Nathansohn N, Schachter J, Gutman H. Patterns of Recurrence in Patients With Melanoma After Radical Lymph Node Dissection. Arch Surg. 2005;140(12):1172-1177. doi:10.1001/archsurg.140.12.1172
Copyright 2005 American Medical Association. All Rights Reserved. Applicable FARS/DFARS Restrictions Apply to Government Use.2005
Previous interventions (excisional biopsy, incomplete dissection) in the regional basin that drain a melanoma site prior to definitive surgical procedures significantly increase the risk of melanoma recurrence in the surgical field.
Tertiary care referral center.
One hundred forty-one consecutive patients who underwent radical lymph node dissection (RLND) either in the groin or the axilla owing to malignant melanoma were followed up for a median period of 41 months.
All of the 141 patients received either elective or therapeutic RLND. Their medical records were analyzed for demographic data, disease history, previous treatments, recurrence patterns, and survival.
Main Outcome Measures
Patterns of first recurrence after RLND and survival.
Radical lymph node dissection was performed on 148 lymph node basins (141 patients; 86 axillae and 62 groins). Nineteen patients (13%) received previous open interventions in the lymph node basin (tampering) other than radical dissection. Radical lymph node dissection was performed prophylactically in 38 basins (26%), for palpable disease in 75 (51%), and for a positive sentinel node in 35 (24%). There were 74 failures (52%) of RLND: 51 patients (70%) with systemic disease, 12 (16%) with recurrence in the surgical field, 9 (11%) with in-transit metastases, and 2 (3%) with local recurrence. On multivariate analysis, the only significant predictors of recurrence after RLND were Breslow thickness of greater than 4 mm (P = .02), tampering (P = .01), and lymph node capsular invasion (P = .001). Tampering was the only independent prognosticator of failure in the surgical field, as tampering was noted in 10 (83%) of 12 patients with failure in the surgical field as compared with 6 (10%) of 62 patients with other types of first failures (P<.001). This effect did not translate into a survival difference (P = .54). Failure in the surgical field was not detected in any of the patients who underwent sentinel lymph node biopsy.
Previous interventions (excisional biopsy, incomplete dissection) in the regional basin that drain a melanoma site prior to definitive surgical procedures significantly increase the risk of melanoma recurrence in the surgical field, and they should be avoided. Fine-needle aspiration and sentinel node biopsy, performed with strict surgical oncologic techniques, are safe with regard to failure in the surgical field.
Malignant melanoma poses a significant and growing public health burden in the United States and worldwide. The American Cancer Society, Atlanta, Ga, estimated that 54 200 new cases of invasive malignant melanoma and 37 700 new cases of melanoma in situ would be diagnosed in the United States in 2003.1 New cases of invasive malignant melanoma account for 5% of all of the new cases of invasive cancer in men and women.2
First failures during the first 2 years after primary excision of cutaneous metastatic melanoma are usually regional (50%-85%) whereas first failures after 5 to 10 years usually occur at distant sites.3- 8 The incidence of clinical involvement of the regional nodes in patients under observation following treatment of the primary site varies with the microstage of the primary lesion, from 10% for lesions that are 1 to 2 mm thick to 30% for lesions that are 3 to 4 mm thick.6
Prophylactic lymphadenectomy of the regional lymph node basins draining the site of the primary tumor was introduced in an attempt to improve locoregional control and prolong survival. In Europe, however, most centers adhere to a “wait-and-see” policy for lymphatic basins at risk; elective lymph node dissection for suspected occult lymph node metastases, once a common measure, is no longer routinely applied.3,9,10 For clinically involved (palpable) nodes, lymphadenectomy is recommended unequivocally.7,11,12
The described incidence of failure in the surgical field after regional lymphadenectomy for malignant melanoma ranges from 0% to 52%.3,5,13 Associated factors include location of the dissected nodal basin, number of involved nodes, presence of extracapsular nodal disease, and patient age.3,13,14 These recurrences are associated with a 5-year survival rate of 9% to 43% (as compared with 86% for patients without recurrence) and a 10-year survival rate of 5% to 35%.3
On the basis of studies showing an orderly progression of tumor cells in the initial phase of the metastatic process within the lymphatic system, in 1992, Morton et al15 introduced the practice of sentinel lymph node mapping and biopsy as part of the therapeutic procedure. Multiple studies have since validated their results.10,16- 23 Using this procedure, patients with melanoma can be carefully selected for surgical clearance of microscopically involved basins or for observational follow-up alone.24
Patients with lymph node metastases, either palpable or microscopic, currently undergo radical lymph node dissection (RLND) of the involved basin. The reported 5-year survival rate ranges from 19% to 43%.6,11,13,25 Better outcome was noted for patients who received surgical procedures for microscopic lymph node disease13 and for patients with unknown primary melanoma.13,25,26 Being older than 50 years, tumor thickness, extracapsular extension of the nodal metastases, and number of involved nodes were found to be independent prognosticators of reduced survival and reduced recurrence-free survival.
