The patient is an otherwise healthy male transferred from an outside hospital with a newly diagnosed melanoma from an unknown primary presenting as a large, left axillary mass.
Approximately 1 month before presentation to an outside hospital, this 40-year-old gentleman noted a firm, mildly painful mass in the left axilla measuring approximately 3 cm. A brief trial of antibiotics was unsuccessful in shrinking the mass, and he subsequently underwent an excisional biopsy. The pathology from the biopsy was inconclusive, showing a lymph node with hyperplasia. The postoperative course was complicated by significant bleeding, and the mass rapidly returned.
Approximately 3 weeks after the initial biopsy, a second procedure was performed. This time an 8-cm mass was removed, demonstrating a "large cell undifferentiated malignant neoplasm" by outside pathology review. The residual mass continued to grow rapidly, and 1 week after the second operation a computed tomography (CT) scan was obtained, revealing a 5.7 X 6.7 cm left axillary mass with possible invasion into the chest wall. No other sites of metastatic disease were identified. Eight days after the second procedure, a third attempt was made to obtain tissue for diagnosis. At that time, a 10-cm mass was removed and pathologically found to be S100- and MART1-positive and most consistent with malignant melanoma.
The patient's postoperative course was complicated by significant bleeding, requiring numerous transfusions. The mass continued to grow very rapidly and a follow-up CT scan, done 3 weeks after the third surgery, showed the left axillary mass to now measure 9.9 X 9.5 cm (Figure 1A). A positron-emission tomography (PET) scan at that time showed hypermetabolic uptake only in the left axilla (Figure 1B).
The man was treated by his outside physicians with a taxane analog as part of a clinical trial. Despite two cycles of the experimental therapy, the mass continued to grow rapidly. He presented to an outside hospital with acute bleeding from the growing left axillary mass. On physical exam, he was found to be febrile and pale, and the mass now clinically measured 25 X 12 X 30 cm. Laboratory studies showed a white blood cell (WBC) count of 78 X 109/L with predominate neutrophils and a hemoglobin value of 8.8 g/dL. The patient underwent embolization of the bleeding mass in an attempt to control the bleeding. He was subsequently transferred to the University of Colorado Health Sciences Center for further evaluation and care.
On physical exam, the patient was pale but in no acute distress. He was febrile to 38.2°C. His cardiac, pulmonary, and abdominal exams were unremarkable. His skin exam revealed multiple clinically dysplastic-appearing nevi. No lesion suggestive of a primary melanoma was identified. No lymphadenopathy was palpable in the cervical, supraclavicular, or inguinal regions. There was no right axillary adenopathy. The left axilla contained a large (approximately 30 cm) fungating mass with serosanguinous drainage (Figure 2).
Laboratory studies revealed a WBC count of 16.7 X 109/L with predominant neutrophils, a hemoglobin of 7.1 g/dL, elevated lactate dehydrogenase at 853 U/L (normal: 98-192 U/L), albumin of 1.4 g/dL, and normal kidney and liver function tests. Chest x-ray showed no evidence of intrapulmonary disease and a large amorphous soft-tissue density in the left axilla. We were unable to obtain a CT or magnetic resonance imaging (MRI) scan, as the abduction of his arm from the large mass prevented him from fitting into the scanning machine.
Establishing a Diagnosis
Dr. Karl Lewis: Dr. Fitzpatrick, what is the gold standard for establishing the diagnosis of poorly differentiated malignancies that are suspected of being malignant melanoma?
Dr. James Fitzpatrick: At present, there is no gold standard that is universally accepted by all pathologists and dermatopathologists, and different cases require different evaluations. This case illustrates several issues and pitfalls. In most cases of malignant melanoma, the diagnosis is established by correlating the cytologic detail with the architecture (nested growth pattern) and location of the tumor (nests of cells at the dermoepidermal junction) with the production of melanin and appropriate immunohistochemical profile.
In this case, since this involves metastasis to the lymph node, the tumor lacks the expected relationship to the epidermis that is used to support the diagnosis in primary malignant melanomas. The tumor is also composed of alternating sheets of atypical cells and areas of necrosis and lacks the characteristic nested-growth pattern seen in most malignant melanomas (Figure 3). Moreover, the tumor lacks melanin, another important distinguishing feature. In summary, from a purely histologic standpoint, the only features on routine stains that support the diagnosis of malignant melanoma are the large atypical epithelioid cells, and therefore, the diagnosis will be contingent on the results of the special stains and immunoperoxidase profile.
