The contemporary role of lymph node dissection in the management of renal cell carcinoma

Abstract

The appropriate role of lymph node dissection (LND) in the management of patients with renal cell carcinoma (RCC) is still a matter of debate. There is ample evidence that LND is the most accurate modality for staging the regional lymph nodes (LNs), which may harbor metastatic disease in greater than one-third of patients with high-risk RCC. The presence of LN metastases is an independent negative prognostic factor in this disease and accurate determination of LN status not only helps with patient counselling regarding prognosis and tailoring of postoperative surveillance schedules, but it also identifies patients at high risk of systemic disease recurrence who may qualify for clinical trials of adjuvant systemic therapies. Meanwhile, the therapeutic value of LND has been brought into question by a randomized trial (European Organisation for Research and Treatment of Cancer; EORTC 30881) that showed no difference in progression-free or overall survival between patients who were treated with radical nephrectomy (RN) and LND and those treated with RN alone. Given that most patients enrolled in this trial had small renal masses and therefore were at low risk for LN metastases, the question of whether patients with high-risk tumors derive a therapeutic benefit from a standardized, extended LND remains unanswered.

Introduction

Lymph node dissection (LND) is widely accepted as an indispensable component of the management of multiple genitourinary malignancies, especially cancers of the testis, penis and bladder. In contrast, the appropriate role of LND in the treatment of renal cell carcinoma (RCC) continues to be debated. In the original description of radical nephrectomy (RN) by Robson and colleagues, LND was deemed an integral part of the operation and its use was cited as a possible explanation for the superior outcomes of patients in that cohort compared with historical series.¹ Since then, the value of LND in the management of RCC has been put into question by multiple studies that reported no significant benefit of LND in terms of improving cancer-specific survival (CSS). The most impactful of these was the European Organisation for Research and Treatment of Cancer (EORTC) randomized trial 30881 that reported no difference in progression-free survival (PFS) or overall survival (OS) between patients undergoing RN and LND and those treated with RN alone.² However, because this study enrolled predominantly low-risk patients, with only 28% having locally advanced (⩾pT3) tumors, its findings do not rule out the possibility that an extended LND in patients at high risk of harboring microscopic or gross lymph node (LN) metastases may be of therapeutic value. With this in mind, we set out to summarize the current evidence pertaining to the value of LND as both a staging tool and a therapeutic intervention in patients with RCC.

Prevalence, distribution and pathophysiology of lymph node metastases from renal cell carcinoma

The true prevalence and distribution of LN metastases among patients diagnosed with RCC is currently unknown, since an unselected cohort of such patients has never been subjected to an extended, template-based LND.³ The reported prevalence of LN metastases has been shown to differ significantly according to the pathologic characteristics of the primary tumor, the extent of the dissection and the method of histopathologic evaluation. Prior studies have identified pathologic T stage or tumor size,^4–6 tumor grade^4,5 and the presence of tumor necrosis and sarcomatoid differentiation^4,5 as factors associated with a higher risk of LN metastases. Among a cohort of 169 patients with high-risk tumors, as defined by the presence of two or more negative prognostic factors (primary tumor stage pT3–T4, tumor size >10 cm, Fuhrman grade III or IV, sarcomatoid differentiation, tumor necrosis) on intraoperative pathologic examination, the incidence of LN metastases was reported to be 38%.⁵ Nevertheless, given that in the absence of a preoperative or intraoperative biopsy these tumor characteristics will only be known after surgery, the clinical utility of these prognostic factors in preoperatively identifying patients who would most likely benefit from LND is limited. To our knowledge, only one prior study has assessed exclusively preoperative factors associated with LN metastases and identified the presence of enlarged LN on preoperative cross-sectional imaging, Eastern Cooperative Oncology Group (ECOG) performance status, presence of local symptoms and elevated serum lactate dehydrogenase (LDH) levels as significant predictors of LN status.⁷ Patients undergoing an extended LND are also more likely to be found to have LN metastases, with one study reporting a prevalence of 21% among patients with ⩾13 LNs removed compared with 10% among those with <13 LNs removed.⁸ Finally, it is apparent that a significant proportion of micrometastases are missed on routine hematoxylin and eosin (H&E) staining-based histopathologic examination, with a recent study of 15 LNs from five patients initially deemed to be pN0 reporting that occult metastases could be identified using immunohistochemical staining for cytokeratin 18, carbonic anhydrase IX (CAIX) and cadherin 6 in eight of these LNs.⁹

The lymphatic drainage of the kidney is more complex and less predictable than that of other genitourinary organs such as the testis and penis. The kidneys drain into lymphatic channels located between the first and fifth lumbar vertebrae, which in turn drain into the thoracic duct.¹⁰ LN metastases from right-sided tumors will typically be found in the paracaval, precaval and interaortocaval LNs, whereas those from left-sided tumors will be found in the paraaortic, preaortic and interaortocaval LNs.^11,12 Although the hilar LNs are the most commonly excised LNs at the time of RN, up to 46% of patients with paraaortic or paracaval LN involvement will have negative hilar LNs.¹³ Aberrant patterns of spread are not uncommon, with one study reporting an incidence of isolated supraclavicular or pelvic metastases of 43%.¹⁴ Contralateral spread, however, is thought to be rare.

