Determining in whom to perform pelvic lymphadenectomy

The preoperative parameters conventionally used to assess the need for pelvic lymph node dissection at the time of radical prostatectomy focus on determining the probability of finding positive lymph nodes. The greater the predicted likelihood of lymph node involvement, the stronger the recommendation to perform lymphadenectomy. However, current predictive tools do not consider possible therapeutic benefit. The predominant variables used by all preoperative prediction algorithms for lymph node disease are prostate-specific antigen (PSA) at diagnosis, clinical stage, and Gleason score on biopsy. In the past few years, several investigators have combined these 3, in varying ways, and have been able to achieve predictive accuracies ranging from 75% to 88%. When these factors are augmented by other variables, such as the number of positive biopsy cores or an endorectal magnetic resonance imaging (MRI) of the pelvis, accuracy can be further improved to approximately 90%. Interestingly, Karam and colleagues showed close to 98% predictive accuracy of lymph node involvement by adding preoperative serum endoglin level, a cell surface protein highly expressed in endothelial cells, to the 3 traditional variables ( Table 1 ).

Each statistical model, whether a nomogram, a table, or something else, is limited by the data used to create it. Predicative accuracy can only be determined by final pathology, which, in this situation, is largely a function of surgical template. Almost all predictive models are formulated from data obtained by a standard or more limited pelvic lymph dissection (see later discussion). One might surmise then that the models derived from an extended pelvic lymph node dissection would possess higher predictive accuracy. Briganti and colleagues were unable to show a clinically significant benefit in predictive accuracy for detecting lymph node positive disease from the data garnered from pelvic lymph nodes removed within an extended template.

Although it seems clear that PSA, clinical stage, and Gleason score on biopsy are vital to the prediction model, an improvement in the ability to predict the presence of lymph node disease would require the addition of other informative variables (eg, radiography, biomarkers) to the equation.

The 2009 National Comprehensive Cancer Network guidelines state that “a pelvic lymph node dissection can be excluded in patients with <7% predicted probability of nodal metastasis by nomograms, although some patients with lymph node metastases will be missed.” Depending on the definition used, the probability of lymph node disease in men deemed to possess low preoperative risk of lymph node metastasis is anywhere between less than 1% and 8%. Data from clinical practice since 2001 obtained from Cancer of the Prostate Strategic Urologic Research Endeavor database seem to reveal acceptance of this recommendation by practitioners; more than 90% of men deemed high risk received a pelvic lymph node dissection, whereas less than 75% of those at low risk did so. If the goal is to improve the detection rate of men with lymph node positive disease, confining pelvic lymph node dissection to higher-risk men will yield better results. Whether this logic is true when the aim is therapeutic, however, is open to question and is discussed later.

Evaluating the role of imaging in identifying lymph node invasion

If a radiological technique that was able to detect lymph node metastases with high sensitivity and specificity existed, the need for pelvic lymph node dissection for staging purposes would be obviated. Current computed tomography (CT) and MRI modalities only have a sensitivity of approximately 35% for detecting lymph node involvement. This is because of the poor performance of these modalities in detecting subcentimeter lesions and the high prevalence of micrometastases in the PSA era. Several innovations in imaging technology have dramatically improved the ability to identify lymph node disease preoperatively. The addition of lymphotropic contrast agents, such as paramagnetic iron oxide nanoparticles and ferumoxtran-10, can increase the sensitivity and specificity to 80% to 90% and 96% to 98%, respectively. Imaging based on a lymphotropic agent relies on the tendency of these compounds to concentrate and flow within the lymphatic system. The absence of lymphotropic contrast agent in a visualized lymphatic chain may occur if the lymph node in question is infiltrated with cancer, or less commonly, if a benign process is at work (eg, fibrosis or lipomatosis). Thoeny and colleagues reported an exciting development in lymph node imaging using ultrasmall superparamagnetic particles of iron oxide coupled with diffusion-weighted MRI. These authors found that this technique was capable of identifying lymph node invasion with 92% accuracy in patients with bladder and prostate cancer.

Positron emission tomography/CT, although useful in advanced testicular cancer, has demonstrated marginal utility for detecting lymph node disease in men with prostate cancer. Advancements using modifications of MRI technology or perhaps delivery of agents such as indocyanine green, a lymphotropic contrast agent, into the lymphatic system hold promise for improved lymph node staging in the future.

