Lymph node mapping in patients with bladder cancer undergoing radical cystectomy and lymph node dissection to the level of the inferior mesenteric artery


  • Jørgen B. Jensen,

    1. Department of Urology, Aarhus University Hospital, Skejby, and Institute of Pathology, Aarhus University Hospital, Aarhus Sygehus NBG, Denmark
    Search for more papers by this author
  • Benedicte P. Ulhøi,

    1. Department of Urology, Aarhus University Hospital, Skejby, and Institute of Pathology, Aarhus University Hospital, Aarhus Sygehus NBG, Denmark
    Search for more papers by this author
  • Klaus M.-E. Jensen

    1. Department of Urology, Aarhus University Hospital, Skejby, and Institute of Pathology, Aarhus University Hospital, Aarhus Sygehus NBG, Denmark
    Search for more papers by this author

Jørgen Bjerggaard Jensen, Department of Urology, Aarhus University Hospital, Skejby, Denmark. e-mail:


Study Type – Therapy (case series)
Level of Evidence 4


To evaluate extended lymph node dissection (LND) as a nodal staging tool in the treatment of invasive carcinoma of the urinary bladder and to suggest a reasonable proximal limit of the dissection.


In all, 170 patients underwent radical cystectomy with extended LND up to the level of the inferior mesenteric artery. Specimens were evaluated as 13 separate packages from pre-designated anatomical locations. The number of LNs and presence of positive LNs (LN+) at each location was prospectively registered.


The median (range) number of LNs removed was 24 (6–62). In all, 25.3% of the patients had LN+. The median (range) number of LN+ was 2 (1–20). Advanced T-stage was correlated with a higher risk of LN+ but not to the specific location of the LN+. Two patients had LN+ above the common iliac bifurcation with no LN+ more distally located within the pelvic region. All other patients with LN+ above the common iliac bifurcation had more distally located LN+. There were no skip lesions to LNs above the aortic bifurcation.


Extended LND above the common iliac bifurcation including the presacral area provides a more accurate LN staging compared with a standard pelvic LND. Extending the limits above the aortic bifurcation is not necessary from a staging perspective.


radical cystectomy


(pelvic) lymph node (dissection)


positive LNs


transurethral resection of the bladder


squamous cell carcinomas.


Radical cystectomy (RC) remains the most efficient treatment in patients with invasive carcinoma of the urinary bladder with no distant metastases upon diagnosis [1]. Since Marshall and Whitmore in 1949 [2] and later Leadbetter and Cooper in 1950 [3] suggested dissection of the regional lymph nodes (LNs) during RC, discussions have centred on the extent of the LN dissection (LND) [4]. Whether the LND is purely a staging method of LN involvement (N-stage) or if it should be considered part of intended curative surgery in selected patients with positive LNs (LN+) remains a controversial issue. Skinner [5] published in 1982 a series of patients with long-term survival despite LN metastases, the general perception has been that patients with only a few LN+ can be cured by surgery alone. Agreement upon the ‘extent’ of the extended LND still remains a controversial issue [6].

Limited pelvic LND (PLND) restricted to the obturator fossae has proven to be insufficient for accurate LN staging [6–10]. Standard PLND therefore consists of removal of all lymphatic and fatty tissue up to the level of the bifurcation of the common iliac artery, thus including the obturator, internal iliac and external iliac LNs bilaterally. A better nodal staging is thought to be provided by an extended LND including the LNs above the common iliac bifurcation. However, almost the same patients have been used for arguments ‘for’[8] and ‘against’[7] a ‘sentinel node’ region located in the endopelvic region.

The present study sought to describe routes of lymphatic metastases to evaluate the staging aspect of extended LND compared with a more limited PLND. The proximal limit of extended LND is discussed.


From January 2004 to January 2009, 282 patients underwent RC with LND at the Department of Urology, Aarhus University Hospital. In all, 32 patients with previous oncological treatment (radiotherapy or systemic chemotherapy), previous radical prostatectomy with LND and patients with gross metastatic tissue left behind from widespread LN metastases found preoperatively, were excluded from the study.