The natural history of RCC differs from that of other genitourinary malignancies and is characterized by early hematogenous spread to distant sites as opposed to the stepwise progression from the primary tumor to regional LNs and then to visceral sites or bone. In a seminal study by Johnsen and colleagues of 554 patients who were found to have kidney cancer on autopsy, LN metastases were identified in only 80 patients (14%), with just five having LN metastases in the absence of visceral or bone metastases.¹⁵ This may be partially explained by the fact that in up to one-third of patients, renal lymphatics may drain directly into the thoracic duct without first traversing regional LNs.¹⁶ The early systemic dissemination of RCC has important implications for the therapeutic potential of LND, as further discussed below.

Contemporary trends in the utilization of lymph node dissection in patients undergoing surgery for renal cell carcinoma

It is evident that LND is performed infrequently in patients undergoing surgery for RCC. In a population-based study using the Surveillance, Epidemiology and End Results (SEER) registry, Kates and colleagues reported that only 20% of patients who underwent RN had any LNs examined.¹⁷ Because hilar LNs are often attached to the kidney specimen, patients could have a few LNs examined but not actually undergo any kind of intentional LND. In fact, the proportion of patients who had five or more LNs removed, which was the definition of a formal LND used in the study, was only 6.6%. LND may be performed more frequently at high-volume academic centers, with an analysis of the ASSURE trial of adjuvant tyrosine kinase inhibitor (TKI) therapy showing that all patients with cN+ disease and 30% of those with cN0 disease underwent LND¹⁸ and a recent publication from a consortium of European institutions reporting use of LND in 43% of patients.¹⁹ The latter study reported that younger patients, those with larger and locally advanced tumors and those undergoing RN as opposed to partial nephrectomy were more likely to undergo LND.

Multiple studies have shown that the utilization of LND has been decreasing over time.^17,19,20 This trend is likely at least partly explained by the contemporaneous uptake of laparoscopic renal surgery, since patients undergoing laparoscopic surgery have been shown to have a lower probability of undergoing LND than those having an open RN.^19,21 In skilled and experienced hands, robotic-assisted laparoscopic LND is feasible;^22–24 nevertheless, it is likely that most surgeons who perform either a pure laparoscopic or robotic RN are not comfortable with dissecting near the great vessels, which is a prerequisite for a thorough retroperitoneal LND. Conversely, one institution performing predominantly open renal surgery has reported increasing use of LND in the setting of locally advanced disease.²⁵

Morbidity of lymph node dissection

Multiple studies have reported that the incidence of perioperative complications among patients undergoing RN with LND is similar to that among patients undergoing RN alone.²⁶ Importantly, most such studies employed limited dissections, such that their reported complication rates may not be generalizable to patients undergoing a more extended LND. A post-hoc analysis of the EORTC randomized trial of LND in patients undergoing RN showed no statistically significant difference in the incidence of complications among those who received LND compared with those who did not, with the reported incidences of bleeding, venous thromboembolism and infection being 6.7% and 4.6%, 2.0% and 0.4% and 5.1% and 5.8%, respectively.²⁶ Similarly, the ASSURE trial reported no increase in the risk of surgical complications with the addition of LND to RN.¹⁸ It is important to note that these studies were not adequately powered to assess differences in complication rates, such that a small but clinically significant difference in major complications cannot be excluded. The most significant complication is major hemorrhage, which usually occurs due to injury to lumbar vessels, the inferior vena cava (IVC), or the aorta and can potentially be life-threatening. The risk of developing chylous ascites has also shown to be greater among patients undergoing RN with LND than those managed with RN alone.²⁷

