Anatomy

Unfortunately, the anatomy of the lymph drainage of UTUC has not been well studied or described, and templates used from institution to institution have varied widely. This lack of uniformity, coupled with the differing drainage patterns observed depending on where the tumor originates, only muddies the debate.

The lymph drainage of the kidneys/renal pelvises appears to predominantly collect in a predictable pattern, following the blood supply. Kondo and colleagues mapped the location of regional lymph node metastasis and found that right renal pelvis tumors most often metastasize to the hilar nodes, followed by equal proclivity for the paracaval and retrocaval nodes. Tumors of the upper- and mid-right ureter most often go to the retrocaval and interaortocaval nodes, and tumors of the lower ureter on either side go to their respective common iliac, obturator, and internal iliac nodes. Left-sided renal pelvic tumors metastasize to the left hilar nodes, followed by the para-aortic nodes, followed by interaortocaval nodes. This mapping study was one of the most complete studies to date, but was limited to a total of 42 patients, of which only 23 were actually confirmed pathologically. Although the study was small, it follows the patterns well recognized in lymph drainage, namely the drainage follows the blood supply. These authors should also be commended for accurate communication about their dissection borders. Based on these findings, they proposed that LND for right-sided renal pelvic and mid- and upper-ureteral tumors should include the hilar nodes, the paracaval, interaortocaval, and retrocaval lymph nodes, from the level of the renal hilus to the aortic bifurcation. For left-sided renal pelvic tumors and mid- and upper ureters, dissection should include para-aortic and hilar nodes from the renal hilus to the aortic bifurcation. This is a similar dissection scheme proposed by others ; however, it should be mentioned that there are widely variable templates and practices and no standard dissection. It seems clear that if an LND is to be performed, the hilar nodes must be included. Others have confirmed that the hilar nodes are the primary site of metastasis for renal pelvic tumors, accounting for between 30% and 50% of nodal metastasis. The other regional sites (such as para-aortic, retrocaval, paracaval and interaortocaval) are involved less frequently, although some have found higher lymph node (LN) involvement to these regions than even the hilar nodes. Various authors recommend varying degrees of aggressiveness from templates, nearly mirroring RPLND templates for testis cancer on 1 extreme, to less aggressive templates where they only remove hilar nodes. These data, or lack thereof, indicate an opportunity for improvement with more lymph node mapping studies, clear definitions of anatomic boundaries used in LND, and larger series.

The overall incidence of LN involvement in patients with UTUC is between 12% and 25%. LN involvement increases with pathologic stage and grade. For example, 1 study showed that as pathologic T stage increased from T1, T2, T3 to T4, the frequency of nodal involvement was 4.5%, 8.9%, 28.7%, and 70.6%, respectively. Although the percentages were not identical, a similar pattern was observed in another large population-based study with + LNs in 1.8%, 5.7%, 19%, and 20% of patients, respectively.

Lymphadenectomy

To build a rational argument for LND, one must be convinced that the extra effort, time, and increased risk of complications associated with LND are rewarded by some measurable benefit to the patient. These benefits may be realized by better staging, better local control, reduced risk of local or regional nodal relapse, or ultimately better cancer-specific or overall survival. The remainder of this article will address each of these points in turn.

Lymphadenectomy

To build a rational argument for LND, one must be convinced that the extra effort, time, and increased risk of complications associated with LND are rewarded by some measurable benefit to the patient. These benefits may be realized by better staging, better local control, reduced risk of local or regional nodal relapse, or ultimately better cancer-specific or overall survival. The remainder of this article will address each of these points in turn.

Improved staging

There appears to be little debate that LND in patients with UTUC will be staged better than those without LND. The knowledge of LN metastases allows more complete patient counseling because it is recognized that patients with positive nodes (pN1/2) have a 2.5- to 3.0-fold worse survival compared with pN0 (pathologically negative lymph node) patients. Computed tomography (CT) Imaging of LNs often misses between 40% and 50% of small lymph node metastases in patients with UTUC. Even with modern imaging techniques such as positron emission tomography (PET), CT, and PET magnetic resonance imaging (MRI), about 20% of patients with nodal metastases will still be missed on radiographic imaging.

Routine pathologic staging may not even identify all metastases. Recent data from a multi-institutional study found that the more lymph nodes removed in pN0 patients, the better the cancer-specific survival (CSS). The authors suggested that there may have been micrometastatic disease removed during the LND that contributed to the survival advantage. This phenomenon has also been observed in bladder cancer, strengthening the argument. Further evidence to support this theory comes from Japan, where investigators found that 14% of patients with negative nodes by hematoxylin and eosin (H&E) staining were positive on immunohistochemical staining.

It seems fair to conclude that LND will undoubtedly allow for the most accurate predictions of survival, remove micrometastatic disease not identified by routine H&E staining, and identify those who may benefit from immediate systemic therapy. However, if LND does not improve patient outcomes, it is not clear that this extra information is worth the increased operative time, effort, and complications associated with LND.

Local control/nodal relapse

To answer the question of whether LND prevents locoregional relapse is somewhat challenging, although the answer based on biologic plausibility alone is probably yes. The challenge in answering this question stems from the quality of the studies. In general, the studies are relatively few and often underpowered (and only univariate); the methodology of the existing studies is not rigorous, and the application of LND is not universal (which introduces confounding). The various studies report subset analyses and use differing definitions of survival, and the dissection templates are not uniform. The end result is a mixture of arguments for and against LND preventing local/nodal relapse in all patients with UTUC.

In patients with lower-risk UTUC (ie, those with Ta, Tis and T1 disease), the risk of nodal involvement at initial surgery is quite low, probably between 0% and 5% ( Table 1 ) and 82% lower than the risk of nodal involvement in patients with invasive disease. Although the authors were unable to find the risk of local/nodal recurrence in this low-risk cohort, it stands to reason that the rates would be so low that no study would be powered to detect a difference given the rarity of this disease. Furthermore, the 5-year cancer-specific survival in this cohort is 92%.

In higher-risk patients, there seems to be less ambiguity. Abe and colleagues found in a subset analysis of 214 patients with T1 or higher UTUC, the risk of local/regional relapse was nearly fourfold higher for those not undergoing LND compared with those who did and was found to be pN0 (hazard ratio [HR] 3.96, 95% CI:1.6–11.2). However, they did not compare the risk of relapse for those without an LND versus those with an LND in all patients with UTUC. Other groups were unable to demonstrate that any LND prevents local/nodal relapse. In one of these studies however, the crude rates were lower in those patients undergoing a compete LND versus no LND (6.7% vs 9.0%), and the authors hypothesized that an incomplete LND may be no better than no LND.