Fax: (713) 794-5573
Adjuvant irradiation for cervical lymph node metastases from melanoma
Article first published online: 18 MAR 2003
Copyright © 2003 American Cancer Society
Volume 97, Issue 7, pages 1789–1796, 1 April 2003
How to Cite
Ballo, M. T., Bonnen, M. D., Garden, A. S., Myers, J. N., Gershenwald, J. E., Zagars, G. K., Schechter, N. R., Morrison, W. H., Ross, M. I. and Kian Ang, K. (2003), Adjuvant irradiation for cervical lymph node metastases from melanoma. Cancer, 97: 1789–1796. doi: 10.1002/cncr.11243
- Issue published online: 18 MAR 2003
- Article first published online: 18 MAR 2003
- Manuscript Accepted: 19 NOV 2002
- Manuscript Received: 8 OCT 2002
- National Cancer Institute, U.S. Department of Health and Human Services. Grant Number: CA 06294
- Gilbert H. Fletcher Chair
- head and neck neoplasm;
- lymph nodes;
The risk of regional disease recurrence after surgery alone for lymph node metastases from melanoma is well documented. The role of adjuvant irradiation remains controversial.
The medical records of 160 patients with cervical lymph node metastases from melanoma were reviewed retrospectively. Of these, 148 (93%) presented with clinically palpable lymph node metastases. All patients underwent surgery and radiation to a median dose of 30 grays (Gy) at 6 Gy per fraction delivered twice weekly. Surgical resection was either a selective neck dissection in 90 patients or local excision of the lymph node metastasis in 35 patients. Only 35 patients underwent a radical, modified radical, or functional neck dissection.
At a median follow-up of 78 months, the actuarial local, regional, and locoregional control rates at 10 years were 94%, 94%, and 91%, respectively. Univariate analysis of patient, tumor, and treatment characteristics failed to reveal any association with the subsequent rate of local or regional control. The actuarial disease-specific (DSS), disease-free, and distant metastasis-free survival (DMFS) rates at 10 years were 48%, 42%, and 43%, respectively. Univariate and multivariate analyses revealed that patients with four or more involved lymph nodes had a significantly worse DSS and DMFS. Nine patients developed a treatment-related complication requiring medical management, resulting in a 5-year actuarial complication-free survival rate of 90%.
Adjuvant radiotherapy resulted in a 10-year regional control rate of 94%. Complications for all patients were rare and manageable when they did occur. The authors recommend adjuvant irradiation for patients with extracapsular extension, lymph nodes measuring 3 cm in size or larger, the involvement of multiple lymph nodes, recurrent disease, or any patient having undergone a selective therapeutic neck dissection. Cancer 2003;97:1789–96. © 2003 American Cancer Society.
Adjuvant radiotherapy for lymph node metastases from melanoma is generally reserved for patients with clinicopathologic features indicating a high risk for subsequent regional recurrence after surgery alone. Although debated in the literature, generally accepted high-risk features include extracapsular extension (ECE), large lymph nodes, the involvement of multiple lymph nodes with metastatic disease, clinically palpable adenopathy (i.e., therapeutic dissection rather than elective dissection), and cervical lymph node location.1–7 Failure rates ranging from 30% to 50% have been reported when these high-risk clinicopathologic features are present.1–7 Despite this well documented pattern of disease recurrence, routine irradiation is often avoided because of an unsubstantiated belief that radiotherapy is ineffective for melanoma as well as an equally unsubstantiated concern regarding long-term sequelae of radiation.
To address these issues, a study was initiated at the University of Texas M. D. Anderson Cancer Center (MDACC) to evaluate the role of radiotherapy as an adjuvant to surgery for patients with cervical lymph node metastases.8 After reporting the results of this study in 1994, patients with American Joint Committee on Cancer Stage III disease or cervical lymph node recurrence without distant metastases were routinely administered adjuvant irradiation. In the current series, we update the outcome of patients with cervical lymph node metastases from malignant melanoma managed with surgical resection and adjuvant radiation.
