Desmoplastic melanoma is an uncommon variant of cutaneous melanoma that mimics soft tissue sarcoma both clinically and morphologically. An activating point mutation of the BRAF oncogene has been identified in a high proportion of conventional cutaneous melanomas, but its frequency in the desmoplastic subtype is not known.
The authors tested 12 desmoplastic melanoma specimens for the thymine (T)→adenine (A) missense mutation at nucleotide 1796 of the BRAF gene using a newly developed assay that employs a novel primer extension method. They also tested 57 vertical growth phase cutaneous nondesmoplastic melanoma specimens.
The 1796 T→A mutation was detected in 23 of the 57 conventional cutaneous melanoma specimens but in none of the 12 desmoplastic melanoma specimens (40% vs. 0%; P = 0.0006, Fisher exact 2-tailed test).
Desmoplastic melanoma is a rare variant of malignant melanoma that differs from conventional melanoma in several respects.1 Compared with other types of cutaneous melanomas, desmoplastic melanomas 1) involve an older group of patients2, 3; 2) are more deeply invasive at the time of diagnosis2; 3) more frequently arise in sun-exposed sites, particularly the skin of the head, neck, and upper back1–3; 4) demonstrate a greater propensity for extension along nerves, resulting in higher rates of local disease recurrence3; and 5) exhibit a tendency to metastasize hematogenously to the lungs and other distant sites, bypassing regional lymph nodes.1 Indeed, this high incidence of local disease recurrence, low incidence of lymph node metastases, and proclivity for distant metastasis to the lungs more closely parallel the behavioral profile of soft tissue sarcoma than malignant melanoma.1
This close parallel with soft tissue sarcoma and divergence from conventional melanoma are maintained at the microscopic level. The microscopic picture is dominated by a vertical growth phase of pleomorphic spindle cells, the radial growth phase is often inconspicuous and sometimes absent, and melanin production is usually not present. By immunohistochemistry, the majority of desmoplastic melanomas do not express specific markers of melanocytic differentiation such as HMB-45, melan A, or microopthalmia transcription factor.4–6 Differentiating desmoplastic melanoma from benign and malignant nerve sheath tumors and other types of mesenchymal lesions can present a daunting diagnostic dilemma.
Limited studies suggest that differences in the clinicopathologic profiles of conventional melanoma and its uncommon variants can be correlated with distinct patterns of genetic alterations.7, 8 With regard to the desmoplastic variant, Gutzmer et al.9 observed a disproportionately high frequency of loss of heterozygosity at the NF1 tumor suppressor gene locus compared with conventional melanomas. Indeed, at this one tumor suppressor gene locus, desmoplastic melanoma appears more closely allied to a nerve sheath tumor than a malignant melanoma.
BRAF encodes a serine-threonine kinase that acts in the mitogen-activated protein kinase (MAPK) pathway.10 Activating BRAF mutations induce constitutive activation of the signal transduction pathway, providing a potent promitogenic force that drives malignant transformation.11, 12 The role of BRAF mutational activation in the development of cutaneous melanocytic tumorigenesis appears to be substantial. BRAF mutations have been identified in ≤ 82% of cutaneous melanocytic nevi13 and in 66% of primary melanomas.12 The majority of these mutations represent a single nucleotide change of thymine (T) → adenine (A) at nucleotide 1796, resulting in a valine-to-glutamic acid change at residue 599. However, to our knowledge little is known regarding the frequency of this V599E mutation in the desmoplastic variant.
