Thick melanomas are described as Breslow thickness ≥4 mm. These patients have a significant risk of regional metastases (60–70 %) but an equally high risk of systemic disease (70 %). The high risk of systemic disease in this population has therefore led many authors to question the utility of sentinel lymph node biopsy in patients with melanomas greater than 4 mm in depth rationalizing that their prognosis is more strongly linked to progression to stage IV illness than to lymph node status [30, 31]. Because the survival of these patients is poor overall, Balch et al. [32, 33] initially hypothesized that locoregional management via nodal dissection was unlikely to confer survival benefit. However, two recent studies advocated that sentinel lymph node status—even in patients with thick melanoma—was found to be an independent predictor of survival [30, 31]. Gershenwald et al. [31] looked at 116 patients with melanoma >4 mm thick and found that sentinel lymph node status was still the most powerful predictor of overall survival by univariate and multivariate analyses. Ferrone et al. [34] likewise looked at 126 patients with thick melanoma and found comparable rates of positive SLN (30 % vs. 39 %) and 3-year recurrence-free survival (76 % vs. 72 %). In the presence of conflicting data, many institutions routinely perform SLN biopsy even in those with thick primary tumors. With the advent of more robust therapeutic options for visceral disease and investigations demonstrating promising early results for genetic profiling of tumors, the role of surgical staging via sentinel lymph node biopsy becomes more ambiguous, generating a greater series of questions. The opportunity for early intervention in high-risk patients without the potential morbidity of surgical lymphadenectomy may pose a better alternative than the current paradigm in a subset of patients.

The MSLT-1 trial has proven that early detection of regional metastases improves survival (Fig. 1); however, because the ability to select patients to undergo sentinel lymph node biopsy is impeded by the limitations of using histologic characteristics to determine biology, the procedure itself does not afford an overall survival benefit for all comers (intervention can only impact survival in the 17 % of patients who actually have nodal disease). This is compounded by the significant heterogeneity observed in the node positive group, which–with increasing recognition of microscopic and immunohistochemically detected disease, is likely to become more diverse. Survival in this group can range from 64 to 91 % depending on the population [35]. The resulting limitation of the sentinel lymph node procedure therefore, is that even in patients who are at the highest risk for having nodal metastases, nearly two-thirds will be undergoing a procedure that they do not need and, therefore, can derive no benefit. Within this context, the optimal improvement in this procedure will not be a technical one, but rather an intervention which aids in improving selection. Unfortunately, the ability to make this improvement will likely rely upon techniques other than histology such as genetic analyses or similar. Likewise, improvments in selection will be dependent on an ability to accrue large numbers of patients in order to discern even small differences in study groups. There is still much work to be done to define the group most likely to benefit from sentinel node biopsy.

To work this direction, we could use large database studies, using propensity scoring to match those undergoing sentinel- lymph node biopsy with similar controls. While pooling of clinical data from multiple institutions has been done by the American Joint Committee on Cancer (AJCC), using a larger data set and standardizing the pathology variables (in which there was previously wide variability), could help us find an answer. Decision analyses could also assist in this process, and could effectively summarize the costs, benefits, and probabilities of each branch point. Though it may be difficult to assign accurate probabilities to some of the more qualitative outcomes, such as quality of life and patient satisfaction, application of the decision sciences could be very productive in defining the proper use of sentinel lymph node biopsies.

The advent of sentinel node biopsy has made the management of nodal metastases even more controversial than the detection of nodal metastases. It is well recognized that in many diseases, management of regional disease is controversial (breast cancer, gastric cancer) and melanoma is similarly challenged. Using conventional histologic evaluation, only 20 % of patients undergoing CLND will have additional nodal disease, implying that as many as 80 % of SLN positive patients may be exposed to the risks of a second, more morbid procedure with potential long-term effects, without benefit. There are differences in the detection of metastatic deposits in sentinel nodes versus nodes in a completion lymphadenectomy specimen given the more rigorous examination with immunohistochemistry and serial sectioning applied to sentinel nodes. One could argue, that without the more intense scrutiny of the completion lymph nodes that is applied to sentinel nodes, the true incidence of nodal involvement in CLND specimens is not known. As a result, regional disease management after a positive sentinel lymph node remains a discussion point.

