- Email: [email protected]
Perioperative Morbidity of Lymph Node Dissection for Renal Cell Carcinoma: A Propensity Score–based Analysis
EURURO-7602; No. of Pages 7 E U R O P E A N U R O L O G Y X X X ( 2 0 17 ) X X X – X X X
available at www.sciencedirect.com journal homepage: www.europeanurology.com
Kidney Cancer
Perioperative Morbidity of Lymph Node Dissection for Renal Cell Carcinoma: A Propensity Score–based Analysis Boris Gershman a,*, Daniel M. Moreira b, R. Houston Thompson c, Stephen A. Boorjian c, Christine M. Lohse d, Brian A. Costello e, John C. Cheville f, Bradley C. Leibovich c a
Division of Urology, Rhode Island Hospital and The Miriam Hospital, Providence, RI, USA; b Department of Urology, University of Illinois, Chicago, IL, USA;
c
Department of Urology, Mayo Clinic, Rochester, MN, USA;
d
Department of Health Sciences Research, Mayo Clinic, Rochester, MN, USA; e Department of
Oncology, Mayo Clinic, Rochester, MN, USA; f Department of Pathology, Mayo Clinic, Rochester, MN, USA
Article info
Abstract
Article history: Accepted October 23, 2017
Background: There are little data regarding the morbidity of lymph node dissection (LND) for renal cell carcinoma (RCC) to assess its risk–benefit ratio. Objective: To evaluate the association of LND with 30-d complications among patients undergoing radical nephrectomy (RN) for RCC. Design, setting, and participants: A total of 2066 patients underwent RN for M0 or M1 RCC between 1990 and 2010, of whom 774 (37%) underwent LND. Intervention: RN with or without LND. Outcome measurements and statistical analysis: Associations of LND with 30-d complications were examined using logistic regression with several propensity score techniques. Extended LND, defined as removal of 13 lymph nodes, was examined in a sensitivity analysis. Results and limitations: A total of 184 (9%) patients were pN1 and 302 (15%) were M1. Thirty-day complications occurred in 194 (9%) patients, including Clavien grade 3 complications in 81 (4%) patients. Clinicopathologic features were well balanced after propensity score adjustment. In the overall cohort, LND was not statistically significantly associated with Clavien grade 3 complications, although there was an approximately 40% increased risk of any Clavien grade complication that did not reach statistical significance. Likewise, LND was not significantly associated with any Clavien grade or Clavien grade 3 complications when separately evaluated among M0 or M1 patients. Extended LND was not significantly associated with any Clavien grade or Clavien grade 3 complications. LND was not associated with length of stay or estimated blood loss. Limitations include a retrospective design. Conclusions: LND is not significantly associated with an increased risk of Clavien grade 3 complications, although it may be associated with a modestly increased risk of minor complications. In the absence of increased morbidity, LND may be justified in a predominantly staging role in the management of RCC. Patient summary: Lymph node dissection for renal cell carcinoma is not associated with increased rates of major complications. © 2017 European Association of Urology. Published by Elsevier B.V. All rights reserved.
Associate Editor: Giacomo Novara Statistical Editor: Andrew Vickers Keywords: Lymph node dissection Complications Nephrectomy Renal cell carcinoma Propensity score
* Corresponding author. Rhode Island Hospital and The Miriam Hospital, Warren Alpert Medical School of Brown University, 195 Collyer Street, Suite 201, Providence, RI 02904, USA. Tel. +1 401 272 7799; Fax: +1 401 453 9078. E-mail address: [email protected] (B. Gershman). https://doi.org/10.1016/j.eururo.2017.10.020 0302-2838/© 2017 European Association of Urology. Published by Elsevier B.V. All rights reserved.
Please cite this article in press as: Gershman B, et al. Perioperative Morbidity of Lymph Node Dissection for Renal Cell Carcinoma: A Propensity Score–based Analysis. Eur Urol (2017), https://doi.org/10.1016/j.eururo.2017.10.020
EURURO-7602; No. of Pages 7 2
E U R O P E A N U R O L O G Y X X X ( 2 0 17 ) X X X – X X X
1.
Introduction
The role of lymph node dissection (LND) in the surgical management of renal cell carcinoma (RCC) remains uncertain. Although there are little data to support an oncologic benefit to LND at the time of radical nephrectomy (RN) for RCC, it provides pathologic assessment of nodal stage [1–7]. This may provide valuable prognostic information given the association of lymph node metastasis with survival [6,8–10], and it may identify candidates for consideration of adjuvant therapies [11]. In this context, proponents have argued that LND may have a role in the surgical management of RCC by providing valuable staging information, and guidelines have recommended LND in the setting of radiographic lymphadenopathy [ 1,12,13]. Moreover, it has been suggested that LND may provide a survival benefit to a small subset of patients with lymph node metastases [14–17]. An accurate assessment of the risks of LND is critical to determine its risk–benefit ratio. Yet despite numerous studies examining a therapeutic benefit to LND, there is a paucity of data regarding its morbidity [2,3,6,18,19]. We, therefore, examined the perioperative morbidity of LND among patients undergoing RN for RCC. Specifically, in this study, we characterized 30-d complications among patients undergoing RN with or without LND, and we evaluated the associations of LND with 30-d complications using a propensity score–based approach to adjust for nonrandom treatment allocation.
2.3.
Statistical methods
Clinicopathologic and radiographic features were summarized with medians and interquartile ranges (IQRs) or frequency counts and percentages, and were compared for patients with and without LND using Wilcoxon rank sum, chi-square, and Fisher exact tests. Propensity scores (PSs) for treatment with LND were estimated using a logistic regression model with LND as the outcome and the features listed in Table 1 as covariates, except body mass index (BMI), estimated blood loss (EBL), and length of stay. BMI was excluded because of the amount of missing data and its lack of a statistically significant association with LND. Year of surgery and age at surgery were categorized into four groups of approximately equal patient numbers since there was evidence that the associations of these features with LND were not linear. Forty-nine patients missing covariate data (16 with and 33 without LND) were excluded from the PS analytic cohort, as were 91 patients with a PS that did not fall within the common range (50 with and 41 without LND). A total of 1926 patients formed the cohort for PS analyses, including 708 (37%) who underwent LND and 1218 (63%) who did not. The associations of LND with any complication or a Clavien grade 3 complication were evaluated with logistic regression models using several PS techniques, and summarized with odds ratios (ORs) and 95% confidence intervals (CIs) [21]. Patients with LND were matched 1:1 to patients without LND using the caliper method based on the logit of the PS and a caliper width of 0.2 times the standard deviation of the logit [22]. Using this approach, 457 patients with LND were matched to 457 patients without LND. To incorporate all 1926 patients in the analyses, we conducted PS analyses reweighting by stabilized inverse probability weights (IPWs). Additional PS techniques were employed including adjustment for PS quintile and stratification by PS quintile (Supplementary Material). ORs for the stratified analyses were obtained by pooling estimates obtained within each PS quintile. ORs for the subset of 457 matched pairs were obtained by stratifying on the matched sets.
Patients and methods
2.
Stabilized IPWs were truncated by setting weights below the first percentile to the value of the first percentile, and weights above the 99th
2.1.
Patient population
percentile to the value of the 99th percentile [23]. Models were further separately adjusted for surgeon and BMI with similar results. We further
After obtaining Institutional Review Board approval, we identified
examined the association of LND with 30-d complications among M0
2103 patients treated with RN for sporadic, unilateral, M0 or M1 RCC
and M1 patients by constructing logistic regression models that included
between 1990 and 2010. Thirty-seven patients lacked data regarding
an indicator variable for performance of LND, an indicator for M1 status,
complications within 30 d of surgery and were excluded. The final cohort
and the interaction between the two.
included 2066 patients for study, of whom 774 (37%) underwent LND.
Given heterogeneity in the extent of LND, we conducted a sensitivity
LND was performed at the surgeon's discretion, and a standardized
analysis to examine associations of extended LND, defined as removal of
template was not utilized.
13 LNs [24] (vs no LND or LND with <13 LNs removed), with complications. PSs for treatment with extended LND were obtained
2.2.
Clinicopathologic features
The clinicopathologic and radiographic features in Table 1 were assessed as described in the Supplementary Material. All pathologic features were reviewed by one urologic pathologist (J.C.C.) without knowledge of patient outcome. Indicator variables were employed to denote missing data for the radiographic features studied.
using a logistic regression model, and associations of extended LND with any complication or a Clavien 3 complication were evaluated using the PS techniques described above. Statistical analyses were performed using SAS version 9.4 (SAS Institute, Cary, NC, USA) and R version 3.1.1 (R Foundation for Statistical Computing, Vienna, Austria). All tests were two sided, and p < 0.05 was considered statistically significant.