MATERIALS AND METHODS
Sample Selection and DNA Extraction
Study approval was obtained from The Johns Hopkins institutional review board. Twelve patients with cutaneous desmoplastic melanoma were identified from a search of the surgical pathology files of The Johns Hopkins Hospital (Baltimore, MD). After initial patient identification, the original histologic slides were reviewed by an experienced dermatopathologist (T.L.B.) for diagnostic confirmation. Desmoplastic melanoma was defined as a vertical growth phase (i.e., invasive) melanoma composed of atypical spindled cells in a collagenous stromal background (Fig. 1). All 12 samples were immunoreactive for S100. As a reference group, we also included 57 conventional malignant melanoma specimens from a variety of cutaneous sites. For both groups of melanomas, an appropriate paraffin block was selected and sectioned, sections were microdissected to obtain > 80% neoplastic cells from the vertical growth phase component, and DNA was extracted using standard protocols as previously published.14
Detection of BRAF Mutations
All tumor specimens were analyzed for the presence of a T→A missense mutation at nucleotide 1796 in the BRAF gene. This hot spot was chosen because the reported BRAF-activating mutations in cutaneous nevi and cutaneous melanoma occur almost exclusively at this position.12, 13 Admittedly, the targeted assay employed in the current study does not permit detection of the small proportion of non-V599E mutations that cluster in and around V599 of exon 15, or of mutations that involve exon 11. Polymerase chain reaction (PCR) primer sequences were designed to amplify a 102-base pair fragment of exon 15 (5′-GAA GAC CTC ACA GTA AAA ATA GGT GA-3′, and 5′- CCA CAA AAT GGA TCC AGA CA-3′). PCR amplification was performed using 100 ng of tumor sample DNA as template. The PCR reactions were performed in a 96-well thermocycler. Cycling conditions were as follows: a denaturation step at 95 °C for 5 minutes was followed by 2 cycles of denaturation at 95 °C for 1 minute, annealing at 60 °C for 1 minute, primer extension at 72 °C for 1 minute, 2 cycles of denaturation at 95 °C for 1 minute, annealing at 58 °C for 1 minute, primer extension at 72 °C for 1 minute, 35 cycles of denaturation at 95 °C for 1 minute, annealing at 56 °C for 1 minute, primer extension at 72 °C for 1 minute, and 1 final extension at 72 °C for 5 minutes. Amplified fragments were separated on an agarose gel and visualized by ethidium bromide staining.
Analysis of the PCR products for a BRAF mutation at nucleotide position 1796 was performed using the Mutector assay (TrimGen, Sparks, MD).15, 16 In brief, the unpurified PCR product was incubated with a proprietary detection primer that is linked covalently to the surface of a reaction well. The detection primer is designed to hybridize with the BRAF antisense strand exactly 1 base 3′ of the expected mutation site. The primer extension reaction buffer includes a mixture of Taq DNA polymerase and nucleotide triphosphates including ddTTP and 1.6 μM biotinylated dATP. Hybridization, 25 cycles of primer extension, and colorimetric detection of biotinylated detection primers were performed under conditions specified by TrimGen. The colorimetric signal was measured with a 405-nm absorbance filter on a Sunrise plate reader (Tecan, Maennedorf, Switzerland) and reported as an optical density. A positive result is indicated when the specimen optic density/negative control optic density was greater than two.16 When we compared BRAF detection for a large number of human tumors in a previous study, we found a 100% correlation between the Mutector assay and direct sequencing.17, 18 As a positive control for the BRAF T1796A mutation, we tested the cutaneous melanoma cell line HTB72. The cervical carcinoma cell line ME180 served as a negative control.
A summary of the clinicopathologic features is shown in Table 1. The mean age of patients with desmoplastic melanomas was 54 years. One-half of these patients were male. Five of the desmoplastic melanomas represented disease recurrences, likely reflecting referral bias. Eleven of the desmoplastic melanomas arose from the skin of the head and neck, and one arose from the skin of the trunk. All demonstrated Level IV or V invasion. The mean tumor thickness was 14.1 mm (range, 4.5–50 mm). The reference group of patients with conventional cutaneous melanomas consisted of 30 males and 27 females. The mean age of this group was 59 years. These nondesmoplastic melanomas arose from the skin of the head and neck (n = 20), trunk (n = 17), extremities (n = 17), and acral regions (n = 3). All were vertical growth phase melanomas with a mean tumor thickness of 3.1 mm (range, 0.5–15.0 mm).
