Abstract
Background
Primary melanoma of the external ear (PMEE) is rare and therefore well-suited for large population-based registry analysis. The objective of this study was to utilize the Surveillance, Epidemiology, and End Results (SEER) set of cancer registries to determine the incidence, treatment, and survival characteristics of PMEE.
Methods
A retrospective cohort analysis of SEER data from 2004 to 2013 identified all cases of PMEE stage I-IV by AJCC 7th edition guidelines. Population-based incidence was calculated. Cancer-specific survival data by stage was assessed using Kaplan-Meier analysis and the relative effects of tumor characteristics were analyzed with Cox regression models.
Results
A total of 5481 patients were analyzed (mean age 66.7 years, 86.5% male, 93.6% non-Hispanic white). The incidence of PMEE was 1.91 per 100,000 persons-per-year. At diagnosis, 68.1% were stage I, 15.2% were stage II, 4.7% were stage III, 1.5% were stage IV, and 10.8% were unknown. The five-year overall and cancer-specific survival was 78.8% and 90.0%, and, according to AJCC stage, was 85.7% and 95.3% for stage I ( n = 2287), 64.6% and 81.1% for stage II ( n = 453), 50.8% and 57.0% for stage III ( n = 154), 17.2% and 20.5% for stage IV ( n = 34), and 71.0% and 87.1% for unknown stage ( n = 330), respectively. The multivariable Cox model identified tumor characteristics that were independently associated with survival.
Conclusions
This is the first study to characterize the epidemiology, presentation and outcome of PMEE using the SEER registries. Older age, increasing Breslow thickness, stage, presence of ulceration, positive lymph nodes and distant metastasis each independently predicted time to cancer-specific death.
1
Introduction
The incidence of cutaneous melanoma continues to rise in the United States with an estimated 76,380 new cases diagnosed in 2016 . Though melanoma accounts for only 1% of all skin cancer cases, it is responsible for the vast majority of skin cancer deaths . Approximately 25% of all cutaneous melanomas involve the head and neck region, and primary melanoma of the external ear (PMEE) accounts for 7–20% of all head and neck melanoma . Overall, PMEE accounts for 1–4% of all cutaneous melanomas .
The literature on PMEE prognosis is replete with controversy. Early series suggested that PMEE carries a worse prognosis compared to other head and neck subsites due to inherently more aggressive tumor biology, the presence of thin skin in an area of high levels of ultraviolet exposure, the variable lymphatic drainage pathways of the ear, and surgical mismanagement of prioritizing aesthetic outcome over sound oncologic treatment . However, more recent series have contradicted the theory that PMEE portends a worse prognosis and have shown that wider margins do not necessarily improve overall survival .
Since PMEE is an uncommon clinical entity, the literature on its management and prognosis is largely limited to relatively small, single institution case series. To circumvent these limitations, the current study utilizes a national population-based tumor registry to better define the incidence and survival characteristics of PMEE.
2
Materials and methods
The Surveillance, Epidemiology and End Results (SEER) 18 database includes a set of 20 population-based tumor registries covering approximately 28% of the United States. For this study, all patients with a first time diagnosis of histopathologically confirmed invasive PMEE recorded in the SEER registries between 2004 and 2013 were analyzed. The year 2004 was chosen since this was the first year for which Collaborative Stage was catalogued in SEER, which includes site-specific factors relevant to melanoma that capture the basis of the information used for the T, N, M and Stage data. Cases of melanoma were identified using International Classification of Diseases for Oncology (ICD O 3) codes for melanoma of external ear (C44.2) with the histology codes of 8720, 8721, 8722, 8730, 8740, 8742, 8743, 8744, 8745, 8761, 8770, 8771, 8772. Staging was performed as per the American Joint Committee on Cancer (AJCC) 7th edition. Select cases between 2004 and 2009 that could not be reclassified according to the AJCC 7th edition system, were classified as per the AJCC 6th edition system. Exclusion criteria included patients with melanoma in-situ (stage 0, n = 5349). For survival analysis we excluded those with subsequent cancer diagnoses ( n = 2216) and those diagnosed on death or autopsy ( n = 7).