Although the LND was intended to include all locations in all patients, tissue from locations above the bifurcation of the aorta was removed in only 170 patients (68%). Similarly in only 216 of the patients (86%) was tissue actually removed from locations above the bifurcation of the common iliac artery. In all, only 34 patients (14%) underwent a LND similar to a standard template. The reasons for these incomplete dissections were various: surgical technical problems in adipose patients, extensive atherosclerotic disease, major bleeding, fibrous tissue difficult to resect, anaesthetic problems requiring short operation time and, in a few cases, nerve-sparing technique where omission of dissection of the presacral and most proximal LNs was chosen based on the operating surgeons preference. Because of the design of the project as a LN mapping study and not a survival study, we decided it was methodologically more correct to exclude patients with an insufficient LND rather than including them in the analyses. Thus, 170 patients (128 male and 42 female) were prospectively included in the present study.

Indication for RC was invasive carcinoma in 165 patients and treatment-refractory carcinoma in situ in five patients. Three of these five patients had invasive carcinoma in the RC specimen. The median (range) patient age at surgery was 64 (39–80) years. Routine CT of the abdomen and CT or X-ray of the chest showed no evidence of metastases at the time of surgery in all patients. Final tumour stage (T-stage) was the highest of the before RC stage (pathologically verified transurethral resection of the bladder [TURB] stage) and the pathological stage (pT) of the RC specimen.

The upper limit of the extended LND was at the level of the inferior mesenteric artery. The lower limits were the inguinal ligaments and the pelvic floor. The genitofemoral nerves made the lateral limitations. Specimens were sent for pathological examination from 12 pre-designated locations as separate packages numbered 1–12. LNs within the RC specimen were registered as location 13 (Fig. 1).

Figure 1.

The 13 anatomical locations of the extended LND. i.c.v., inferior caval vein; i.m.a., inferior mesenteric artery; c.i.a., common iliac artery; I.I.a, internal iliac artery; e.i.a, external iliac artery; g.f.n, genitofemoral nerve; o.n., obturator nerve; cyst., RC specimen.

Pathological examination included meticulous palpation in bright light and sectioning of the tissue into thin slices if required to identify LNs. This method has been proved to identify 95% of the LNs in separate package specimens [11]. No fat-clearing or LN-revealing solutions were used in this material. The number of LNs and number of LNs with metastases (LN+) were prospectively recorded from each location separately. Two collaborating pathologists evaluated all specimens.

In 25 of the most recently operated patients, the LNDs were timed to estimate the additional duration of surgery required for the extended LND. In this way, the duration of dissection of LNs included in a standard PLND and the duration of dissection of the more proximal LNs (above the bifurcation of the common iliac artery) were recorded separately.

Comparison of incidences was assessed using Fisher’s exact test or chi-square test where appropriate. P values were based on two-sided testing at a 5% significance level.


Histologically, 165 (97%) of the tumours were TCCs, four (2%) squamous cell carcinomas (SCCs) and one (0.6%) an adenocarcinoma. Adenocarcinoma of the prostate was present in 34% of male patients. One patient had LN+ in the left common iliac LNs (location 6) with metastases from a prostate adenocarcinoma with no evidence of metastases from the TCC. No other metastatic spread from prostate carcinomas was identified in any of the patients.

A median (range) of 24 (6–62) LNs were identified [mean (sd) 25.4 (8.44)]. Figure 2 shows the approximation to a normal distribution of the number of removed LNs.

Figure 2.

The number of LNs recovered in the 170 extended LND specimens. The normal distribution is shown.

LN metastases from carcinoma of the bladder were present in 43 patients (25.3%) with a median (range) 2 (1–20) LN+[mean (sd) 4.2 (4.60)].

The total number of retrieved LNs was not significantly different comparing patients with LN+ and patients with no LN metastases, at a mean (sd) of 27.4 (9.0) and 24.8 (8.2), respectively (P = 0.08).

Total number of LNs was independent of patient age comparing the youngest and oldest half of the cohort whereas the number was slightly higher in male patients compared with female patients (mean 26.7 and 21.4, respectively). However, this difference was not statistically significant (P = 0.3).

Incidence of LN+ was independent of gender. It was higher in the oldest half of the cohort (29.4% vs 21.2%). This difference was not significant in univariate analysis (P = 0.29) and even less significant in multivariate analyses adjusted for tumour stage.

The patients were divided into quartiles according to time period of surgery. There were no statistically significant variations in the number of LNs retrieved or frequency of LN+ between these quartiles and there was no variation between surgeon in these variables.

