Need for and extent of lymph node dissection for upper tract urothelial carcinoma: an updated review in 2023

Purpose of review

Although lymph node dissection (LND) during radical nephroureterectomy (RNU) is recommended for high-risk nonmetastatic upper tract urothelial carcinoma (UTUC), adherence to guidelines remains insufficient in clinical practice. Therefore, this review aims to comprehensively summarize the current evidence regarding the diagnostic, prognostic, and therapeutic impact of LND during RNU in UTUC patients.

Recent findings

Clinical nodal staging using conventional CT scan has low sensitivity (25%) and diagnostic accuracy [area under the curve (AUC): 0.58] in UTUC, suggesting the importance of LND for obtaining accurate N-staging. Patients with pathological node-positive (pN+) disease have poor disease-free survival (DFS), cancer-specific survival (CSS), and overall survival (OS) compared with those with pN0. In addition, population-based studies showed that patients who underwent LND improved CSS and OS than those who did not, even in patients who received adjuvant systemic therapy. The number of lymph nodes removed has also been shown to be associated with improved CSS and OS, even in pT0 patients. Template-based LND should be performed as the extent of lymph node is more important than the number of lymph nodes. Robot-assisted RNU may facilitate performing a meticulous LND compared with a laparoscopic approach. Postoperative complications such as lymphatic and/or chylous leakage are increased but adequately manageable. However, the current evidence is not supported by high-quality studies.

Summary

Based on the published data, LND during RNU is a standard procedure for high-risk nonmetastatic UTUC, owing to its diagnostic, staging, prognostic, and, potentially, therapeutic benefits. Template-based LND should be offered to all patients who are planned for RNU for high-risk nonmetastatic UTUC. Patients with pN+ disease are optimal candidates for adjuvant systemic therapy. Robot-assisted RNU may facilitate meticulous LND compared with laparoscopic RNU.

INTRODUCTION

Nonmetastatic upper tract urothelial carcinoma (UTUC) is managed with minimally invasive therapies, such as kidney-sparing endoscopic surgery for low-risk tumors (i.e. single, tumor size <20 mm, low-grade, and ≤cT2) [1–5]. For high-risk UTUC, however, radical nephroureterectomy (RNU) is still the guideline-recommended standard treatment [6,7]. Based on continuously growing evidence, template-based lymph node dissection (LND) during RNU is recommended for high-risk UTUC patients, to improve lymph node staging and prognostication in order to ensure adjuvant treatment when needed [6,8]. Indeed, pathologically confirmed lymph node metastasis is an indication for adjuvant systemic therapies [9,10].

However, UTUC is a relatively rare urological malignancy and potentially heterogeneous disease, making the establishment of reliable and evidence-based recommendations challenging because of the lack of high-level evidence [6]. Indeed, adherence to guideline recommendations is limited in the management of UTUC in real-world practice, including diagnostic tests as well as LND [11–14]. Therefore, this review aims to comprehensively summarize the current evidence regarding the diagnostic, staging, prognostic, and therapeutic impact of LND during RNU in UTUC patients with focus on the recent literature. 

Box 1.

no caption available

EVIDENCE ACQUISITION

A literature search on PubMed/Medline databases was performed in December 2022 to identify studies investigating the oncologic impact of LND, including nodal status and extent of LND, during RNU for UTUC. The detailed search strategy was ‘(urothelial) AND (cancer or carcinoma) AND (upper tract) AND (lymph node dissection OR lymphadenectomy)’. Studies lacking original patient data, letters, editorial comments, replies from authors, case reports, and articles not written in English were excluded. After the full-text investigation, two authors independently extracted the relevant data, such as the first author's name, publication year, study design, number of patients, surgical approach, extent of LND, pathologic outcomes, additional systemic therapies, follow-up duration, and oncologic outcomes.

To compare oncologic outcomes across lymph node status, we extracted 5-year disease-free survival (DFS), cancer-specific survival (CSS), and overall survival (OS). Subsequently, hazard ratios and 95% confidence intervals (CIs) for DFS, CSS, and OS were extracted from Cox regression multivariable analysis. In addition, we only relied on studies with study populations of more than 250 patients for evaluating oncologic outcomes across lymph node status, owing to the limited number of patients who underwent LND and those with pathological node-positive disease.

