* Objective.-Primary mucosal melanomas are rare tumors. We compare melanomas arising in 2 histologically different mucosa, the stratified oral squamous mucosa and pseudostratified sinonasal respiratory mucosa, to investigate the clinicopathologic influence of native mucosal histology on the tumor.
Methods.-Clinicopathologic features of 36 melanomas arising in the squamous mucosa of the oral cavity were compared with 59 melanomas arising in the sinonasal respiratory mucosa.
Results.-The median age of patients was 61 and 63 years for oral and sinonasal melanomas, respectively, with the squamous and respiratory mucosa covering the maxilla being most frequently involved (68.7% and 66%, respectively). The former had a remarkable male predilection (28 men, 8 women), while the latter affected both sexes equally (29 men, 30 women). The oral melanomas were more likely to be detected in the early in situ or microinvasive stage (4 cases vs none, P = .008) and were more frequently amelanotic (14 vs 12, P = .049) than sinonasal melanomas. The sinonasal melanomas were frequently thicker (median thickness, 9 vs 2.6 mm), polypoid (29 vs none), ulcerated (57 vs 20), and necrotic (57 vs 14) than oral melanoma (P
Conclusion.-Sinonasal melanomas demonstrated aggressive morphologic features significantly more frequently than oral melanomas; however, prognosis remained similar in both groups.
The vast majority (>90%) of melanomas in the head and neck are cutaneous, followed in decreasing order of frequency by ocular and mucosal melanoma (MM). Head and neck MMs constitute 0.7% of malignant melanomas arising in all sites and involve (in decreasing order of frequency) the sinonasal cavity, oral cavity, pharynx, larynx, and upper esophagus.1,2 Oral MMs constitute 0.5% of all oral malignancies, whereas sinonasal MMs constitute less than 7% of all sinonasal neoplasms.3 Their etiology in UV light-protected sites is not fully understood. The role of inhaled and ingested carcinogens in their pathogenesis has been suggested, similar to squamous cell carcinoma,4 but no precursor lesions (eg, melanocytic dysplasia) have been described for primary MM. The relative inaccessibility of the mucosa to self-examination often delays diagnosis, resulting in late detection and poor survival. At presentation, approximately 13% to 19% of patients have lymph node metastases, and another 16% to 20% are likely to develop metastases subsequently. The 5-year survival rate reportedly varies from 0% to 55%.3,5-9
Although the squamous and ciliated columnar respiratory mucosa of the upper aerodigestive tract are both derived from ectoderm, the former is a stratified epithelium, similar to epidermis, whereas the latter is pseudostratified. To investigate whether histologic and microanatomic characteristics of native integument are associated with site-specific clinicopathologic differences in primary MM, we performed a comparative clinicopathologic analysis of 95 primary head and neck MMs.
MATERIALS AND METHODS
Ninety-five cases of primary MM of head and neck (Figure 1, A and B) for which sufficient pathologic material was available for review were identified from the archives (1956-1999) of the Department of Pathology at Memorial Sloan-Kettering Cancer Center, New York, NY. The diagnosis of melanoma was previously established in 62 patients by immunohistochemistry for melanocytic differentiation markers, S100 protein, HMB-45, melan-A, tyrosinase, and microphthalmia transcription factor and by presence of melanin pigment.10 The primary nature of the melanomas was supported by review of the patients' charts to rule out mucosal metastasis in an advanced-stage disseminated cutaneous melanoma and by the presence of melanoma in situ in the adjacent mucosa seen in 31 melanomas arising in squamous mucosa (SQMM) and 42 melanomas arising in respiratory mucosa (RMM) (Figure 1, C and D). The histologic slides from 89 initial tumors, 4 recurrent mucosal tumors, and 2 metastatic deposits of primary MM to lymph node were reviewed. Tumors localized to the mucosa were considered stage I, whereas tumors with regional lymph node involvement and with distant metastasis were considered stages II and III, respectively.8,9
Tumor size and gross appearance were obtained from the pathology report or from clinical examination, as noted in patients' medical charts. The tumor thickness was measured with an ocular micrometer from the most superficial layer of mucosal epithelium, ulcer base, or granular layer of squamous mucosa, to the deepest invasive tumor cell. Presence of in situ melanoma and depth of invasion were noted. Microinvasion was defined similar to cutaneous melanoma as single or clusters of no more than 10 atypical cells present in the lamina propria within less than 0.3 mm of the basement membrane showing no mitoses.9 The tumor cell morphology, architecture, presence of ulceration, necrosis, melanin pigment, and vascular and perineural invasion were noted. Statistical analysis using 2-sided [chi]^sup 2^ analysis and the Fisher exact test was done to compare SQMM and RMM with [alpha] level set at
Malignant melanoma originated in the squamous mucosa in 36 (SQMM) and respiratory mucosa in 59 (RMM) patients. The race distribution (known in only 67 patients) was as follows: 55 white (82%), 6 Asian (9%), 3 Hispanic (4.5%), and 3 African American (4.5%) patients. Malignant melanoma developed in the radiation field in 3 patients (4.5 years after radiotherapy for squamous cell carcinoma of the tongue in 1 patient and 40 years after radiotherapy for acne in 2 patients).