Complications within 30 d of RN were assessed by a trained nurse abstractor using standardized definitions and graded according to the
3.
Results
Clavien classification [20]. Complication categories included bleeding (requiring reoperation or angioembolization) or hematoma (on crosssectional imaging), deep vein thrombosis (DVT), pulmonary embolism (PE), myocardial infarction (MI), wound infection (requiring antibiotics or drainage) or dehiscence (partial or complete), abscess, sepsis (bacteremia with hypotension or requiring admission to intensive care unit), acute renal failure with or without dialysis, pneumothorax, any other complication not specified above, and death. Patients were typically evaluated at 3 mo after surgery and then according to the recommended surveillance protocol based on surgical pathology.
A total of 2066 patients formed the study cohort, including 774 (37%) who underwent LND. Overall, 184 (9%) patients were pN1 and 302 (15%) had distant metastases at the time of RN. The median number of LNs removed during LND was 6 (IQR 2–13). Clinicopathologic and radiographic features stratified, by performance of LND, are summarized in Table 1. Patients who underwent LND had more advanced disease and more aggressive pathologic characteristics,
Please cite this article in press as: Gershman B, et al. Perioperative Morbidity of Lymph Node Dissection for Renal Cell Carcinoma: A Propensity Score–based Analysis. Eur Urol (2017), https://doi.org/10.1016/j.eururo.2017.10.020
EURURO-7602; No. of Pages 7 3
E U R O P E A N U R O L O G Y X X X ( 2 0 17 ) X X X – X X X
Table 1 – Clinicopathologic and radiographic features stratified by performance of LND in the overall cohort (N = 2066) Feature
Median (IQR) or number (%) No LND (N = 1292)
Year of surgery 1990–1994 1995–1999 2000–2004 2005–2010 Age at surgery (yr) 18–54 55–64 65–69 70 Sex Female Male Charlson score BMI (kg/m2; N = 1907) Tumor size (cm; N = 2056) Symptoms (N = 2065) Constitutional symptoms (N = 2065) Smoking status (N = 2036) Never Current Former ECOG performance status (N = 2065) 0 1 2 3 4 cN1 Renal vein involvement on CT/MRI IVC involvement on CT/MRI Radiographic evidence of Hemorrhage Necrosis Calcification Extrarenal extension Neovascularity Adrenal involvement Cystic or indeterminate cysts Neoadjuvant therapy Neoadjuvant systemic therapy Surgical approach Open radical nephrectomy Laparoscopic radical nephrectomy Vascular bypass during RN Concurrent metastasectomy Other concurrent surgery Estimated blood loss (cm; N = 1893) Length of stay (d) RCC histologic subtype Clear cell Papillary or chromophobe Collecting duct or NOS All others 2010 pT (N = 2052) pT1a pT1b pT2a pT2b pT3a pT3b pT3c pT4 M stage M0
p value LND (N = 774)
357 (28) 344 (27) 356 (28) 235 (18)
121 (16) 130 (17) 226 (29) 297 (38)
<0.001
289 (22) 331 (26) 219 (17) 453 (35)
224 (29) 216 (28) 141 (18) 193 (25)
<0.001
449 (35) 843 (65) 1 (0–2) 28.3 (25.2–31.9) 6.0 (4.0–8.5) 713 (55) 264 (20)
259 (33) 515 (67) 1 (0–6) 28.0 (24.8–31.7) 9.5 (7.0–12.0) 583 (75) 266 (34)
0.55
550 (43) 205 (16) 518 (41)
307 (40) 156 (20) 300 (39)
0.043
1093 (85) 153 (12) 24 (2) 18 (1) 4 (<1) 24 (2) 116 (9) 68 (5)
650 (84) 90 (12) 20 (3) 13 (2) 0 161 (21) 257 (33) 158 (20)
0.7
26 (2) 93 (7) 89 (7) 44 (3) 86 (7) 20 (2) 267 (21) 18 (1) 8 (1)
21 (3) 123 (16) 81 (10) 133 (17) 122 (16) 31 (4) 95 (12) 27 (3) 13 (2)
0.3 <0.001 0.004 <0.001 <0.001 <0.001 <0.001 0.002 0.020
1,098 (85) 194 (15) 17 (1) 32 (2) 175 (14) 250 (125–500) 5 (4–7)
735 (95) 39 (5) 29 (4) 65 (8) 200 (26) 600 (300–1200) 6 (4–7)
<0.001
1,037 (80) 233 (18) 18 (1) 4 (<1)
653 (84) 98 (13) 16 (2) 7 (1)
0.003
338 (26) 360 (28) 164 (13) 74 (6) 259 (20) 52 (4) 14 (1) 19 (1)
38 (5) 96 (12) 94 (12) 77 (10) 295 (28) 117 (15) 20 (3) 35 (5)
<0.001
1176 (91)
588 (76)
<0.001
0.025 0.6 <0.001 <0.001 <0.001
<0.001 <0.001 <0.001
<0.001 <0.001 <0.001 <0.001 <0.001
Please cite this article in press as: Gershman B, et al. Perioperative Morbidity of Lymph Node Dissection for Renal Cell Carcinoma: A Propensity Score–based Analysis. Eur Urol (2017), https://doi.org/10.1016/j.eururo.2017.10.020
EURURO-7602; No. of Pages 7 4
E U R O P E A N U R O L O G Y X X X ( 2 0 17 ) X X X – X X X
Table 1 (Continued ) Feature
Median (IQR) or number (%) No LND (N = 1292)
M1 Grade 1 2 3 4 Coagulative tumor necrosis (N = 2,065) Sarcomatoid differentiation Number of metastatic sites 0 1 2 3 4 5 Location of metastases Lung Liver Bone Brain
p value LND (N = 774)
116 (9)
186 (24)
80 (6) 593 (46) 521 (40) 98 (8) 376 (29) 51 (4)
12 (2) 160 (21) 405 (52) 197 (25) 448 (58) 77 (10)
<0.001
1176 (91) 76 (6) 29 (2) 7 (1) 4 (<1) 0
588 (76) 114 (15) 49 (6) 17 (2) 4 (1) 2 (<1)
<0.001
73 (6) 9 (1) 33 (3) 10 (1)
117 (15) 26 (3) 39 (5) 8 (1)
<0.001 <0.001 0.003 0.5
<0.001 <0.001
BMI = body mass index; CT = computed tomography; ECOG = Eastern Cooperative Oncology Group; IQR = interquartile range; IVC = inferior vena cava; LND = lymph node dissection; MRI = magnetic resonance imaging; NOS = nitric oxide synthase; RCC = renal cell carcinoma; RN = radical nephrectomy.
