Elective Lymph Node Dissection in MelanomaIt isn’t Really Therapeutic, is It?
Abstract & Commentary
Synopsis: In a study of more than 700 patients with stage I and II melanoma, the Intergroup Melanoma Surgical Program found that immediate elective regional lymph node dissection (ELND) has no therapeutic value compared to delayed therapeutic dissection for patients with intermediate thickness (1-4 mm) melanomas. Although post-hoc subgroup analysis once again identified a subset of patients (those younger than 60 years, 1-2 mm thick, with no ulceration) who may benefit from ELND, the overall analysis of the trial does not support this conclusion.
Source: Balch CM, et al. Ann. Surg 1996;224:255-263.
Elective regional lymph node dissection (elnd) has long been a controversial issue. Retrospective analyses of individual surgical series have not yielded a clear answer. Two large randomized trials by the World Health Organization and the Mayo Clinic did not find any advantage for ELND, but this may not preclude a benefit for some patient subsets. If you believe that a delayed therapeutic lymph node dissection performed at recurrence is curative for some patients, it would seem reasonable to assume that this same subgroup of patients might be cured by ELND. The big question is whether ELND cures additional patients by resecting tumor-involved nodes before distant spread has occurred. The authors of the current study tried to identify subsets who might benefit by studying outcomes of patient sets from the University of Alabama at Birmingham and Sydney Melanoma Unit. The analysis led to a hypothesis that ELND benefited patients with intermediate thickness tumors (1-4 mm), and a clinical trial was generated using the Intergroup Melanoma Surgical Program to achieve a sample size sufficient for statistical validity.
The clinical trial entered 786 patients with stage I and II melanomas, excluding the lentigo maligna subtype. Patients were prerandomized before their initial biopsy to several options for local excision and randomized to ELND vs. observation with therapeutic lymph node resection at the time of recurrence. About 10% of the patients did not receive the assigned treatment. Surgical and pathology quality were monitored throughout the trial. The randomization achieved equal distribution of tumor characteristics such as thickness, presence of ulceration, site of lesion, and age and gender of host; factors that are known to have prognostic significance.
There was no difference in five-year survival between the two treatment groups; however, statistically significant differences were observed between subgroups. For example, five-year survival of patients less than 61 years old was 88% with ELND compared to 81% with observation (P = 0.04). Patients older than 60 years tended to have a worse survival with ELND (74% vs 86%; P = 0.238) compared to observation. Further subgroup analysis showed a significant increase in the five-year survival of the younger patients without tumor ulceration with ELND (95% vs 84%; P = 0.01); in the younger patients with melanomas 1-2 mm thick (96% vs 86%; P = 0.02); and particularly in the group with both characteristics, lesions from 1-2 mm and no ulceration (97% vs 87%; P = 0.005).
There was a trend for improved survival for the whole group when only tumors 1-2 mm were examined (P = 0.08) that reached statistical significance when survival was based on actual treatment received (P = 0.031). Similarly, when all patients without ulceration randomized to ELND were examined, there was a trend (P = 0.12) that reached significance when the groups were compared by actual treatment received (P = 0.018).
The authors feel that this clinical trial identifies a major subgroup of patients with melanoma whose overall survival is improved by ELND. This group includes patients 60 years of age and younger with intermediate thickness lesions, especially those whose tumors are 1-2 mm thick without ulceration. Obviously, that prediction now requires a prospective evaluation.
A wit once said that all mathematical problems are either insoluble or trivial. Similarly, it often seems that patients with melanoma have problems that are either trivial (thin lesions easily cured) or insoluble (deep lesions or widespread disease with poor prognosis). Fortunately, most patients present with thin lesions. The most important issue in treating patients with melanoma is identifying those patients who will benefit from a specific intervention. The highlighted article states that ELND is an intervention that is beneficial to specific patient groups. The authors identify the main beneficiaries as persons less than 60 years old having lesions 1-2 mm thick with no ulceration.
The current study grew out of the authors’ long interest in melanoma and their numerous previous studies of this question. The authors have expressed their concerns about the previous randomized clinical trials of ELND, the WHO Melanoma Group Trial1 and the Mayo Clinic trial.2 The authors reanalyzed the data from those studies and found that the WHO data suggested that tumor thickness and ulceration were extremely important prognostic variables. This information was not available before the WHO study and, consequently, was not used to stratify patients. The current study was aimed at correcting this deficiency. This is an appropriate use of post-hoc subgroup analysis, namely developing hypotheses for testing in subsequent randomized studies. Unfortunately, in the current study, age was not used to stratify patients, although balance was achieved between the two treatment groups without stratification.
When the data for all 740 patients were analyzed, there was no statistical advantage for the ELND group in the prior identified subsets. This is hardly surprising since it confirms the results of the two previous randomized studies. However, after the data were analyzed, the authors arrived at the age criterion that allowed them to reach the recommendation that a subset of patients indeed may benefit from ELND. This would appear to be the same sin that these authors identified in the WHO study. The paradox is heightened a bit by the observation that the group over age 60 years had a worse prognosis with ELND. This raises the possibility that the observed difference in a patient subset is a mathematical artifact. The authors’ conclusion that this is the first randomized study "to prove" (emphasis mine) the value of surgical treatment for clinically occult regional metastases does not seem justified. Purportedly significant differences detected by post-hoc subgroup analysis don’t prove anything. As Blake Cady commented in the discussion following this paper, " . . .if you torture data long enough, they will confess to anything."
Should we now support a randomized trial of ELND as a treatment for melanoma in younger people? How about in younger people with 1-2 mm lesions and no ulceration? Probably not. The recent publication of an effective adjuvant treatment for patients with positive lymph nodes (interferon-a) seems to have made this a moot question. The role of ELND has shifted from being used with therapeutic intent to being used with the goal of defining patient prognosis. Indeed, the use of ELND has probably increased as a result of the interferon adjuvant study. In breast cancer, we perform node dissections to identify node-positive patients who may benefit from adjuvant therapy. We don’t wait until relapse in the axillary nodes is clinically apparent before we initiate treatment. Breast cancer and melanoma are quite distinct diseases, but the principles of treatment that have evolved from clinical research in breast cancer have been untestable in melanoma because of the absence of an effective adjuvant treatment. Perhaps it would be of interest to examine whether early use of interferon for microscopic node-positive disease is more or less effective than use of interferon after therapeutic lymph node dissection at the time of clinical recurrence. In the absence of a clinical trial, it makes sense to extrapolate from the existing data to recommend staging ELND followed by interferon if the nodes are positive. For those convinced of ELND’s therapeutic potential, perhaps a randomized trial of ELND vs. interferon in patients under age 60 with nonulcerating 1-2 mm lesions would be of interest. However, other developments seem likely to reduce the need for ELND. Sentinel node biopsy, imaging techniques, and PCR detection methods may lead to lymph node sampling as the tool to distinguish prognostic subsets rather than removal of all the nodes in an entire region. It is important that we continue to support prospective clinical trials addressing important management questions.
1. Veronesi U, et al. Cancer 1982;49:2420.
2. Sim FH, et al. Cancer 1978;41:948.
3. Kirkwood JM, et al. J Clin Oncol 1995;14:7.