The median age of the patients was 61 years (range, 23-88 years) for SQMM and 63 years (range, 34-93 years) for RMM. The SQMM group included 28 men and 8 women, and the RMM group included 29 men and 30 women. The predominant site of tumors was as follows: nasal cavity, 27; maxillary antrum, 12; sinonasal, 20; upper alveolus, 15; hard palate, 9; lower alveolus, 3; lip, 5; buccal mucosa, 2; and floor of mouth, 1. The most frequent presentation was nasal obstruction in 27 patients, followed by epistaxis in 15, oral mucosal pigmentation in 15, and squamous mucosal ulcer and mass in 7 patients each. The duration of symptoms ranged from 15 days to 4 years in the SQMM group and up to 11 years in the RMM group. The tumor stage at presentation was stage I in 72 patients, stage II in 12, stage III in 5, and unknown in 6 (Table 1).
The gross appearance was known in all 36 SQMM and 55 RMM cases. Most SQMMs were flat, macular, pigmented, or erythematous mucosal lesions (Figure 1, A). Six tumors presented as mucosal masses. In contrast, more than half of the RMM cases (n = 29) were exophytic polypoid masses, 5 were flat pigmented lesions with or without ulceration, and the remaining (n = 21) were endophytic mucosal nodules (Figure 1, B; Table 2). The rumor size was available in 24 SQMMs and varied from 0.15 mm to 32 mm, whereas size ranged from 15 to 70 mm in the 23 RMMs for which size was known. Extensive tumors involving multiple paranasal sinuses were seen in 18 patients, while only 3 tumors extensively involved the oral mucosa. Two tumors involved both the squamous and respiratory mucosa by directly extending through the hard palate into the adjacent nasal cavity and maxillary antrum.
Mucosal melanomas were composed of either predominantly epithelioid or spindle cells or showed a mixed morphology, including multinucleated tumor giant cells (Figure 1, C through F). Five SQMMs and 24 RMMs contained an undifferentiated small cell or poorly differentiated multinucleated giant cell component (Table 2, Figure 2). Progression from melanotic epithelioid and spindle cell morphology to amelanotic undifferentiated small round cell neoplasm over 4 years was seen in 1 SQMM. A pseudopapillary growth pattern was seen exclusively in RMM and consisted of multiple layers of undifferentiated cells around fibrovascular cores amidst extensive necrosis (Figure 2, B). The architecture of the spindle and mixed tumors resembled sarcoma in 6 SQMM and 8 RMM cases, most frequently malignant fibrous histiocytoma, fibrosarcoma, malignant peripheral nerve sheath tumor, and epithelioid angiosarcoma in an RMM. Six SQMMs and 1 RMM were desmoplastic. Fourteen SQMMs and 12 RMMs were amelanotic.