including a greater incidence of radiographic lymphadenopathy and distant metastases at RN. However, baseline characteristics were well balanced after PS adjustment. Specifically, no feature was statistically significantly different between patients with and without LND after adjusting for PS quintile (data not shown), the 457 matched pairs were well balanced (data not shown), and no feature was statistically significantly different between patients with and without LND after reweighting by stabilized IPWs (Supplementary Table 1). Notably, although patients who underwent LND had greater EBL and length of stay in unadjusted comparisons (Table 1), there were no statistically significant differences in these perioperative end points after reweighting by stabilized IPWs (Supplementary Table 1). Complications were diagnosed within 30 d of surgery in 194 (9%) patients, including Clavien grade 3 complications in 81 (4%) patients (Table 2). The most common complication categories were acute renal failure (3%), bleeding (2%), and wound complications (2%). Among patients who underwent LND, complications occurred more frequently in patients with pN1 disease—any grade complications occurred in 66 (11%) pN0 and 40 (22%) pN1 patients (p < 0.001), while Clavien 3 complications occurred in 25 (4%) pN0 and 18 (10%) pN1 patients (p = 0.004). In unadjusted analyses, performance of LND was associated with increased rates of any 30-d complication (14% vs 7%, p < 0.001) and Clavien grade 3 complications (6% vs 3%, p = 0.003). However, after reweighting by stabilized IPWs, there were no statistically significant differences in rates of any complication (11% vs 8%, p = 0.09) or Clavien grade 3 complications (4% vs 3%, p = 0.8; Supplementary Table 2). Notably, LND was associated with an increased rate of Clavien grade 2 complications (6% vs 3%, p = 0.04). This appeared driven by a small but
statistically significant increase in the rates of DVT (1% vs <1%, p = 0.04) and MI (2% vs <1%, p = 0.02). Next, we examined the associations of LND with 30-d complications in the overall cohort (Table 3 and Supplementary Table 3). Here, LND was not statistically significantly associated with Clavien grade 3 complications using any of the PS techniques, although there was a modestly increased risk of any Clavien grade complication (OR ranging from 1.41 to 1.46) that did not reach statistical
Table 2 – Thirty-day complications after RN stratified by performance of LND in the overall cohort All No LND LND p value (N = 2066) (N = 1292) (N = 774) Clavien grade 1 complication 67 (3) 81 (4) Clavien grade 2 complication 29 (1) Clavien grade 3 complication Clavien grade 4 complication 46 (2) 20 (1) Clavien grade 5 complication Clavien grade 3 complication 81 (4) Any complication 194 (9) Complication category 37 (2) Bleeding or hematoma DVT 26 (1) 23 (1) PE 7 (<1) MI Wound infection or dehiscence44 (2) 8 (<1) Abscess 14 (1) Sepsis Acute renal failure 66 (3) 17 (1) Dialysis 18 (1) Pneumothorax Other 19 (1) 20 (1) Death
36 (3) 27 (2) 16 (1) 20 (2) 8 (1) 38 (3) 88 (7)
31 (4) 54 (7) 13 (2) 26 (3) 12 (2) 43 (6) 106 (14)
0.1 <0.001 0.4 0.007 0.036 0.003 <0.001
17 (1) 6 (<1) 7 (1) 2 (<1) 18 (1) 6 (<1) 9 (1) 31 (2) 8 (1) 12 (1) 6 (<1) 8 (1)
20 (3) 20 (3) 16 (2) 5 (1) 26 (3) 2 (<1) 5 (1) 35 (5) 9 (1) 6 (1) 13 (2) 12 (2)
0.035 <0.001 0.001 0.11 0.003 0.7 0.9 0.008 0.2 0.7 0.005 0.036
vein thrombosis; LND = lymph node DVT = deep MI = myocardial infarction; PE = pulmonary embolism; nephrectomy.
dissection; RN = radical
Please cite this article in press as: Gershman B, et al. Perioperative Morbidity of Lymph Node Dissection for Renal Cell Carcinoma: A Propensity Score–based Analysis. Eur Urol (2017), https://doi.org/10.1016/j.eururo.2017.10.020
EURURO-7602; No. of Pages 7 E U R O P E A N U R O L O G Y X X X ( 2 0 17 ) X X X – X X X
Table 3 – Associations of LND with any Clavien grade and Clavien grade 3 30-d complications PS technique
Any Clavien grade OR (95% CI)
IPW Matched pairs
p value
1.44 (0.95–2.19) 1.46 (0.93–2.27)
0.09 0.10
OR (95% CI) 1.07 (0.59–1.92) 1.54 (0.77–3.09)
p value
0.8 0.2
Table 4 – Associations of LND with any Clavien grade and Clavien grade 3 30-d complications among M0 and M1 patients using IPW Clavien grade 3
p value
OR (95% CI) M0 M1
1.50 (0.94–2.38) 1.18 (0.48–2.90)
0.09
OR (95% CI) 1.04 (0.55–1.98) 1.24 (0.28–5.45)
0.7
p value
0.9 0.8
CI = confidence interval; IPW = inverse probability weight; LND = lymph node dissection; OR = odds ratio. Models included an indicator for LND, an indicator for M1 status, and the interaction term between the two.
Table 5 – Associations of extended LND with any Clavien grade and Clavien grade 3 30-d complications PS technique
Any Clavien grade OR (95% CI)
IPW Matched pairs
1.66 (0.81–3.40) 1.53 (0.80–2.94)
p value
0.2 0.2
these analyses, extended LND was again not associated with any complication or Clavien grade 3 complications.
Clavien grade 3
CI = confidence interval; IPW = inverse probability weight; LND = lymph node dissection; OR = odds ratio; PS = propensity score.
Any Clavien grade
5
Clavien grade 3 OR (95% CI) 0.95 (0.36–2.46) 1.43 (0.54–3.75)
p value
0.9 0.5
CI = confidence interval; IPW = inverse probability weight; LND = lymph node dissection; OR = odds ratio.
significance. In analyses further adjusted for surgeon, results were similar (Supplementary Table 4). We further evaluated the association of LND with 30-d complications among M0 and M1 patients by including an interaction term in the regression models (Table 4 and Supplementary Table 5). In these analyses, LND was not significantly associated with any complication or Clavien grade 3 complications in either M0 or M1 patients. Moreover, the p values for the interaction terms in models examining any complication or Clavien grade 3 complications were not statistically significant (p > 0.5 for each), suggesting that there was no effect of heterogeneity according to M stage. Given heterogeneity in the extent of LND, we conducted a sensitivity analysis examining extended LND, defined as removal of 13 LNs (Table 5 and Supplementary Table 6). In
4.
Discussion
In this large, single-institutional cohort, LND at the time of RN was associated with a low overall rate of any (9%) and Clavien grade 3 (4%) complications. More importantly, LND was not statistically significantly associated with an increased risk of Clavien grade 3 complications, although there was an approximately 40% increased risk of any Clavien grade complication that did not reach statistical significance. This appears to be driven by a small increase in Clavien grade 2 complications related to DVT and MI, which may have clinical relevance despite low Clavien grade. Taken together, these results suggest that LND can be performed without major perioperative morbidity, although it may be associated with a modest increase in minor postoperative complications. Moreover, subset analyses indicate that the morbidity of LND does not differ between patients with nonmetastatic or metastatic RCC. Furthermore, performance of a more extensive LND was similarly not associated with increased complication rates. There are remarkably little data regarding the incremental morbidity of LND at the time of RN, either for localized or for metastatic RCC. Indeed, the majority of studies evaluating the oncologic impact of LND do not report any data on perioperative morbidity [4,5,25– 29]. Moreover, although systematic and narrative reviews on the role of LND in the management of RCC echo a consensus opinion that LND is not associated with increased morbidity, all cite the European Organization for Research and Treatment of Cancer (EORTC) 30881 trial as the only source of primary data to inform this topic [3,18,19]. In that randomized trial of RN with or without LND, complication rates were overall low and the authors comment that “lymph-node dissection had no real impact on the complication rate” [2]. However, when examining complication categories, LND was associated with a higher rate of bleeding (9.4% vs 6.5%) and embolism (2.2% vs 1.1%), compared with RN alone, although statistical comparison was not provided for these data. In the only other study, to our knowledge, that provides comparative data regarding perioperative morbidity of LND, Pantuck et al [6] describe perioperative outcomes of RN with or without LND. The authors report that LND was not associated with statistically significant differences in operative times, hospital stay, or perioperative complications, among either M0 or M1 patients. Notably, the authors present only unadjusted results, and it is unclear if these conclusions would be altered after multivariable adjustment. Our propensity score–adjusted analyses support the lack of incremental morbidity to LND in both M0 and M1 patients. It is important to note that the overall complication rate of 9% in the present study is lower than that reported in prior literature. Specifically, the overall complication rate in the EORTC 30881 trial was 24% [2]. In the study by Pantuck et al [6], the overall complication rates were 17% (no LND)
Please cite this article in press as: Gershman B, et al. Perioperative Morbidity of Lymph Node Dissection for Renal Cell Carcinoma: A Propensity Score–based Analysis. Eur Urol (2017), https://doi.org/10.1016/j.eururo.2017.10.020
EURURO-7602; No. of Pages 7 6
E U R O P E A N U R O L O G Y X X X ( 2 0 17 ) X X X – X X X
versus 23% (LND) among M0 patients, and 24% (no LND) versus 25% (LND) among M1 patients. The lower overall rate of complications in the current study may reflect differences in study populations, surgical technique, complications assessed/ascertainment methods, and treatment era. The role of LND in the surgical management of RCC has been controversial. While early studies suggested a potential survival benefit to LND [6,25–27], more rigorous recent investigations have not reported improved oncologic outcomes with LND [2,4,5,7,30]. Notwithstanding the possibility of a small, yet-to-be defined subset of patients for whom LND may provide a survival benefit, as suggested in several studies of patients with lymph node metastases [14–17], the contemporary role of LND appears to be a predominantly staging one. Indeed, pathologic nodal involvement is a poor prognostic indicator among both M0 and M1 settings, and LND may accordingly provide actionable staging information [6]. However, if LND is to be justified as a predominantly staging procedure, it must have a favorable morbidity profile in the absence of therapeutic benefit. To this end, the present study suggests that LND may be performed without an increase in major perioperative morbidity. This study has several limitations. First, it is retrospective and it is possible that unmeasured confounding may impact the observed results. In addition, neither the decision to perform LND nor the extent of LND was standardized, although the PS model included previously identified predictors of lymph node involvement, and further adjustment for surgeon did not alter results. LND dissection templates were not standardized or captured, and extended LND was defined according to the number of lymph nodes removed. Moreover, these data reflect the experience of a high-volume academic center and may not reflect outcomes in different settings. Finally, we did not examine other measures of perioperative morbidity, such as operative time, readmission rates, or certain complications that were not captured such as intraoperative injury to adjacent organs, lymphocele formation, or chylous ascites. Accordingly, the conclusions of this study must be qualified within the end points examined. Despite these limitations, the present study benefits from a large cohort size, rigorous assessment of 30-d postoperative complications, and propensity score adjustment for nonrandom treatment allocation.