LN+ correlated with final T-stage is shown in Table 1. As expected, the incidence of LN+ was higher in locally advanced disease (≥T3a) compared with disease limited to the bladder wall (≤T2; P < 0.001). In patients with perivesical tumour invasion the risk of LN+ was higher in patients with macroscopic (pT3b) rather than just microscopic (pT3a) invasion (51.4% vs 20.0%). However, this difference was not statistically significant (P = 0.06).

Table 1.  Incidence and location of LN metastases correlated with T-stage
T-stageNo. patientsNo. patients with LN+ (%)No. patients with LN+ above the common iliac bifurcation (%)No. patients with LN+ above the aortic bifurcation (%)
  • *

    Final T-stage is highest of the before RC T-stage (pathologically verified TURB stage) and pathological (RC) pT-stage. CIS, carcinoma in situ.

Before RC:
 Ta/CIS  5 0 00
 T1 35 5 (14.3) 2 (6.3)2 (6.3)
 T2 +12435 (28.2)15 (12.1)5 (4.0)
 T4a  6 3/6 1/60
RC pT-stage
 pT0 44 4 (9.1) 2 (4.5)1 (2.3)
 pTa/pTIS 13 2/13 1/131/13
 pT1 21 2 (9.5) 1 (4.8)1 (4.8)
 pT2 24 5 (20.8) 1 (4.2)0
 pT3a 15 3/15 2/150
 pT3b 3518 (51.4) 7 (20.0)1 (2.9)
 pT4a 18 9 (50.0) 4 (22.2)3 (16.7)
Final T-stage*    
 Ta/CIS  2 0 00
 T1 32 3 (9.4) 2 (6.3)2 (6.3)
 T2 6810 (14.7) 3 (4.4)1 (1.5)
 T3a 15 3/15 2/150
 T3b 3518 (51.4) 7 (20.0)1 (2.9)
 T4a 18 9 (50.0) 4 (22.2)3 (16.7)
Total17043 (25.3)18 (10.6)7 (4.1)

In all, 44 patients (25.9%) had no remaining tumour in the RC specimen (pT0). These patients had a 9.1% risk of LN+. This risk was not significantly different comparing patients with muscle-invasive disease (T2) and patients with less invasive disease limited to the lamina propria (T1) in the preceding TURB specimens (P = 1.00).

LN+ above the common iliac bifurcation were present in 18 patients (11% of all patients, 42% of patients with LN+). LN+ above the aortic bifurcation were present in seven patients (4% of all patients, 16% of patients with LN+). LN+ above the iliac bifurcation were more common in more advanced tumour stages. However, there were no significant differences between the different tumour stages and presence of LN+ above the aortic bifurcation. This made it unreliable to try to predict risk of LN+ and the required extension of the LND based on the preoperative presumed T-stage. Even non-muscle-invasive tumours had a risk of distant LN metastases (Table 1).

Table 2 shows the percentage of specimens where no LNs could be identified in the retrieved specimens, the number of LNs identified from the different locations and the percentage of patients with LN+ at different locations. The highest numbers of LNs were located in the proximity of the bladder and especially in the obturator fossae (locations 8 and 11) and laterally in the external iliac artery (locations 7 and 12). LN+ were more common in the lower locations except for the perivesical tissue (locations 9 and 10, and the RC specimen). The presacral area (location 5) was involved in five patients (3% of all patients, 12% of patients with LN+).

Table 2.  Specimens without LNs, number of LNs removed and patients with LN+ stratified by location
LocationSpecimens with no LNs, %Mean (range) no. LNs removedPatients with LN+, n (%)% patients with LN+ of LN+ patients only
1 Para-caval32.01.2 (0–4) 3 (1.8) 5.0
2 Inter-aortocaval15.31.5 (0–7) 3 (1.8) 5.0
3 Para-aortic32.11.1 (0–5) 4 (2.4) 6.7
4 Right common iliac 3.02.3 (0–8) 8 (4.7)13.3
5 Presacral16.01.8 (0–7) 5 (2.9)10.0
6 Left common iliac22.41.6 (0–6) 7 (4.1)13.3
7 Right external iliac 4.14.0 (0–14)14 (8.2)30.0
8 Right obturator fossa 1.83.8 (0–15)20 (11.8)43.3
9 Right internal iliac45.90.9 (0–6) 2 (1.2) 3.3
10 Left internal iliac49.00.9 (0–6) 4 (2.4)15.0
11 Left obturator fossa 1.23.5 (0–12)16 (9.4)41.7
12 Left external iliac 3.63.5 (0–13)12 (7.1)28.3
13 Peri-vesical68.90.8 (0–5) 8 (4.7)15.0

