Controversy exists as to whether patients with thick (Breslow depth >4 mm), clinically lymph node-negative melanoma require sentinel lymph node (SLN) biopsy. The authors examined the impact of SLN biopsy on prognosis and outcome in this patient population.
A review of the authors' institutional review board-approved melanoma database identified 293 patients with T4 melanoma who underwent surgical excision between 1998 and 2007. Patient demographics, histologic features, and outcome were recorded and analyzed.
Of 227 T4 patients who had an SLN biopsy, 107 (47%) were positive. The strongest predictors of a positive SLN included angiolymphatic invasion, satellitosis, or ulceration of the primary tumor. Patients with a T4 melanoma and a negative SLN had a significantly better 5-year distant disease-free survival (DDFS) (85.3% vs 47.8%; P < .0001) and overall survival (OS) (80% vs 47%; P < .0001) compared with those with metastases to the SLN. For SLN-positive patients, only angiolymphatic invasion was a significant predictor of DDFS, with a hazard ratio of 2.29 (P = .007). Ulceration was not significant when examining SLN-positive patients but the most significant factor among SLN-negative patients, with a hazard ratio of 5.78 (P = .02). Increasing Breslow thickness and mitotic rate were also significantly associated with poorer outcome. Patients without ulceration or SLN metastases had an extremely good prognosis, with a 5-year OS >90% and a 5-year DDFS of 95%.
The preliminary results from the Multicenter Selective Lymphadenectomy Trial-I represent the first randomized prospective clinical trial to reveal a potential survival advantage to performing sentinel lymph node (SLN) biopsy in patients with malignant melanoma. However, these results were limited to patients with intermediate-thickness melanoma.1 Controversy continues to surround the use of SLN biopsy in patients with both thin (<1.0 mm in Breslow thickness) and thick melanoma (≥4.0 mm in Breslow thickness). For patients with thin melanoma, the discussion has centered around identifying a subset of these patients who may still benefit from SLN biopsy.2-8 For patients with thick melanoma, there remains controversy as to whether SLN biopsy is beneficial.
The arguments for performing SLN biopsy at the time of wide local excision in any patient with melanoma include prognostic and staging information, implications of SLN status for adjuvant therapy decisions, and the potential therapeutic impact of completion lymph node dissection in patients with a positive SLN. However, many clinicians contend that SLN biopsy may not be necessary in patients with thick melanoma, because these patients have a high rate of both regional and systemic occult disease at the time of presentation.9 Thus, SLN biopsy may not provide significant prognostic information beyond Breslow thickness, nor would completion lymph node dissection have any significant impact on outcome should the SLN biopsy be positive. In regard to adjuvant therapy, high-dose interferon alpha-2b was approved by the US Food and Drug Administration (FDA) in 1995 for adjuvant therapy of both lymph node-positive and T4 lymph node-negative disease,10 so these patients are already candidates for treatment based on their tumor thickness. Thus, SLN biopsy may not be necessary for adjuvant therapy decisions. To shed additional light on this controversy, we examined the impact of SLN biopsy among patients with melanoma ≥4.0 mm in Breslow thickness.
MATERIALS AND METHODS
This study was approved by the University of Michigan Institutional Review Board. Our prospectively collected melanoma database was queried for patients with primary cutaneous melanoma ≥4.0 mm in Breslow thickness (American Joint Committee on Cancer [AJCC] stage T4) who underwent definitive surgical treatment at the University of Michigan between January of 1997 and December of 2007. After excluding patients with clinical or radiographic evidence of distant disease, 293 consecutive cases were identified. Our SLN biopsy technique and method of SLN evaluation has been described previously.3 All primary lesions and SLN biopsy slides were reviewed by 1 of 4 dermatopathologists at the University of Michigan, who generated a 14-point melanoma profile for the primary lesion. Survival information was obtained from computerized medical records and from the Tumor Registry. The local and disease-free survival (DDFS) were calculated from the time of definitive surgery. Locoregional recurrence was defined as any local recurrence, in-transit recurrence, or regional recurrence within a primary draining lymph node basin.
Patient and tumor characteristics were tested for significant association with a positive SLN using the chi-square and t test for categorical and continuous characteristics, respectively. Overall survival (OS) and DDFS were calculated for each patient beginning at the date of surgery. For OS patients known to be alive were censored on their last contact date. For DDFS, patients were followed until date of documented distant disease or death. Patients known to be alive without intervening distant failure were also censored on their last contact date. The product limit method of Kaplan and Meier was used to estimate survival probabilities, with the population stratified by the positivity of the sentinel lymph node basin. Cox proportional hazard models were used to estimate the bivariate association between survival endpoints and the available patient and tumor characteristics. Those characteristics found to be significantly associated in bivariate models were simultaneously modeled, to create a best multivariate model for both endpoints separately. The best multivariate model was defined as the model that resulted from application of a backward elimination algorithm, whereby the multivariate model was pared down iteratively, removing sequentially the characteristic with the highest Wald-type P value, until only significant characteristics remained. This model schema was also applied to those cases with negative sentinel lymph nodes and those with at least 1 positive sentinel lymph node, separately. Significance for all statistical tests was defined as a P value <5%.