Although one might extrapolate from the MSLT trial that there may be survival benefit to clearing the nodal basin with completion lymphadenectomy (early detection of disease impacts outcome which would imply some potential benefit to regional disease control), other studies have not been so forthcoming. Outside of MSLT, there fails to be a proven survival advantage for undergoing a completion lymphadenectomy. Van der Ploeg et al. [36] recently evaluated 1,174 patients with sentinel node positive melanoma, 61 of whom did not undergo completion lymph node dissection. Completion lymphadenectomy did not show any influence on survival (HR 0.86, 0.46–1.61; P = 0.640). Another multicenter trial examined 134 SLN positive patients at 16 centers who did not undergo completion lymphadenectomy and compared them against a cohort of patients from Memorial Sloan Kettering Cancer Center who had a positive SLN and underwent completion lymphadenectomy [37]. There was no difference in nodal recurrence-free survival between the groups (P = 0.07) or in the disease-specific survival between the groups (P = 0.65). Other studies have documented similar results with no difference in recurrence-free or disease-specific survival [4, 38].

Other studies have attempted to delineate characteristics of the sentinel lymph node which may predict involvement of nonsentinel lymph nodes. Additional factors such as tumor burden, depth of invasion from capsule, microanatomic location, and maximum diameter of the largest tumor have all been considered as predictors of additional lymph node disease [4, 39–41]. Nagaraj recently published a metaanalysis to determine clinicopathologic variables most predictive of nonsentinel node metastases in the setting of a positive sentinel node [42]. There were nine factors including ulceration, satellitosis, neurotropism, >1 positive SLN, angiolymphatic invasion, extensive locations, macrometasases >2 mm, extranodal extension, and capsular involvement which predicted additional positive nodes beyond the sentinel lymph node. Unfortunately, this makes for a complex set of prognosticators when attempting to decide on completion lymphadenectomy and may not be as useful practically. Van der Ploeg et al. published a more straightforward study demonstrating that the burden of disease and allocation of tumor within the sentinel lymph node influences melanoma-specific survival. Patients with metastases <0.1 mm and found in the subcapsular area had a 5-year overall survival of 91 %. Nonsentinel lymph node rates for these patients was 2 %. The study concluded that completion lymph node dissection in these patients may be overtreating patients who have a survival that is equivalent to SLN negative patients [43].

The current guidelines of the 2008 National Comprehensive Cancer Network recommend either CLND or participation in a clinical trial for a positive sentinel lymph node. Despite these recommendations, only 50 % of patients in the National Cancer Data Base with a positive sentinel lymph node actually undergo CLND, clearly indicating some disconnect between recommendations and practice [44]. Reasons for lack of CLND in SLN positive patients are variable. Kingham et al. looked at over 2,000 patients who had undergone SLN biopsy, where 317 patients had positive SLN followed by lymphadenectomy and 42 patients with positive SLN did not. The patients not undergoing CLND were older (median age 70 vs. 56 years, p < 0.01) and had a trend toward thicker melanoma (Breslow 3.5 vs. 2.8 mm, p < 0.06). Additionally, as expected, there were a higher percentage of lower extremity melanomas in the group that did not undergo CLND (40 % vs. 13 %; <0.01) since many surgeons avoid groin dissections secondary to their high risk of complications and lymphedema. Bilimoria et al. and Cormier et al. [44, 45] found similar reasons for lower than expected rates of CLND such as older age, lower extremity melanoma, thin melanomas, and African-American race.

Perhaps most importantly, these studies, when analyzed in the context of the others, provide an ambiguity to the overall benefit to the patient undergoing completion lymphadenectomy. A survival benefit has not been proven, morbidity from many lymphadenectomies is high, and few patients in the sentinel node era actually recur in the nodal basin making palliative interventions a low priority. It is a challenge to demonstrate an overall comparative effectiveness to completion lymphadenectomy and it will be a long time before any data is available to provide any insight. Obstacles to our understanding this in greater detail include inherent bias toward the MSLT-2 trial (patients are only referred for possible observation if they are perceived as “lower risk”), low incidence of events in this population necessitating a large patient population with extended follow up, and finally a long “tail” in which events can occur before data is conclusively determined to represent a comprehensive review. With these obstacles, there will be a considerable delay before the questions surrounding lymphadenectomy can be answered.

Similar to the issues regarding the optimal use of sentinel lymph node biopsy, the uncertainty around completion lymph node dissections need to be explored using alternative research methods. Studies that make use of large databases and pooled multi-institutional clinical data will help us avoid the bias inherent in MSLT-2 and the long follow-up time required for meaningful results. Decision analyses can also help us examine how we should guide our patient through the process of choosing a completion lymphadenectomy or not.