Prognostic value of lymph node dissection in patients with renal cell carcinoma

Accurate determination of the presence or absence of regional LN metastases in patients with RCC is important as it not only helps clinicians counsel patients regarding the prognosis of their disease and tailor postoperative surveillance schedules but can also identify patients who may benefit from participation in clinical trials of adjuvant targeted therapies and the newly introduced checkpoint inhibitors. LND is the gold standard modality for staging the regional LNs in patients with RCC. CT and magnetic resonance imaging (MRI) are routinely used to screen for LN metastases prior to surgery, but their sensitivity and specificity are suboptimal. False negatives occur due to the inability to detect foci of micrometastatic disease, while false positives occur because of LN hyperplasia due to inflammation. In a series of 2954 patients with RCC treated at a single institution, 14% of patients had LNs ⩾1 cm in size identified on preoperative contrast-enhanced CT but only 29% of these were confirmed to have LN metastases after undergoing LND.²⁸ Conversely, among 861 cN0 patients who underwent LND, LN metastases were identified in only 4.2%. This translates into a sensitivity, specificity and positive and negative predictive values of 77%, 73%, 29% and 96%, respectively. The sensitivity of cross-sectional imaging in detecting LN metastases increases as the cutoff value for the short-axis diameter of imaged LN is lowered,²⁹ although this comes at a cost of decreased specificity. Lymphotrophic nanoparticle (ferumoxtran-10)-enhanced MRI showed superior sensitivity and specificity (100% and 96%, respectively) in a pilot study,³⁰ although development of this imaging modality has stalled. PET, using iodine-124-labelled antibodies against CAIX, also showed significant promise as a modality for identifying clear cell RCC,³¹ although its role as a LN staging modality was not defined and production of this antibody has since ceased.

Pathologic LN status is one of the strongest predictors of prognosis in patients with clinically localized RCC undergoing nephrectomy. The presence of LN metastases portends a poor prognosis that is similar to that of patients with visceral or bone metastases.³² An analysis of the SEER registry reported 5-year CSS probabilities of 84% and 38% among pN0 and pN1 patients, respectively.³³ Importantly, the detrimental impact of LN metastases was found to be greatest among patients with localized and low-grade tumors. Single-institution series have reported similarly dismal outcomes among patients with pN1 RCC. Among a cohort of 138 patients treated at the Mayo Clinic over a 30-year period, 10-year CSS was only 21%, with a median time from surgery to the development of metastases of only 4 months.³⁴ Only two patients recurred in the retroperitoneum in the absence of distant metastases, further supporting the notion that the vast majority of patients without clinically overt distant metastases who are found to have regional LN metastases at the time of nephrectomy in fact have occult systemic disease. Among patients with distant metastases undergoing cytoreductive nephrectomy, the presence of LN metastases likewise portends a worse prognosis.^35–37 For example, among a cohort of 187 patients who underwent LND at the time of cytoreductive nephrectomy, the median CSS was 11 months among those with LN involvement compared with 24 months among those with negative LNs.³⁷ This finding that may reflect a higher burden of systemic disease among LN-positive patients or a more aggressive disease biology among tumors that have a propensity for lymphogenous dissemination.

The prognostic significance of LN disease burden among patients with pN1 RCC remains unclear as previously published studies yielded conflicting results. For example, while a SEER registry analysis reported that LN density (proportion of LNs removed that are positive) ⩾60% predicted worse CSS independent of grade and stage,³⁸ other studies found no association between measures of LN disease burden and survival.³⁴ Importantly, no study has assessed whether the prognostic significance of a measure such as LN density varies according to the extent of LND performed. Furthermore, to our knowledge, no study has assessed the prognostic significance of LN micrometastases detected by way of immunohistochemistry or polymerase chain reaction, despite evidence from other disease sites that the prognosis of patients with micrometastatic disease within regional LNs may differ from that of patients with larger metastases that can be detected on routine H&E staining.³⁹

Therapeutic value of lymph node dissection in patients with renal cell carcinoma

In other genitourinary malignancies, namely germ cell tumors of the testis and squamous cell carcinoma of the penis, LND is almost universally accepted to have therapeutic value. Importantly, these diseases demonstrate a predictable pattern of metastatic spread from the primary tumor to the regional LN and then distant sites, which, as outlined above, likely differs from the natural history of RCC. Conceptually, LND can only be therapeutic if there is a disease state in which LN metastases occur in the absence of any extranodal, systemic metastatic disease. Whether such a disease state exists in patients with RCC is currently unknown, particularly since some patients are disease free for decades before experiencing late relapses in unusual sites such as the thyroid and pancreas. Although multiple studies have identified clinical and pathologic factors that predict LN status in patients undergoing LND, as outlined previously, all of these factors also predict the development of distant metastases and therefore are not useful in identifying those patients with isolated LN metastases who would be expected to derive a therapeutic benefit from LND. Furthermore, given the possible association between LN disease burden and risk of systemic disease progression, it is possible that LND may only be therapeutic in patients with micrometastases who are classified as pN0 on routine H&E staining.