The median survival of patients with recurrent regional nodal metastases ranges from 5 to 14 months after redissection, and the overall 5-year survival rate is 19% to 38%.14,25 Previous tampering with the lymph node basin (ie, excisional biopsy) does not seem to adversely affect these rates.25,27
The aim of the present study was to identify the factors closely associated with recurrence of melanoma after radical dissection of metastatic lymphatic basins, with particular focus on recurrence in the surgical field.
Between 1990 and 2000, 141 patients with malignant melanoma underwent RLND by a single surgical oncologist (H.G.) at our tertiary care, university-affiliated center. They were followed up by the surgeon and a single oncologist (J.S.), thereby eliminating interoperator and interobserver variability. Patients after inguinal, axillary, or popliteal RLND, which was either elective or therapeutic, including patients with an unknown primary melanoma location, were enrolled in the study. Patients who had received cervical lymph node dissection and patients who were followed up for less than 12 months were excluded.
From 1990 to 1994, lymph node dissection was performed in our center either prophylactically for intermediate-thickness (1.5- to 4.0-mm) malignant melanoma or therapeutically for palpable metastases. Thereafter, preoperative lymphoscintigraphy, intraoperative lymphatic mapping, and sentinel lymph node biopsy were performed in all of the patients meeting 1 of the following criteria: intermediate-thickness melanoma; melanoma measuring 0.75 to 1.50 mm with additional risk factors (high mitotic rate, microscopic satellites, ulceration, primary tumor at the perianal or genital area, 2 synchronous primary tumors, delayed referral, discordance between the Clark level and Breslow thickness); melanoma measuring 4 to 7 mm; or melanoma of undetermined thickness (owing to extensive regression or technical failure).21 Radical lymph node dissection was performed in patients with positive sentinel nodes and mapping failures with a high-risk primary tumor. Therapeutic lymph node dissection was performed in all of the patients with palpable metastatic melanoma proven by fine-needle aspiration or by outside open biopsy. Deep (iliac and obturator) lymph node dissection was performed for all of the Cloquet nodes, positive or bulky superficial nodes, or pelvic lymphadenopathy per computed tomography.
All of the clinical and pathological data for this study were obtained from our prospective melanoma database and were verified and completed by reviewing the records of the departments of surgery and oncology and the surgical oncology clinic files.
The impact of the following factors on the risk of recurrence overall and in the surgical field was evaluated: patient sex and age at diagnosis of melanoma; primary tumor location (trunk vs extremities); Breslow thickness; regression; ulceration and mitotic rate in the primary tumor; presence of satellite or in-transit metastases; interval from diagnosis of melanoma to lymph basin surgical procedures and to first failure; lymph node capsular invasion at the time of basin surgical procedures; and the fine-needle aspiration procedure before RLND.
