By continuing to browse this site you agree to us using cookies as described in About Cookies
Wiley Online Library is migrating to a new platform powered by Atypon, the leading provider of scholarly publishing platforms. The new Wiley Online Library will be migrated over the weekend of February 24 & 25 and will be live on February 26, 2018. For more information, please visit our migration page: http://www.wileyactual.com/WOLMigration/
Lymph lymph node metastasis from melanoma ≤0.50 mm (ultrathin) is an infrequent event. However, because many newly diagnosed melanomas are ultrathin, a significant proportion of patients who present with lymph node disease have an ultrathin melanoma. The authors hypothesized that ultrathin melanomas that present with lymph node metastasis represent biologically aggressive lesions with a worse prognosis.
The Surveillance, Epidemiology, and End Results registry data were queried to identify patients with cutaneous melanoma who presented with lymph node metastasis diagnosed between 1998 and 2008. Hazard ratios (HRs) from Cox proportional hazards regression models were used to compare disease-specific survival (DSS) between various tumor depths.
In total, 6134 patients with lymph node-positive melanoma were identified and stratified according to tumor depth, including 588 (10%) with a tumor depth ≤0.50 mm, 519 (8%) with a tumor depth from 0.51 to 1.00 mm, 1669 (27%) with a tumor depth from 1.01 to 2.00 mm, 1871 (31%) with a tumor depth from 2.01 to 4.00 mm, and 1487 (24%) with a tumor depth >4.00 mm; and the respective 5-year DSS rates were 63%, 76%, 75%, 60%, and 43%. Multivariable analysis confirmed a similar trend in HRs for DSS: The HR was 1.00 for a tumor depth ≤0.50 mm (reference category) and 0.64 (P < .001), 0.65 (P < .001), 0.95 (P = .57), and 1.42 (P < .001) for tumor depths of 0.51 to 1.00 mm, 1.01 to 2.00 mm, 2.01 to 4.00 mm, and >4.00 mm, respectively. This association of tumor depth with DSS persisted for N1 and N2 disease but not for N3 disease.
Thickness of the primary tumor and presence of lymph node metastasis are the 2 most important prognostic factors for cutaneous melanoma.1, 2 Increasing thickness of the primary is directly correlated with an increased risk of lymph node metastasis, and the extent of lymph node metastasis is correlated with survival.3-5 Survival rates for patients with lymph node-positive or stage III melanoma range between 40% and 78%, depending primarily on the number of lymph nodes with metastases.6
Melanomas with a thickness ≤1 mm have a low risk of developing lymph node metastasis.7, 8 However, because 70% of newly diagnosed melanomas are ≤1 mm, in absolute terms, these melanomas account for 25% of melanoma lymph node metastases and 30% of melanoma-related deaths.9-11 The incidence of melanoma is rising and is largely attributable to the rise of melanomas ≤1 mm thick, of which approximately half are ≤0.50 mm thick (ultrathin).7, 9, 11-13 The clinical significance of ultrathin melanomas that present with lymph node metastasis has not been well defined.
A primary ultrathin melanoma that metastasizes to lymph nodes is an uncommon event—we hypothesize that these melanomas represent biologically aggressive disease associated with a poor prognosis. By using a large, population-based tumor registry, we sought to evaluate whether patients who present with lymph node metastasis from ultrathin melanomas have worse survival compared with patients who present with lymph node metastasis from thicker melanomas.
MATERIALS AND METHODS
Surveillance, Epidemiology, and End Results Database
The National Cancer Institute's Surveillance, Epidemiology, and End Results (SEER) tumor registry was used for this study. SEER contains over 3 million cases from 17 geographic sites, covering approximately 26% of the US population. The database is designed to reflect the overall characteristics of the US population and is regarded as a model population-based cancer database. Quality control is an important component of the SEER Program; the current standard for accuracy of data is an error rate <5%.14 Registries that are part of SEER routinely collect data on patient demographics, tumor characteristics, surgery, vital status, and survival. The 2010 update was used for this study, providing information from the years 1973 to 2008. Because this was a population-based study with no patient identifiers involved, our study was exempt from institutional research board review.