Evaluating the role of imaging in identifying lymph node invasion

If a radiological technique that was able to detect lymph node metastases with high sensitivity and specificity existed, the need for pelvic lymph node dissection for staging purposes would be obviated. Current computed tomography (CT) and MRI modalities only have a sensitivity of approximately 35% for detecting lymph node involvement. This is because of the poor performance of these modalities in detecting subcentimeter lesions and the high prevalence of micrometastases in the PSA era. Several innovations in imaging technology have dramatically improved the ability to identify lymph node disease preoperatively. The addition of lymphotropic contrast agents, such as paramagnetic iron oxide nanoparticles and ferumoxtran-10, can increase the sensitivity and specificity to 80% to 90% and 96% to 98%, respectively. Imaging based on a lymphotropic agent relies on the tendency of these compounds to concentrate and flow within the lymphatic system. The absence of lymphotropic contrast agent in a visualized lymphatic chain may occur if the lymph node in question is infiltrated with cancer, or less commonly, if a benign process is at work (eg, fibrosis or lipomatosis). Thoeny and colleagues reported an exciting development in lymph node imaging using ultrasmall superparamagnetic particles of iron oxide coupled with diffusion-weighted MRI. These authors found that this technique was capable of identifying lymph node invasion with 92% accuracy in patients with bladder and prostate cancer.

Positron emission tomography/CT, although useful in advanced testicular cancer, has demonstrated marginal utility for detecting lymph node disease in men with prostate cancer. Advancements using modifications of MRI technology or perhaps delivery of agents such as indocyanine green, a lymphotropic contrast agent, into the lymphatic system hold promise for improved lymph node staging in the future.

Examining the importance of the anatomic template

One of the fundamental issues that has fueled the controversy on pelvic lymphadenectomy in prostate cancer has been the matter of which nodes need to be removed. Anatomic studies in men without known cancer done with labeling agents injected directly into the prostate have demonstrated that the draining lymphatics typically involve 3 main areas: (1) along the pelvic side wall comprising the internal and external iliac lymphatics up to their junction with the common iliac lymphatic chain, (2) inferior to the prostate to the perineal floor to the lymphatics at and below the level of the internal iliac artery, and (3) the presacral area. The extent of pelvic lymph node dissection can be termed either standard or extended ( Fig. 1 , Table 2 ). Most investigators agree that the standard template is limited to the bladder (medially), the pelvic side wall (laterally), the femoral canal (distally), the bifurcation of the common iliac vessels (proximally), the external iliac vein (superiorly), and the obturator nerve (inferiorly). Definitions of an extended dissection differ but invariably include the area bounded by the standard template with the addition of removal of the fibrofatty tissue in the hypogastric region, with possible inclusion of the presacral, common iliac, and nodal tissue lateral to the external iliac vein. Golimbu and colleagues observed in their series in the pre-PSA era that among men with lymph node positive disease, the presacral and presciatic regions were involved as commonly as the external iliac and obturator nodes. From the same time period, Fujioka and colleagues found that lymph node disease was always present in the hypogastric region.

Anatomic boundaries of lymphadenectomy. Standard lymphadenectomy includes the nodal tissue from the external iliac vein and obturator fossa, whereas an extended lymphadenectomy also includes the tissue overlying the internal iliac (hypogastric) and common iliac vessels as well as the presacral area. ( Reprinted from Burkhard FC, Schumacher M, Studer UE. The role of lymphadenectomy in prostate cancer. Nat Clin Pract Urol 2005;2:336–42; with permission.)

With regard to prostate cancer lymph node detection, numerous investigators have shown that an extended template is much more likely to find cancer than a more limited one. For example, Bader and colleagues noted in their cohort that close to 20% of positive lymph nodes were located exclusively in the hypogastric region. In addition, Heidenreich and colleagues noted in their series that approximately 40% of positive lymph nodes were located outside of the standard template. Further, Allaf and colleagues, who compared the outcomes between men treated with extended dissection and those treated with standard dissection, observed a significantly higher lymph node detection rate in men who received a wider dissection (detection rate, 3.1% vs 1.1%; P <.01). In addition, they noted that among men with less than 15% lymph node density, the 5-year PSA progression-free rate for the extended lymph node group was 43% versus 10% for men in the more limited lymph node dissection group ( P = .01). These findings have been corroborated by several investigators, as has the intuitive notion that, in general, an extended template will yield more lymph nodes than a standard one. The obvious conclusion that can be drawn from these data is that the wider the template, the more the lymph nodes will be removed and the more likely lymph node metastases will be detected.

