The management of regional lymph nodes in patients with clinically lymph node-negative (NN) cutaneous melanoma has changed over the past 15 years since the description and rapid international adoption of sentinel lymph node biopsy (SNB), first published in 1992 by Morton et al.1 Elective lymphadenectomy for NN melanoma came to be replaced by selective lymphadenectomy (SNB) with or without completion lymphadenectomy (CLND), which was reserved only for patients who harbored melanoma deposits in sentinel lymph node(s). There is general or universal agreement on the following: 1) Approximately 18% to 20% of patients with clinically NN melanoma will have positive sentinel lymph nodes; 2) sentinel lymphadenectomy will identify 95% of lymph node-positive (NP) patients (false-negative rate, approximately 5%) with truncal and extremity melanoma; 3) only 10% to 20% of patients will have melanoma identified in nonsentinel lymph nodes removed at CLND by hematoxylin and eosin staining; and 4) no patient benefits from the excision of negative lymph nodes. Therefore, the majority of patients who undergo CLND will receive all of the morbidity with no added benefit. Because SNB can accurately stage the lymph nodes with limited morbidity, what is the role of excising additional lymph nodes in sentinel lymph node-positive patients? Potential benefits of removing additional positive lymph nodes earlier rather than later include enhanced regional control and the removal of regional metastases, which themselves may be the source of distant metastases, therefore improving long-term survival. The Multicenter Selective Lymphadenectomy Trial 1 (MSLT-I), an international, randomized controlled trial of selective lymphadenectomy versus observation, hoped to show that early lymph node intervention was better than delayed intervention in patients who had regional lymph node metastases. The primary endpoint of that trial was melanoma-specific survival, and the third interim analysis results indicated that there was no difference between the observation group and the SNB study group (86.6% and 87.1%, respectively).2 An additional planned subgroup analysis of NP patients who underwent SNB (n = 122) compared with patients who became NP in the observation group (n = 78) revealed a 20% improvement in survival in the SNB group and a higher number of positive lymph nodes in the observational group that underwent therapeutic lymph node dissection [TLND].3 Critics have noted that such subgroup analyses after randomization have limited validity and that SNB may be removing and identifying false-positive, limited-burden lymph node disease that is not clinically relevant4, 5; however, it is exactly these NP patients who we need to evaluate.
Long-term follow-up of over 3300 patients with pathologically NP, American Joint Committee on Cancer (AJCC) stage III melanoma from the 2008 AJCC Melanoma Staging Database indicates that there has been a near flattening of the survival curves at 10 years, with a 25% to 70% 10-year survival rate,6, 7 suggesting that long-term control or cure is possible with the removal of positive lymph nodes in a substantial percentage of patients. Intuitively, it has always made sense to remove clinically relevant, positive lymph nodes when the tumor burden is lowest. The number of positive lymph nodes is related inversely to survival, and several studies have reported an increase in the number of positive lymph nodes in patients who undergo delayed TLND or undergo CLND after a false-negative SNB. In this issue of Cancer, Pasquali et al8 report a nonsignificant 18% higher 5-year overall survival of their patients (total number = 190) who underwent CLND over that of patients who underwent TLND. Then, they present a meta-analysis of 6 studies (5 were retrospective, including their study) that met strict criteria to include only patients with lymph node metastases who underwent CLND after a positive SNB or who underwent TLND for clinically detectable disease that was identified during follow-up and had sufficient survival measured from primary diagnosis to death. The pooled data from these 2633 patients favored CLND (hazard ratio, 1.60 for TLND compared with CLND) and were statistically significant using the fixed-effects model and several other sophisticated statistical models to account for heterogeneity and inconsistency across studies and to eliminate any dominant study. The bottom line is that this study still has the limitations of retrospective, nonrandomized data that leave us with inferences without definite conclusions.
The standard of care in the United States is completion lymphadenectomy for nearly all patients after a positive SNB. A very rational alternative is the MSLT-II trial currently underway, in which there is planned accrual of 1925 sentinel lymph node-positive patients for randomization over 7 years to determine whether there is improved melanoma-specific survival with CLND over observation (including regional ultrasound evaluation) with TLND reserved for patients who have clinically or radiologically identified regional recurrence. Because regional lymphadenectomy can lead to extensive morbidity—primarily life-long lymphedema after groin and axillary lymph node dissections—it would be an important clinical advance to identify sentinel lymph node-positive patients who can be followed safely with observation. Again, approximately 80% of these patients will have negative lymph nodes on CLND and will not benefit from having these lymph nodes removed. The MSLT-II trial is not designed as a noninferiority trial. If the results do not support the hypothesis that CLND has a survival advantage over delayed TLND, then we still may not have the answer to this very important question. However, this trial should be supported fully to give us the largest dataset in this important NP subgroup after SNB.
An additional unresolved question is how to define a lower limit of clinically significant lymph node positivity in patients who undergo SNB. van Akkooi et al9 reported no difference in the survival of 36 patients who had sentinel lymph node metastases <0.1 mm compared with a sentinel lymph node-negative cohort. However, Ollila et al10 reported disease recurrence in 24% (n = 33) vs 11% (n = 488) of patients, respectively, in these same subgroups. The AJCC seventh edition Melanoma Task Force considered all current evidence-based studies and concluded that, currently, there is no definition for a lower threshold of clinically insignificant lymph node metastases.6 Therefore, even positive cells identified only by immunohistochemistry in a sentinel lymph node will be staged as N1a (micrometastases). The significance of sentinel lymph nodes identified by reverse transcriptase-polymerase chain reaction was not included, and the most definitive data on this investigation will come from the Sunbelt Melanoma Trial, which has completed accrual. However, a related question is whether there is a lower limit of sentinel lymph node burden at which a CLND is not indicated. The incidence of additional positive lymph nodes after CLND in the studies by van Akkooi et al and Ollila et al was 1 of 36 patients (3%)9 and 1 of 30 patients (3%),10 respectively. It is possible that a lower threshold of lymph node tumor burden may reach significance as larger datasets are reported and CLND can be omitted in these patients. In the interim, accrual to the MSLT-II trial is strongly encouraged.
Subsequent to the widespread application of SNB, surgical progress to enhance survival or decrease morbidity in patients with melanoma has been nearly nonexistent. SNB is the best staging tool for patients who have clinically and radiologically negative lymph nodes and will become even more important in stratifying patients when systemic treatment becomes effective. Regional lymphadenectomy remains a ball-and-chain we carry and hang on our patients because we have not found anything better.