EVIDENCE SYNTHESIS

Incidence and predictive factors of lymph node metastases in upper tract urothelial carcinoma patients

Up to 25% of UTUC patients harbor regional lymph node metastases at diagnosis, depending on clinical tumor stage and/or tumor grade [15]. For example, Kondo et al.[16] reported that the incidence of lymph node involvement (either clinical or pathological) increased with higher pathological T stages. In this study, they found 5% of pT2, 24% of pT3, and 84% of pT4 tumors to harbor lymph node metastases, whereas patients with pTa/is/1 had no lymph node involvement [16]. T-stage, tumor grade, tumor necrosis, presence of carcinoma in situ (CIS), lymphovascular invasion, and sessile tumor architecture have been reported to be predictive of lymph node metastases [16–18]. Conversely, tumor location (i.e. renal pelvis vs. ureter) was not associated with the presence of lymph node metastases [16,19]. Therefore, experts recommend LND (i.e. template-based LND) in all patients with high-risk UTUC during RNU [20,21].

Diagnostic issues of clinical staging of the nodes

For clinical staging of UTUC, computed tomography (CT) remains a standard radiographic tool with a high pooled overall sensitivity (92%) and specificity (95%) [22]. For the detection of lymph node involvement, Millán-Rodríguez et al.[23] reported, in 1999, that CT had a good diagnostic ability with a sensitivity of 87.5% and specificity of 98% using 93 tumors from 82 patients. Since then, traditionally, clinical lymph node staging has relied on cross-sectional imaging, specifically on CT. However, recently, Pallauf et al. conducted a multicenter observational study with 865 UTUC patients treated with RNU and LND and reported that conventional cross-sectional imaging yielded a good specificity with 91% but with a very low sensitivity (i.e. 25%), resulting in a poor diagnostic accuracy [area under the curve (AUC): 0.58] [24^▪▪]. The authors suggested to use conventional cross-sectional imaging as a rule-in but not a rule-out tool for predicting lymph node metastases.

Voskuilen et al.[25] conducted a retrospective multicenter study to assess the diagnostic value of 18F-Fluorodeoxglucose PET/CT (FDG-PET/CT) for the detection of lymph node metastasis in 117 surgically treated UTUC patients. The authors reported that FDG-PET/CT yielded a promising diagnostic performance with a sensitivity of 82% and a specificity of 84% for predicting lymph node metastasis [25]. Despite improvements of preoperative lymph node staging thanks to modern imaging modalities [26], LND remains the standard to ensure true N-staging. Furthermore, the POUT trial, a phase III randomized controlled trial, showed the significant DFS benefit from adjuvant chemotherapy in UTUC patients with pathologically advanced features in the RNU specimen, including positive lymph node [10]. Therefore, accurate N-staging is, indeed, essential for selecting optimal candidates for adjuvant therapy.

Oncologic impact of pN+ versus pN0

Tables 1 and 2 summarize the patient demographics and oncologic outcomes of eligible studies included in the following sections, ‘Oncologic impact of pN+ versus pN0’ and ‘Therapeutic impact of LND’. Several retrospective cohort studies evaluated the differential oncologic outcomes in cN0 UTUC patients who underwent RNU with pathological node-positive (pN+) versus negative (pN0) disease. Among the patients who underwent LND in these studies, the pooled pN+ rate was 21.5% (range: 6.2–34%).

Table 1.

Patient demographics of eligible studies assessing oncologic outcomes stratified lymph node status with large cohort (more than 250 patients included)

