Moreover, although the SEER database is a well-maintained and highly regulated cancer registry, coding misattribution and entry error may have occurred.

In conclusion, the findings of the present study failed to corroborate an associated survival benefit when performing LND at nephrectomy. The handling of missing data using methods such as imputation may result in statistically significant associations, which may be biased. Such considerations are non-negligible with respect to the interpretation of results for investigators who rely on large administrative cohorts.

Acknowledgements

  1. Top of page
  2. Abstract
  3. Introduction
  4. Patients and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Conflict of Interest
  9. References

Pierre I. Karakiewicz is partially supported by the University of Montreal Health Centre Urology Specialists, Fonds de la Recherche en Santé du Québec, the University of Montreal Department of Surgery and the University of Montreal Health Centre (CHUM) Foundation.

Conflict of Interest

  1. Top of page
  2. Abstract
  3. Introduction
  4. Patients and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Conflict of Interest
  9. References

The authors declare that there are no conflicts of interest.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Patients and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Conflict of Interest
  9. References
  • 1
    Blom JHM, van Poppel H, Maréchal JM et al. Radical nephrectomy with and without lymph-node dissection: final results of European Organization for Research and Treatment of Cancer (EORTC) randomized phase 3 trial 30881. Eur Urol 2009; 55: 2834
  • 2
    Joslyn SA, Sirintrapun SJ, Konety BR. Impact of lymphadenectomy and nodal burden in renal cell carcinoma: retrospective analysis of the National Surveillance, Epidemiology, and End Results database. Urology 2005; 65: 675680
  • 3
    Trinh Q-D, Schmitges J, Bianchi M et al. Node-positive renal cell carcinoma in the absence of distant metastases: predictors of cancer-specific mortality in a population-based cohort. BJU Int 2011; 110: E2127
  • 4
    Pantuck AJ, Zisman A, Dorey F et al. Renal cell carcinoma with retroperitoneal lymph nodes: role of lymph node dissection. J Urol 2003; 169: 20762083
  • 5
    Whitson JM, Harris CR, Reese AC et al. Lymphadenectomy improves survival of patients with renal cell carcinoma and nodal metastases. J Urol 2011; 185: 16151620
  • 6
    Schafer J. Multiple imputation: a primer. Stat Methods Med Res 1999; 8: 315
  • 7
    Horton N, Lipsitz S. Multiple imputation in practice: comparison of software packages for regression models with missing variables. Am Stat Assoc 2001; 55: 244258
  • 8
    Jeldres C, Sun M, Liberman D et al. Can renal mass biopsy assessment of tumor grade be safely substituted for by a predictive model? J Urol 2009; 182: 25852589
  • 9
    Chen S, Bilchik A. More extensive nodal dissection improves survival for stages I to III of colon cancer. Ann Surg 2006; 244: 602610
  • 10
    Le Voyer T, Sigurdson E, Hanlon AL et al. Colon cancer survival is associated with increasing number of lymph nodes analyzed: a secondary survey of intergroup trial INT-0089. J Clin Oncol 2003; 21: 29122919
  • 11
    Shirotake S, Kikuchi E, Matsumoto K et al. Role of pelvic lymph node dissection in lymph node-negative patients with invasive bladder cancer. Jpn J Clin Oncol 2010; 40: 247251
  • 12
    Minervini A, Lilas L, Morelli G et al. Regional lymph node dissection in the treatment of renal cell carcinoma: is it useful in patients with no suspected adenopathy before or during surgery? BJU Int 2001; 88: 169172
  • 13
    Trinh Q-D, Saad F, Lattouf J-B. The current management of small renal masses. Curr Opin Support Palliat Care 2009; 3: 180185
  • 14
    Pizzocaro G, Piva L, Salvioni R. Lymph node dissection in radical nephrectomy for renal cell carcinoma: is it necessary? Eur Urol 1983; 9: 1012
  • 15
    Schafhauser W, Ebert A, Brod J et al. Lymph node involvement in renal cell carcinoma and survival chance by systematic lymphadenectomy. Anticancer Res 1999; 19: 15731578
  • 16
    Canfield SE, Kamat AM, Sánchez-Ortiz RF et al. Renal cell carcinoma with nodal metastases in the absence of distant metastatic disease (clinical stage TxN1-2M0): the impact of aggressive surgical resection on patient outcome. J Urol 2006; 175: 864869
  • 17
    Lane BR, Babineau D, Kattan MW et al. A preoperative prognostic nomogram for solid enhancing renal tumors 7 cm or less amenable to partial nephrectomy. J Urol 2007; 178: 429434
  • 18
    National Comprehensive Cancer Network. Kidney cancer. NCCN Clinical Practice Guidelines in Oncology, 2011. Available at: www.nccn.org Accessed 20 July 2012
  • 19
    Ljungberg B, Cowan NC, Hanbury DC et al. EAU guidelines on renal cell carcinoma: the 2010 update. Eur Urol 2010; 58: 398406