Impact of Sentinel Node Status and Other Risk Factors on the Clinical Outcome of Head and Neck Melanoma Patients

Objective  To determine the impact of sentinel lymph node (SLN) status and other risk factors on recurrence and overall survival in head and neck melanoma patients.

Design  The SLN Working Group, based in San Francisco, Calif, with its 11 member centers, the John Wayne Cancer Institute, and The University of Texas M. D. Anderson Cancer Center pooled data on 629 primary head and neck melanoma patients who had selective sentinel lymphadenectomy. A total of 614 subjects were analyzable. All centers obtained internal review board approval and adhered to the Health Insurance Portability and Accountability Act of 1996 regulations. A Cox proportional hazards model was used to identify factors associated with overall and disease-free survival.

Setting  Tertiary care medical centers.

Main Outcome Measure  Clinical outcome of head and neck melanoma patients undergoing selective sentinel lymphadenectomy.

Results  Overall, 10.1% (n = 62) of the subjects had at least 1 positive node. Subjects with positive SLN status had significantly thicker tumors (mean thickness, 2.8 vs 2.1 mm; P<.001), and were more likely to have ulcerated tumors (P = .004). During the median follow-up of 3.3 years, the overall mortality from head and neck melanoma was 10%, with more than 20% experiencing at least 1 recurrence. Multivariate analysis showed that tumor site was an independent predictor of mortality; location on the scalp had a more than 3-fold (P<.001) greater mortality than tumors on the face. Tumor thickness was also an independent predictor of overall survival, and SLN status was the most important predictor of disease-free survival in the multivariate model (P<.001). Tumors on the scalp had the highest rate of recurrence, while those on the neck had the lowest. Tumor ulceration was the significant predictor of time to recurrence or disease-free survival (P<.001).

Conclusion  In this multicenter study, SLN status and other risk factors have an effect on recurrence and/or overall survival.

During the past 3 decades, the incidence of cutaneous melanoma, including those of the head and neck (H&N), have increased significantly.^1,2 At the same time, the metastatic rate has decreased, probably because of earlier diagnosis and surgical treatment.³ Before selective sentinel lymphadenectomy was widely used,⁴ nodal status from elective lymph node dissection was the most important prognostic predictor of recurrence and death.^5,6 In the sentinel lymph node (SLN) era, SLN status has been shown to be an important prognosticator for recurrence and death.^7,8

Several articles^9-21 have shown the feasibility of harvesting SLNs in H&N melanoma and squamous cell carcinoma. However, the effect of SLN alone and with other high-risk factors in predicting H&N melanoma outcome has not been well-defined. To determine the effect of SLN and other risk factors on melanoma outcome, we established a multicenter database exclusively for H&N melanoma by pooling databases from numerous melanoma centers. The greater number of patients permits more robust statistical analysis and overcomes the bias from single institutional databases. We hypothesize that SLN status and other risk factors influence the clinical outcome of patients with H&N melanoma.

The SLN Working Group was formed in 2003 with a mission to build a robust database for melanoma. It consists of more than 30 members and is based in San Francisco, Calif. The 11 member centers contributing 316 patients in this study obtained approval from their respective institutional review boards and adhered to the Health Insurance Portability and Accountability Act regulations. Two additional centers, the John Wayne Cancer Institute and The University of Texas M. D. Anderson Cancer Center, also participated in this study by contributing data on 227 and 86 patients, respectively, with primary H&N melanoma who underwent selective sentinel lymphadenectomy following their institutional review board approval.

The standard technique of using blue dye, technetium Tc 99m sulfur colloid, or both was adopted by different melanoma centers.²² A total of 614 subjects were analyzable. Fifteen patients were excluded because of missing SLN status. To correlate SLN status and other melanoma risk factors with disease-free and overall survival, we analyzed our data using the Cox proportional hazards model.

Overall, 10.1% (n = 62) of the 614 patients in the database had at least 1 positive node. Table 1 shows the clinical and pathological features. There was no difference in SLN status by age or sex. The mean age of those with a positive node (n = 62) was 56.3 years, and of those with a negative node (n = 550), 58.2 years (P = .40). Patients with positive SLN status (n = 62) had significantly thicker tumors than those with negative SLN status (n = 525) (mean thickness, 2.8 vs 2.1 mm; P<.001), and were more likely to have ulcerated tumors (P = .004). During the median follow-up of 3.3 years, the overall mortality from H&N melanoma was 10%, with more than 20% experiencing at least 1 recurrence.

For disease-free survival, SLN status was the most important predictor in the multivariate model (hazard ratio, 2.8; 95% confidence interval, 1.7-4.6; P<.001) (Figure 1). Tumors on the scalp had the highest rate of recurrence, whereas those on the neck had the lowest. Tumor ulceration was the only other significant predictor of time to recurrence (hazard ratio, 2.3; 95% confidence interval, 1.5-3.7; P<.001) (Table 2).

In a univariate model, mortality was significantly correlated with SLN status (hazard ratio, 2.6; 95% confidence interval, 1.4-4.8; P = .002) (Figure 2). When other factors were considered in a multivariate model, positive SLN status was no longer an independent predictor of mortality. In the multivariate model, tumor site was significantly associated with mortality; tumors on the scalp had a greater than 3-fold increased mortality over tumors on the face (Figure 3). Tumors located on the ear and neck were associated with greater mortality than those on the face, but the difference was not statistically significant. Tumor thickness was also an independent predictor of overall survival in the multivariate model (Table 3).

