Mutations driving melanoma growth have diagnostic, prognostic, and therapeutic implications. Traditional classification systems do not correlate optimally with underlying melanoma growth–promoting mutations. Our objective was to determine whether unique dermoscopic growth patterns directly correlate with driving mutations.
We evaluated common driving mutations in 4 different dermoscopic patterns (rhomboidal, negative pigmented network, polygonal, and dark homogeneous streaks) of primary cutaneous melanomas; 3 melanomas per pattern were tested. Three of the 4 patterns lacked common mutations in BRAF, NRAS, KIT, GNAQ, and HRAS. One pattern, the dark homogeneous streaks pattern, had unique KIT mutations in the second catalytic domain of KIT in exon 17 for all 3 samples tested. Two tumors with the dark homogeneous streaks pattern turned out to be different primary melanomas from the same patient and had different sequence mutations but had an impact on the same KIT domain.
Conclusions and Relevance
While future study is required, these results have multiple implications. (1) The underlying melanoma-driving mutations may give rise to specific dermoscopic growth patterns, (2) BRAF/NRAS mutations in early melanomas may not be as common as previously thought, and (3) patients may be predisposed to developing specific driving mutations giving rise to melanomas or nevi of similar growth patterns.
Primary cutaneous melanomas are typically categorized by clinicohistopathologic features, namely, superficial spreading melanoma (SSM), lentigo maligna melanoma, acral lentiginous melanoma, and nodular melanoma (NM).They have also been subcategorized based on the degree of sun damage: chronic sun damage, intermittent sun damage, or nonchronic sun damage.1 Molecular studies have revealed BRAF, NRAS, and KIT are the common and major growth-promoting mutated genes involved in melanoma pathogenesis.1- 5HRAS and GNAQ mutations have been reported in cutaneous melanoma, but they are much less frequent.5,6 Several publications have shown that BRAF and NRAS mutations are present in up to 50% and 20%, respectively, of cutaneous melanomas.2,3BRAF and NRAS are more prevalent in the vertical growth phase and in fast-growing cutaneous melanomas, especially NRAS.7,8 These mutated genes can be partially correlated with the current clinicohistopathologic classification system in which BRAF mutation is more likely to be present in SSM, and NRAS in NM.2,3BRAF and KIT mutations have also been shown to predominate in melanomas arising in intermittently sun damaged and chronically sun damaged skin, respectively.1,4 While there is a correlation of the driving mutation with the current classification scheme, the correlation overall is modest, and for a considerable fraction of melanomas, the growth-promoting mutation is unknown.
Growth-promoting mutations activate the mitogen-activated protein kinase/extracellular signal-regulated kinase (MAPK/ERK) pathway, which promotes DNA replication, cell cycle progression, and, ultimately, tumor growth. Only certain mutations seem to be able to activate these proteins. All the reported NRAS gene mutations are located in exon 2 and exon 3 (previously termed exon 1 and exon 2); most are present in exon 3 at codon 61, where the common amino acids substitutions are Q61K and Q61R. BRAF gene mutations are located in codon 600, usually converting valine to glutamic acid.2,3 Mutations in KIT are more widely distributed and frequently affect exons 11 and 13, but mutations in exon 17 and 18 have been reported. The usual amino acid substitutions for KIT are L576P, K642E, D816H, and A829P.1,4 Although HRAS and GNAQ mutations are more likely to be found in Spitz nevi and blue nevi, respectively, there have been 3 reported cases of HRAS (Q61H/L/K) mutations and 2 reported cases of GNAQ (Q209L/P) mutations in cutaneous melanoma.5,6 In particular, GNAQ mutations were present within melanomas with chronically sun-damaged areas.6
While the commonly identified melanoma growth–promoting mutations all drive the MAPK/ERK pathway, they also interact with other regulatory pathways, which may alter their growth pattern. Dermoscopy allows for improved visualization of tumor structures and increased diagnosis of thin or early cutaneous melanomas. We hypothesize that tumors with similar “dermoscopic” surface growth patterns will have similar underlining growth-promoting mutations and that ultimately dermoscopic growth patterns may serve to better segregate melanoma subtypes and genotypes than the current classification system.