Extracted demographic and tumor-specific data included age at diagnosis, sex, race, tumor size, stage, histologic subtype, lymph node status, Breslow depth, Clark’s level, ulceration, serum lactate dehydrogenase (LDH), and primary tumor mitotic rate. Histologic subtypes were grouped into four main categories: “superficial spreading melanoma”, “lentigo maligna melanoma”, “nodular melanoma”, and “other/unknown”. The latter included melanoma not otherwise specified, and subtypes representing a small minority of the population including amelanotic, desmoplastic, or spindle cell melanoma.
Population-based annual incidence for 2004–2013 was calculated. Cancer-specific survival, stratified by stage, was assessed using the Kaplan-Meier method and compared using the log-rank test. Patients were censored at non-cancer death or loss to follow-up. Multivariable Cox proportional hazards models were constructed to examine the relative effects of tumor characteristics on the primary endpoint of “time to cancer specific death.” The first model examined the impact of age, sex, histopathological subtype, Breslow thickness, ulceration (absent, present, or unknown), lymph node status (negative, positive, or unknown) and distant metastasis on time to cancer-specific death. The second model evaluated the impact of age, sex, histopathological subtype, and stage on the primary endpoint. These two models were designed to limit overlapping features. Serum LDH and primary tumor mitotic rate were excluded from the model due to the large amount of missing data for these two variables. The Mayo Clinic Institutional Review Board has deemed analyses of SEER data exempt from review. All analyses were performed using the SAS software package (version 9.4; SAS Institute Inc., Cary, North Carolina), and p values < 00.05 were considered statistically significant.
2
Materials and methods
The Surveillance, Epidemiology and End Results (SEER) 18 database includes a set of 20 population-based tumor registries covering approximately 28% of the United States. For this study, all patients with a first time diagnosis of histopathologically confirmed invasive PMEE recorded in the SEER registries between 2004 and 2013 were analyzed. The year 2004 was chosen since this was the first year for which Collaborative Stage was catalogued in SEER, which includes site-specific factors relevant to melanoma that capture the basis of the information used for the T, N, M and Stage data. Cases of melanoma were identified using International Classification of Diseases for Oncology (ICD O 3) codes for melanoma of external ear (C44.2) with the histology codes of 8720, 8721, 8722, 8730, 8740, 8742, 8743, 8744, 8745, 8761, 8770, 8771, 8772. Staging was performed as per the American Joint Committee on Cancer (AJCC) 7th edition. Select cases between 2004 and 2009 that could not be reclassified according to the AJCC 7th edition system, were classified as per the AJCC 6th edition system. Exclusion criteria included patients with melanoma in-situ (stage 0, n = 5349). For survival analysis we excluded those with subsequent cancer diagnoses ( n = 2216) and those diagnosed on death or autopsy ( n = 7).
Extracted demographic and tumor-specific data included age at diagnosis, sex, race, tumor size, stage, histologic subtype, lymph node status, Breslow depth, Clark’s level, ulceration, serum lactate dehydrogenase (LDH), and primary tumor mitotic rate. Histologic subtypes were grouped into four main categories: “superficial spreading melanoma”, “lentigo maligna melanoma”, “nodular melanoma”, and “other/unknown”. The latter included melanoma not otherwise specified, and subtypes representing a small minority of the population including amelanotic, desmoplastic, or spindle cell melanoma.
Population-based annual incidence for 2004–2013 was calculated. Cancer-specific survival, stratified by stage, was assessed using the Kaplan-Meier method and compared using the log-rank test. Patients were censored at non-cancer death or loss to follow-up. Multivariable Cox proportional hazards models were constructed to examine the relative effects of tumor characteristics on the primary endpoint of “time to cancer specific death.” The first model examined the impact of age, sex, histopathological subtype, Breslow thickness, ulceration (absent, present, or unknown), lymph node status (negative, positive, or unknown) and distant metastasis on time to cancer-specific death. The second model evaluated the impact of age, sex, histopathological subtype, and stage on the primary endpoint. These two models were designed to limit overlapping features. Serum LDH and primary tumor mitotic rate were excluded from the model due to the large amount of missing data for these two variables. The Mayo Clinic Institutional Review Board has deemed analyses of SEER data exempt from review. All analyses were performed using the SAS software package (version 9.4; SAS Institute Inc., Cary, North Carolina), and p values < 00.05 were considered statistically significant.