Author	Year	Periods	Institution	Pts.	Surgical approach [n (%)]	Median number of LNs removed	Decision and extent of LND	Additional systemic therapies [n (%)]	Follow-up periods (months)	pN stage [n (%)]
Retrospective cohort study
Roscigno	2009	1987– 2007	Multi (International)	1130	Open: 924 (82) Lap: 206 (18)	NR	Surgeon's discretion (template)	AC: 187 (17)	Median (range) 45 (1–250)	pN0: 412 (36) pN+: 140 (13) pNx: 578 (51)
Burger	2011	1987– 2008	Multi (International)	785	Open: 715 (91) Lap: 70 (9)	3 (IQR: 2–6)	Surgeon's discretion (Hilar/regional)	AC: 69 (9)	Median (IQR) 34 (15–65)	pN0: 136 (17) pN+: 54 (7) pNx: 595 (76)
Mason	2012	1990– 2010	Multi (Canada)	1029	Open: 583 (57) Lap: 446 (43)	4.3 (mean)	Surgeon's discretion	AC: 112 (11)	Median (IQR) 20 (7.2–54)	pN0: 199 (20) pN+: 77 (7) pNx: 753 (73)
Yoo	2017	1998– 2012	Single (Korea)	418	Open: 184 (44) Lapa: 234 (56)	7 (IQR: 3–10)	Surgeon's discretion (template)	NR	Median 69	pN0: 116 (29) pN+: 16 (3) pNx: 286 (68)
Ikeda	2017	1985–2013	Multi (Japan)	399	Open: 296 (74) Lap: 103 (26)	6 (IQR: 3–10)	Surgeon's discretion	AC: 74 (19)	Median (IQR) 43 (7.2–54)	pN0: 182 (46) pN+: 40 (10) pNx: 177 (44)
Inokuchi	2017	1995–2009	Multi (Japan)	2037	Open: 1234 (61) Lap: 787 (39)	6 (IQR: 3–11)	Surgeon's discretion	NAC: 71 (3) AC: 329 (16)	Median (IQR) 46 (22–76)	pN0: 955 (47) pN+: 223 (11) pNx: 859 (42)
Li	2021	2001–2021	Multi (Taiwan)	1340	Hand-assisted:741 (55) Lap: 458 (34) Robotic: 141 (11)	NR	Surgeon's discretion	NR	NR	pN0: 278 (21) pN+: 58 (4) pNx: 1004 (75)
Hsieh	2022	2000–2015	Single (Taiwan)	520	NR	NR	Surgeon's discretion	AC: 107 (21) NAC was excluded	Mean ± SD 48 ± 29	pN0: 303 (58) pN+: 20 (3.8) pNx: 197 (38)
Leeb	2022	2001–2021	Multi (Taiwan)	658	NR	4	Surgeon's discretion	NR	Median LND (-): 33 LND (+): 24	pN0: 159 (24) pN+: 36 (5.4) pNx: 463 (70)
Hakimi	2022	2006–2019	Multi (ROBUUST)	877	All robotic	Mean ± SD pN0: 9.8 ± 9.5 pN+: 10.2 ± 9.5	Surgeon's discretion	NAC: 77 (8.8) AC: 89 (10)	Mean ± SD pN0 : 15.6 ± 18 pN+:10.9 ± 13.6 pNx:13.7 ± 16.7	pN0: 285 (32) pN+: 73 (8.3) pNx: 519 (59)
Population-based study
Lughezzani	2010	1988–2004	SEER database	2842	NR	NR	Surgeon's discretion	NR	Median (range) 43 (1–203)	pN0: 1835 (64) pN+: 242 (9) pNx: 747 (26)
Chappidi	2016	2004–2012	SEER database	2862	NR	2 (IQR: 1–5)	Surgeon's discretion	NR	Median (IQR) 28 (11–53)	pN0: 466 (16) pN+: 255 (8.9) pNx: 2141 (75)
Lenis	2018	2010–2013	NCDB	3116	Open: 969 (32) Lap: 1385 (44) Robotic: 762 (24)	3 (IQR: 1–7)	Surgeon's discretion	NAC: 60 (1.9) AC: 400 (13)	NR	pN0: 765 (25) pN+: 257 (8.3) pNx: 2094 (67)
Dongc	2019	2004–2014	SEER database	2731	NR	2 (IQR: 1–5)	Surgeon's discretion	AC: 345 (13)	Median 31	pN0: 491 (18) pNx: 2240 (82)
Zhai	2019	2004–2015	SEER database	7278	NR	NR	Surgeon's discretion	NR	NR	pN0: 1296 (18) pN+: 665 (9.1) pNx: 5317 (73)
Lec (1)	2021	2006–2013	NCDB	1421 (RNU)	NR	3 (IQR: 1–6)	Surgeon's discretion	NAC: 23 (1.6) AC: 167 (12)	Median (IQR) 33 (20–56)	pN0: 258 (18) pN+: 20 (1.4) pNx: 1143 (80)
Lec (2)	2021	2006–2013	NCDB	541 (SU)	NR	5 (IQR: 2–8)	Surgeon's discretion	NAC: 7 (1.3) AC: 74 (14)	Median (IQR) 32 (18–52)	pN0: 144 (27) pN+: 15 (2.8) pNx: 382 (71)

AC, adjuvant chemotherapy; IQR, interquartile range; Lap: laparoscopy; LN, lymph node; NAC, neoadjuvant chemotherapy; NCDB, National Cancer Database; NR, not reported; Pts. patients; RNU, radical nephroureterectomy; SD, standard deviation; SEER: Surveillance, Epidemiology, and End Results; SU, segmental ureterectomy.

^aIncluding laparoscopic, hand-assisted laparoscopic, robot-assisted radical nephroureterectomy.

^bOnly included >pT2 patients.

^cOnly included clinical or pathological node-negative patients.

Table 2.