Although selective sentinel lymphadenectomy for H&N melanoma has been noted to be more challenging than other anatomic sites such as trunk and extremities, numerous studies have shown that selective sentinel lymphadenectomy is a feasible and reliable procedure for the staging of H&N melanoma.^9-16,18-21 To date, to our knowledge, this study is the largest study of H&N melanoma patients from different melanoma centers to determine the effect of SLN status and other risk factors on clinical outcome. As noted in the “Results” section, more than 20% of the patients developed at least 1 recurrence over the median follow-up period of 3.3 years, whereas the overall mortality from H&N melanoma was only 10%, indicating that mortality would probably rise with longer follow-up. The SLN status was the most important predictor in the multivariate model for the disease-free survival. With respect to the overall survival, SLN status shows a significant trend toward overall survival impact (P = .08). Longer follow-up may result in its significance. Other factors to predict recurrence include tumor site, with melanoma of the scalp having the highest rate of recurrence and tumor ulceration. Furthermore, tumor site and thickness significantly predict overall survival. In the pre-SLN era, the biological behavior of H&N melanoma was established to be more aggressive than melanoma at other sites.^5,6,23 A pre-SLN combined study of the H&N melanoma database of 534 clinical stage I patients from the University of Alabama at Birmingham (n = 117) and the Sydney Melanoma Unit (n = 417) found, by multivariate analysis,⁶ that growth patterns, tumor thickness, ulceration, and anatomic subsites were risk factors for local recurrence and survival. In another study⁵ combining the H&N melanoma database of Duke University, Durham, NC, and the Sydney Melanoma Unit, the 10-year survival was 54% when compared with rates of 61% for trunk, 71% for lower extremity, and 76% for upper extremity melanomas (P<.001). Multivariate analysis found that age, ulceration, and American Joint Committee on Cancer stage were significant prognostic factors. Also, the combined databases showed that scalp lesions had significantly poorer prognosis than other skin sites, such as the ear, face, and neck.⁵

The predominance of males with H&N melanoma that we have found in this study is consistent with other studies.^5,24 In our study, patients with positive SLN status had significantly thicker tumors than those with negative SLN status, and were more likely to have ulcerated tumors. For the scalp lesions, the prognosis is significantly poorer than that of other sites of the H&N and of the trunk and extremities (data not shown). This may suggest that H&N melanoma, particularly of the scalp, may have early bloodborne metastasis. Recent breakthroughs in the reverse-transcriptase polymerase chain reaction method to detect functional RNAs in archival paraffin blocks^25,26 may allow us to decipher the molecular signatures of the melanoma lesions from different anatomic sites. If subgroups with different clinical outcomes can be further classified by molecular signatures, then adjuvant therapy can be targeted specifically to a more biologically aggressive subgroup. Thus, the more favorable group can be spared a regimen with significant side effects.

In a multicenter database study of 614 patients with H&N melanoma, SLN status was the most important predictor of recurrence. Tumors on the scalp had the highest mortality, whereas tumors on the face had the lowest mortality. Disease-free survival also differed by tumor location. Tumor ulceration was a significant predictor of disease-free survival but not overall survival, while tumor thickness was a significant predictor of overall survival but not disease-free survival.

Correspondence: Stanley P. L. Leong, MD, Department of Surgery, University of California, San Francisco, 1600 Divisadero, Campus Box 1674, San Francisco, CA 94115 (leongs@surgery.ucsf.edu).

Submitted for Publication: May 2, 2005; accepted July 8, 2005.

Financial Disclosure: None.

Funding/Support: This study was supported in part by an educational grant from Schering Plough Oncology, Kenilworth, NJ.

Role of the Sponsor: The funding body had no role in data extraction and analyses, in the writing of the manuscript, or in the decision to submit the manuscript for publication.

Previous Presentation: This study was presented at the Annual American Head and Neck Society meeting; May 15, 2005; Boca Raton, Fla.

Acknowledgment: We thank our data managers, Marilyn Florero, University of California, San Francisco, and Virginia Stell, Carolinas Medical Center, for their effort to keep an active database. We also thank Jorge Arteaga, University of California, San Francisco, for his effort in the preparation of the manuscript.

Stanley P. L. Leong, MD, University of California, San Francisco; Neil A. Accortt, PhD, and Seng-Jaw Soong, PhD, University of Alabama at Birmingham; Barbara Pockaj, MD, Mayo Clinic, Scottsdale, Ariz; Harald J. Hoekstra, MD, Groningen University Medical Center, Groningen, the Netherlands; Carlos Garberoglio, MD, Loma Linda University, Loma Linda, Calif; Richard L. White, Jr, MD, Carolinas Medical Center, Charlotte, NC; David Chu, MD, City of Hope Medical Center, Duarte, Calif; Merrill Biel, MD, ENT Specialty Care of Minnesota, PA, Minneapolis; Kim Charney, MD, St Joseph Hospital, Orange, Calif; Harold Wanebo, MD, Roger Williams Medical Center, Providence, RI; Eli Avisar, MD, University of Miami, Coral Gables, Fla; John Vetto, MD, Oregon Health and Sciences University, Portland.