The patient characteristics of the study population are listed in Table 1. The mean age of patients was 58 years, with a range of 4 to 94 years, including 7 patients under the age of 16 years. The median Breslow thickness was 5.6 mm, with a range of 4 to 24 mm. Primary tumor distribution was relatively well distributed between the trunk, head and neck, and extremities. Six patients had subungual melanomas. The most common melanoma histologic pattern subtype was nodular (38%), followed by superficial spreading (20%), desmoplastic (9%), acral lentigo (6%), and lentigo maligna (5%). Ulceration was present in over half of the cases (53%), and angiolymphatic invasion was present in nearly 1 quarter (24%).
Table 1. Characteristics of 293 Patients With Thick Primary Melanoma
Head and neck
Breslow thickness, mm
Mitotic rate, mitoses/mm2
Clinical nodal involvement
Of the 293 patients, 18 (6%) presented with clinically evident regional metastases. Of the remaining 275 patients, SLN biopsy was performed in 227 (83%). Of the 48 clinically lymph node-negative patients who did not have a SLN biopsy, 21 did not undergo the procedure based on the patient's elderly age and comorbidities. In 17 cases, the patient and physician opted not to perform SLN biopsy for undocumented reasons. In 8 patients, a SLN biopsy was attempted but failed. Two patients had an elective lymph node dissection (ELND). For patients who had a successful SLN biopsy, the melanoma mapped to a single basin in 161 patients, 2 basins in 65, and 3 basins in 2. The median number of harvested SLNs per basin was 2 (range, 1-10).
The SLN was positive in 107 (47%) patients. Factors associated with a positive SLN are presented in Table 2. There was no association between the year of surgery and the presence of a positive SLN. Histologic pattern subtype was significant primarily because patients with a desmoplastic melanoma had a significantly lower risk of SLN metastases (11%), an observation we have previously reported.11, 12 Location of the primary melanoma did impact the likelihood of a positive SLN, with a lower likelihood in melanoma on the head and neck, although this too may be driven by the high percentage of desmoplastic melanoma in this region. Among 57 head and neck melanomas undergoing SLN biopsy, 13 (23%) were desmoplastic, with a SLN positivity rate of 15% compared with 35% for nondesmoplastic melanoma. Satellitosis (P = .009), angiolymphatic invasion (P = .001), and ulceration (P = .007) were all significantly more likely to be associated with a positive SLN, whereas sex, mitotic rate, regression, and neurotropism were not predictive. Although Breslow thickness is strongly predictive of SLN positivity, among patients with thick (≥4.0 mm) melanomas, the incidence of a positive SLN did not continue to rise significantly with increasing thickness.
Table 2. Association of Patient Characteristics With a Positive SLN Among Patients With T4 Melanoma
With SLN Biopsies
Chi- Square Test P
SLN indicates sentinel lymph node; SD, standard deviation.
*Unknown/missing group is excluded from the statistical test.
Sex, No. (%)
Patient age at surgery, mean y (SD)
Head and neck
Breslow depth, mean mm (SD)
Mitotic rate, mean mitoses/mm2 (SD)
Regression, No. (%)
Satellitosis, No. (%)
Angiolymphatic invasion, No. (%)
Ulceration, No. (%)
Neurotropism, No. (%)
Sixty-eight patients received high-dose adjuvant interferon therapy (46 lymph node positive, 18 lymph node negative, and 2 with an unknown nodal status). Four recurred while receiving treatment, and therapy was discontinued secondary to severe side effects in 12. Five patients received an experimental vaccine, and 2 underwent biochemotherapy. One hundred eighty-three patients received no interferon treatment. Adjuvant therapy data was not available for 37 patients.
The median follow-up was 43 months. The locoregional recurrence rate was 22%. For patients with a negative SLN, the locoregional recurrence rate was 11%, compared with 34% with a positive SLN. Among the clinically lymph node-negative patients who did not have an SLN biopsy, the locoregional recurrence rate was 21%. Among the 17 cases where the patient and/or physician opted not to perform SLN biopsy not based on age or comorbidities, the locoregional recurrence rate was slightly higher (29%).