Oncologic outcomes stratified by lymph node status of studies with large cohort (more than 250 patients included)

						5-year survival rates			HR (95% CI, or P value) from multivariable analysis
Author	Year	Periods	Institution	Pts.	pN stage [n (%)]	DFS (%)	CSS (%)	OS (%)	DFS		CSS		OS
Retrospective cohort study
Roscigno	2009	1987–2007	Multi (International)	1130	pN0: 412 (36)	71	77	NR	Ref.
					pN+: 140 (13)	29	35	NR	2.40 (P < 0.001)		2.39 (P < 0.001)		NR
					pNx: 578 (51)	66	69	NR	1.16 (P = 0.008)		1.46 (P = 0.007)		NR
Burger	2011	1987–2008	Multi (International)	785	pN0: 136 (17)	72	79	NR	Ref.
					pN+: 54 (7)	21	27	NR	2.0 (1.2–3.4)		2.1 (1.1–3.8)		NR
					pNx: 595 (76)	77	77	NR	1.1 (0.7–1.7)		1.3 (0.8–2.2)		NR
Mason	2012	1990–2010	Multi (Canada)	1029	pN0: 199 (20)	91a	72	66	Ref.	NR	Ref.	NR	Ref.	NR
					pN+: 77 (7)	80a	30	22	2.01 (1.18–3.42)	2.03 (1.31–3.14)	2.94 (1.32–6.55)	2.83 (1.54–5.18)	2.97 (1.47–6.01)	2.70 (1.56–4.69)
					pNx: 753 (73)	71a	75	66	1.08 (0.8–1.43)	Ref.	0.81 (0.51–1.28)	Ref.	1.04 (0.69–1.56)	Ref.
Yoo	2017	1998–2012	Single (Korea)	418	pN0: 116 (29)	73	NR	80	1.38 (0.91–2.09)		NR		1.06 (0.74–1.51)
					pN+: 16 (3)	13	NR	13
					pNx: 286 (68)	76	NR	72	Ref.
Ikeda	2017	1985-2013	Multi (Japan)	399	pN0: 182 (46)	78	85	NR	Ref.
					pN+: 40 (10)	33	44	NR	3.25 (1.90–5.55)		3.56 (1.90–6.69)		NR
					pNx: 177 (44)	62	73	NR	1.91 (1.23–2.96)		2.28 (1.31–3.95)		NR
Inokuchi	2017	1995-2009	Multi (Japan)	2037	pN0: 955 (47)	NR	NR	69	NR		1.13 (0.89–1.42)		1.34 (1.00–1.79)
					pN+: 223 (11)	NR	NR	30	NR
					pNx: 859 (42)	NR	NR	61	NR		Ref.		Ref.
Li	2021	2001-2021	Multi (Taiwan)	1340	pN0: 278 (21)	NR	NR	NR	Ref.
					pN+: 58 (4)	NR	NR	NR	NR		3.08 (1.96–4.84)		4.41 (2.56–7.59)
					pNx: 1004 (75)	NR	NR	NR	NR		1.10 (0.84–1.44)		1.14 (0.78–1.67)
Hsieh	2022	2000-2015	Single (Taiwan)	520	pN0: 303 (58)	76	85	77	Ref.
					pN+: 20 (3.8)	29	26	26	1.68 (0.86–3.28)		2.01 (1.00–4.39)		1.55 (0.79–3.02)
					pNx: 197 (38)	65	86	64	1.21 (0.84–1.76)		0.97 (0.56–1.69)		1.32 (0.90–1.93)
Leec	2022	2001-2021	Multi (Taiwan)	658	pN0: 159 (24)	60	75	69	0.89 (0.64–1.24)	Ref.	1.07 (0.68–1.67)	Ref.	1.09 (0.74–1.59)
					pN+: 36 (5.4)					P < 0.001		P = 0.01
					pNx: 463 (70)	64	77	68	Ref.		Ref.		Ref.
Hakimi	2022	2006-2019	Multi (ROBUUST)	877	pN0: 285 (32)	61b	91b	86b	1.01 (0.75–1.35)		1.58 (0.75–3.33)		1.60 (0.93–2.76)
					pN+: 73 (8.3)	35b	45b	42b	1.80 (1.20–2.71)		2.74 (1.34–5.61)		2.77 (1.59–4.84)
					pNx: 519 (59)	53b	90b	80b	Ref.
Population-based study
Lughezzani	2010	1988–2004	SEER database	2842	pN0: 1835 (64)	NR	81	NR	Ref.
					pN+: 242 (9)	NR	34	NR	NR		2.54 (P < 0.001)		NR
					pNx: 747 (26)	NR	78	NR	NR		0.99 (P = 0.9)		NR
Chappidi	2016	2004-2012	SEER database	2862	pN0: 466 (16)	NR	75	NR	Ref.
					pN+: 255 (8.9)	NR	34	NR	NR		2.33 (1.50–3.28)		2.01 (1.48–2.72)
					pNx: 2141 (75)	NR	72	NR	NR		0.97 (0.78–1.42)		1.12 (0.91–1.39)
Lenis	2018	2010–2013	NCDB	3116	pN0: 765 (25)	NR	NR	NR	Ref.
					pN+: 257 (8.3)	NR	NR	NR	NR		NR		1.87 (1.47–2.37)
					pNx: 2094 (67)	NR	NR	NR	NR		NR		1.03 (0.85–1.25)
Dongd	2019	2004–2014	SEER database	2731	pN0: 491 (18)	NR	NR	NR	Ref.
					pNx: 2240 (82)	NR	NR	NR	NR		1.27 (1.04–1.55)		1.28 (1.09–1.51)
Zhai	2019	2004-2015	SEER database	7278	pN0: 1296 (18)	NR	77	52	Ref.
					pN+: 665 (9.1)	NR	40	21	NR		2.49 (2.06–3.01)		2.20 (1.93–2.50)
					pNx: 5317 (73)	NR	74	47	NR		1.32 (1.13–1.53)		1.19 (1.08–1.30)
Lec (1)	2021	2006-2013	NCDB	1421 (RNU)	pN0: 258 (18)	NR	NR	NR	Ref.
					pN+: 20 (1.4)	NR	NR	NR	NR		NR		2.41 (1.29–4.47)
					pNx: 1143 (80)	NR	NR	NR	NR		NR		NR
Lec (2)	2021	2006-2013	NCDB	541 (SU)	pN0: 144 (27)	NR	NR	NR	Ref.
					pN+: 15 (2.8)	NR	NR	NR	NR		NR		1.43 (0.67–3.07)
					pNx: 382 (71)	NR	NR	NR	NR		NR		NR