Attempts at delineating the therapeutic value of LND have focused on comparing survival among patients who received LND at the time of nephrectomy with those who did not or assessing the association between the extent of LND and survival. All but one of these studies relied on observational data. EORTC 30881 was a phase III randomized trial that assigned 772 patients without clinical evidence of regional LNs or distant metastases to RN with LND or RN alone.² For right-sided tumors, LND was intended to encompass the paracaval, precaval, retrocaval and interaortocaval LN from the crus of the diaphragm to the bifurcation of the IVC, whereas patients with left-sided tumors were meant to undergo resection of the paraaortic, preaortic, retroaortic and interaortocaval LNs. At a median follow-up time of 12.6 years, there was no difference in PFS or OS between the two treatment groups. Criticism of this study has centered on the fact that only 4% of patients who underwent LND were found to have LN metastases. This is likely because the majority of patients had low-risk disease, with 72% of tumors being staged as pT1 or pT2, in the setting of which the prevalence of LN metastases is expected to be low. Another potential explanation for the low prevalence of LN metastases in the EORTC trial is poor adherence to the extended template mandated by the protocol, a possibility that cannot be ruled out given that no data regarding average LN yield was reported in the published study.

Observational studies have generally not shown any benefit of LND in terms of decreasing the risk of cancer-specific mortality (CSM) in patients undergoing RN. In a recent analysis of the SEER registry, Marchioni and colleagues reported worse CSM among patients with pT2–T3 RCC undergoing RN and LND compared with RN alone.² Similarly, Feuerstein and colleagues reported lower 5-year recurrence-free survival (64% versus 77%) among patients with renal tumors ⩾7 cm in size treated with concomitant LND.⁴⁰ These studies highlight the pitfalls of relying on observational data to assess the therapeutic benefit of LND. Specifically, such studies are likely to be undermined by confounding due to true LN disease status, with patients who underwent LND being those who were more likely to have LN metastases based on preoperative imaging or intraoperative findings, the same patients who would then be at higher risk of disease progression. In the largest study of LND in patients with RCC performed to date, Gershman and colleagues reported on 1797 patients with M0 RCC treated with nephrectomy, 606 (34%) of whom underwent concomitant LND. After a median follow-up time of 10.6 years, there was no significant difference between patients who did and did not undergo LND in terms of CSS or OS after adjustment for patient and tumor characteristics using propensity-score methods.⁴¹ Importantly, there was no evidence that the relationship between LND and any of these outcomes differed according to the presence of enlarged LN on preoperative imaging or the probability of harboring LN metastases as determined by pathologic grade and stage. Nevertheless, the possibility of residual confounding secondary to true LN status remains as in many cases the decision to proceed with LND is made based on an intraoperative finding of enlarged or otherwise suspicious LNs, a factor that is difficult to ascertain and therefore control for in an observational study.

No randomized trials have assessed the impact of the extent of LND performed on survival, whereas several observational studies have suggested a possible positive association between higher LN yield and survival in specific patient subgroups. For example, Whitson and colleagues reported an improvement in CSS with an increasing number of LNs removed among pN1 but not pN0 patients.⁴² Similarly, Capitanio and colleagues reported that a higher LN yield was associated with better CSS among patients with pT2–pT4 disease and those with sarcomatoid differentiation.⁴³ These findings are consistent with the hypothesis that patients at highest risk of harboring LN metastases are those who are most likely to benefit from LND. There are no published studies comparing outcomes among patients treated with different LND templates.

While the presence of LN metastases in patients with M1 RCC confers a worse prognosis, there is no evidence that performing LND in this setting is of any therapeutic value. Among 305 patients treated with cytoreductive nephrectomy at the Mayo Clinic, the majority of whom were diagnosed in the pre-TKI era, there was no improvement in CSS or OS in patients who underwent LND.⁴⁴ Again, there was no evidence that the association between LND and survival differed according to the probability of harboring LN metastases. Other studies reported worse CSS among patients who underwent cytoreductive nephrectomy and concomitant LND compared with nephrectomy alone,^45,46 which again likely reflects the fact that LND is more likely to be performed in patients with lymphadenopathy, which is associated with worse outcomes in patients with metastatic RCC.

Isolated retroperitoneal LN recurrence in the absence of concomitant visceral or bone metastases in a patient with previously surgically resected RCC is a rare and challenging clinical scenario. The largest series of 50 such patients reported 3 and 5-year PFS probabilities of 41% and 35%, respectively, after local resection.⁴⁷ Time to local disease recurrence less than 12 months was an independent predictor of subsequent disease progression. This demonstrates that carefully selected patients with isolated retroperitoneal LN recurrence after nephrectomy may be cured with surgery alone, although it must be noted that prolonged disease-free intervals after resection in such patients may simply reflect a more indolent disease biology.

Conclusion

LND is the most accurate method for staging regional LNs in patients with RCC and should be performed in those patients who are deemed to be at high risk of having LN metastases based on clinical, radiologic and pathologic criteria. There is level I evidence for a lack of therapeutic benefit of LND in patients with low-risk RCC, but a randomized trial enriched for high-risk patients has yet to be performed. As such, it remains to be determined whether patients with RCC who are at high risk of regional LN metastases stand to benefit from a standardized, extended LND.