A ‘sentinel node area’ was estimated based on 21 patients with LN+ where only one location was involved. Figure 3 shows that this ‘sentinel node area’ and in all but two patients was located in LNs inferior to the common iliac bifurcation. One patient with a T3a TCC tumour located in the right side of the bladder had one LN+ located in the presacral LNs (location 5) with no LN+ at other locations. One patient with a pT3a TCC tumour located to the left side of the bladder had one LN+ located in the left common iliac LNs (location 6) also with no evident LN+ at more distally located LNs. All the remaining patients with LN+ above the bifurcation of the common iliac artery had more distally located LN+.

Figure 3.

Estimation of ‘sentinel node area’: Location of LN+ in 21 patients with metastases in one location only.

All seven patients with LN+ above the aortic bifurcation (locations 1–3) also had more distally located LN+.

In all, 21 patients with LN+ and a strictly unilateral tumour in the bladder were analysed to evaluate the degree of contralateral LN+: 16 patients (76%) had only ipsilateral LN+, four (19%) had bilateral LN+ and one (5%) had only contralateral LN+. The contralateral LN+ were located as distally as the obturator fossa. Thus, location of the primary tumour cannot be used as a predictor of the side of LN+.

The material was analysed to assess whether the extended LND had provided a more accurate N-stage than a more limited PLND would have made.

A limited PLND restricted to the obturator fossae and perivesical tissue would only have diagnosed 70% of the present LN+ patients as such. Furthermore, in 28 patients (65% of the LN+ patients, 16% of all patients) LN+ would have been left behind due to LN+ located outside the limited template.

If a standard PLND template had been used (upper limit at the bifurcation of the common iliac artery and excluding the presacral area) two patients with LN+ would have been diagnosed N0 while the remaining 168 patients (99%) would have remained unchanged regarding N-stage. However, LN+ would have been left behind in 18 patients (11% of all patients).

Setting the upper limit of the LND at the aortic bifurcation would not have changed the N-stage in any of the patients. With this limit, LN+ would have been left behind in seven patients (4% of all patients).

Timing of the procedures showed that the dissection of locations 7–12 took a median (range) of 29 (16–50) min. The dissection of locations 1–6 took additional median (range) of 25 (14–48) min.


Bochner et al.[12] first described how submission of lymphadenectomy specimens as separate packages optimizes the pathological evaluation of the number of LNs compared to en bloc submission of the specimens. The number of LNs registered by this technique is thought to be close to the true number retrieved in RC specimens resulting in a higher total number of LNs [13].

Other studies of extended LND where separate package submission has been applied are available [7–10,13–15]. All studies had a somewhat identical ratio of LN+ patients. (22–30%). The number of LN+ also seems comparable with a median of 2–3 LN+. However, the median number of LNs per patient is more variable between the studies (range 22–68). One obvious explanation to this could be the different extension of an ‘extended’ LND. Fleischmann et al.[15] reported a median number of 22 LNs using a template with the proximal limit where the ureter crosses the iliac vessels. Less LNs can therefore be expected when comparing their findings with other mapping studies where the upper limit is at or even above the aortic bifurcation. But, even in studies with identical templates the median number of removed LNs varies. A true difference in patient anatomy or differences in surgical quality could cause this variation. Another explanation could be from how the pathologist defines a LN [16]. In the present study, we recorded only LNs with fibrous capsules and lymphatic nodules as LNs. Lymphocyte accumulations in the fatty tissue with no evident capsule were considered not to be LNs and therefore not recorded as such.