Statistical significance is represented as italic font. Values are described as rounded up at the first decimal point in survival rates and at the third decimal point in hazard ratio. AC, adjuvant chemotherapy; CI, confidence interval; CSS, cancer-specific survival; DFS, disease-free survival; HR, hazard ratio; IQR, interquartile range; Lap, laparoscopy; LN, lymph node; NAC, neoadjuvant chemotherapy; NCDB, National Cancer Database; NR, not reported; OS, overall survival; Pts. patients; Ref., referent; RNU, radical nephroureterectomy; SD, standard deviation; SEER, Surveillance, Epidemiology, and End Results; SU, segmental ureterectomy.

^aNot adjusting covariates using univariable analysis.

^bTwo-year survival outcomes.

^cOnly included >pT2 patients.

^dOnly included clinical or pathological node-negative patients.

In 2009, Roscigno et al.[27] conducted an international multicenter study with 1130 patients to assess the impact of pN-status as well as LND on oncologic outcomes. The authors showed inferior DFS (hazard ratio 2.40, P < 0.001) and CSS (hazard ratio 2.39, P < 0.001) in pN+ patients compared with pN0 patients using multivariable Cox regression analysis [27]. As shown in Table 1, compared with pN0, the negative prognostic impact of pN+ with regard to DFS and CSS has been supported by several studies, such as Burger et al.[28] in 2011 (hazard ratio for DFS: 2.0, hazard ratio for CSS: 2.1, n = 785), Mason et al.[29] in 2012 (hazard ratio for DFS: 2.01, hazard ratio for CSS: 2.94, n = 1029), Ikeda et al. in 2017 (hazard ratio for DFS: 3.25, hazard ratio for CSS: 3.56, n = 399) [30], Li et al. in 2021 (hazard ratio for CSS: 3.08, hazard ratio for CSS: 4.41, n = 1340) [31], Hsieh et al.[33] in 2022 (hazard ratio for CSS: 2.01, n = 520) [32], Lee et al.[34^▪▪] in 2022 (P < 0.001 in DFS and P = 0.01 in CSS, n = 658), and Hakimi et al. in 2022 (HR for DFS: 1.80, HR for CSS: 2.74, n = 877). In addition, two studies showed that pN+ was associated with worse OS in UTUC patients compared to pN0; for example, Mason et al. in 2012 (HR for OS: 2.70, n = 1029) and Hakimi et al. in 2022 (hazard ratio for OS: 2.77, n = 877) [29,31]. The negative prognostic impact of pN+ on CSS and OS has been confirmed in population-based studies with 5-year DFS, CSS, and OS ranging from 13 to 33%, 26 to 44%, and 13 to 30% in pN+ patients and 71 to 78%, 72 to 85%, and 52 to 80% in pN0 patients, respectively [35,36^▪,37–39]. Taken together, these data highlight the negative prognostic impact of lymph node metastasis at RNU on survival endpoints.

