Lymph node density: surrogate marker for quality of resection in bladder cancer?

Radical cystectomy with pelvic lymph node dissection is the accepted ‘standard of care’ therapy, which provides the best cancer control for patients with muscle invasive bladder cancer [1]. Since up to 25% of patients undergoing radical cystectomy have nodal metastases and 30% of patients with node-positive disease may be cured by surgery, complete removal of disease burden requires a thorough nodal dissection [1,2]. Further support for a thorough node dissection at cystectomy is demonstrated by the fact that even N0 patients demonstrate improved survival if more (> nine in this reference) nodes are removed, presumably through the elimination of micrometastatic disease [3]. Surprisingly though, analysis of the Surveillance Epidemiology and End Results (SEER) registry revealed 40.3% of patients who undergo cystectomy do not undergo a lymph node dissection and, overall, 67.9% have fewer than six nodes removed [3]. This raises the following questions:

•	If lymph node dissection offers potential for cure in node-positive and -negative disease, why aren’t all patients undergoing lymphadenectomy?
•	Are all surgeons capable of performing or trained to perform an adequate node dissection at radical cystectomy?
•	Is there a surrogate end point for surgical quality control that can be applied to the lymph node dissection performed during a radical cystectomy?

In an attempt to answer these types of question, Stein [4] and Herr [5] in 2003, put forward the idea of ‘lymph node density’ (LND). Herein, we examine whether LND may be a surrogate end point for the quality of node dissection at cystectomy.

The concept of LND is a composite of two variables: number of positive lymph nodes, divided by total number of nodes removed (or examined, depending on one’s trust in one’s pathologist). Thus, LND takes two known prognostic factors for recurrence and combines them into one composite variable (a percentage) [4–6]. The numerator (number of positive nodes) is one component of the American Joint Committee on Cancer (AJCC) staging criteria and incorporating this variable makes intuitive sense as the greater the extent of disease the worse the prognosis will be. The denominator (total number of nodes obtained) is more controversial since it depends on the thoroughness and anatomic extent of dissection, but is also influenced by the technique whereby the nodal tissue is processed, and later evaluated, by the pathologist. To further confound matters, the anatomic extent of node dissection and number of nodes retrieved varies widely between surgeons and institutions. Limited node dissections occur despite the well-documented potential for cure and evidence suggesting that complete node dissection may have a greater impact on survival than adjuvant chemotherapy [3]. In one prospective study, only the anatomic extent of node dissection (limited vs standard vs extended) was associated with significant differences in nodal yield [7]. Our limits of dissection for pelvic lymphadenectomy include the node of Cloquet distally, genitofemoral nerve laterally, hypogastric nodes inferiorly/posteriorly, and the aortic bifurcation as the proximal boundary. In high-risk patients, we will extend the dissection to incorporate the paracaval and para-aortic nodes up to the inferior mesenteric artery as well as the presacral lymph nodes anterior to the sacral promontory. There is mounting evidence that a limited template or ‘substandard’ removal of lymphatic tissue may lead to understaging as well as increase the likelihood of recurrence and death from bladder cancer. At a recent conference, Dhar from Switzerland presented a provocative study comparing patients undergoing radical cystectomy at the Cleveland Clinic (where the standard practice is to perform a limited node dissection), with those undergoing surgery at University of Bern (where the practice is to perform an extended node dissection up to the common iliac vessels). When matched for primary tumor (pT) stage, more node-positive patients recurred at the Cleveland Clinic than in Bern (93 vs 65% recurrence, respectively) suggesting that the patients who were treated with an extended template fared better [8].

Experienced surgeons at high-volume bladder cancer centers have lower mortality rates and fewer perioperative complications [9,10]. However, there are variations in outcome, even among high-volume, experienced surgeons, suggesting that surgical technique rather than experience may independently affect outcome. In a multi-institutional study, improved surgical quality (margin status and > ten lymph nodes obtained) significantly correlated with better survival than pathological stage, nodal status, patient age and chemotherapy status [11]. One expert panel went so far as to conclude that, to be considered ‘proficient’ at radical cystectomy, surgeons should perform at least ten radical cystectomies per year with a minimum of 10–14 nodes obtained [12]. We believe that this number should be even higher since the variation in surgical volume is such that this definition would consider those centers that perform more than 150 radical cystectomies per year in the same category as those performing ten per year.