Some investigators have emphasized the number of lymph nodes retrieved, irrespective of the template used. Surgeons who routinely obtain a higher lymph node yield typically perform a more extended dissection. The total lymph node count may have diagnostic and potentially therapeutic implications, as discussed later. Nevertheless, the surgeon can only control the template, and the ultimate lymph node count is, to some degree, dependent on the pathologist performing the analysis. In radical cystectomy for bladder cancer, for example, investigators have shown that submitting individual packets of fibrofatty tissue for analysis increases the number of lymph nodes identified by the pathologist versus en bloc submission. Other factors, such as the technique of surgical specimen processing and examination, and the skill of the histopathologist can also influence the total lymph node count.

Taking into account the surgical approach

The traditional open approach to radical retropubic prostatectomy allows excellent exposure and access for the standard and extended templates. The size of the operative workspace in the open surgical field permits a meticulous lymphadenectomy to be performed in a safe and expeditious manner. The past decade has seen a rapid emergence in the adoption of minimally invasive surgical techniques for radical prostatectomy and bilateral pelvic lymph node dissection. Laparoscopic and robotic-assisted approaches enable the surgeon to visualize tissues under high magnification; enhanced lighting and increased intracorporeal pressures facilitate lymph node dissection. Two specific issues are germane to the discussion vis-à-vis minimally invasive pelvic lymphadenectomy. The first is port placement. Placing the midline port for the laparoscope too inferiorly to enhance visualization of the prostatic apex and vesicourethral anastomosis may hinder access to the more proximal lymphatic tissues in the extended template. The second is the peritoneal approach. The extraperitoneal robotic-assisted technique, by virtue of its anatomic position, makes visualization and access beyond the standard template challenging. In contrast, the transperitoneal robotic-assisted approach has no such encumbrances.

Assessing the rationale for pelvic lymph node dissection

As discussed earlier, there are essentially 2 reasons to perform a pelvic lymphadenectomy at the time of radical prostatectomy: (1) for staging and prognostic purposes and (2) for possible therapeutic benefit. No one can argue that performing a pelvic lymph node dissection (standard or extended) gives greater information for pathologic staging than not doing one at all. Improved pathologic staging translates into better prognostic information that can aid the decision-making process regarding adjuvant therapies (eg, the survival benefit seen with the use of early androgen deprivation therapy for men with lymph node positive disease ). Confounding issues are (1) the current low prevalence of lymph node positive disease in men undergoing radical prostatectomy and (2) the lack of uniformity in prognosis among men with lymph node metastases.

In large contemporary series from Johns Hopkins (n = 5744) and a multiinstitutional group led by Memorial Sloan-Kettering (n = 7014), the overall lymph node positive incidence rate was found to be 5% and 3.7%, respectively. As discussed earlier, preoperative prediction tools can be used to increase this proportion by identifying men at higher risk.