Therapeutic impact of lymph node dissection

To date, there is no head-to-head randomized controlled trial, assessing the oncologic impact of LND versus no LND. Several retrospective studies have assessed the potential impact of LND on oncologic outcomes, comparing patients who underwent LND (pN0 and/or pN+) with those who did not (pNx). When comparing oncologic outcomes of pN0 with pNx patients, data from retrospective studies are contradictory, with half of them showing worse oncologic outcomes for pNx patients and the other half not [27–32]. For example, Roscigno et al.[27] and Ikeda et al.[30] reported that, compared with pN0 patients, pNx patients had worse DFS and CSS (Table.1). In addition, Dong et al. and Zhai et al. confirmed worse OS in patients with pNx than in those with pN0 using the surveillance, epidemiology, and end results (SEER) database [39,40]. Theoretically, the pNx cohort includes both pN0 and pN+ patients if they had undergone LND; therefore, when compared with pN0 patients, worse oncologic outcomes in pNx patients (no LND) would be expected. On the contrary, Burger et al., Li et al., Hsieh et al., Lughezzani et al., Chappidi et al., and Lenis et al. showed no differences in DFS, CSS, and OS between patients with pNx and pN0, suggesting that survival outcomes are affected by patient selection, as patients harboring advanced disease associated with potential lymph node metastasis being more likely to undergo LND [28,31,32,35,37,38].

Zhai et al.[39] assessed the impact of performing LND during RNU on CSS and OS using 7278 patients from the SEER database. In multivariable Cox regression analyses, performing LND was associated with a higher OS or CSS in the entire cohort; however, LND did not improve OS in patients with pT1 and pT2 disease [39]. In addition, Dong et al. evaluated the survival impact of LND, including only clinical or pathological N0 patients using the SEER database [40]. After propensity score matching, CSS and OS in cNx-pN0 patients were significantly better than those in cN0-pNx patients [40]. When limiting the analysis to patients who received adjuvant treatment, the authors found that performing LND improved CSS and OS (hazard ratio for OS: 0.57, P = 0.013, hazard ratio for CSS: 0.61, P = 0.046) [40]. These findings from studies using SEER database suggest that LND could improve survival outcomes in patients with clinically node-negative UTUC, especially for those with advanced diseases (T3–4 stages) even in patients with pN0 disease likely by removing micrometastasis. Further investigation is needed to establish more robust evidence regarding the potential benefits of LND on survival, specifically taking into consideration the effect of anatomical extent of LND impact on oncologic outcomes.

Lymph node dissection in clinical node-positive patients

Given the high prevalence of lymphatic spread in UTUC, clinical node-positive (cN+) UTUC patients should be considered for systemic platinum-based combination chemotherapy followed by RNU and LND; however, there is no guideline endorsement of this treatment strategy due to a lack of data to support this strategy that is used in urothelial cancer of the bladder [6,41]. Regarding the optimal timing of chemotherapy for cN+ UTUC patients who were candidates for RNU, Shigeta et al.[42] recently published the results from a retrospective multicenter study comparing oncologic outcomes of systemic induction chemotherapy followed by RNU versus RNU followed by adjuvant chemotherapy. The authors reported that systemic induction chemotherapy followed by RNU was associated with significantly better DFS and CSS compared with RNU followed by adjuvant chemotherapy [42]. However, although this data reflected clinical reality, there was severe selection bias that patients who did not respond to induction chemotherapy received second-line therapy (e.g. immunotherapy) instead of surgical intervention [43]. This is in line with the concept that cN+ is to be considered as systemic disease that requires systemic therapy. Indeed, most studies assessing the impact of LND during RNU excluded cN+ patients, thereby, making the assessment of the natural history of this disease difficult. Recently, Piontkowski et al.[44^▪] conducted a retrospective study assessing the impact of LND in 423 patients with cN+ UTUC using the National Cancer Database (NCDB). The authors found that LND does not impact OS (hazard ratio 0.93, 95% CI 0.70–1.24) after adjusting for the effects of covariates in multivariable analysis, irrespective of clinical N stage (i.e. cN1 vs. cN2–3) [44^▪]. Therefore, to date, the oncologic benefit of performing LND and its optimal extent for cN+ UTUC remain controversial. Based on the biology of cN+ disease, timely and effective systemic therapy is key to the success of the treatment for patients with cN+; LND may have a limited impact on oncologic outcomes. However, accurate N-staging is also essential for the decision on additional systemic therapy in this clinical setting [9,10].