Capitanio et al.[17] suggested in a statistical model based on a multicentre study that a 90% threshold for detection of nodal metastasis occurs with ≈45 LNs, which can only be accomplished with a more extensive LND. However, the model was based on material with several different extensions of LND where the lowest number of dissected LNs was one LN in one patient. The number of removed LNs is very variable between patients and we think that the extension and thoroughness of LND is more important for the patient than the number itself. However, in the present cohort we also had patients with very few LNs removed despite a very extended LND. Thus a number of six LNs in a patient would make one question whether a true mobilization of the vessels, that is necessary to reveal the tissue lateral and posterior to these structures, had been made in the patient. However, in the present study, in patients with few LNs removed the LNs these tended to be larger than the average LNs in patients with a substantial number of LNs removed. This indicates that a fixed amount of lymphatic tissue could be present rather than a fixed number of LNs.

However, the true thoroughness of LND of all locations in the present study can be questioned based on the numbers in Table 2. Thus, there were fewer LNs in the more inaccessible left common iliac than in the right common iliac location. No potentially LN-bearing tissue was deliberately left undissected but proper exposure especially of the most proximal locations might have been a problem leading to the high number of LN packages with no LNs in these locations.

The high number of packages with no LNs in the immediate perivesical locations (internal iliac LNs) is more likely due to surgical technique, where most of this tissue was removed before the LND during the RC itself and thus included in the RC specimen (location 13: perivesical). The RC specimen is not subjected to the same minute search for LNs as the LN packages. LNs can therefore more easily be missed in the RC specimen.

The concept of LN density was suggested by Stein et al.[18] as a strong prognostic marker after RC taking both LN+, metastatic burden and extent and completeness of LND into consideration.

However, the variation in total number of LNs could make it somewhat unreliable to compare LN density in different RC series because the LN density is defined as number of LN+ divided by total number of LNs removed. However, different studies can still be considered comparable for mapping of metastatic spread in patients with LN+.

Standard PLND in the present patients would have had a nodal staging sensitivity of 99% in LN+ patients compared with extended LND, but LN+ would have been left behind in 18 patients (42% of LN+ patients). Other studies have proved long-time survival of LN+ patients with an additional advantage of extending the procedure to LNs above the common iliac bifurcation for both LN+ patients as well as supposedly LN-negative patients [19–21]. The explanation for a more favourable prognosis of LN-negative patients is suggested to be removal of micrometastases missed by regular pathological examination of the LNs [22].

Wishnow et al.[23] advocated in 1987 for a restricted PLND limited to the obturator fossae and perivesical tissue in patients with no evident LN+ upon gross examination and even suggested a unilateral dissection in patients with unilateral tumour only. However, this strategy results in an incorrect nodal staging in many patients and leaves behind LN+ in even more patients [7–9,22]. This is also emphasized in the present study where a limited PLND would have resulted in an incorrect N-staging in 30% of the LN+ patients and LN+ would have been left behind in 65% of the LN+ patients. Furthermore, there were several contralateral LN+ in patients with unilateral tumours. This is also consistent with earlier findings [7,24].

Before dissection of the presacral LNs, mobilization of the sigmoid mesentery is required. This manoeuvre creates a wide berth for the left ureter to cross over to the right side of the pelvis for urinary diversion and facilitates the necessary exposure to remove the presacral and medial common iliac LNs.

Stein [25] reported in a survey study the risk of having LN+ in the presacral area to be ≈6%. In the present study there were presacral LN+ in 2.9% of the patients and one patient had a solitary LN+ in this location. This emphasizes the need to include the presacral nodes in an extended LND. Interestingly this area is often left untreated in patients undergoing curative intended radiation therapy to avoid radiation damage to the rectum.

The cranial limitation of the extended LND still remains controversial. As mentioned above, some studies define the upper limit as the level where the ureter crosses the iliac vessels whereas others define the limit according to vascular structures only. Although the present study used separate package submission, it was not possible to determine whether LN+ at the level of the common iliac artery were situated below or above the crossing of the ureter. However, in our opinion LNs at this level are more constantly and safely defined by the vascular structures than by the ureter. A limit at or above the aortic bifurcation secures sufficient presacral dissection and possibly a safer negative margin as far as concerns nodal disease, i.e. proximally located LNs containing prospective micrometastases will be removed.

Whether the optimal proximal limit of LND should be the bifurcation of the aorta or at the level of the inferior mesenteric artery has not been determined yet [6]. No randomized studies concerning prognostic value of this issue are available. Decisions are therefore made based on nodal mapping studies.