The level of invasion was known in 29 SQMMs and 55 RMMs. The SQMM was pure melanoma in situ without any invasion in 2 patients, one of which was a lentigo maligna melanoma involving the mucosa and adjacent skin at the vermilion border of the upper lip arising in a background of pronounced solar damage. The other tumor involved the hard palate and buccal mucosa extensively. Two patients had melanoma in situ with microinvasion extensively involving the oral mucosa. In contrast, none of the RMMs were pure in situ or microinvasive, and most (60%) showed invasion of deep tissue, such as skeletal muscle, bone, or cartilage. The lamina propria was involved in 12 SQMMs and 22 RMMs. Tumor thickness could be measured in 23 SQMMs and 55 RMMs and was up to 7.9 mm (median 2.6 mm) in the former and up to 20 mm in the latter. Mucosal ulceration was noted in 20 SQMMs and 57 RMMs. Tumor necrosis was noted in 14 SQMMs and 57 RMMs. Vascular invasion was seen in 8 SQMMs and 22 RMMs, and perineural invasion was seen in 11 tumors in each group (Table 2).
Management and Outcome
Follow-up was available in 33 SQMM and 55 RMM patients (range, 1 month to 19.5 years; median, 1.5 years). Surgical resection formed the crux of therapy in 82 patients (32 SQMMs, 50 RMMs) with cervical lymph node dissection in 19 (11 SQMMs, 8 RMMs). Six patients received definitive and 13 patients received adjuvant radio-therapy. Six patients received adjuvant chemotherapy including dimethyltriazenoimidazolecarboxamide (DTIC), nimustine, and vincristine, and 5 were treated with adjuvant immunotherapy consisting of Bacillus Calmette-Guerin vaccine or [alpha]-interferon.
Local mucosal recurrence occurred in 17 of 28 SQMMs and 31 of 46 RMMs 1 month to 8 years after surgery. Distant metastases were noted in 18 of 29 tumors in the SQMM group and in 17 of 32 in the RMM group and involved (in order of frequency) lung, brain, soft tissue, gastrointestinal tract, bone, and skin. At the conclusion of this study, 20 patients with SQMM and 32 with RMM died with disease. Five SQMM and 8 RMM patients were alive and free of disease. Five SQMM and 6 RMM patients were alive with disease. Three SQMM and 9 RMM patients died of other causes. The overall median survival was 2.6 and 1.7 years, and the disease-specific median survival was 3.0 and 2.8 years (P = .74) for SQMM and RMM, respectively (Figure 3). The overall survival was 49% (95% confidence interval [CI], 29%-66%) at 3 years and 33% (95% CI, 16%51%) at 5 years, and the disease-specific survival was 52% (95% CI, 31%-70%) at 3 years and 35% (95% CI, 17%-54%) at 5 years for SQMM patients. For RMM patients, the 3-year overall and disease-specific survival rates were 40% (95% CI, 26%-54%) and 50% (95% CI, 33%-64%), respectively, and the 5-year overall and disease-specific survival rates were 28% (95% CI, 16%-42%) and 37% (95% CI, 22%-53%), respectively (Table 1).
We hypothesized that there would be no site-specific clinicopathologic differences in primary MMs arising in the squamous and respiratory mucosa of the head and neck since both mucosa were ectodermally derived, UV light protected, most likely exposed to similar inhaled and ingested carcinogens, and since tumors in both sites shared the cell of origin, that is, the neuroectodermally derived mucosal melanocyte. However, several significant morphologic and some subtle clinical differences were noted between SQMM and RMM.
Most melanomas arising in the oral squamous mucosa were flat and relatively thin lesions, in contrast to the melanomas of the respiratory mucosa, which were large and polypoid. The RMMs had a median thickness of 9 mm, more than 3 times the median thickness of SQMMs, which was 2.6 mm. This difference may be partly explained by the oral hygiene habits of most people (eg, brushing, flossing, and dental visits), which may lead to frequent examination of the oral cavity and earlier detection of oral as compared to sinonasal tumors. In contrast to oral melanomas, none of the sinonasal melanomas were detected in early stage, and most (60%) were found infiltrating skeletal muscle, cartilage, or bone at the time of surgery. Exophytic polypoid MMs were limited to the upper respiratory tract. This finding was in keeping with other polypoid tumors arising in this location, for example, embryonal rhabdomyosarcoma, esthesioneuroblastoma, and squamous cell carcinoma, among others. The polypoid appearance of these tumors may be due to the loose and richly vascular subepithelial connective tissue in the respiratory mucosa and appears to be responsible for the significantly high incidence of ulceration and necrosis in RMM compared to SQMM.