The majority of patients were alive without known melanoma recurrence at the time of last follow-up (174 patients, 59%). Eighty-two (28%) patients developed distant recurrence. No data regarding DDFS was available for 10 patients. Seventy-one (24%) patients died of melanoma, 33 (11%) died without disease, 11 (4%) were alive with known recurrence, and died of unknown cause 4 (1%).
The associations between prognostic factors and both OS and DDFS were studied among patients with thick melanomas who underwent successful SLN biopsy. Several factors were associated with OS on univariate analysis, noted in Table 3, including increasing patient age (P = .004), increasing Breslow thickness (P = .006), increasing mitotic rate (P = .004), and the presence of satellitosis (P = .02), angiolymphatic invasion (P > .001), ulceration (P = .002), and at least 1 positive SLN (P < .001). On multivariate analysis, the most significant predictors of OS in patients with thick primary melanomas were increasing patient age (P = .03), increasing Breslow thickness (P = .003), the presence of at least 1 positive SLN (P < .001), and the presence of angiolymphatic invasion (P = .006) (Table 4). Figure 1 demonstrates the Kaplan-Meier OS analysis for patients with thick primary melanoma stratified by the SLN status. The 5-year OS for SLN-positive patients was 47% (95% confidence interval [CI], 35.2-57.8) compared with 80% (95% CI, 70.1-87; P < .0001) in SLN-negative patients.
Table 3. Univariate Associations With Overall Survival, SLN Biopsy Population
We next stratified patients by their SLN status and examined separately which factors were most prognostic for SLN-negative patients and SLN-positive patients (Table 4). On multivariate analysis for SLN-positive patients, both increasing Breslow thickness and the presence of angiolymphatic invasion were significant. In contrast, the most significant predictors of outcome for SLN-negative patients with thick melanoma were increasing patient age, increasing Breslow thickness, and the mitotic rate.
We also examined the impact of these prognostic factors on DDFS. As shown in Table 5, DDFS was significantly associated with location of the primary melanoma (P = .003), increasing mitotic rate (P = .001), increasing Breslow thickness (P = .004), presence of angiolymphatic invasion (P < .0001), presence of ulceration (P = .0002), and SLN positivity (P < .0001). Although patient age was a significant predictor of OS, it was not significant when looking at DDFS. On multivariate analysis, both angiolymphatic invasion (P = .005) and ulceration (P = .009) maintain significance; the most significant predictor of DDFS is SLN positivity, with a hazard ratio of 3.95 for at least 1 positive SLN. Figure 2 demonstrates the Kaplan-Meier DDFS analysis for patients with thick primary melanoma stratified by SLN status. The 5-year DDFS for SLN-positive patients was 47.8% (95% CI, 35.9-58.8), compared with 85.3% in SLN-negative patients (95% CI, 75.1-91.6; P < .0001).
Table 5. Univariate Associations With Distant Failure-Free Survival, SLN Biopsy Population
Again, we performed separate analyses for lymph node-negative and lymph node-positive disease DDFS (Table 6). For SLN-positive patients, only angiolymphatic invasion was a significant predictor of DDFS, with a hazard ratio of 2.29 (P = .007). Ulceration, which was significant when looking at the entire group, was not significant when examining SLN-positive patients. When analyzing only the thick, SLN-negative patients, ulceration was the most significant factor, with a hazard ratio of 5.78 (P = .02). Increasing Breslow thickness and mitotic rate were also significantly associated with poorer outcome. Figure 3 shows the Kaplan-Meier DDFS for SLN-negative patients with thick melanoma stratified by the presence of ulceration, showing the significant impact of ulceration on outcome (log-rank P value of .0034). Figure 4 represents DDFS for SLN-positive patients with thick melanoma, demonstrating the poor survival regardless of ulceration (P = .4).
Table 6. Multivariate Cox Proportional Hazards Model for Distant Failure-Free Survival, SLN Biopsy Population
95% Confidence Interval
Hazard ratio for a 1-unit change in the covariate.
These results represent a large single-institution experience with thick melanoma. SLN biopsy was offered to the majority of clinically lymph node-negative patients and was positive in almost half (47%). In this population, angiolymphatic invasion, satellitosis, and ulceration were predictive of SLN positivity. With a median follow-up of 43 months, patients with a T4 melanoma and a negative SLN had a significantly better 5-year DDFS (85.3% vs 47.8%; P < .0001) and OS (80% vs 47%; P < .0001) compared with those with metastases to the SLN. The SLN status was the most prognostic of outcome on multivariate analysis. In addition to a positive SLN, angiolymphatic invasion and ulceration were also predictive of DDFS, whereas increasing age, Breslow thickness, and angiolymphatic invasion were associated with OS. However, when SLN-positive and SLN-negative patients were analyzed separately, a slightly different picture emerges. Patient age is only a significant predictor of OS in SLN-negative patients, and had no impact on DDFS. Among patients with thick melanomas, angiolymphatic invasion was a significant predictor of both OS and DDFS among SLN-positive patients, whereas mitotic rate was a significant predictor of both OS and DDFS among SLN-negative patients. Among SLN-negative patients with T4 melanoma, ulceration was the most significant predictor of outcome. Despite the Breslow thickness, patients without ulceration or SLN metastases had an extremely good prognosis, with a 5-year OS >90% and a 5-year DDFS of 95%.