We also took into account previous tampering of the basin that received the surgical procedures. Tampering was defined as any open intervention in the lymph node basin carried out before performing a type of RLND that was the target of this study, namely, incomplete dissection of the lymphatic basin or excisional or incisional lymph node biopsy. A previous dissection was deemed incomplete if the present surgical procedure revealed anatomicopathological structures (ie, saphenous vein, deep fascia of thigh, lymph nodes) that should have been removed according to acceptable standard surgical oncologic practices.28,29 Radiotherapy was delivered to all of the patients with extracapsular extension of the lymph node metastases. High-dose interferon was offered only during the second half of the study period, after the US Food and Drug Administration had recognized it as an adjuvant to surgical procedures for patients with melanoma at high risk of recurrence.
Overall survival and recurrence-free survival were calculated from the date of RLND. The overall survival was measured to the point of death from melanoma; recurrence was defined as any new metastasis detected after RLND.
Survival probabilities were calculated with the Kaplan-Meier method. Survival was compared between subgroups with the Cox-Mantel log-rank test, and recurrence patterns were compared between subgroups with χ2 analysis; a P value of less than .05 was considered statistically significant.
The clinical characteristics of the 141 patients and the histologic characteristics of their 130 known primary tumors are listed in Table 1. In 7 patients, 2 basins received surgical procedures, for a total of 148 lymph node basins. The characteristics of the radical lymph node basin surgical procedures are listed in Table 2. Postoperative adjuvant therapy included radiotherapy (14 patients), high-dose interferon (37 patients), or radiotherapy plus high-dose interferon (42 patients).
Follow-up ranged from 12 to 252 months (median, 41 months). Seventy-four patients (52%) had some form of relapse after the surgical procedure for the basin. The pattern of first post-RLND relapse is depicted in Table 3.
The median disease-free interval (DFI) after RLND in this subgroup was 14.1 months. The median DFI to relapse that was limited to the surgical field (without concomitant systemic failure) (n = 12; median, 51.5 months; range, 6.4-152.5 months) was significantly longer than the median DFI for all of the other types of failures combined (median, 20.4 months; P<.001). The only other factor associated with shortened DFI was the presence of lymph node metastases in the basin that received the surgical procedure (ie, therapeutic vs elective RLND).
The factors that were significantly associated with any form of recurrence after RLND are listed in Table 4. On multivariate analysis using the logistic regression model, the only significant predictors of recurrence after RLND were Breslow thickness of greater than 4 mm (P = .02), previous tampering (P = .005), and lymph node capsular invasion (P<.001).
One might consider that tampering, per se, is a variable depending on tumor burden, ie, patients who have undergone lymph node basin tampering owing to higher tumor burden (palpable disease). However, such a relationship was not found: tampering was evident in 6 (9%) of 67 patients without palpable disease and in 13 (18%) of 74 patients with palpable disease (P = .15 by Fisher exact test).
Previous tampering was recorded in 10 of 86 axillary basins and in 11 of 62 inguinal basins. The difference was not statistically significant (P = .34).
In 17 patients, the first relapse after RLND (excluding patients with in-transit–only relapse) occurred in the surgical field: 12 of the 17 patients had no systemic metastases, and 5 patients had regional failure as part of systemic disease. Previous tampering was noted in 15 patients (88%) in this subgroup, including 10 (83%) of the 12 patients with pure regional failure. These 10 patients accounted for a significantly greater proportion of the tampering subgroup (67%) than patients with other types of failure (n = 5; P<.001). Previous tampering was the only factor that was significantly associated with relapse in the surgical field (P<.001). Relapse in the surgical field was not associated with any of the temporal variables tested, namely, DFI from diagnosis to first failure, DFI after RLND, or patient age, and it was not significantly associated with sex, characteristics of the primary lesion, the basin that received the surgical procedure, sentinel node biopsy, and elective vs therapeutic RLND.
Nine patients had in-transit metastases after RLND, with systemic disease in 1 of the patients. Significant factors associated with pure in-transit metastases (without systemic disease) were nodular melanoma (4 of 8 patients; P<.001), groin dissection (7 of 8 patients; P<.05), and Breslow thickness greater than 2 mm (P<.05). Neither previous tampering nor fine-needle aspiration were associated with in-transit metastases.