MATERIALS AND METHODS
Patients were identified through a search of the MDACC Department of Radiation Oncology database. Between 1983 and 1998, 160 patients were identified with cervical lymph node metastases from melanoma who were treated with macroscopic total surgical resection and radiation. No patient was reported to have visceral metastases at the time of irradiation.
There were 133 males and 27 females. The median age of the patients at the time of cervical lymph node dissection was 52 years (range, 16–86 years). The primary site was the head and/or neck in 126 patients and unknown in the remaining 34 patients. All patients underwent wide local excision of the primary lesion, when known, with negative microscopic resection margins. Tumor characteristics at the time of initial melanoma diagnosis are shown in Table 1. The mean and median Breslow thickness in patients with a known primary tumor were 2.9 mm and 2.3 mm, respectively (range, 0.1–13 mm).
|AJCC stage at initial melanoma diagnosisa|
|Primary site known|
|Breslow thickness (mm)|
Seventy-one patients were lymph node positive at initial diagnosis and the remaining 89 patients had lymph node recurrence after previous primary surgery. One hundred forty-eight patients (93%) presented with palpable lymph node disease. The median time from diagnosis of the primary to the lymph node recurrence was 12.8 months (range, 1.2–331 months). For the 150 patients in whom the largest involved lymph nodes were documented, the mean and median were 2.2 cm and 2.0 cm, respectively (range, 0.1–13 cm). Pathologic review of the surgical specimens ascertained the involvement of a median of 2 lymph nodes (range, 1–58 involved lymph nodes), whereas a median of 23 lymph nodes (range, 1–107 lymph nodes) were dissected.
The surgical procedure for lymph node disease included radical, modified radical, or functional neck dissection for 35 patients; selective neck dissection for 90 patients; or local excision of macroscopic disease without formal neck dissection for 35 patients. Fourteen of the patients in this last group underwent superficial (13 patients) or total (1 patient) parotidectomies for lymph node disease without formal neck dissection. The selective neck dissection group consisted primarily of patients undergoing either supraomohyoid or posterolateral neck dissections. Only 12 of the 160 patients (7.5%) had lymph node disease discovered at the time of elective neck dissection. Of these 12 patients, 7 underwent selective neck dissection and 4 underwent lymph node sampling without formal neck dissection. One patient underwent elective radical neck dissection.
Patients with cervical lymph node metastases at initial diagnosis (Stage III disease) or cervical lymph node recurrence were routinely offered adjuvant irradiation. Despite these broad indications, based on analyses from this and other institutions, the majority of patients are believed to have been at high risk for subsequent local or lymph node recurrence after surgery alone. To illustrate, all but 43 patients were referred because their neck dissections revealed 1 or more of the following findings: ECE (78 patients), a lymph node 3 cm or larger (34 patients), the involvement of four or more lymph nodes (41 patients), and/or recurrent disease after initial surgical dissection for pathologically confirmed neck disease (9 patients). Thirty-eight patients had two or more of these factors. ECE was defined as either microscopic or macroscopic (matted lymph nodes) evidence of growth outside the lymph node capsule. For 33 of the 43 patients (77%) without 1 of these high-risk features, the surgical procedure was selective neck dissection or local excision of macroscopic disease only without formal neck dissection.
Radiotherapy portals were designed to cover the primary site and ipsilateral cervical lymphatics including the supraclavicular fossa while respecting the normal tissue tolerance of the spinal cord and temporal lobe (Fig. 1). The majority of patients were immobilized in an “open neck” position and radiation was delivered using multiple low-energy electron beams. In general, if a patient presented with a parotid lymph node metastasis, the field of irradiation would encompass the preauricular and postauricular regions and the entire ipsilateral neck including the supraclavicular fossa. If a patient presented with a cervical metastasis, the field of irradiation would encompass the neck and supraclavicular region only, unless the parotid gland was included in the dissection. The primary site was included in the field of irradiation as long as the time between primary diagnosis and irradiation was 1 year or less. Seventeen patients were treated preoperatively, but were included in this analysis because surgery was performed within a median of 4 days after the end of radiation. The prescription dose was generally 30 Gy delivered twice weekly (i.e., Monday and Thursday or Tuesday and Friday) at 6 Gy per fraction over 2.5 weeks, specified as a maximal dose. The junction between abutting electron fields was moved after the second and fourth fraction. In eight patients, a single boost dose (6 Gy) was delivered to a reduced field. Ten of the 17 patients treated preoperatively received a dose of 24 Gy in four fractions. One patient was treated at 2 Gy per fraction to a total dose of 60 Gy. A beveled tissue equivalent bolus (usually 2-cm thick) was placed above a line drawn between the lateral canthus and mastoid to limit the dose to the temporal lobe. An additional tissue equivalent bolus was placed over the larynx.