Study concept and design: Gershman, Moreira, Thompson, Boorjian, Lohse, Costello, Cheville, Leibovich. Acquisition of data: Thompson, Boorjian, Lohse, Leibovich. Analysis and interpretation of data: Gershman, Lohse, Thompson, Leibovich. Drafting of the manuscript: Gershman, Lohse, Leibovich. Critical revision of the manuscript for important intellectual content: Gershman, Moreira, Thompson, Boorjian, Lohse, Costello, Cheville, Leibovich. Statistical analysis: Lohse. Obtaining funding: None. Administrative, technical, or material support: Thompson, Leibovich. Supervision: Thompson, Leibovich. Other: None. Financial disclosures: Boris Gershman certifies that all conflicts of interest, including specific financial interests and relationships and affiliations relevant to the subject matter or materials discussed in the manuscript (eg, employment/affiliation, grants or funding, consultancies, honoraria, stock ownership or options, expert testimony, royalties, or patents filed, received, or pending), are the following: None. Funding/Support and role of the sponsor: None.
Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at https://doi.org/10.1016/j. eururo.2017.10.020. References [1] Motzer RJ, Jonasch E, Agarwal N, et al. Kidney cancer, version 2.2017, NCCN clinical practice guidelines in oncology. J Natl Compr Canc Netw 2017;15:804–34. [2] Blom JH, van Poppel H, Marechal 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;55:28–34. [3] Capitanio U, Becker F, Blute ML, et al. Lymph node dissection in renal cell carcinoma. Eur Urol 2011;60:1212–20. [4] Feuerstein MA, Kent M, Bazzi WM, Bernstein M, Russo P. Analysis of lymph node dissection in patients with >/ = 7-cm renal tumors. World J Urol 2014;32:1531–6. [5] Feuerstein MA, Kent M, Bernstein M, Russo P. Lymph node dissection during cytoreductive nephrectomy: a retrospective analysis. Int J Urol 2014;21:874–9. [6] Pantuck AJ, Zisman A, Dorey F, et al. Renal cell carcinoma with retroperitoneal lymph nodes: role of lymph node dissection. J Urol
5.
Conclusions
2003;169:2076–83. [7] Gershman B, Thompson RH, Moreira DM, et al. Radical nephrectomy
Performance of LND at the time of RN was not significantly associated with an increased risk of the major 30-d complications captured in this study, although it may be associated with a modestly increased risk of minor complications. In the absence of increased morbidity, LND may be justified in a predominantly staging role in the management of RCC.
with or without lymph node dissection for nonmetastatic renal cell carcinoma: a propensity score-based analysis. Eur Urol 2017;71:560–7. [8] Pantuck AJ, Zisman A, Dorey F, et al. Renal cell carcinoma with retroperitoneal lymph nodes. Impact on survival and benefits of immunotherapy. Cancer 2003;97:2995–3002. [9] Giuliani L, Giberti C, Martorana G, Rovida S. Radical extensive surgery for renal cell carcinoma: long-term results and prognostic factors. J Urol 1990;143:468–73, discussion 73-4. [10] Robson CJ, Churchill BM, Anderson W. The results of radical nephrectomy for renal cell carcinoma. J Urol 1969;101:297–301.
Author contributions: Boris Gershman had full access to all the data in
[11] Ravaud A, Motzer RJ, Pandha HS, et al. Adjuvant sunitinib in
the study and takes responsibility for the integrity of the data and the
high-risk renal-cell carcinoma after nephrectomy. N Engl J Med
accuracy of the data analysis.
2016;375:2246–54.
Please cite this article in press as: Gershman B, et al. Perioperative Morbidity of Lymph Node Dissection for Renal Cell Carcinoma: A Propensity Score–based Analysis. Eur Urol (2017), https://doi.org/10.1016/j.eururo.2017.10.020
EURURO-7602; No. of Pages 7 E U R O P E A N U R O L O G Y X X X ( 2 0 17 ) X X X – X X X
7
[12] Margulis V, Wood CG. The role of lymph node dissection in renal cell
[22] Austin PC. Optimal caliper widths for propensity-score matching
carcinoma: the pendulum swings back. Cancer J 2008;14:308–14.
when estimating differences in means and differences in propor-
[13] Ljungberg B, Bensalah K, Canfield S, et al. EAU guidelines on renal cell carcinoma: 2014 update. Eur Urol 2015;67:913–24. [14] Delacroix Jr SE, Chapin BF, Chen JJ, et al. Can a durable disease-free
tions in observational studies. Pharm Stat 2011;10:150–61. [23] Xu S, Ross C, Raebel MA, Shetterly S, Blanchette C, Smith D. Use of stabilized inverse propensity scores as weights to directly estimate
survival be achieved with surgical resection in patients with patho-
relative risk and its confidence intervals. Value Health 2010;13:273–7.
logical node positive renal cell carcinoma? J Urol 2011;186:1236–41.
[24] Terrone C, Guercio S, De Luca S, et al. The number of lymph nodes
[15] Trinh QD, 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 2012;110(2 Pt 2):E21–7.
examined and staging accuracy in renal cell carcinoma. BJU Int 2003;91:37–40. [25] Vasselli JR, Yang JC, Linehan WM, White DE, Rosenberg SA, Walther MM. Lack of retroperitoneal lymphadenopathy predicts survival of
[16] Karakiewicz PI, Trinh QD, Bhojani N, et al. Renal cell carcinoma with
patients with metastatic renal cell carcinoma. J Urol 2001;166:68–72.
nodal metastases in the absence of distant metastatic disease:
[26] Whitson JM, Harris CR, Reese AC, Meng MV. Lymphadenectomy
prognostic indicators of disease-specific survival. Eur Urol 2007;
improves survival of patients with renal cell carcinoma and nodal
51:1616–24. [17] Gershman B, Moreira DM, Thompson RH, et al. Renal cell carcinoma
metastases. J Urol 2011;185:1615–20. [27] Herrlinger A, Schrott KM, Schott G, Sigel A. What are the benefits of
with isolated lymph node involvement: long-term natural history
extended dissection of the regional renal lymph nodes in the
and predictors of oncologic outcomes following surgical resection.
therapy of renal cell carcinoma. J Urol 1991;146:1224–7.
Eur Urol 2017;72:300–6.
[28] Minervini A, Lilas L, Morelli G, et al. Regional lymph node dissection
[18] Bekema HJ, MacLennan S, Imamura M, et al. Systematic review of
in the treatment of renal cell carcinoma: is it useful in patients with
adrenalectomy and lymph node dissection in locally advanced renal
no suspected adenopathy before or during surgery? BJU Int
cell carcinoma. Eur Urol 2013;64:799–810.
2001;88:169–72.
[19] MacLennan S, Imamura M, Lapitan MC, et al. Systematic review of
[29] Capitanio U, Suardi N, Matloob R, et al. Extent of lymph node
perioperative and quality-of-life outcomes following surgical man-
dissection at nephrectomy affects cancer-specific survival and met-
agement of localised renal cancer. Eur Urol 2012;62:1097–117. [20] Dindo D, Demartines N, Clavien PA. Classification of surgical com-
astatic progression in specific sub-categories of patients with renal cell carcinoma (RCC). BJU Int 2014;114:210–5.
plications: a new proposal with evaluation in a cohort of
[30] Gershman B, Thompson RH, Moreira DM, et al. Lymph node dissec-
6336 patients and results of a survey. Ann Surg 2004;240:205–13.
tion is not associated with improved survival among patients
[21] D’Agostino Jr RB. Propensity score methods for bias reduction in the
undergoing cytoreductive nephrectomy for metastatic renal
comparison of a treatment to a non-randomized control group. Stat
cell carcinoma: a propensity score based analysis. J Urol 2017;
Med 1998;17:2265–81.
197(3 Pt 1):574–9.