Review of the SEER database identified patients who presented with a known melanoma primary and synchronous lymph node metastasis diagnosed from 1998 to 2008. Delayed lymph node metastasis is not captured by the SEER database, therefore, in the current study, we report only on those patients who presented with primary cutaneous melanoma and who, at the same time, were diagnosed with lymph node metastasis: ie, patients with stage III melanoma. Exclusion criteria were age <18 years at diagnosis, unknown lymph node status, unknown lymph-node surgery, stage IV disease, no histologic confirmation of diagnosis, and patient identification from autopsy report. Patient information (age, sex, race), clinicopathologic data (tumor depth, ulceration status, primary surgery site, lymph node category, date of death, cause of death, date of last follow-up) and surgery information (lymph node surgery method) were collected. The American Joint Commission on Cancer staging for cutaneous melanoma was used to classify lymph node category as N1 (1 metastatic lymph node), N2 (2-3 metastatic lymph nodes), or N3 (≥4 metastatic lymph nodes).6 Tumor depth was classified into the 5 following categories of ≤0.50 mm, 0.51 to 1.00 mm, 1.01 to 2.00 mm, 2.01 to 4.00 mm, and >4.00 mm.
Chi-square tests and Kruskal-Wallis rank-sum tests were used to compare patient, tumor, and surgery characteristics between the 5 tumor depth categories in the overall sample. The Kaplan-Meier method was used to estimate disease-specific survival after surgery, censoring at the date of last follow-up in patients who did not die secondary to melanoma. In both the overall sample and separately according to lymph node category, Cox proportional hazards regression models were used to evaluate the associations of tumor depth with disease-specific survival. Hazards ratios (HRs) and 95% confidence intervals (CIs) were estimated. Single-variable models were used in an exploratory analysis, whereas multivariable models were adjusted for the prespecified potentially confounding variables of age, sex, race, primary surgery site, ulceration, lymph node surgery method, and lymph node category (overall sample analysis only). Lymph node category and tumor depth also were evaluated for interaction. To graphically display a potential nonlinear association between tumor depth as a continuous variable and disease-specific survival, we used Cox proportion hazards regression models (in the overall sample and in each lymph node category) using a natural spline transformation with knots at the 25th, 50th, and 75th percentiles of tumor depth; expected 5-year disease-specific survival was then calculated across the range of tumor depths along with point-wise 95% CIs.15 A Bonferroni correction was used to account for multiple testing when testing for differences between lymph node groups, and P values ≤.0125 were considered significant after 4 pair-wise comparisons within a given patient group. Statistical analyses were performed using SAS (version 9.2; SAS Institute Inc., Cary, NC) and R statistical software (version 2.11.0; R Foundation for Statistical Computing, Vienna, Austria).
Our study population consisted of 6944 patients who presented with a known cutaneous melanoma primary and simultaneous lymph node metastasis diagnosed between 1998 and 2008. In total, 810 patients were excluded for any of the following reasons: 84 patients (1.2%) had duplicate entries, 91 (1.3%) were <18 years of age at diagnosis, 423 (6.0%) had unknown lymph node category, 45 (0.6%) had follow-up of 0 months, 24 (0.3%) had primary surgery listed as “none,” and 174 (2.5%) had lymph node surgery listed as “none.” In total, 6134 patients (88.3%) were included in the analysis. Patients who were excluded because of unknown lymph node category (n = 423) were more likely to be older (median age, 74 years vs 56 years; P < .001), to have a thicker melanoma (>2.0 mm: 65% vs 55%; P < .001), to have a primary melanoma of the head and neck (34% vs 16%; P < .001) and were less likely to have ulceration (33% vs 34%; P = .006).
Table 1 summarizes patient, tumor, and surgery characteristics in the overall sample and according to tumor depth. Statistically significant differences were observed for age, sex, race, primary surgery site, ulceration, lymph node category, and lymph node surgery. Clinically significant findings were reported for ulceration and lymph node surgery. Ulceration of the primary was less likely for melanomas 0.51 to 1.00 mm thick (13.8%), equally likely for melanomas 1.01 to 2.00 mm thick (21.8%), and more likely for melanomas 2.01 to 4.00 mm thick (37.5%) and >4.00 mm thick (57.4%) compared with melanomas ≤0.50 mm thick (24.3%). Regional lymph node dissection (RLND) for presumed palpable lymph nodes was less likely for melanoma 0.51 to 1.00 mm thick (49.5%), 1.01 to 2.00 mm thick (46.5%), 2.01 to 4.00 mm thick (47.2%) and >4.00 mm thick (52.8%) compared with melanoma ≤0.50 mm thick (58.3%).