Another issue that further confounds matters is where pathology reports from two different institutions, both of which have surgeons who have been trained by similar mentors, report nodal yield differently – for example, one institution routinely reports 50–60 lymph nodes removed while the other averages a more modest (and common) 20–30 nodes removed, even though both surgeons report the exact same anatomic extent of dissection. Even within known centers of excellence, changing the number of lymph-node packets sent to the pathologist, despite the same surgical template, gives a widely varying return of nodal counts [7,13]. This is where the current variability in handling and reporting of lymph-node packets must be standardized. A national consensus is needed to address the methods used by pathologists in locating, dissecting, defining and reporting nodes. Certainly, such consensus is needed before pathologists can agree upon whether one lymph node measuring 3 cm should be reported as n = 1 or cut into sections and reported as n = 3 or even higher.

Let us assume, for a moment, that all these issues have been sorted out and we have a ‘true’ LND. LND has been demonstrated to be directly correlated with recurrence-free survival [3–6]. In a recent review, LND was found to be a better surrogate end point than total number of positive nodes in four out of five studies in the literature evaluated by Herr [14]. Since LND is more indicative of completeness of disease resection than the absolute number of positive lymph nodes, it is not surprising that it affects survival rates such that if LND is less than 20%, then patients have a 5-year survival of 64% compared with only 8% if LND is more than 20% [5]. As such, this is a compelling argument for inclusion of LND in the AJCC staging criteria, especially since this would include a surgical quality control component in the staging system. This is even more relevant today, since some low-volume hospitals are acquiring robotic technology where the potential for substandard lymph-node dissections is increased. High-volume robotic centers are starting to perform robotic cystectomy with acceptable nodal yields [15]; however, this procedure has a steep learning curve and is best left to surgeons who have extensive experience with open radical cystectomies as well as robotic experience. Additionally, it remains to be established whether an extended node dissection can be thoroughly performed robotically and we are in the process of performing a prospective trial to evaluate this issue. Full excision of the lymphatics from the fossa of Marcille (where the obturator nerve enters the psoas) and presacral plexus can be challenging in inexperienced hands. This highlights the need for adoption of a standard node-dissection template and a surrogate end point that can be used as quality control, and comparison of outcomes across surgeons and techniques.

Even with these factors in mind, there remain several obstacles to using LND as a surrogate marker for quality of surgical care. First, the arbitrary LND cut offs of 20–25% were assigned retrospectively, and represent statistical cut points [4,5] that would benefit from prospective validation. The effect of chemotherapy on LND also needs further evaluation although, in our series, it appears to hold true [Kamat et al., Unpublished observation]. As previously mentioned, both the surgical ‘template of dissection’ and the techniques of pathologic isolation, review and reporting need standardization. Once these factors are standardized, we believe that a minimum value for the denominator must be prescribed to ensure an adequate dissection by all surgeons.

The real utility of LND, rather than simply focusing on the prognostication, may be to serve as a yardstick whereby to measure the ‘completeness’ of nodal dissection in a particular patient. For example, if a patient has ten positive nodes and 50 were removed by his surgeon, his LND (20%), and hence prognosis, might be improved compared with someone whose surgeon removed only those ten nodes that were positive (LND = 100%). If surgical templates and pathologic processing are standardized, LND would also more accurately reflect the potential for cure (therapeutic value) of a pelvic lymph node dissection in a particular patient and might correctly identify which patients may benefit from adjuvant therapy versus observation, even with nodal disease. As a surgical community, we need to develop objective measurements of quality of care. LND with standardized technique may serve as the quality-control measure for pelvic lymph-node dissection since it can reflect not only the thoroughness of a nodal dissection but also the intraoperative judgment of the operating surgeon.