Please cite this article in press as: Gershman B, et al. Perioperative Morbidity of Lymph Node Dissection for Renal Cell Carcinoma: A Propensity Score–based Analysis. Eur Urol (2017), https://doi.org/10.1016/j.eururo.2017.10.020
available at www.sciencedirect.com journal homepage: www.europeanurology.com
Kidney Cancer
Perioperative Morbidity of Lymph Node Dissection for Renal Cell Carcinoma: A Propensity Score–based Analysis Boris Gershman a,*, Daniel M. Moreira b, R. Houston Thompson c, Stephen A. Boorjian c, Christine M. Lohse d, Brian A. Costello e, John C. Cheville f, Bradley C. Leibovich c a
Division of Urology, Rhode Island Hospital and The Miriam Hospital, Providence, RI, USA; b Department of Urology, University of Illinois, Chicago, IL, USA;
c
Department of Urology, Mayo Clinic, Rochester, MN, USA;
d
Department of Health Sciences Research, Mayo Clinic, Rochester, MN, USA; e Department of
Oncology, Mayo Clinic, Rochester, MN, USA; f Department of Pathology, Mayo Clinic, Rochester, MN, USA
Article info
Abstract
Article history: Accepted October 23, 2017
Background: There are little data regarding the morbidity of lymph node dissection (LND) for renal cell carcinoma (RCC) to assess its risk–benefit ratio. Objective: To evaluate the association of LND with 30-d complications among patients undergoing radical nephrectomy (RN) for RCC. Design, setting, and participants: A total of 2066 patients underwent RN for M0 or M1 RCC between 1990 and 2010, of whom 774 (37%) underwent LND. Intervention: RN with or without LND. Outcome measurements and statistical analysis: Associations of LND with 30-d complications were examined using logistic regression with several propensity score techniques. Extended LND, defined as removal of 13 lymph nodes, was examined in a sensitivity analysis. Results and limitations: A total of 184 (9%) patients were pN1 and 302 (15%) were M1. Thirty-day complications occurred in 194 (9%) patients, including Clavien grade 3 complications in 81 (4%) patients. Clinicopathologic features were well balanced after propensity score adjustment. In the overall cohort, LND was not statistically significantly associated with Clavien grade 3 complications, although there was an approximately 40% increased risk of any Clavien grade complication that did not reach statistical significance. Likewise, LND was not significantly associated with any Clavien grade or Clavien grade 3 complications when separately evaluated among M0 or M1 patients. Extended LND was not significantly associated with any Clavien grade or Clavien grade 3 complications. LND was not associated with length of stay or estimated blood loss. Limitations include a retrospective design. Conclusions: LND is not significantly associated with an increased risk of Clavien grade 3 complications, although it may be associated with a modestly increased risk of minor complications. In the absence of increased morbidity, LND may be justified in a predominantly staging role in the management of RCC. Patient summary: Lymph node dissection for renal cell carcinoma is not associated with increased rates of major complications. © 2017 European Association of Urology. Published by Elsevier B.V. All rights reserved.
Associate Editor: Giacomo Novara Statistical Editor: Andrew Vickers Keywords: Lymph node dissection Complications Nephrectomy Renal cell carcinoma Propensity score
* Corresponding author. Rhode Island Hospital and The Miriam Hospital, Warren Alpert Medical School of Brown University, 195 Collyer Street, Suite 201, Providence, RI 02904, USA. Tel. +1 401 272 7799; Fax: +1 401 453 9078. E-mail address: [email protected] (B. Gershman). https://doi.org/10.1016/j.eururo.2017.10.020 0302-2838/© 2017 European Association of Urology. Published by Elsevier B.V. All rights reserved.
Please cite this article in press as: Gershman B, et al. Perioperative Morbidity of Lymph Node Dissection for Renal Cell Carcinoma: A Propensity Score–based Analysis. Eur Urol (2017), https://doi.org/10.1016/j.eururo.2017.10.020
EURURO-7602; No. of Pages 7 2
E U R O P E A N U R O L O G Y X X X ( 2 0 17 ) X X X – X X X
1.
Introduction
The role of lymph node dissection (LND) in the surgical management of renal cell carcinoma (RCC) remains uncertain. Although there are little data to support an oncologic benefit to LND at the time of radical nephrectomy (RN) for RCC, it provides pathologic assessment of nodal stage [1–7]. This may provide valuable prognostic information given the association of lymph node metastasis with survival [6,8–10], and it may identify candidates for consideration of adjuvant therapies [11]. In this context, proponents have argued that LND may have a role in the surgical management of RCC by providing valuable staging information, and guidelines have recommended LND in the setting of radiographic lymphadenopathy [ 1,12,13]. Moreover, it has been suggested that LND may provide a survival benefit to a small subset of patients with lymph node metastases [14–17]. An accurate assessment of the risks of LND is critical to determine its risk–benefit ratio. Yet despite numerous studies examining a therapeutic benefit to LND, there is a paucity of data regarding its morbidity [2,3,6,18,19]. We, therefore, examined the perioperative morbidity of LND among patients undergoing RN for RCC. Specifically, in this study, we characterized 30-d complications among patients undergoing RN with or without LND, and we evaluated the associations of LND with 30-d complications using a propensity score–based approach to adjust for nonrandom treatment allocation.
2.3.
Statistical methods
Clinicopathologic and radiographic features were summarized with medians and interquartile ranges (IQRs) or frequency counts and percentages, and were compared for patients with and without LND using Wilcoxon rank sum, chi-square, and Fisher exact tests. Propensity scores (PSs) for treatment with LND were estimated using a logistic regression model with LND as the outcome and the features listed in Table 1 as covariates, except body mass index (BMI), estimated blood loss (EBL), and length of stay. BMI was excluded because of the amount of missing data and its lack of a statistically significant association with LND. Year of surgery and age at surgery were categorized into four groups of approximately equal patient numbers since there was evidence that the associations of these features with LND were not linear. Forty-nine patients missing covariate data (16 with and 33 without LND) were excluded from the PS analytic cohort, as were 91 patients with a PS that did not fall within the common range (50 with and 41 without LND). A total of 1926 patients formed the cohort for PS analyses, including 708 (37%) who underwent LND and 1218 (63%) who did not. The associations of LND with any complication or a Clavien grade 3 complication were evaluated with logistic regression models using several PS techniques, and summarized with odds ratios (ORs) and 95% confidence intervals (CIs) [21]. Patients with LND were matched 1:1 to patients without LND using the caliper method based on the logit of the PS and a caliper width of 0.2 times the standard deviation of the logit [22]. Using this approach, 457 patients with LND were matched to 457 patients without LND. To incorporate all 1926 patients in the analyses, we conducted PS analyses reweighting by stabilized inverse probability weights (IPWs). Additional PS techniques were employed including adjustment for PS quintile and stratification by PS quintile (Supplementary Material). ORs for the stratified analyses were obtained by pooling estimates obtained within each PS quintile. ORs for the subset of 457 matched pairs were obtained by stratifying on the matched sets.
Patients and methods
2.
Stabilized IPWs were truncated by setting weights below the first percentile to the value of the first percentile, and weights above the 99th
2.1.
Patient population
percentile to the value of the 99th percentile [23]. Models were further separately adjusted for surgeon and BMI with similar results. We further
After obtaining Institutional Review Board approval, we identified
examined the association of LND with 30-d complications among M0
2103 patients treated with RN for sporadic, unilateral, M0 or M1 RCC
and M1 patients by constructing logistic regression models that included
between 1990 and 2010. Thirty-seven patients lacked data regarding
an indicator variable for performance of LND, an indicator for M1 status,
complications within 30 d of surgery and were excluded. The final cohort
and the interaction between the two.
included 2066 patients for study, of whom 774 (37%) underwent LND.
Given heterogeneity in the extent of LND, we conducted a sensitivity
LND was performed at the surgeon's discretion, and a standardized
analysis to examine associations of extended LND, defined as removal of
template was not utilized.
13 LNs [24] (vs no LND or LND with <13 LNs removed), with complications. PSs for treatment with extended LND were obtained
2.2.
Clinicopathologic features
The clinicopathologic and radiographic features in Table 1 were assessed as described in the Supplementary Material. All pathologic features were reviewed by one urologic pathologist (J.C.C.) without knowledge of patient outcome. Indicator variables were employed to denote missing data for the radiographic features studied.
using a logistic regression model, and associations of extended LND with any complication or a Clavien 3 complication were evaluated using the PS techniques described above. Statistical analyses were performed using SAS version 9.4 (SAS Institute, Cary, NC, USA) and R version 3.1.1 (R Foundation for Statistical Computing, Vienna, Austria). All tests were two sided, and p < 0.05 was considered statistically significant.