Table 1. Summary of Clinicopathologic Characteristics of the Patients
No. of patients
Mean age (yrs)
Mean depth of invasion (mm)
BRAF mutation (%)
Head and neck
Head and neck
Mutations were detected by a strong color reaction with the Mutector assay as determined visually and confirmed by an optical density reading greater than twice the value of the negative control (Fig. 2). A BRAF mutation was noted in the HTB72 positive control, but a mutation was not detected in the ME180 negative control. BRAF mutations were detected in 23 of the 57 conventional cutaneous melanoma specimens, but in none of the 12 desmoplastic melanoma specimens (40% vs. 0%; P = 0.0006, Fisher exact 2-tailed test) (Table 1). In the nondesmoplastic melanoma specimens, BRAF mutations were more likely to be encountered in melanomas arising from intermittently sun-exposed skin (i.e., on the trunk and extremities) than in melanomas arising from skin with chronic sun exposure (i.e., the head and neck) or low sun exposure (acral regions; 53% vs. 25%; P < 0.05).
Malignant transformation of a melanocyte (or its progenitor) may differentially engage alternative genetic pathways, and pathway selection may account, in large part, for deviations in melanoma phenotype and behavior.7, 8 For example, constitutive activation of the MAPK pathway resulting from mutational activation of the BRAF oncogene is not a consistent route of tumorigenesis across all types of melanoma: BRAF mutations are observed in 29–66% of cutaneous melanomas12, 19, but they are encountered rarely in uveal melanomas20, 21 and mucosal melanomas of the upper respiratory tract.22, 23 The differential activation of BRAF in cutaneous and noncutaneous melanomas has suggested certain epidemiologic variables such as sun exposure as a contributing factor to the variable rates of BRAF mutations.22
Maldonado et al.24 recently observed that the relation between sun exposure and BRAF mutations is complex. BRAF mutations are commonly encountered in melanomas arising from intermittently sun-exposed skin, but they are not commonly encountered at the extremes of sun exposure (i.e., non–sun-exposed and chronically sun-exposed skin). In our group of nondesmoplastic melanomas, we also observed an unequal distribution of BRAF mutations with a much higher mutation rate in melanomas of the trunk and extremities (sites of intermittent sun exposure) relative to the mutation rate in melanomas of the head and neck (sites of chronic sun exposure). Of the 12 desmoplastic melanoma specimens, none harbored an activating mutation of the BRAF oncogene. The absence of BRAF mutations in desmoplastic melanoma—a subtype of melanoma with a strong predilection for chronically sun-exposed skin of the head and neck—seems to underscore this preferential selection of pathways driving melanocytic tumorigenesis depending on the type and duration of genotoxic exposure.
Given its spindle cell morphology, deeply infiltrative growth, inconspicuous radial growth phase, and absence of melanin deposition, distinguishing desmoplastic melanoma from a soft tissue sarcoma can be challenging.25 This diagnostic dilemma is not resolved readily using contemporary panels of immunohistochemical stains.4–6 Detection of tumor-specific genetic alterations is now providing novel ways to solve difficult diagnostic predicaments. As just one example, the detection of a BRAF mutation provides an effective strategy to differentiate the sarcomatoid variant of anaplastic thyroid carcinoma from true sarcomas of the thyroid.26 A novel strategy based on tumor-specific genetic alterations, however, does not appear to be forthcoming when it comes to differentiating desmoplastic melanomas from sarcomas. The absence of BRAF mutations in desmoplastic melanoma mirrors the low frequency of BRAF mutations detected in sarcomas.12
The unequal distribution of BRAF mutations across subtypes of melanomas carries practical therapeutic implications as well. Activation of the MAPK pathway via BRAF mutations provides attractive targets for pathway inhibition whether by antisense oligonucleotides or small molecule inhibitors. Drugs targeting Raf proteins have already entered clinical trials and show promising signs of anticancer efficacy.27–29 The promise of anti-BRAF therapy, however, must be balanced against an awareness that fundamental differences exist in the genetic pathways of melanocytic tumorigenesis. Anti-BRAF therapy may not be uniformly effective across all subtypes of melanoma. Melanoma subtypes need to be viewed individually, not collectively, as novel strategies are designed that target specific genetic alterations.