Table 1. Patient Demographic, Clinical, and Surgical Characteristics According to Tumor Depth Category
There were 1772 patients (28.9%) who died because of melanoma. The median follow-up was 2.5 years (range, 0-10.9 years). Kaplan-Meier estimates of disease-specific survival according to tumor depth for all patients and for patients with N1, N2, and N3 disease are displayed in Figure 1A-D, respectively. The 5-year disease specific survival rates for patients who had tumor depths of ≤0.50 mm, 0.51 to 1.00 mm, 1.01 to 2.00 mm, 2.01 to 4.00 mm, and >4.00 mm were 63%, 76%, 60%, and 43%, respectively, for all patients; 68%, 83%, 78%, 66%, and 54%, respectively, for patients with N1 disease; 61%, 67%, 71%, 56%, and 35%, respectively, for patients with N2 disease; and 34%, 34%, 48%, 35%, and 25%, respectively, for patients with N3 disease.
Cox regression analyses of tumor depth with disease-specific survival are displayed in Table 2. There was strong evidence of an overall difference in single-variable analysis for disease-specific survival between tumor depth categories in the overall group of patients and separately in the N1, N2, and N3 groups in single-variable analysis (all P ≤ .003). Multivariable analysis of the overall cohort of patients indicated that disease-specific survival, compared with to the ≤0.50 mm tumor depth group, was better for patients with melanomas 0.51 to 1.00 mm thick (HR, 0.64; 95% CI, 0.50-0.82; P < .001) and 1.01 to 2.00 mm thick (HR: 0.65, 95% CI, 0.54-0.78; P < .001), similar for patients with melanomas 2.01 to 4.00 mm thick (HR, 0.95; 95% CI, 0.80-1.13; P = .57), and worse for patients with melanomas >4.00 mm thick (HR, 1.42; 95% CI, 1.20-1.69; P < .001). Multivariable analysis confirmed similar findings for patients with N1 and N2 disease, but the association for those with N3 disease was no longer significant after adjustment for multiple testing (P = .029). Again, differences in patients with melanomas ≤0.50 mm thick were most pronounced and significant in the N1 and N2 groups, with similar though weaker and nonsignificant trends observed in patients with N3 disease. There was no statistically significant evidence of a formal interaction between lymph node category and tumor depth in relation to disease-specific survival in single-variable (P = .21) or multivariable (P = .30) analyses.
Table 2. Association Between Tumor Depth and Disease-Specific Survival in the Overall Sample and According to Lymph Node Category Stratified by Tumor Deptha
HRs and P values were calculated using Cox proportional hazards regression models and are given in relation to the ≤0.50 mm tumor depth group.
Multivariable models were adjusted for age, sex, race, primary surgery site, ulceration, lymph node surgery method, and lymph node category (overall sample analysis only). P values ≤ .0125 were considered significant after adjustment for multiple testing. Patients who had a primary surgery site of overlapping (n = 2) were excluded from the multivariable analysis.
All patients, N = 6134
Test of overall difference
N1 patients, N = 3846
Test of overall difference
N2 patients, N = 1664
Test of overall difference
N3 patients, N = 624
Test of overall difference
The expected 5-year disease-specific survival rates according to tumor depth as a continuous variable for the overall cohort and for patients with N1, N2, and N3 disease are displayed in Figure 2A-D, respectively. Similar to the results from Cox regression analysis using predefined cut-off points when categorizing tumor depth, a nonlinear relation was observed between 5-year disease-specific survival and tumor depth. Expected 5-year disease-specific survival was better in patients with a tumor depth between 0.5 mm and 2.0 mm and worse in patients with a tumor depth ≤0.50 mm and >2.00 mm.