There were significant differences in tumor cell morphology and growth pattern in the 2 groups of MM. Although the tumor cells displayed a morphologic spectrum, the undifferentiated small round cells were seen predominantly in RMM. Franquemont and Mills6 noted the small round cell morphology in 8 of 14 sinonasal MMs in their series, highlighting an important diagnostic pitfall. The differential diagnosis in tumors composed predominantly of undifferentiated cells includes non-Hodgkin lymphoma, Ewing sarcoma/peripheral neuroectodermal tumor, esthesioneuroblastoma, and alveolar rhabdomyosarcoma. The demonstration of S100 protein expression and melanocyte differentiation markers by immunohistochemistry with HMB-45, A103, tyrosinase, and the nuclear fusion protein microphthalmia transcription factor are very helpful, as is ultrastructural demonstration of premelanosomes.10,11-13 Melanoma in situ, which is seen frequently in MM, is an important diagnostic clue. Olfactory neuroblastoma may have a similar morphology and occasionally demonstrates an in situ component. Unlike melanoma, the tumor cells in olfactory neuroblastoma do not express S100 protein and vimentin, and may show at least focal sustentacular cell pattern even in high-grade (Hyam grade III-IV) tumors lacking neurofibrillary background and Homer-Wright rosettes.
Awareness of the morphologic diversity in MM is important for diagnosis, which may be further compounded by frequent lack of melanin pigment, especially in oral melanoma. Predominantly epithelioid cell melanoma needs to be distinguished from undifferentiated carcinomas, such as sinonasal undifferentiated carcinoma, nasopharyngeal carcinoma, poorly differentiated squamous cell carcinoma, respiratory cell carcinoma or small cell neuroendocrine carcinoma, and anaplastic large cell lymphoma.14 Epithelial malignancies demonstrate fairly strong and diffuse cytokeratin expression and fail to express S100 protein. Anaplastic large cell lymphomas usually express CD30 and anaplastic large cell lymphoma kinase proteins. Melanomas composed of spindle cells raise the possibility of spindle cell carcinoma and sarcoma, for example, malignant fibrous histiocytoma, malignant peripheral nerve sheath tumor, synovial sarcoma, and when the tumor cells are bland, fibrosarcoma. Diffuse expression of S100 protein in a malignant spindle cell neoplasm favors melanoma. Although malignant peripheral nerve sheath tumors may express S100 protein, the reaction is usually focal. Furthermore, malignant peripheral nerve sheath tumors do not express other melanocytic differentiation markers.10
A pseudopapillary growth pattern was seen in almost a quarter of RMMs and none of the SQMMs, usually associated with extensive necrosis and small round undifferentiated cells. Another peculiarity between the 2 MMs was the higher frequency of desmoplastic MMs in the SQMMs. Thus, we found significant morphologic differences between SQMM and RMM. Respiratory mucosal melanomas demonstrated several pathologic features that are well-established negative prognostic predictors in cutaneous melanoma. We have shown that undifferentiated tumor cells, pseudopapillary and sarcomatoid growth pattern, deep invasion, necrosis, and vascular invasion are significantly associated with poor survival in head and neck melanoma.8-10,15,16 These features would suggest a poorer clinical outcome for RMM as compared to SQMM, since several were more frequent in the RMM group. However, the overall and disease-specific survival rates were similar in RMM and SQMM. The incidence of local recurrence and metastasis was comparable in the 2 MMs. This finding is similar to the experience of Guzzo et al17 from the Istituto Nazionale Tumori in Italy. The incidence may be partially explained by the earlier clinical stage at presentation in RMMs. Most (85%) RMMs presented in stage I, despite their large size, and fewer of these tumors presented with lymph node metastases as compared to SQMM. Moreover, we found no significant difference in vascular invasion, an important predictor of outcome.9 Thus, RMMs did not behave more aggressively, as indicated by the incidence of recurrence, metastasis, or death due to disease.