Complications within 30 d of RN were assessed by a trained nurse abstractor using standardized definitions and graded according to the
3.
Results
Clavien classification [20]. Complication categories included bleeding (requiring reoperation or angioembolization) or hematoma (on crosssectional imaging), deep vein thrombosis (DVT), pulmonary embolism (PE), myocardial infarction (MI), wound infection (requiring antibiotics or drainage) or dehiscence (partial or complete), abscess, sepsis (bacteremia with hypotension or requiring admission to intensive care unit), acute renal failure with or without dialysis, pneumothorax, any other complication not specified above, and death. Patients were typically evaluated at 3 mo after surgery and then according to the recommended surveillance protocol based on surgical pathology.
A total of 2066 patients formed the study cohort, including 774 (37%) who underwent LND. Overall, 184 (9%) patients were pN1 and 302 (15%) had distant metastases at the time of RN. The median number of LNs removed during LND was 6 (IQR 2–13). Clinicopathologic and radiographic features stratified, by performance of LND, are summarized in Table 1. Patients who underwent LND had more advanced disease and more aggressive pathologic characteristics,
Please cite this article in press as: Gershman B, et al. Perioperative Morbidity of Lymph Node Dissection for Renal Cell Carcinoma: A Propensity Score–based Analysis. Eur Urol (2017), https://doi.org/10.1016/j.eururo.2017.10.020
EURURO-7602; No. of Pages 7 3
E U R O P E A N U R O L O G Y X X X ( 2 0 17 ) X X X – X X X
Table 1 – Clinicopathologic and radiographic features stratified by performance of LND in the overall cohort (N = 2066) Feature
Median (IQR) or number (%) No LND (N = 1292)
Year of surgery 1990–1994 1995–1999 2000–2004 2005–2010 Age at surgery (yr) 18–54 55–64 65–69 70 Sex Female Male Charlson score BMI (kg/m2; N = 1907) Tumor size (cm; N = 2056) Symptoms (N = 2065) Constitutional symptoms (N = 2065) Smoking status (N = 2036) Never Current Former ECOG performance status (N = 2065) 0 1 2 3 4 cN1 Renal vein involvement on CT/MRI IVC involvement on CT/MRI Radiographic evidence of Hemorrhage Necrosis Calcification Extrarenal extension Neovascularity Adrenal involvement Cystic or indeterminate cysts Neoadjuvant therapy Neoadjuvant systemic therapy Surgical approach Open radical nephrectomy Laparoscopic radical nephrectomy Vascular bypass during RN Concurrent metastasectomy Other concurrent surgery Estimated blood loss (cm; N = 1893) Length of stay (d) RCC histologic subtype Clear cell Papillary or chromophobe Collecting duct or NOS All others 2010 pT (N = 2052) pT1a pT1b pT2a pT2b pT3a pT3b pT3c pT4 M stage M0
p value LND (N = 774)
357 (28) 344 (27) 356 (28) 235 (18)
121 (16) 130 (17) 226 (29) 297 (38)
<0.001
289 (22) 331 (26) 219 (17) 453 (35)
224 (29) 216 (28) 141 (18) 193 (25)
<0.001
449 (35) 843 (65) 1 (0–2) 28.3 (25.2–31.9) 6.0 (4.0–8.5) 713 (55) 264 (20)
259 (33) 515 (67) 1 (0–6) 28.0 (24.8–31.7) 9.5 (7.0–12.0) 583 (75) 266 (34)
0.55
550 (43) 205 (16) 518 (41)
307 (40) 156 (20) 300 (39)
0.043
1093 (85) 153 (12) 24 (2) 18 (1) 4 (<1) 24 (2) 116 (9) 68 (5)
650 (84) 90 (12) 20 (3) 13 (2) 0 161 (21) 257 (33) 158 (20)
0.7
26 (2) 93 (7) 89 (7) 44 (3) 86 (7) 20 (2) 267 (21) 18 (1) 8 (1)
21 (3) 123 (16) 81 (10) 133 (17) 122 (16) 31 (4) 95 (12) 27 (3) 13 (2)
0.3 <0.001 0.004 <0.001 <0.001 <0.001 <0.001 0.002 0.020
1,098 (85) 194 (15) 17 (1) 32 (2) 175 (14) 250 (125–500) 5 (4–7)
735 (95) 39 (5) 29 (4) 65 (8) 200 (26) 600 (300–1200) 6 (4–7)
<0.001
1,037 (80) 233 (18) 18 (1) 4 (<1)
653 (84) 98 (13) 16 (2) 7 (1)
0.003
338 (26) 360 (28) 164 (13) 74 (6) 259 (20) 52 (4) 14 (1) 19 (1)
38 (5) 96 (12) 94 (12) 77 (10) 295 (28) 117 (15) 20 (3) 35 (5)
<0.001
1176 (91)
588 (76)
<0.001
0.025 0.6 <0.001 <0.001 <0.001
<0.001 <0.001 <0.001
<0.001 <0.001 <0.001 <0.001 <0.001
Please cite this article in press as: Gershman B, et al. Perioperative Morbidity of Lymph Node Dissection for Renal Cell Carcinoma: A Propensity Score–based Analysis. Eur Urol (2017), https://doi.org/10.1016/j.eururo.2017.10.020
EURURO-7602; No. of Pages 7 4
E U R O P E A N U R O L O G Y X X X ( 2 0 17 ) X X X – X X X
Table 1 (Continued ) Feature
Median (IQR) or number (%) No LND (N = 1292)
M1 Grade 1 2 3 4 Coagulative tumor necrosis (N = 2,065) Sarcomatoid differentiation Number of metastatic sites 0 1 2 3 4 5 Location of metastases Lung Liver Bone Brain
p value LND (N = 774)
116 (9)
186 (24)
80 (6) 593 (46) 521 (40) 98 (8) 376 (29) 51 (4)
12 (2) 160 (21) 405 (52) 197 (25) 448 (58) 77 (10)
<0.001
1176 (91) 76 (6) 29 (2) 7 (1) 4 (<1) 0
588 (76) 114 (15) 49 (6) 17 (2) 4 (1) 2 (<1)
<0.001
73 (6) 9 (1) 33 (3) 10 (1)
117 (15) 26 (3) 39 (5) 8 (1)
<0.001 <0.001 0.003 0.5
<0.001 <0.001
BMI = body mass index; CT = computed tomography; ECOG = Eastern Cooperative Oncology Group; IQR = interquartile range; IVC = inferior vena cava; LND = lymph node dissection; MRI = magnetic resonance imaging; NOS = nitric oxide synthase; RCC = renal cell carcinoma; RN = radical nephrectomy.
including a greater incidence of radiographic lymphadenopathy and distant metastases at RN. However, baseline characteristics were well balanced after PS adjustment. Specifically, no feature was statistically significantly different between patients with and without LND after adjusting for PS quintile (data not shown), the 457 matched pairs were well balanced (data not shown), and no feature was statistically significantly different between patients with and without LND after reweighting by stabilized IPWs (Supplementary Table 1). Notably, although patients who underwent LND had greater EBL and length of stay in unadjusted comparisons (Table 1), there were no statistically significant differences in these perioperative end points after reweighting by stabilized IPWs (Supplementary Table 1). Complications were diagnosed within 30 d of surgery in 194 (9%) patients, including Clavien grade 3 complications in 81 (4%) patients (Table 2). The most common complication categories were acute renal failure (3%), bleeding (2%), and wound complications (2%). Among patients who underwent LND, complications occurred more frequently in patients with pN1 disease—any grade complications occurred in 66 (11%) pN0 and 40 (22%) pN1 patients (p < 0.001), while Clavien 3 complications occurred in 25 (4%) pN0 and 18 (10%) pN1 patients (p = 0.004). In unadjusted analyses, performance of LND was associated with increased rates of any 30-d complication (14% vs 7%, p < 0.001) and Clavien grade 3 complications (6% vs 3%, p = 0.003). However, after reweighting by stabilized IPWs, there were no statistically significant differences in rates of any complication (11% vs 8%, p = 0.09) or Clavien grade 3 complications (4% vs 3%, p = 0.8; Supplementary Table 2). Notably, LND was associated with an increased rate of Clavien grade 2 complications (6% vs 3%, p = 0.04). This appeared driven by a small but
statistically significant increase in the rates of DVT (1% vs <1%, p = 0.04) and MI (2% vs <1%, p = 0.02). Next, we examined the associations of LND with 30-d complications in the overall cohort (Table 3 and Supplementary Table 3). Here, LND was not statistically significantly associated with Clavien grade 3 complications using any of the PS techniques, although there was a modestly increased risk of any Clavien grade complication (OR ranging from 1.41 to 1.46) that did not reach statistical
Table 2 – Thirty-day complications after RN stratified by performance of LND in the overall cohort All No LND LND p value (N = 2066) (N = 1292) (N = 774) Clavien grade 1 complication 67 (3) 81 (4) Clavien grade 2 complication 29 (1) Clavien grade 3 complication Clavien grade 4 complication 46 (2) 20 (1) Clavien grade 5 complication Clavien grade 3 complication 81 (4) Any complication 194 (9) Complication category 37 (2) Bleeding or hematoma DVT 26 (1) 23 (1) PE 7 (<1) MI Wound infection or dehiscence44 (2) 8 (<1) Abscess 14 (1) Sepsis Acute renal failure 66 (3) 17 (1) Dialysis 18 (1) Pneumothorax Other 19 (1) 20 (1) Death
36 (3) 27 (2) 16 (1) 20 (2) 8 (1) 38 (3) 88 (7)
31 (4) 54 (7) 13 (2) 26 (3) 12 (2) 43 (6) 106 (14)
0.1 <0.001 0.4 0.007 0.036 0.003 <0.001
17 (1) 6 (<1) 7 (1) 2 (<1) 18 (1) 6 (<1) 9 (1) 31 (2) 8 (1) 12 (1) 6 (<1) 8 (1)
20 (3) 20 (3) 16 (2) 5 (1) 26 (3) 2 (<1) 5 (1) 35 (5) 9 (1) 6 (1) 13 (2) 12 (2)
0.035 <0.001 0.001 0.11 0.003 0.7 0.9 0.008 0.2 0.7 0.005 0.036
vein thrombosis; LND = lymph node DVT = deep MI = myocardial infarction; PE = pulmonary embolism; nephrectomy.