The duration of follow-up for the 77 patients alive at last contact ranged from 6 to 224 months with a median of 78 months (6.5 years). Disease recurrence was scored if there was any clinical or radiographic evidence of tumor regrowth. Actuarial data for local control (LC), regional control (RC), disease-free (DFS), distant metastasis-free (DMFS), disease-specific (DSS), overall, and complication-free survival curves were calculated using the Kaplan–Meier method9 and tests of significance were based on the log rank statistic. The significance of differences between proportions was tested with the chi-square statistic or with the Fisher exact test, as appropriate.10 Multivariate analysis was done with the Cox proportional hazards model.9 The completion of radiotherapy was used as Time zero. Complications were retrospectively classified according to symptomatology as follows: Grade 1, asymptomatic finding noted at the time of follow-up physical examination; Grade 2, symptomatic finding requiring any form of medical therapy (e.g., prescription for pain medications); and Grade 3, finding requiring surgical intervention.
At the time of analysis, only 13 patients had developed locoregional recurrence of disease that resulted in a 10-year locoregional control (LRC) rate of 91% (Fig. 2). The median time to development of locoregional disease recurrence in these 13 patients was 9.4 months (range, 2.3–21.4 months). The actuarial 10-year LC and RC rates were 94% and 94%, respectively. Five of the locoregional disease recurrences were local only, five were regional (within the ipsilateral neck only), and three were both local and regional. Six of the regional disease recurrences were within the field of irradiation, whereas two were marginal. Univariate analysis revealed no association between ECE, lymph node size, number of involved lymph nodes, disease presentation (primary vs. recurrent and therapeutic vs. elective dissection), type of surgical procedure, age, gender, Clark level, Breslow thickness, or time from diagnosis of the primary lesion to the development of lymph node metastases with the rate of LRC, LC, or RC. Specifically, the 5-year RC rates for select characteristics were as follows: ECE present versus absent, 95% and 93%,respectively; size 3 cm or larger versus smaller than 3 cm, 97% and 94%, respectively; 4 or more lymph nodes involved versus fewer than 4 lymph nodes involved, 91% and 95%, respectively; recurrent disease versus primary disease, 100% and 94%, respectively; therapeutic versus elective dissection, 95% and 91%, respectively; and local excision only versus other surgical procedure, 97% and 94%, respectively. None of these differences were statistically significant. Figure 3 shows the RC curves according to the presence or absence of ECE.