Primary MM may occur anytime after the age of 30 years. The median age of 61 years with a peak in the seventh decade, as well as the gender distribution, correlates well with results of other studies and is older than that for cutaneous melanomas.18-20 A striking male predilection was noted in SQMM, with men being affected 3 and a half times more frequently than women, in contrast to RMM, which affected both sexes equally, similar to previous observations.1,2,5,6,18,21-24 A predilection for mucosa associated with maxilla was noted in both groups of tumors. An explanation may be that during early embryogenesis, the oral and nasal cavities are confluent, lined by a common ectodermally derived epithelium. The palate develops as 1 median, and 2 lateral palatine processes from the maxilla fuse to separate the nasal from the oral cavity. This intimate sharing of epithelium in the upper oral and inferior sinonasal regions during embryogenesis may predispose to similar risk of developing melanoma in the mucosa related to the maxilla.
In conclusion, this study demonstrates significant morphologic differences between MMs arising in the oropharyngeal stratified squamous and sinonasal pseudostratified, ciliated columnar respiratory mucosa of the head and neck. The oral melanomas are more likely to be flat and thin, amelanotic, and may be detected at a noninvasive or early invasive stage. They are also more frequently desmoplastic and neuroinvasive than melanomas arising in respiratory mucosa. The latter are usually pigmented, large, thick, polypoid, ulcerated, necrotic, and deeply infiltrative tumors with a slightly higher tendency to infiltrate blood vessels. They frequently contain an undifferentiated, small, round, blue cell component and display pseudopapillary architecture. However, the ultimate clinical outcome was slightly but not significantly worse in sinonasal melanomas. Thus, although the native microenvironment significantly influences the morphology of primary MMs, the clinical behavior of SQMMs and RMMs appears to be similar.
The authors thank James Woodruff, MD, and Victor Reuter, MD, pathologists at Memorial Sloan-Kettering Cancer Center, New York, NY, for contributing several cases with follow-up information.
1. Chang AE, Karnell LH, Menck HR. The National Cancer Data Base report on cutaneous and noncutaneous melanoma: a summary of 84,836 cases from the past decade. Cancer. 1998;83:1664-1678.
2. Nandapalan V, Roland NJ, Helliwell TR, Williams EM, Hamilton JW, Jones AS. Mucosal melanoma of the head and neck. Clin Otolaryngol. 1998;23:107-116.
3. Matias C, Corde J, Soares J. Primary MM of the nasal cavity. J Surg Oncol. 1988;39:29-32.
4. Holmstrom M, Lund VJ. Malignant melanoma of the nasal cavity after occupational exposure to formaldehyde. Br J Ind Med. 1991;48:9-11.
5. Chiu NT, Weinstock MA. Melanoma of the oronasal mucosa: population based analysis of occurrence and mortality. Arch Otolaryngol Head Neck Surg. 1996;122:985-988.
6. Franquemont DW, Mills SE. Sinonasal malignant melanoma: a clinicopathologic and immunohistochemical study of 14 cases. Am J Clin Pathol. 1991;96: 689-697.
7. Freedman HM, DeSanto LW, Devine KD, Weiland LH. MM of the nasal cavity and paranasal sinuses. Arch Otolaryngol. 1973;97:322-325.
8. Patel SG, Prasad ML, Estrig M, et al. Primary mucosal malignant melanoma of the head and neck. Head Neck. 2002;24:247-257.
9. Prasad ML, Patel S, Hoshaw-Woodard S, et al. Prognostic factors for malignant melanoma of the squamous mucosa of the head and neck. Am J Surg Pathol. 2002;26:883-892.