Surrogates of melanoma biology have become increasingly important in predicting patient outcome. Patients with lymph node-positive melanoma have highly variable survival outcomes, and ulceration of the primary melanoma is currently the only biologic surrogate used to determine prognosis in the setting of stage III disease.6 In the current study, we identify a potential additional biologic marker for poor survival in the setting of lymph node metastasis: an ultrathin primary. We demonstrate that, for patients with melanoma who present with lymph node metastasis, ie, stage III disease, those whose have a primary melanoma thickness ≤0.50 mm have disease-specific survival similar to that of patients who have a primary melanoma thickness of 2.00 to 4.00 mm and have worse disease-specific survival than patients who have a primary melanoma thickness of 0.51 to 2.00 mm. Counterintuitively, increasing tumor thickness is initially associated with improving survival until a certain threshold depth is reached, at which point, further increase in thickness is associated with worsening survival, as expected. The data reflect the interaction between tumor burden and tumor biology. A melanoma >4.00 mm that metastasizes to lymph nodes is an expected event that likely reflects overwhelming tumor burden; conversely, a melanoma ≤0.50 mm that metastasizes to lymph nodes is an unexpected event (<5%) that likely reflects aggressive tumor biology. It is important to note that we report on patients with melanoma who had lymph node metastasis at the time of presentation and not on patients with melanoma who developed lymph node metastasis in a delayed manner.
In a recent study, 66% of all newly diagnosed melanomas were ≤1.00 mm thick, and 52% of those melanomas were ≤0.50 mm thick.7 The outcomes for the patients with these melanomas are excellent, because the risk of lymph node metastasis is very low. Yet, because approximately 33% of all newly diagnosed melanomas are ≤0.50 mm thick, in absolute terms, these ultrathin melanomas account for a significant proportion of patients who are originally diagnosed with lymph node metastasis. In the current study, ultrathin melanomas accounted for approximately 10% of all patients who originally presented with lymph node-positive disease. Patients who had ultrathin melanomas and lymph node metastasis at the time of their original diagnosis were more likely to present with ulceration and to undergo RLND. These patients were treated in the era of the sentinel lymph node biopsy, where a RLND is typically reserved for palpable metastatic lymph nodes. The higher rate of ulceration and palpable metastatic lymph nodes associated with metastatic ultrathin melanomas may reflect aggressive tumor biology.
An ultrathin primary that presents with lymph node metastases represents a low tumor burden state in which a biologic factor likely initiated a metastatic cascade, which led not only to regional metastases but also to the early development of distant metastasis and patient death. The significant association between ultrathin melanomas and poor survival was observed in patients with N1 and N2 disease; and a similar but nonsignificant trend was observed in those with N3 disease. Survival for patients with N3 disease was poor in all tumor depth categories. The weakened trend in survival observed with N3 disease likely reflects the negative prognostic effect of a large tumor burden observed with ≥4 metastatic lymph nodes.
Increasing age, increasing tumor depth, ulceration, and the mitotic rate have been reported as clinicopathologic factors associated with metastasis in the setting of a primary melanoma ≤1.00 mm thick.8 However, a better understanding of which molecular factors promote melanoma metastasis is needed to successfully identify and treat these patients. Recent studies have identified promising mechanisms, such as loss of LKB1 (serine/threonine kinase 11), for melanoma metastasis.16-18 It has been demonstrated that neuroblastoma v-ras oncogene (NRAS) mutation status predicts central nervous system involvement and worse survival.19 Whether NRAS or other biologic factors are responsible for the development of metastasis from an ultrathin melanoma remains to be studied.
The poor prognosis observed among patients with melanoma who present with lymph node metastasis and an ultrathin primary has clinical implications. First, patients with lymph node-positive ultrathin melanoma represent a high-risk group and, thus, may need greater consideration for adjuvant therapy and recruitment into clinical trials. Second, patients with lymph node-positive ultrathin melanomas may require closer follow-up to detect recurrence. Site-specific recurrence is not reported by the SEER registry, so it is unclear whether patients who have lymph node-positive ultrathin melanoma are prone to develop skin, lung, or other visceral metastases. Third, patients with lymph node-positive ultrathin melanomas may need to be risk-stratified in clinical trials to improve trial design and reported outcomes.