Eighty-three of the 160 patients had developed disease recurrence and 73 patients had died of disease. The 5 and 10-year overall survival rates were 48% and 39%, respectively. The 10-year DSS and DMFS rates were 48% and 43%, respectively (Fig. 2). The 10-year overall DFS rate was 42%. The median time to development of distant disease recurrence was 9.4 months (range, 0.69–139 months). The first site of distant disease recurrence was the lung in 22 patients, the skin or subcutaneous tissue in 16 patients, the liver in 16 patients, the brain in 10 patients, distant lymph nodes in 9 patients, the bone in 4 patients, and other sites in 4 patients. Univariate analysis revealed no association between patient or tumor characteristics and the 5-year DFS rate, although it did decrease as the number of involved lymph nodes increased (1–3, 50%; 4–10, 40%; and > 10, 33%; P = 0.07) and when Clark Level IV or V disease was noted histologically (41% vs. 57%; P = 0.07). Univariate analysis did reveal an inferior actuarial 5-year DSS and DMFS when 4 or more lymph nodes were involved (DSS: 42% vs. 57% [P = 0.03] and DMFS: 37% vs. 52% [P = 0.04]). Grouping patients according to increasing number of positive lymph nodes (i.e., 1–3 vs. 4–10 vs. > 10) also correlated with a progressive decline in the 5-year DSS and DMFS rates (DSS: 1–3, 57%; 4–10, 48%; and > 10, 32% [P = 0.02] and DMFS: 1–3, 52%; 4–10, 40%; and > 10, 33% [P = 0.04]). There was a trend toward an association between a lower DSS and lymph nodes 3 cm or larger (43% vs. 58%; P = 0.09), Clark Level IV or V disease (46% vs. 64%; P = 0.08), and therapeutic dissection (51% vs. 78%; P = 0.08). Age, gender, lymph node disease presentation (primary or recurrent), ECE, Breslow thickness, and time from diagnosis of the primary lesion to the development of lymph node metastases did not appear to impact the rates of DSS, DMFS, or DFS. The association between the number of positive lymph nodes and an inferior DSS and DMFS remained significant on multivariate analysis (DSS, P = 0.01; DMFS, P = 0.04). The association between the number of positive lymph nodes and an inferior DFS was of only borderline significance on multivariate analysis (P = 0.06).
Treatment-related complications were documented in 27 patients. Given the retrospective nature of this review, it was difficult to categorize complications as specifically related to surgery or radiation. Eighteen complications were regarded as Grade 1 (asymptomatic finding noted at the time of follow-up physical examination) and nine were Grade 2 (symptomatic finding requiring any form of medical therapy). There were no Grade 3 complications (requiring surgery). These results yielded 5-year actuarial complication rates of 12% and 10% for Grade 1 and Grade 2 complications, respectively (Fig. 4). The Grade 1 complications all involved the irradiated skin or mucous membranes and corresponded to a Radiation Therapy Oncology Group/European Organization for Research and Treatment of Cancer late radiation morbidity score of Grade 1, which is defined as slight induration, loss of subcutaneous fat, or slight atrophy. The Grade 2 complications comprised ipsilateral decreased hearing in three patients, clinical hypothyroidism in two patients, wound breakdown in two patients, and bone exposure and mild ear pain in one patient each. One of the more severe Grade 2 complications occurred when a patient applied a heating pad to his surgical scar and burned his irradiated skin, resulting in hospitalization. The mean time to development of a Grade 2 complication was 27.3 months (range, 2.76–73.6 months).
Adjuvant irradiation is routinely considered at this institution for patients with lymph node metastases from melanoma when cervical dissection reveals clinicopathologic features believed to predict a high risk for subsequent regional disease recurrence. Defining what constitutes a high-risk feature, however, has been challenging. Disease presentations may range from microscopic burden detected by sentinel lymph node biopsy to palpable disease requiring extensive lymph node dissection (both Stage III disease). In addition, patients may present years after the diagnosis of Stage I or II disease with clinically detected lymph node recurrence. Certainly, we cannot assume that the risk of subsequent disease recurrence in these patient groups is equal. Also, we must consider competing risks, such as the development of distant metastases, when making treatment recommendations. To complicate matters further, for a single patient, the recommended surgical procedure might vary from radical neck dissection to selective dissection of macroscopic disease only. This decision is often institution specific.
These points aside, there are a few consistently reported factors associated with an elevated risk of locoregional disease recurrence after surgery alone for lymph node metastases. The most consistently reported factor has been ECE. Two early surgical series from MDACC, one including 287 patients with cervical lymph node metastases and one including 1001 patients with any lymph node metastasis, revealed significantly higher lymph node recurrence rates after surgery alone when lymph nodes were described as matted (e.g., macroscopic ECE).2, 11 In the series confined to cervical metastases, the overall lymph node recurrence rate was only 15%, but rose significantly to 44% when macroscopic ECE was found. Investigators from Roswell Park Cancer Institute also reported a significantly elevated regional disease recurrence rate in the presence of ECE.1 In their series of 338 patients, including 56 neck dissections, the lymph node recurrence rate was 63% when ECE was noted microscopically. This association remained significant on multivariate analysis. In a recent series from the John Wayne Cancer Institute, lymph node ECE was the single most important predictor of subsequent regional disease recurrence after surgery alone.5 In their series of 217 patients, the overall regional disease recurrence rate was only 14%, but rose to 31% in the presence of ECE (P = 0.004). Although other factors, such as the status of disease (therapeutic vs. elective dissection) and the number of positive lymph nodes, did not reach statistical significance in the John Wayne Cancer Institute series, a number of other larger series have substantiated the prognostic value of additional characteristics that must be considered when evaluating a patient for adjuvant radiation.