In all available mapping studies, except one, no patients with skip lesions to the LNs above the aortic bifurcation have been identified [7–9,14]. However, Wiesner et al.[10] reported on 152 patients undergoing extended LND up to the level of the inferior mesenteric artery, where three patients had solitary LN+ above the aortic bifurcation. However, they still questioned the clinical relevance of removing para-aortic and para-caval LNs.

There are various reasons why only one study has reported patients with proximally located solitary lesions. It is probably a consequence of the rarity of this condition and the fact that not all patients in the mapping studies actually have had tissue from all these locations removed.

In the present study group with 250 patients, only 170 patients (68%) had tissue from these locations removed. For similar reasons that we omitted this dissection, Vazina et al.[9] only dissected tissue up to the bifurcation of the common iliac artery in 18 of 176 patients (10%) in their study . In the multicentre study reported by Leissner et al.[7] 211 of 290 patients (73%) had tissue removed from all anatomical locations including the para-aortic, para-caval and inter-aortocaval. Also, in that study a large proportion of the tumours were SCCs (18.6%), which was even more pronounced in a single centre sub-analysis of 200 of these patients (34% SCC) [8]. SCC is well known to metastasize without skip lesions, which is used in, e.g. the sentinel node concept of penile SCC [26]. This characteristic is not thought to be as pronounced in TCC. The dilution of the number of patients with TCC and with a complete LND above the aortic bifurcation, in combination with the infrequency of skip lesions, therefore seems to be a plausible explanation to why only one study has identified these lesions.

In the sentinel node study by Liedberg et al.[27] the proximal limit of the LND was the bifurcation of the aorta. Thus, single metastases to para-aortic nodes would have been missed in this otherwise excellent study.

However, it is important to stress the rarity of these prospective skip lesions. LN+ at these locations might also be an expression of more widespread disease and therefore not just a few resectable LN+ in potentially curable patients.

From a staging perspective, dissection of LNs above the aortic bifurcation therefore seems unnecessary. However, LN+ would have been left behind in seven patients (4% of all patients) in the present study if the proximal limit had been at the aortic bifurcation. Whether removal of these LN+ has an influence on cancer-specific survival will hopefully be apparent after a relevant follow-up period.

Extended LND is inevitably more time-consuming than a PLND limited to the pelvis. Leissner et al.[7] estimated that the extended LND took up to 60 min longer than a limited PLND. Brössner et al.[28] described a retrospective, non-randomized series of 92 patients where 46 underwent limited PLND and 46 underwent extended LND. The mean operative duration in the extended LND group was 63 min longer than the limited PLND group. However, it is not clear whether this additional time was exclusively used to make the extended LND and the limited and extended LNDs were performed by two different surgeons, respectively. The present study showed a prolongation of operative duration with a median of 25 min compared with a standard template.

The time needed to extend the LND to include the lower para-aortic LNs was estimated by El-Shazli et al.[29] to be a median (range) of 15 (10–20) min.

Brössner et al.[28] could not demonstrate an increase in perioperative complications from the extended LND. Similarly other studies have been unable to report increased morbidity from the extended LND procedure [13,29]. Conversely, we had one patient who required re-operation within the first postoperative day due to extensive bleeding from a branch of the inferior caval vein. This lesion was made during dissection of the most cranially placed LNs of the LND. However, complications were not systematically prospectively registered in the present material.

Despite a presumed somewhat equal morbidity of the extended LND compared with the standard PLND, it is associated with a prolongation of the operation time. Therefore suggestions have been made to restrict the extended LND to high-risk patients. But as shown in the present material, LN+ are, although more frequent in high T-stages, difficult to predict from preoperative tumour characteristics. Extended LND is therefore indicated in all patients undergoing RC, even in patients with non-muscle-invasive disease.

In conclusion, extended LND above the common iliac bifurcation is more accurate as a staging procedure compared with standard PLND restricted to the pelvic region. The extended LND should at least include dissection up to the aortic bifurcation and the presacral LNs. Extending the limits above the aortic bifurcation does not improve nodal staging, but might be indicated from an intended curative perspective.


This study was supported by the Danish Cancer Society.

A part of data in this article was presented at the 24th annual European Association of Urology congress in Stockholm, March 2009. The presentation was selected as the Best Poster Presentation in the Poster Session: ‘Cystectomy: technical variations – functional and oncological outcome’.


None declared.