The first recurrence after RLND was systemic in 52 patients, including 6 patients with concurrent regional relapse (5 of whom had relapse in the surgical field and 1 of whom had in-transit failure). Comparison of these 52 patients with the patients with relapse in the surgical field alone (n = 12), isolated in-transit failure (n = 8), or local failure (n = 2) on univariate analysis failed to identify any risk factor that significantly predicted the occurrence of systemic failure as opposed to all of the types of isolated failures combined.
At the end of the follow-up period, 65 (46%) of the 141 patients were alive with no evidence of disease, 14 (10%) were alive with melanoma metastases, 53 (38%) had died of melanoma, 8 (6%) had died of other causes, and 1 (1%) was lost to follow-up. Median overall survival after RLND for the whole study group was 43.2 months (range, 1.3-251.6 months). Median survival of the 74 patients with recurrence was 31.1 months (mean, 44.7 months; 95% confidence interval, 34.5-54.8), which was significantly shorter than the median survival of 54.9 months (mean, 57.5 months; 95% confidence interval, 49.7-65.3) in the patients without recurrence (P = .05). The factors found to be significantly associated with overall survival after RLND on univariate analysis (using the Kaplan-Meier procedure) are presented in Table 5. On multivariate analysis using the Cox regression model, only pure regional failure retained its statistical significance (P = .02), predicting longer overall post-RLND survival than all of the other types of failures.
This study demonstrates the pattern of first relapse in patients with melanoma who are undergoing RLND, with emphasis on relapse in the surgical field. The rate of failure in the surgical field was 8%; its only independent prognosticator was previous tampering in the lymph node basin. In 1994, Warso and Das Gupta14 described a regional failure rate of 2.7% after lymph node dissection, and they found the number of positive nodes to be a significant prognostic factor. Lee et al,4 in a series of 338 patients after complete lymph node dissection, noted an overall nodal basin recurrence rate of 30% at 10 years. Extracapsular extension, therapeutic vs elective dissection, lymph node size, and the number of involved lymph nodes were independent prognostic factors for recurrence in the surgical field. Our study, however, did not confirm the significance of these factors. This discrepancy may be explained by the larger number of patients, higher percentage of recurrences in the surgical field, and longer follow-up in the study by Lee and colleagues. Lymph node size was not entered into our database, but we detected no single case of relapse in the surgical field in patients undergoing RLND for microscopically involved sentinel (or other) nodes. White et al13 documented the site of first recurrence in 2422 patients after lymph node dissection. They found a nodal recurrence rate of 12%, which is comparable to that found in our study. Regional recurrences are reportedly more common after inguinal or iliac lymph node dissection than after axillary lymph node dissection. Our data show similar rates of relapse in the axilla (7 [8.1%] of 86 basins), and groin (5 [8.1%] of 62 basins) surgical fields, but this difference did not reach statistical significance (P>.99). White and colleagues also observed that local, in-transit, and regional recurrences accounted for a larger proportion of first recurrences during the first 2 years after lymph node dissection as compared with recurrences occurring later. Only 5 of the 12 failures isolated in the surgical field in our study were detected within 24 months of RLND. The failures in the surgical field that were part of systemic metastases were detected earlier, after a median follow-up of 4.4 months (range, 2-30 months).