10. Prasad ML, Jungbluth AA, Iversen K, Huvos AG, Busam KJ. Expression of melanocytic differentiation markers in malignant melanomas of the oral and sinonasal mucosa. Am J Surg Pathol. 2001;25:782-787.
11. Busam KJ, Jungbluth AA. Melan-A: a new melanocytic differentiation marker. Adv Anat Pathol. 1999;6:12-18.
12. Jungbluth AA, Iversen K, Coplan K, et al. T311: an anti-tyrosinase monoclonal antibody for the detection of melanocytic lesions in paraffin tissues. Pathol Res Pract. 2000;196:459-165.
13. King R, Weilbaecher KN, McGill G, Cooley E, Mihm M, Fisher D. Microphthalmia transcription factor: a sensitive and specific melanocyte marker for melanoma diagnosis. Am J Pathol. 1999;155:731-738.
14. Auerbach MJ, Adair CF, Kardon D, et al. Respiratory epithelial carcinoma: a clinicopathologic study. Mod Pathol. 2002;15:215A.
15. Shah JP, Huvos AG, Strong EW. Mucosal melanomas of the head and neck. Am J Surg. 1977;134:531-535.
16. Prasad ML, Patel SG, Hoshaw-Woodard S, et al. Prognostic factors for primary mucosal malignant melanoma of the head and neck. Am J Clin Pathol. 2001;116:604.
17. Guzzo M, Grandi C, Licitra L, Podrecca S, Cascinelli N, Molinari R. Mucosal malignant melanoma of head and neck: forty-eight cases treated at Istituto NazionaleTumori of Milan. Eur J Surg Oncol. 1993;19:316-319.
18. Barker BF, Carpenter WM, Daniels TE, et al. Oral mucosal melanomas: the WESTOP Banff workshop proceedings. Oral Surg Oral Med Oral Pathol Oral Radiol Endod. 1997;83:672-679.
19. Chaudhry AP, Hampel A, Gorlin RJ. Primary malignant melanoma of the oral cavity: a review of 105 cases. Cancer. 1958;11:923-928.
20. Takagi M, Ishikawa G, Mori W. Primary malignant melanoma of the oral cavity in Japan with special reference to mucosal melanosis. Cancer. 1974;34: 358-370.
21. Ostman J, Anneroth G, Gustaffson H, Tavelin B. Malignant oral tumors in Sweden 1960-1989: an epidemiological study. Oral Oncol Eur J Cancer. 1995; 31B:106-112.
22. Manolidis S, Donald PJ. Malignant mucosal melanoma of the head and neck: review of literature and report of 14 cases. Cancer. 1997;80:1373-1386.
23. Lopez-Graniel CM, Ochoa-Carrillo FJ, Meneses-Garcia A. Malignant melanoma of the oral cavity: diagnosis and treatment experience in a Mexican population. Oral Oncol. 1999;35:425-430.
24. Hicks MJ, Flaitz CM. Oral mucosal melanoma: epidemiology and pathobiology. Oral Oncol. 2000;36:152-169.
Manju L Prasad, MD; Klaus J. Busam, MD; Snehal C. Patel, MD; Stacy Hoshaw-Woodard, PhD; Jatin P. Shah, MD; Andrew G. Huvos, MD
Accepted for publication April 1, 2003.
From the Departments of Pathology (Dr Prasad) and Biostatistics (Dr Hoshaw-Woodard), Ohio State University Medical Center, Columbus; and the Department of Pathology (Drs Prasad, Busam, and Huvos) and the Head and Neck Service (Drs Patel and Shah), Memorial Sloan-Kettering Cancer Center, New York, NY.
Presented as an abstract at the 90th Annual Meeting of the United States and Canadian Academy of Pathology, Atlanta, Ga, March 2001.
Reprints: Manju L. Prasad, MD, Department of Pathology, Ohio State University Medical Center, E 418 Doan Hall, 410 W 10 Ave, Columbus, OH 43210 (e-mail: firstname.lastname@example.org).
Copyright College of American Pathologists Aug 2003
Provided by ProQuest Information and Learning Company. All rights Reserved