Optimal extent of lymph node dissection

The therapeutic benefit of LND may depend on the number of lymph node removed. As such, Lec et al.[36^▪] reported that LND was associated with improved OS in RNU patients when more than three lymph nodes were removed (hazard ratio 0.58, 95% CI 0.39–0.89). In addition, population-based studies showed that a lower number of lymph nodes removed was associated with worse CSS and OS, even in pN0 patients [35,37]. In 2009, Roscigno et al. assessed the optimal cut-off value of how many lymph node should be removed prognosticating CSS using a multicentric cohort constituting 522 patients [15]. The authors demonstrated that eight lymph nodes were the best cut-off in predicting CSS (hazard ratio 0.42; P = 0.004) [15]. Of note, analyzing in patients with pN0, a higher number of lymph nodes removed remained associated with better disease recurrence (hazard ratio 0.97; P = 0.03) and CSS (hazard ratio 0.96; P = 0.04), whereas there was no association between the number of lymph nodes removed and disease recurrence and CSS in pN+ patients [15]. However, in previous studies, the majority of patients undergoing RNU had less than eight lymph nodes removed (Table 1); therefore, standardized LND templates are needed to ensure that the lymph nodes are adequately removed in the right anatomic area.

In 2010, Kondo et al.[45] advocated template-based LND depending on the primary tumor site. This template was based on their preliminary study assessing the association between location of lymph node metastasis and the primary tumor site [16]. The authors demonstrated that template-based LND affects survival outcomes and is superior to just assessing the number of removed lymph nodes [45]. Since then, several studies have confirmed that performing an anatomical template-based LND is the best approach/concept from a diagnostic and oncologic perspective [46,47,48^▪,49]. A systematic review showed that template-based LND and the level of completeness of LND both improve CSS in patients with muscle-invasive disease and reduce the risk of local recurrence [21]. In addition, Kanno et al.[48^▪] recently conducted a multicenter retrospective study to compare oncologic outcomes in patients who underwent template-based LND during RNU and those who did not. After propensity score matching using confounders such as, age, sex, presence of hydronephrosis, and clinical T stage, the authors showed that the estimated 5-year recurrence-free survival was higher in patients who underwent template-based LND (86.8%) compared with those who did not (64.2%) (P = 0.014) [48^▪]. Therefore, a template-based LND should be offered to all patients who are planned for RNU for invasive nonmetastatic UTUC. The modified template for LND based on Kondo et al. is described in Fig. 1. This template includes renal hilar, paracaval, and retrocaval lymph nodes for tumors located in right renal pelvis or upper 2/3 of the ureter, and renal hilar and para-aortic lymph nodes for tumors located on the left-side. For tumors located in the lower 1/3 of the ureter, common iliac, external iliac, internal iliac, and obturator lymph nodes are included in the template, irrespective of tumor side. A multicenter study by Matin et al.[47], that included 73 patients, confirmed the validity of this template. In patients with pelvic or upper ureteral tumors, the template without interaortocaval lymph nodes covered 82.9% in right-side tumors and 86.9% in left-side tumors [47]. However, when including interaortocaval lymph nodes, it increased to 95.8% in right-side tumors and 90.2% in left-side tumors [47]. This suggests that, in the case of proximal/upper ureteral tumors, interaortocaval nodes should be removed (Fig. 1).

FIGURE 1.

Modified anatomical lymph node templates depending on tumor location based on the studies by Kondo et al. and Matin et al. Data from [45,47].

Lymph node density

Other than the number of lymph nodes removed, lymph node density, defined as the number of positive lymph nodes divided by the number of lymph node removed, has been considered a prognostic factor for oncologic outcomes [29,50]. In 2009, Bolenz et al.[50] assessed the impact of lymph node density on oncologic outcomes in 432 patients with a median of four lymph nodes removed. The authors showed that a lymph node density of more than 30% was significantly associated with worse recurrence and CSS [50]. Mason et al.[29] assessed the prognostic utility of lymph node density in 276 patients with a mean of 4.3 lymph nodes removed. The authors showed that a lymph node density of more than 20% was associated with worse DFS and OS, despite the number of positive lymph nodes; the number of lymph nodes removed was not associated with any survival outcomes [29]. Lymph node density seems a promising prognostic marker for oncologic outcomes after LND; further investigation with template-based lymph node density is needed to understand the complete prognostic value of LND.