Pidhorecky et al3 specifically addressed the risk factors for recurrence in the surgical field after lymphadenectomy for melanoma. In their review3 of the 338 patients who were studied by Lee et al,4 the overall incidence of regional nodal recurrence in a dissected basin was 25%, as compared with 8% in our series, and the median time to nodal basin failure was 12 months, as compared with 51 months in our series. This may be an incidental finding, or, as we speculate, the difference may be attributable to the more homogenous nature of our sample. All of the patients were treated by the same highly dedicated surgeon and were closely followed up. The risk factors associated with nodal recurrence in the study by Pidhorecky and colleagues were advanced age, primary lesion in the head and neck region, thickness of the primary lesion, number of involved lymph nodes, and extracapsular extension. For each nodal basin, the elective dissection group had a significantly lower incidence of recurrence than the therapeutic dissection group. By contrast, in our series, there was no difference in the rate of failure in the surgical field between the groups: failure in the surgical field occurred in 3 (9%) of 33 patients who underwent elective RLND and in 9 (8%) of 108 patients who underwent therapeutic RLND. Again, the difference between the series may be explained by our 1-surgeon design. We also included no patients with head and neck melanomas, which may have affected the overall rate of recurrence and post-RLND DFI in the earlier study by Pidhorecky and colleagues.
Goydos et al30 found a similar pattern of melanoma recurrence after lymph node dissection in histologically node-positive patients (50% overall recurrence rate) and patients who are histologically negative but are positive by reverse transcriptase–polymerase chain reaction (21% overall recurrence rate). Thickness of the primary tumor and sentinel lymph node status were the most important predictors of recurrence and death. It is noteworthy that the article by Goydos and colleagues refers to microscopic or submicroscopic tumor deposits only, so the findings cannot be directly compared with ours. Nevertheless, in our subgroup of patients with only microscopic disease (31 patients, 33 basins), we found a recurrence rate of 42% (compared with 50% in the study by Goydos and colleagues), no pure regional failure (as in the study by Goydos and colleagues), a 15% rate of in-transit metastases (compared with 47% found by Goydos and colleagues), a 69% rate of systemic failure (compared with 53% found by Goydos and colleagues), and a 15% rate of local recurrence (compared with 0% found by Goydos and colleagues). We could not find any differences in the pattern of failure between patients with microscopic or palpable metastatic spread, and this factor was not a significant prognosticator for recurrence.
The pattern of first recurrence in our patient population is in line with previous descriptions: 12 patients (16%) had failure in the surgical field alone, 8 (11%) had isolated in-transit metastases, 2 (3%) had local recurrence, and 52 (70%) had systemic failure. Overall survival was significantly associated with multiple independent factors (Table 4). Multivariate analysis, however, excluded all but Breslow thickness and the number of positive nodes. On multivariate analysis, Breslow thickness, previous tampering, and capsular invasion were significantly associated with overall recurrence. While Breslow thickness and extracapsular extension are well-documented prognosticators of poor outcome, previous tampering has rarely been mentioned in this context.
The surgical treatment of regional metastases has been described as curing some patients and prolonging the DFI of others. This was confirmed in our study as well. It certainly offers the best chance for regional control of the disease. The impact of the surgeon's training on outcome, a well-established factor in other cancers,31,32 was addressed in melanoma by only 2 studies.14,33 Neither study noted a difference in patients treated by surgeons with different professional backgrounds. In sharp contrast, our data show that previous intervention in the lymph node basin that is affected by melanoma by surgeons who do not strictly adhere to surgical oncologic principles has a significant adverse effect on the overall recurrence rate and on the rate of recurrence in the surgical field. The finding that sentinel node biopsy did not lead to any relapse in the surgical field in our series further supports the assumption that the application of strict surgical oncologic techniques in these cases is of utmost importance. Although the relapses in the surgical field did not translate into a statistically significant impact on overall survival, they represent a surgical failure associated with significant morbidity as well as a negative effect on quality of life. Minimizing preoperative tampering can significantly reduce the risk of relapse in the surgical field after RLND for metastatic melanoma.
Correspondence: Haim Gutman, MD, Department of Surgery B, Rabin Medical Center, Beilinson Campus, Petah Tiqva 49100, Israel (email@example.com).
Accepted for Publication: February 7, 2005.
Author Contributions: Drs Schachter and Gutman contributed equally to this work.
Acknowledgment: We thank Gloria Ginzach and Charlotte Sachs for their editorial assistance.