The introduction of the SLN biopsy in 1992 by Morton et al has substantially changed the surgical management of cutaneous melanoma.13 Although the potential benefits of SLN biopsy have been demonstrated in the interim results from the Multicenter Selective Lymphadenectomy Trial-I, this was limited to patients with melanoma of between 1.2 mm and 3.5 mm Breslow thickness.1 Controversy remains regarding the appropriate selection of patients with thin melanoma for SLN biopsy and the need for SLN biopsy in patients with thick melanoma. Several authors have examined this latter topic with mixed results.
Before the onset of the SLN biopsy, surgeons debated but generally felt that ELND played little role in patients with clinically lymph node-negative melanoma ≥4 mm in Breslow thickness (T4N0), because these patients were at such high risk for distant disease. Whether SLN status has any prognostic or therapeutic value in this T4N0 subset is unknown. SLN status is often used to make decisions regarding adjuvant therapy, specifically the use of high-dose interferon alfa-2b, which is FDA-approved for high-risk melanoma. Although a positive SLN is the most common indication for adjuvant high-dose interferon alpha-2b, high-risk melanoma is also defined as AJCC stage IIB or C (T4a/bN0M0). Therefore, regardless of SLN status, patients with thick melanomas are already candidates for adjuvant therapy, further arguing against the need for SLN biopsy among these patients.
With regard to the prognostic information obtained from SLN biopsy, previous reports have been mixed (Table 7). Jacobs et al14 found a difference in median survival between patients with a negative versus positive SLN, but this was not statistically significant. Cherpelis et al15 also found a nonsignificant difference in survival. Essner et al16 found SLN status to be a predictor of DFS but not OS. These reports raised concern about the prognostic value of SLN status in thick melanoma.
Table 7. Studies Examining SLN Biopsy in Patients With Thick Melanoma
In contrast, several studies found that SLN status was 1 of the most significant predictors of outcome among patients with thick melanoma. Ferrone et al17 described a risk stratification model for this group of patients using SLN status, age, the presence of ulceration, and primary tumor thickness to categorize patients into low-, moderate-, and high-risk groups, and recommended use of their nomogram rather than the current staging system based on the significant survival difference in a subgroup of patients with T4N0 melanoma previously thought to have a poor prognosis. The data substantiated the importance of SLN biopsy in this patient population, as did a study by Gershenwald et al18 involving 131 patients with T4N0 primary melanomas. They reported SLN status and ulceration as being the most powerful indicator of DFS and OS and recommended routine performance of this procedure for patients with thick primary melanoma for risk stratification. Similarly, the Emory group also found ulceration and SLN status to be the most powerful predictor of DFS, with SLN status retaining its significance in predicting OS.19
Our study represents the largest single-institution study of SLN biopsy in patients with thick T4 melanoma and clearly demonstrates the accuracy of the prognostic information gained. The SLN status was the most significant predictor of both DDFS and OS in our population, and ulceration was highly significant among the SLN-negative patients. This information is not purely academic, but should help guide adjuvant therapy decisions. Although in this study the adjuvant use of high-dose interferon alpha-2b was not significantly associated with either an improved DDFS or OS, the retrospective nature of this analysis limits any conclusions regarding the impact of high-dose interferon alpha-2b. However, the 5-year DDFS and OS of T4 patients who are SLN negative and have no ulceration (95% and 90%, respectively), most of whom received no adjuvant therapy, suggests that there would be little benefit of adjuvant high-dose interferon alpha-2b in this population. If the SLN is negative but ulceration is present, the likelihood of distant recurrence is still significant, and high-dose interferon alpha-2b may play a role.
In conclusion, SLN biopsy should be the standard method not only of staging patients with thick melanoma but also of guiding adjuvant therapy decisions. Assuming adequate surgical therapy, patients with a negative SLN have a good prognosis despite the thick primary tumor, and in the absence of ulceration, they have an extremely low rate of distant recurrence and excellent OS. Adjuvant high-dose interferon is clearly not indicated.