Surgical perspective of lymph node dissection

RNU can be performed by an open, laparoscopic, or robotic approach [51]. Laparoscopic RNU has been suggested to result in worse oncologic outcomes in the treatment of aggressive tumors (i.e. pT3-4) compared with open approach [52], while subsequent retrospective studies showed comparable oncologic outcomes between open and laparoscopic procedures [53,54]. To date, the utilization of robot-assisted RNU has increased based on comparative studies showing noninferiority with regard to perioperative and oncologic outcomes [51,55,56]. However, laparoscopic LND is inferior to other approached for LND because of the limited surgical field. Indeed, Peyronnet et al.[52] showed significantly lower LND rates for laparoscopic compared with open approaches. When performing template-based LND using a laparoscopic retroperitoneal approach, an improved technique without patient repositioning has been advocated [57]. In the context of the number of lymph node removed, Hattori et al.[58] initially reported that an open approach was associated with an increased number of lymph node removed compared with a laparoscopic approach, while two other studies showed no differences between the procedures [15,53]. When comparing robotic and laparoscopic approaches, Melquist et al. reported that a robotic approach was associated with improved lymph node yield (median 21 lymph nodes for robotic and 11 lymph nodes for laparoscopic approaches, P < 0.001). Yet, severe selection biases because of retrospective study design should be considered when interpreting these contradictory results. However, several population-based studies reported increased proportion of performing LND in the era of robotic approach, suggesting superior technical feasibility of LND through a robotic approach [37,51,56].

For the assessment of postoperative complications due to LND, Pearce et al.[59] demonstrated that performing LND increased by 30% the risk of postoperative complications compared with not performing LND. Additionally, several studies aimed to assess the safety of extended LND (ipsilateral renal hilar, para-aortic, pelvic lymph node) during RNU. A prospective study by Rao et al.[60], which evaluated 20 patients (open: n = 10, laparoscopic: n = 4, robot-assisted: n = 6), showed that one patient experienced a severe complication (Clavien–Dindo classification: IIIb), such as postoperative lymphatic leakage requiring surgical procedure, while the other eight suffered from minor complications. Another prospective study by Huang et al.[61] assessed the safety of extended LND during laparoscopic extraperitoneal RNU in 39 patients, reporting no severe complications. A recent published retrospective study led by Kanno et al.[62^▪] assessed the perioperative complication and its management after template-based LND during laparoscopic extraperitoneal RNU. The authors reported a high rate of postoperative chylous leakage (14/88), which was conservatively managed and prevented by meticulously clipping each lymphatic vessel [62^▪].

CONCLUSION

LND during RNU is the standard treatment for invasive nonmetastatic UTUC, owing to its diagnostic, staging, prognostic, as well as therapeutic potential. Although a higher number of lymph node removed is associated with better oncologic outcomes, the template-based LND concept is superior to the number of lymph nodes removed. Robot-assisted RNU seems to facilitate performing a meticulous LND compared with the laparoscopic approach. Surgeons should be aware and prevent postoperative complications associated with LND, such as lymphatic and/or chylous leakage.

Acknowledgements

None.

Authors’ contributions: T.Y. contributed to protocol/project development, data collection and management, and manuscript writing/editing. T.K., M.v.D., and E.L. contributed to manuscript writing/editing. T.K. contributed to supervision and manuscript editing. S.F.S. contributed to supervision, protocol/project development/management, and manuscript editing.

Financial support and sponsorship

None.

Conflicts of interest

T.K. is a paid consultant/advisor of Astellas, Bayer, Janssen and Sanofi. S.F.S. received as follows: honoraria – Astellas, AstraZeneca, BMS, Ferring, Ipsen, Janssen, MSD, Olympus, Pfizer, Roche, Takeda; consulting or advisory role – Astellas, AstraZeneca, BMS, Ferring, Ipsen, Janssen, MSD, Olympus, Pfizer, Pierre Fabre, Roche, Takeda; speakers Bureau – Astellas, Astra Zeneca, Bayer, BMS, Ferring, Ipsen, Janssen, MSD, Olympus, Pfizer, Richard Wolf, Roche, Takeda.

REFERENCES AND RECOMMENDED READING

Papers of particular interest, published within the annual period of review, have been highlighted as:

• ▪ of special interest

• ▪▪ of outstanding interest