This study has several limitations, which must be acknowledged. It was performed with a population-based cohort and was observational in nature. Although our data indicate an association between ultrathin melanoma and worse outcomes in lymph node-positive patients, it does not establish causality and must be viewed as hypothesis-generating. SEER data reportedly have 95% accuracy, yet we cannot exclude the possibility of coding errors. An important limitation of this study is the accuracy of the primary tumor thickness. First, it is possible that a shave biopsy alone was used to incorrectly code primary tumor thickness without consideration of additional melanoma discovered at the time of definitive surgery and that ultrathin melanomas, in fact, are thicker melanomas. Second, additional tumor identified at the base of the biopsy may have been lost secondary to cautery effect. Third, accurate assessment of the primary melanoma thickness may not be assessed in the setting of clinically palpable lymph node disease. Fourth, regression is not captured by the SEER registry; therefore, it is plausible that, in the setting of regression, primary tumor thickness was underestimated. Fifth, the presence of ulceration could result in the superficial removal of tumor and lead to an underestimation of tumor thickness. It is possible that some of these explanations could lead to the underestimation tumor thickness and, thus, could preferentially affect ultrathin melanomas. However, many of these explanations also could systematically affect all tumor depth categories and represent nondifferential information bias.
Another limitation is that we were unable to account for the effect of adjuvant therapy, because adjuvant therapy is not reported by the SEER registry. Interferon-α2a is the only US Food and Drug Administration-approved adjuvant treatment for stage III melanoma. It has demonstrated the ability to improve recurrence-free survival, but not overall survival; therefore, the lack of effective adjuvant therapy for patients with stage III melanoma means that the inability to account for adjuvant therapy is unlikely to have influenced our results.20 Prognostic variables like mitotic rate and microsatellitosis are not reported by the SEER registry; therefore, their influence could not be analyzed in this study. The median follow-up was not long (2.5 years), and it is unclear whether a longer follow-up would have an impact on the results. All of these limitations must be weighed against the finding that the large sample size reported in this study came from diverse locations across the United States, allowing for high power to detect differences in outcomes between tumor depth categories in the overall cohort of patients. This is unlikely to be possible with an institutional data set.
Why patients who had ultrathin melanomas underwent sentinel lymph node biopsy is unclear. There are no uniform guidelines for sentinel lymph node biopsy for clinical stage IA melanoma (melanoma ≤1.00 mm thick without ulceration and with a mitotic rate <1 mitosis per mm2). However, the National Comprehensive Cancer Network (NCCN) does suggest sentinel lymph node biopsy should be considered for clinical stage IA melanoma if adverse features are present, such as tumor depth ≥0.75 mm, positive deep margins, lymphovascular invasion, or Clark level IV. In the current study, 42% of patients who had lymph node-positive melanomas with ultrathin primaries underwent a sentinel lymph node biopsy. A review of the National Cancer Database (NCDB) revealed that 13% of patients with clinical stage IA melanoma underwent sentinel lymph node biopsy and that young age, hospital type, and geographic location were factors associated with sentinel lymph node biopsy in these patients. Whether adverse pathologic features, as suggested by the NCCN, or health system factors, as outlined by the NCDB review, were responsible for the decision to proceed with a sentinel lymph node biopsy for an ultrathin primary in the current study is unclear. It is important to note that 58% of lymph node-positive patients who had ultrathin primaries underwent RLND rather than sentinel lymph node biopsy. This strongly suggests that, for the majority of lymph node-positive patients with ultrathin primaries, regional disease was manifested in the form of palpable lymphadenopathy, which supports our findings of an aggressive underlying biology.
In summary, ultrathin melanomas with lymph node metastasis have a poor prognosis that may be secondary to an aggressive biologic subtype. Primary tumor depth ≤0.50 mm may be used as prognostic marker in stage III melanoma; however, prospective data will likely be needed to confirm these findings. Scrutiny of the molecular characteristics of ultrathin melanomas that present with lymph node metastasis may deepen our understanding of what drives the melanoma metastatic cascade and ultimately may provide biomarkers that identify the subset of melanomas with increased metastatic potential. Further studies are needed to determine whether these findings can help inform decisions about adjuvant treatment, surveillance protocols, and risk stratification in clinical trials.