dissection; RN = radical
Please cite this article in press as: Gershman B, et al. Perioperative Morbidity of Lymph Node Dissection for Renal Cell Carcinoma: A Propensity Score–based Analysis. Eur Urol (2017), https://doi.org/10.1016/j.eururo.2017.10.020
EURURO-7602; No. of Pages 7 E U R O P E A N U R O L O G Y X X X ( 2 0 17 ) X X X – X X X
Table 3 – Associations of LND with any Clavien grade and Clavien grade 3 30-d complications PS technique
Any Clavien grade OR (95% CI)
IPW Matched pairs
p value
1.44 (0.95–2.19) 1.46 (0.93–2.27)
0.09 0.10
OR (95% CI) 1.07 (0.59–1.92) 1.54 (0.77–3.09)
p value
0.8 0.2
Table 4 – Associations of LND with any Clavien grade and Clavien grade 3 30-d complications among M0 and M1 patients using IPW Clavien grade 3
p value
OR (95% CI) M0 M1
1.50 (0.94–2.38) 1.18 (0.48–2.90)
0.09
OR (95% CI) 1.04 (0.55–1.98) 1.24 (0.28–5.45)
0.7
p value
0.9 0.8
CI = confidence interval; IPW = inverse probability weight; LND = lymph node dissection; OR = odds ratio. Models included an indicator for LND, an indicator for M1 status, and the interaction term between the two.
Table 5 – Associations of extended LND with any Clavien grade and Clavien grade 3 30-d complications PS technique
Any Clavien grade OR (95% CI)
IPW Matched pairs
1.66 (0.81–3.40) 1.53 (0.80–2.94)
p value
0.2 0.2
these analyses, extended LND was again not associated with any complication or Clavien grade 3 complications.
Clavien grade 3
CI = confidence interval; IPW = inverse probability weight; LND = lymph node dissection; OR = odds ratio; PS = propensity score.
Any Clavien grade
5
Clavien grade 3 OR (95% CI) 0.95 (0.36–2.46) 1.43 (0.54–3.75)
p value
0.9 0.5
CI = confidence interval; IPW = inverse probability weight; LND = lymph node dissection; OR = odds ratio.
significance. In analyses further adjusted for surgeon, results were similar (Supplementary Table 4). We further evaluated the association of LND with 30-d complications among M0 and M1 patients by including an interaction term in the regression models (Table 4 and Supplementary Table 5). In these analyses, LND was not significantly associated with any complication or Clavien grade 3 complications in either M0 or M1 patients. Moreover, the p values for the interaction terms in models examining any complication or Clavien grade 3 complications were not statistically significant (p > 0.5 for each), suggesting that there was no effect of heterogeneity according to M stage. Given heterogeneity in the extent of LND, we conducted a sensitivity analysis examining extended LND, defined as removal of 13 LNs (Table 5 and Supplementary Table 6). In
4.
Discussion
In this large, single-institutional cohort, LND at the time of RN was associated with a low overall rate of any (9%) and Clavien grade 3 (4%) complications. More importantly, LND was not statistically significantly associated with an increased risk of Clavien grade 3 complications, although there was an approximately 40% increased risk of any Clavien grade complication that did not reach statistical significance. This appears to be driven by a small increase in Clavien grade 2 complications related to DVT and MI, which may have clinical relevance despite low Clavien grade. Taken together, these results suggest that LND can be performed without major perioperative morbidity, although it may be associated with a modest increase in minor postoperative complications. Moreover, subset analyses indicate that the morbidity of LND does not differ between patients with nonmetastatic or metastatic RCC. Furthermore, performance of a more extensive LND was similarly not associated with increased complication rates. There are remarkably little data regarding the incremental morbidity of LND at the time of RN, either for localized or for metastatic RCC. Indeed, the majority of studies evaluating the oncologic impact of LND do not report any data on perioperative morbidity [4,5,25– 29]. Moreover, although systematic and narrative reviews on the role of LND in the management of RCC echo a consensus opinion that LND is not associated with increased morbidity, all cite the European Organization for Research and Treatment of Cancer (EORTC) 30881 trial as the only source of primary data to inform this topic [3,18,19]. In that randomized trial of RN with or without LND, complication rates were overall low and the authors comment that “lymph-node dissection had no real impact on the complication rate” [2]. However, when examining complication categories, LND was associated with a higher rate of bleeding (9.4% vs 6.5%) and embolism (2.2% vs 1.1%), compared with RN alone, although statistical comparison was not provided for these data. In the only other study, to our knowledge, that provides comparative data regarding perioperative morbidity of LND, Pantuck et al [6] describe perioperative outcomes of RN with or without LND. The authors report that LND was not associated with statistically significant differences in operative times, hospital stay, or perioperative complications, among either M0 or M1 patients. Notably, the authors present only unadjusted results, and it is unclear if these conclusions would be altered after multivariable adjustment. Our propensity score–adjusted analyses support the lack of incremental morbidity to LND in both M0 and M1 patients. It is important to note that the overall complication rate of 9% in the present study is lower than that reported in prior literature. Specifically, the overall complication rate in the EORTC 30881 trial was 24% [2]. In the study by Pantuck et al [6], the overall complication rates were 17% (no LND)
Please cite this article in press as: Gershman B, et al. Perioperative Morbidity of Lymph Node Dissection for Renal Cell Carcinoma: A Propensity Score–based Analysis. Eur Urol (2017), https://doi.org/10.1016/j.eururo.2017.10.020
EURURO-7602; No. of Pages 7 6
E U R O P E A N U R O L O G Y X X X ( 2 0 17 ) X X X – X X X
versus 23% (LND) among M0 patients, and 24% (no LND) versus 25% (LND) among M1 patients. The lower overall rate of complications in the current study may reflect differences in study populations, surgical technique, complications assessed/ascertainment methods, and treatment era. The role of LND in the surgical management of RCC has been controversial. While early studies suggested a potential survival benefit to LND [6,25–27], more rigorous recent investigations have not reported improved oncologic outcomes with LND [2,4,5,7,30]. Notwithstanding the possibility of a small, yet-to-be defined subset of patients for whom LND may provide a survival benefit, as suggested in several studies of patients with lymph node metastases [14–17], the contemporary role of LND appears to be a predominantly staging one. Indeed, pathologic nodal involvement is a poor prognostic indicator among both M0 and M1 settings, and LND may accordingly provide actionable staging information [6]. However, if LND is to be justified as a predominantly staging procedure, it must have a favorable morbidity profile in the absence of therapeutic benefit. To this end, the present study suggests that LND may be performed without an increase in major perioperative morbidity. This study has several limitations. First, it is retrospective and it is possible that unmeasured confounding may impact the observed results. In addition, neither the decision to perform LND nor the extent of LND was standardized, although the PS model included previously identified predictors of lymph node involvement, and further adjustment for surgeon did not alter results. LND dissection templates were not standardized or captured, and extended LND was defined according to the number of lymph nodes removed. Moreover, these data reflect the experience of a high-volume academic center and may not reflect outcomes in different settings. Finally, we did not examine other measures of perioperative morbidity, such as operative time, readmission rates, or certain complications that were not captured such as intraoperative injury to adjacent organs, lymphocele formation, or chylous ascites. Accordingly, the conclusions of this study must be qualified within the end points examined. Despite these limitations, the present study benefits from a large cohort size, rigorous assessment of 30-d postoperative complications, and propensity score adjustment for nonrandom treatment allocation.