What of the outcomes of men with lymph node positive prostate cancer detected at radical prostatectomy? In a large cohort of men spanning the period from 1972 to 1999, Daneshmand and colleagues observed that the overall recurrence-free probability at 10 years for men with nodal disease was 65%, with those with a lymph node density of more than 20% at greatest risk of recurrence. Close to one-third of the men in this study received adjuvant treatment. Masterson and colleagues reported the 5-year and 10-year biochemical recurrence-free probability for men with positive lymph nodes to be approximately 23% and 19%, respectively. These results were found to be unaffected by approximately 15% of men in this study who received neoadjuvant hormone therapy. Among men with N0 disease, there was a trend toward better recurrence-free survival for those with a greater number of lymph nodes removed ( P = .01). In a study from Bern, Switzerland, Bader and colleagues observed a biochemical recurrence-free probability at 5 and 10 years of 25% and 10%, respectively, for all men in their study (no adjuvant therapy) with lymph node invasion, with those men with more than 2 nodes positive having the poorest outcome (5- and 10-year PSA recurrence-free probability of 21% and 7%, respectively). Boorjian and colleagues from the Mayo Clinic reported on a cohort of 507 men with lymph node positive disease treated between 1988 and 2001, of whom close to 90% received adjuvant therapy. They also found that at 5 and 10 years, 69% and 56%, respectively, were free of biochemical recurrence. Two or more lymph nodes positive as well as a pathologic Gleason score of 8 to 10, positive surgical margins, and tumor aneuploid status conferred an increased risk of disease progression. Cheng and colleagues from the Mayo Clinic, also reported on men with lymph node involvement encompassing a slightly earlier timeframe and found that in the 322 men studied, the 5- and 10-year PSA progression-free probability rates were 74% and 64%, respectively. Approximately 90% of these men also received adjuvant therapy and once again men with 1 lymph node positive fared better than those with 2 or more lymph nodes positive. Similar results were reported by Briganti and colleagues in a 2-institution study wherein all men received adjuvant hormone therapy. In a report from Johns Hopkins where no man received adjuvant hormone therapy (n = 143), the authors noted the 5-year PSA-free progression rate to be 27% among all men with lymph node positive disease with a lymph node density of more than 15% distinguishing men with the poorest prognosis (5-year recurrence-free survival of 10%).

In total, these data show that not all men with lymph node positive disease share the same prognosis. The number of lymph nodes that are positive (>2 or lymph node density of >15%–20%) is directly proportional to the risk of recurrence. In addition, there are some men with N1 disease who are free of disease at 10 years follow-up. Among node-negative patients, higher lymph node yield has been associated with improved outcomes in bladder, breast, colon, and lung cancer. Masterson and colleagues observed a similar finding in prostate cancer as mentioned earlier. Moreover, in an analysis of the SEER-Medicare database focusing on men with N0 disease, Joslyn and Konety detected a 15% lower risk of cancer-specific death among men who had more than10 lymph nodes removed compared with those who had fewer nodes sampled. These findings have not been universally corroborated. DiMarco and colleagues from the Mayo Clinic (n = 7036) found no difference in progression or survival outcomes in pN0 men treated during the PSA era, regardless of the number of lymph nodes removed. Similarly, Bhatta-Dhar and colleagues noted no difference in recurrence-free survival in men (pN0, n = 806) with low-risk disease based on the presence or absence of a pelvic lymph node dissection. If a therapeutic benefit does exist for pelvic lymph node dissection in men with N0 disease, it is most likely because of the removal of micrometastases/histologically occult disease.

At the outset of this article, we pointed out that the field is devoid of randomized clinical trials, and hence, the interpretation of the retrospective data should be done with some caution. One particular item that should be kept in mind when reviewing the literature, as is the case for pelvic lymph node dissection, is the Will Rogers phenomenon. This statistical effect occurs when an element from one dataset (set A) is moved to another dataset (set B), and consequently the averages of both datasets increase. For this to occur, the element in question must have a value (eg, outcome) lower than its original dataset (set A) and higher than its new dataset (set B). This phenomenon becomes relevant in oncology when data from 2 distinct groups (eg, pN0 vs pN1 or extended vs standard dissection) are compared. Consider a comparison in outcomes among men treated with an extended lymph node dissection: an extended lymph node dissection can improve the outcomes of men with both N0 and N1 disease by enriching the N0 cohort for true node-negative patients and by adding men with smaller nodal tumor volume (micrometastases) to the N1 group. In this scenario, the N1 group also benefits, as the prognosis of men with minimal burden of disease is better than those with more substantial N1 disease. Men who are pN0 as determined by an extended dissection are more likely to be truly node negative than men who are deemed pN0 via a standard dissection. Hence, men thought to be pN0 treated with a standard lymph node dissection may actually be pN1 and bring down the outcomes of node-negative men treated with a more limited lymph node dissection. Randomized controlled trials eliminate this effect because the probability of the Will Rogers phenomenon occurring is equal in all cohorts. Although Clark and colleagues performed a randomized controlled trial of pelvic lymphadenectomy in men with localized prostate cancer, the study had some methodological issues (ie, underpowered, n = 123; each man received an extended dissection on one side and a standard dissection on the other) that precluded a final conclusion to be drawn from this report.