Specifically, an MDACC series regarding cervical lymph node metastases managed with therapeutic modified neck dissection reported a regional disease recurrence rate of 50% for patients with multiple positive lymph nodes.3 This compared with a regional disease recurrence rate of only 17% when the dissection was performed electively, but lymph nodes were involved. A similar series of 998 patients was reported by the Sydney Melanoma Unit.4 For patients undergoing elective neck dissection, the disease recurrence rate was only 9%, but 34% if the dissection was therapeutic. This association was confirmed yet again in the recent analysis from Roswell Park where the regional disease recurrence rates according to therapeutic or elective dissection were 36% and 16%, respectively.1 Although the significance of therapeutic versus elective dissection did not reach statistical significance in the series from the John Wayne Cancer Institute, patients presenting with palpable lymph node disease did have a regional disease recurrence rate of 20% compared with only 4% if disease was nonpalpable.5
Singletary et al.11 reported regional disease recurrence rates according to the number of involved lymph nodes as follows: 1 lymph node, 9% disease recurrence rate; 2–4 lymph nodes, 10%; 5–10 lymph nodes, 22%; and more than 10 lymph nodes, 26%. Similarly, two additional series have revealed an increasing risk of regional disease recurrence with increasing number of positive lymph nodes. In 1 series, the disease recurrence rates according to the number of positive lymph nodes were 14% and 53% for 1–3 and for 4 or more lymph nodes, respectively (P less than 0.0001).7 Likewise, in the second series, the risk of regional recurrence was 25% for 1–3 lymph nodes, 46% for 4–10 lymph nodes, and 63% for more than 10 lymph nodes, (P = 0.0001).1
Although less frequently reported, lymph node size is also associated with regional disease recurrence rates. In the Roswell Park Cancer Institute series, the risk of lymph node recurrence did increase significantly as the size of the largest lymph node increased (< 3 cm, 25% disease recurrence rate; 3–6 cm, 42%; and > 6 cm, 80% [P < 0.001]).1 Therefore, lymph node size must also be taken into account when considering adjuvant radiation.
One additional factor that has particular bearing on this discussion is the location of the lymph node disease. At least two series have reported that lymph node recurrence rates are higher for cervical disease compared with disease recurrence in the axillary or inguinal basins. Specifically, in the Roswell Park series, the regional disease recurrence rate after dissection alone was 43% for cervical disease compared with 28% and 23% for axillary and inguinal disease, respectively (P = 0.008).1 In another smaller series, the regional disease recurrence rate for cervical disease was 33% compared with 13% and 9% for axillary and inguinal disease, respectively.6
Therefore, although ECE is the single most important factor associated with the risk of regional disease recurrence, review of published series would suggest a clear association with the type of dissection (therapeutic vs. elective), the number of involved lymph nodes, the size of the lymph nodes, and the site of disease. Although no series has directly analyzed regional disease recurrence rates according to disease presentation (i.e., disease recurrence after previous dissection vs. primary presentation), recurrent disease is a well accepted risk factor for regional disease recurrence. The presence of any one of these high-risk features results in (at least) a 30–50% risk of subsequent lymph node recurrence after surgery alone. This well documented fact must be considered when considering a patient for adjuvant radiation. In addition, we must bear in mind that lymph node recurrence, particularly in an anatomically confined structure such as the neck, can be refractory to therapy and can produce substantial morbidity.