In the preimmunoperoxidase era, the only useful special stains in establishing the diagnosis were melanin stains such as the Fontana-Masson stain and the Warthin-Starry stain. These stains have been largely replaced by immunoperoxidase stains and are probably underutilized. However, they are still useful, as the demonstration of melanin within a tumor is very strong evidence that the tumor is a malignant melanoma, and it is more specific than many of the immunoperoxidase markers. The limitation of these stains is that some malignant melanomas are amelanotic, causing the stains to be negative.
Since the advent of reliable immunoperoxidase stains, the use of an initial screening panel of immunoperoxidase markers (eg, S-100, pancytokeratin, CD45) has become routine in poorly differentiated tumors to determine the cell lineage. This is followed by a second panel of immunoperoxidase markers to narrow the possibilities or confirm the diagnosis. In our experience, the most common mistake made in undifferentiated tumors is to use only a single marker to diagnose malignant melanoma. This approach is fraught with potential errors since malignant melanomas can occasionally express unexpected markers.
In some studies, for example, up to 2% of malignant melanomas can express pancytokeratin, which might suggest an epithelial-derived malignancy. Similarly, not all malignant melanomas express all of the commonly used melanoma markers, and a tumor that previously expressed one marker can lose this expression. The problem is clearly demonstrated in this case, where portions of the tumor strongly express S-100, while other portions of the tumor are negative for this marker (Figure 4).
In summary, poorly differentiated malignancies that are suspicious for malignant melanoma are best diagnosed by a screening panel of immunoperoxidase markers followed by a second panel to confirm the diagnosis. Our policy is that poorly differentiated tumors need to express at least two melanoma immunoperoxidase markers (ie, S-100, KBA-62, HMB-45, or melanin-A) before a firm diagnosis of malignant melanoma is rendered.
Mechanisms Underlying an Occult Primary
Dr. Lewis: What is the incidence of patients presenting with metastatic melanoma without evidence of a primary site?
Dr. Whitney High: Most large series of melanoma have documented a 4% to 10% incidence of metastatic disease with an occult primary lesion.[1-3] Since the 1960s, three mechanisms have been postulated to explain such cases, specifically: (1) the "metastatic" disease actually represents malignant transformation of ectopic nests of melanocytes within the involved tissue, (2) the primary lesion, often cutaneous, has undergone complete regression, or (3) the primary lesion, while still present, falls below the threshold for detection.
Imaging technology has greatly expanded over the ensuing years. CT, MRI, and PET scanning have resulted in improved detection, decreasing support for the last explanation. Coincidentally, recognition of ocular melanoma has increased and has shed light on some otherwise occult primary lesions. Of the remaining two possibilities, the concept of a completely regressed primary melanoma is particularly intriguing.
The incidence of histopathologic regression within melanoma is sixfold higher than that of many other cancers. Partial regression is recognized in 10% to 35% of all cutaneous melanoma specimens. Completely regressed cutaneous melanoma, while not extensively reported, is well recognized within the literature.
In the largest meta-analysis to date, we examined the world literature and discovered 38 well-documented cases of completely regressed cutaneous melanoma presenting with nodal or visceral metastases. The majority of such patients reported a change in a preexisting nevus, such as enlargement, friability, bleeding, and eventual regression. These changes occurred 2 months to 14 years prior to the diagnosis of metastatic disease. Careful examination of the body surface identified macules, patches, papules, and plaques, both hypopigmented and hyperpigmented. Histologic examination has reliably detected attenuation of the epidermis, decreased epidermal melanocytes, papillary dermal fibrosis, a chronic inflammatory infiltrate, and increased telangiectasias.
It is likely that complete regression may explain a significant number of cases of metastatic melanoma with an occult primary lesion. Accordingly, in all such cases we advise: (1) a complete and thorough skin examination, (2) a complete ocular examination, (3) a biopsy of any lesion that may represent possible regression with examination by a dermatopathologist familiar with the corresponding histologic features, (3) a thorough history to identify any pigmented lesions undergoing past change, (4) reexamination of tissue from any prior removal of a pigmented lesion, even if the original diagnosis was thought to be benign.