Study concept and design: Gershman, Moreira, Thompson, Boorjian, Lohse, Costello, Cheville, Leibovich. Acquisition of data: Thompson, Boorjian, Lohse, Leibovich. Analysis and interpretation of data: Gershman, Lohse, Thompson, Leibovich. Drafting of the manuscript: Gershman, Lohse, Leibovich. Critical revision of the manuscript for important intellectual content: Gershman, Moreira, Thompson, Boorjian, Lohse, Costello, Cheville, Leibovich. Statistical analysis: Lohse. Obtaining funding: None. Administrative, technical, or material support: Thompson, Leibovich. Supervision: Thompson, Leibovich. Other: None. Financial disclosures: Boris Gershman certifies that all conflicts of interest, including specific financial interests and relationships and affiliations relevant to the subject matter or materials discussed in the manuscript (eg, employment/affiliation, grants or funding, consultancies, honoraria, stock ownership or options, expert testimony, royalties, or patents filed, received, or pending), are the following: None. Funding/Support and role of the sponsor: None.
Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at https://doi.org/10.1016/j. eururo.2017.10.020. References [1] Motzer RJ, Jonasch E, Agarwal N, et al. Kidney cancer, version 2.2017, NCCN clinical practice guidelines in oncology. J Natl Compr Canc Netw 2017;15:804–34. [2] Blom JH, van Poppel H, Marechal 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;55:28–34. [3] Capitanio U, Becker F, Blute ML, et al. Lymph node dissection in renal cell carcinoma. Eur Urol 2011;60:1212–20. [4] Feuerstein MA, Kent M, Bazzi WM, Bernstein M, Russo P. Analysis of lymph node dissection in patients with >/ = 7-cm renal tumors. World J Urol 2014;32:1531–6. [5] Feuerstein MA, Kent M, Bernstein M, Russo P. Lymph node dissection during cytoreductive nephrectomy: a retrospective analysis. Int J Urol 2014;21:874–9. [6] Pantuck AJ, Zisman A, Dorey F, et al. Renal cell carcinoma with retroperitoneal lymph nodes: role of lymph node dissection. J Urol
5.
Conclusions
2003;169:2076–83. [7] Gershman B, Thompson RH, Moreira DM, et al. Radical nephrectomy
Performance of LND at the time of RN was not significantly associated with an increased risk of the major 30-d complications captured in this study, although it may be associated with a modestly increased risk of minor complications. In the absence of increased morbidity, LND may be justified in a predominantly staging role in the management of RCC.
with or without lymph node dissection for nonmetastatic renal cell carcinoma: a propensity score-based analysis. Eur Urol 2017;71:560–7. [8] Pantuck AJ, Zisman A, Dorey F, et al. Renal cell carcinoma with retroperitoneal lymph nodes. Impact on survival and benefits of immunotherapy. Cancer 2003;97:2995–3002. [9] Giuliani L, Giberti C, Martorana G, Rovida S. Radical extensive surgery for renal cell carcinoma: long-term results and prognostic factors. J Urol 1990;143:468–73, discussion 73-4. [10] Robson CJ, Churchill BM, Anderson W. The results of radical nephrectomy for renal cell carcinoma. J Urol 1969;101:297–301.
Author contributions: Boris Gershman had full access to all the data in
[11] Ravaud A, Motzer RJ, Pandha HS, et al. Adjuvant sunitinib in
the study and takes responsibility for the integrity of the data and the
high-risk renal-cell carcinoma after nephrectomy. N Engl J Med
accuracy of the data analysis.
2016;375:2246–54.
Please cite this article in press as: Gershman B, et al. Perioperative Morbidity of Lymph Node Dissection for Renal Cell Carcinoma: A Propensity Score–based Analysis. Eur Urol (2017), https://doi.org/10.1016/j.eururo.2017.10.020
EURURO-7602; No. of Pages 7 E U R O P E A N U R O L O G Y X X X ( 2 0 17 ) X X X – X X X
7
[12] Margulis V, Wood CG. The role of lymph node dissection in renal cell
[22] Austin PC. Optimal caliper widths for propensity-score matching
carcinoma: the pendulum swings back. Cancer J 2008;14:308–14.
when estimating differences in means and differences in propor-
[13] Ljungberg B, Bensalah K, Canfield S, et al. EAU guidelines on renal cell carcinoma: 2014 update. Eur Urol 2015;67:913–24. [14] Delacroix Jr SE, Chapin BF, Chen JJ, et al. Can a durable disease-free
tions in observational studies. Pharm Stat 2011;10:150–61. [23] Xu S, Ross C, Raebel MA, Shetterly S, Blanchette C, Smith D. Use of stabilized inverse propensity scores as weights to directly estimate
survival be achieved with surgical resection in patients with patho-
relative risk and its confidence intervals. Value Health 2010;13:273–7.
logical node positive renal cell carcinoma? J Urol 2011;186:1236–41.
[24] Terrone C, Guercio S, De Luca S, et al. The number of lymph nodes
[15] Trinh QD, 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 2012;110(2 Pt 2):E21–7.
examined and staging accuracy in renal cell carcinoma. BJU Int 2003;91:37–40. [25] Vasselli JR, Yang JC, Linehan WM, White DE, Rosenberg SA, Walther MM. Lack of retroperitoneal lymphadenopathy predicts survival of
[16] Karakiewicz PI, Trinh QD, Bhojani N, et al. Renal cell carcinoma with
patients with metastatic renal cell carcinoma. J Urol 2001;166:68–72.
nodal metastases in the absence of distant metastatic disease:
[26] Whitson JM, Harris CR, Reese AC, Meng MV. Lymphadenectomy
prognostic indicators of disease-specific survival. Eur Urol 2007;
improves survival of patients with renal cell carcinoma and nodal
51:1616–24. [17] Gershman B, Moreira DM, Thompson RH, et al. Renal cell carcinoma
metastases. J Urol 2011;185:1615–20. [27] Herrlinger A, Schrott KM, Schott G, Sigel A. What are the benefits of
with isolated lymph node involvement: long-term natural history
extended dissection of the regional renal lymph nodes in the
and predictors of oncologic outcomes following surgical resection.
therapy of renal cell carcinoma. J Urol 1991;146:1224–7.
Eur Urol 2017;72:300–6.
[28] Minervini A, Lilas L, Morelli G, et al. Regional lymph node dissection
[18] Bekema HJ, MacLennan S, Imamura M, et al. Systematic review of
in the treatment of renal cell carcinoma: is it useful in patients with
adrenalectomy and lymph node dissection in locally advanced renal
no suspected adenopathy before or during surgery? BJU Int
cell carcinoma. Eur Urol 2013;64:799–810.
2001;88:169–72.
[19] MacLennan S, Imamura M, Lapitan MC, et al. Systematic review of
[29] Capitanio U, Suardi N, Matloob R, et al. Extent of lymph node
perioperative and quality-of-life outcomes following surgical man-
dissection at nephrectomy affects cancer-specific survival and met-
agement of localised renal cancer. Eur Urol 2012;62:1097–117. [20] Dindo D, Demartines N, Clavien PA. Classification of surgical com-
astatic progression in specific sub-categories of patients with renal cell carcinoma (RCC). BJU Int 2014;114:210–5.
plications: a new proposal with evaluation in a cohort of
[30] Gershman B, Thompson RH, Moreira DM, et al. Lymph node dissec-
6336 patients and results of a survey. Ann Surg 2004;240:205–13.
tion is not associated with improved survival among patients
[21] D’Agostino Jr RB. Propensity score methods for bias reduction in the
undergoing cytoreductive nephrectomy for metastatic renal
comparison of a treatment to a non-randomized control group. Stat
cell carcinoma: a propensity score based analysis. J Urol 2017;
Med 1998;17:2265–81.
197(3 Pt 1):574–9.
Please cite this article in press as: Gershman B, et al. Perioperative Morbidity of Lymph Node Dissection for Renal Cell Carcinoma: A Propensity Score–based Analysis. Eur Urol (2017), https://doi.org/10.1016/j.eururo.2017.10.020