With regard to the efficacy of cervical irradiation, the majority of reported series to our knowledge combine the cervical, axillary, and inguinal lymph node basins or include patients receiving palliative radiation for surgically unresectable lymph node disease. Several retrospective series, however, suggest significant improvements in RC over the surgical series described earlier.12–16 In a series from the Sydney Melanoma Unit reviewing 143 patients with cervical or parotid lymph node metastases, the 5-year actuarial regional disease recurrence rate was 40% after surgery alone compared with 18% in patients who received adjuvant radiation (P = 0.055).12 This apparent improvement occurred despite the fact that the irradiated patients presented with more extensive lymph node disease. In a recent analysis from MDACC reviewing adjuvant radiation for axillary lymph node metastases, the regional disease recurrence rate was only 13% despite the fact that the majority of patients presented with 1 or more high-risk features.16 Although this analysis suggested that a patient with lymph nodes bigger than 6 cm had inferior axillary control despite radiation, no other patient, tumor, or treatment characteristic significantly changed the RC rate for patients receiving adjuvant radiation. This finding is consistent with the current series where radiation abrogates the poor prognostic significance of the high-risk features.
Based on the high rate of regional disease recurrence when certain characteristics are present and the improvement in RC when radiation is used adjuvantly, we continue to recommend radiation in certain clinical situations. These include the presence of ECE, lymph nodes 3 cm in size or larger, the involvement of multiple lymph nodes, or recurrent disease after initial surgical resection. One might hypothesize that the poor prognostic significance of lymph node size or number is lost when radical neck dissection is performed. To date, however, the data do not support this suggestion. In fact, in the Roswell Park series, the majority of patients underwent either radical or modified radical neck dissection and a regional disease recurrence rate of 43% for cervical lymph node metastases was reported.1, 17 Despite radical surgery, the presence of ECE and cervical location remained predictive of regional disease recurrence on multivariate analysis.1 Because radical surgery does not significantly alter the prognostic significance of these high-risk features, we recommend selective neck dissection, when feasible, and adjuvant radiation. An exception would be a patient who had lymph node involvement at the time of elective sentinel lymph node biopsy where completion limited neck dissection is then performed. Unless ECE is noted when the permanent sections are examined, adjuvant radiation may be deferred.
We continue to advocate the hypofractionated regimen because of patient convenience, the positive experience reported in the current series and the axillary metastasis series,16 and a continued belief that some melanoma cell lines are more sensitive to large-dose-per-fraction irradiation. Although theoretically, small-dose heterogeneities can have large effects on fraction size and total dose that may result in profound changes in late normal tissue responses, we report few long-term complications with this regimen. For example, the 10-year long-term complication rate of 10% in the current series compares favorably with that seen after radical or modified radical lymphadenectomy alone.18 In one series comparing crude complication rates among various surgical procedures, the risk of functional deficit after cervical lymphadenectomy was 7% whereas the incidence of long-term pain was 6%.18 The reported incidence would have been higher had actuarial techniques been used. The 12% incidence of Grade 1 complications in the current series is believed to be clinically insignificant. Given the retrospective nature of this review, it is also very likely that this number underestimates the true incidence of Grade 1 toxicity. Suffice it to say that some patients will have atrophy and loss of subcutaneous fat, a very small price to pay for significant improvements in RC and the avoidance of unmanageable disease recurrence.
Adjuvant radiotherapy for patients with cervical lymph node metastases from malignant melanoma results in a 94% 10-year RC rate. Even in the presence of clinicopathologic factors known to increase the risk of regional disease recurrence, the RC rate with adjuvant radiation was very satisfactory. Radiation-related complications for all patients were rare and manageable when they did occur. We currently recommend adjuvant irradiation for patients with ECE, lymph nodes 3 cm or larger, the involvement of multiple lymph nodes, recurrent disease after previous surgery, or any patient having undergone a selective therapeutic neck dissection or local excision of macroscopic disease only.
- 9Survivorship analysis for clinical studies. New York: Marcel Dekker, 1991: 5–125., .
- 10Practical statistics for medical research. London: Chapman and Hall, 1991..