3
Results
From 2004 to 2013, a total of 5481 cases of PMEE were reported in the SEER database ( Table 1 ). The mean age at diagnosis was 66.7 years (SD 16.5). Men were more commonly affected than women (86.5% vs. 13.5%; p ≤ 0.0001). The most commonly reported ethnicity was non-Hispanic white (93.6%). Reporting sources included hospitals (69.9%), physician offices (15.4%), laboratories (12.7%), radiation treatment centers (0.2%) and other outpatient hospitals (1.7%). The incidence of PMEE gradually rose from 1.66 per 100,000 persons-per-year in 2004 to 2.01 per 100,000 persons-per-year in 2013 ( p = 0.1573). The average incidence of PMEE over the study term was 1.91 per 100,000 persons-per-year.
| Characteristic | Number | % |
|---|---|---|
| Age | ||
| Mean (SD) | 66.7 (16.5) | − |
| Median | 69 | − |
| Q1, Q3 | 57.0, 79.0 | − |
| Range | (4.0–101.0) | − |
| Age group | ||
| 0–49 | 840 | 15.3% |
| 50–59 | 766 | 14.0% |
| 60–69 | 1152 | 21.0% |
| 70–79 | 1380 | 25.2% |
| 80 + | 1343 | 24.5% |
| Gender | ||
| Male | 4741 | 86.5% |
| Female | 740 | 13.5% |
| Race | ||
| Non-Hispanic white | 5128 | 93.6% |
| Hispanic | 147 | 2.7% |
| Non-Hispanic Black | 9 | 0.2% |
| Non-Hispanic Asian or Pacific Islander | 14 | 0.3% |
| Other or Unknown | 183 | 3.3% |
Tumor characteristics for patients with PMEE are reported in Table 2 . At diagnosis, 68.0% were stage I, 15.2% were stage II, 4.7% were stage III, 1.5% were stage IV, and 10.6% were unknown. Histologic subtypes included superficial spreading (20.3%), lentigo maligna (17.4%), nodular (7.7%), and other/non-specified (54.6%) melanoma.
| Characteristic | Number | % |
|---|---|---|
| Tumor size | ||
| < 1 cm | 946 | 17.3% |
| 1–2 cm | 655 | 12.0% |
| 2–3 cm | 175 | 3.2% |
| 3 + cm | 213 | 3.9% |
| Missing | 3492 | 63.7% |
| Stage group | ||
| Stage I | 3730 | 68.0% |
| Stage II | 832 | 15.2% |
| Stage III | 258 | 4.7% |
| Stage IV | 80 | 1.5% |
| Stage unknown | 581 | 10.6% |
| Histologic subtype | ||
| Superficial spreading | 1113 | 20.3% |
| Lentigo maligna | 954 | 17.4% |
| Nodular | 422 | 7.7% |
| Other/NOS | 2992 | 54.6% |
| Breslow thickness | ||
| < 1 mm | 3242 | 59.1% |
| 1–2 mm | 914 | 16.7% |
| 2–3 mm | 404 | 7.4% |
| 3–4 mm | 167 | 3.0% |
| 4–5 mm | 115 | 2.1% |
| 5 + mm | 192 | 3.5% |
| Microinvasion-NOS | 445 | 8.1% |
| Clark level | ||
| II | 1631 | 29.8% |
| III | 1016 | 18.5% |
| IV | 1578 | 28.8% |
| V | 230 | 4.2% |
| Unknown/NOS | 1026 | 18.7% |
| Lymph node status | ||
| Negative | 4831 | 88.1% |
| Positive | 232 | 4.2% |
| Unknown | 418 | 7.6% |
| Distant metastasis | ||
| Negative | 5184 | 94.6% |
| Positive | 80 | 1.4% |
| Unknown | 217 | 4.0% |
| Ulceration | ||
| Absent | 4358 | 79.5% |
| Present | 785 | 14.3% |
| Unknown | 338 | 6.2% |
| Serum lactate dehydrogenase (LDH) | ||
| Within normal limits | 407 | 7.4% |
| < 1.5 upper limit of normal | 71 | 1.3% |
| > 1.5 upper limit of normal | 13 | 0.2% |
| Unknown or not performed | 4990 | 91.0% |
| Primary tumor mitotic count/rate | ||
| < 1 per square mm | 874 | 16.0% |
| Equal to or > 1 per square mm | 919 | 16.8% |
| Missing or not performed | 3688 | 67.3% |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