Our study was approved by the institutional review board of the University of Miami. A collection of 182 dermoscopic images of melanomas from the University of Miami Melanoma Program and Plantation Cutaneous and Surgery Clinic from 2000 to 2011 were reviewed by 3 of us (M.I.S., H.S.R., and J.M.G.). Primary cutaneous melanomas with similar dermoscopic features were grouped together based on dermoscopy features from Argenziano et al,9 Pizzichetta et al,10 Keir,11 and our own experience. At least 10 different primary dermoscopic growth patterns were noted, including disorganized fine network,9 epidermal crease sparing, gray-dotted regression,9 homogeneous,9 multicolor multicomponent,9 perifollicular pigmented,9 rhomboidal (grayish brown pigmentation surrounding the follicular ostia),9 polygonal (irregular grayish brown pigmentation outlining the limits between hair follicles forming multiple-sided figures),11 negatively pigmented network (light areas forming the “cords” of the network and darker areas filling the holes),10 and dark homogeneous streaks patterns (structureless, brownish black pigmentation with bulbous projections or streaks at the edge)9 (Figures 1, 2, and 3 and eFigure 1 and eFigure 2 in Supplement). Four of the 10 melanoma patterns were selected for genetic analysis based on the dermoscopic pattern being relatively uniform throughout the entire lesion (reducing the chance of different mutations in different areas of the lesion) and the availability of at least 3 cases with adequate tissue in the formalin-fixed, paraffin-embedded (FFPE) blocks. The 4 patterns selected for genetic analysis were the rhomboidal (eFigure 1 [Supplement]), negatively pigmented network (Figure 2), polygonal (eFigure 2 [Supplement]), and dark homogeneous streaks pattern (Figure 3). The Breslow thicknesses ranged from in situ to 1.1 mm in the lesions tested. Individually, the depths of the rhomboidal, polygonal pattern, negatively pigmented network, and dark homogeneous streaks patterns ranged from in situ to 0.55 mm, in situ to 0.32 mm, 0.4 to 0.46 mm, and 0.75 to 1.1 mm, respectively (eTable 1 [Supplement]). Histologic reports classified the lesions as melanoma in situ and melanoma for the rhomboid and polygonal patterns. For the negatively pigmented network and dark homogeneous streaks patterns, the histologic reports classified the lesions as melanoma and superficial spreading melanoma, respectively. An associated nevus precursor was not noted for any of the tested melanomas. The specific comments on the morphologic characteristics for each pattern are noted in eTable 1 (Supplement).
Shown are the 6 dermoscopic patterns that were not subjected to genetic analysis. A and B, Disorganized fine network pattern; the insets show a brown network structure with irregular holes and lines. C and D, Epidermal crease sparing pattern; insets show well-defined epidermal creases surrounded by pigmented structureless areas. E and F, Gray dotted–regression pattern; insets show gray dots and areas of regression. G and H, Homogeneous pattern; the images show light and dark brown irregular structureless areas. I and J, Multicolor multicomponent pattern; the images show light brown, dark brown, and black colors with structureless areas, an irregularly pigmented network, and asymmetrical fingerlike projections. K and L, Perifollicular pigmented pattern; inset shows eccentric annular pigmentation around follicular ostia.
Dermoscopic images with negative pigmented network features of the 3 melanomas (A, D, and G) with their corresponding histologic images at lower and high magnifications (B, C, E, F, H, I). All 3 have similar negative network pattern with light areas forming the “cords” of the network and darker areas filling the holes (insets). B, C, E, F, H, I: hematoxylin-eosin; B, E, and H, original magnification ×4; C and F, ×10; I, ×20.
Dermoscopic images with streaks and homogeneous features of the 3 melanomas (A, D, and G) with their corresponding histologic images at lower and high magnifications (B, C, E, F, H, and I). All 3 have a similar pattern of dark homogeneous areas with streaks at the margins (insets). B, C, E, F, H, I: hematoxylin-eosin; B, E, and H, original magnification ×4; C, F, and I, ×20.
Three cases were analyzed for each of the 4 specific dermoscopic growth patterns chosen for study. All of the cases were from unique patients except for 2 of the cases of dark homogeneous streaks pattern, which turned out to be different primary tumors from the same patient. To maximally enrich the specimens for tumor DNA, the slides were deparaffinized and immunohistochemically stained with a cocktail of anti–melan A (Abcam) and anti-S100 (Biocare Medical). NovaRed (Vector Laboratories) was used as a chromagen. Melanoma cells and keratinocytes (as controls) from each case were specifically and separately captured within different 0.5-mL tubes using the Laser Capture microdissection system (Leica Microsystems). Each sample went through DNA extraction using a QIAamp kit (QIAGEN) and amplification with a REPLI-g kit (QIAGEN). Polymerase chain reactions (PCRs) for all the samples were performed using a KOD Hot Start Master Mix (Novagen), primers (Integrated DNA Technologies) (described in eTable 2 [Supplement]). Next, DNA was extracted and amplified on a Master Cycle Pro S device (Eppendorf) using the following parameters: 95°C for 2 minutes, 42 cycles of 95°C for 20 seconds, and an annealing temperature as described in eTable 2 for17 seconds and at 70°C for 17 seconds. The final elongation was at 70°C for 20 seconds. The PCR products were analyzed on agarose gels, and bands were purified with a MinElute gel extraction kit (QIAGEN) if not sufficiently pure for sequencing. The products were directly sequenced in both forward and reverse directions in the Genewiz Laboratories. Sequence output was reviewed and visually analyzed for the presence of secondary peaks indicating mutations.
DNA isolated from rhomboid, negatively pigmented network and polygonal dermoscopic growth pattern groups were wild type for BRAF, NRAS, HRAS, GNAQ, and KIT. DNA isolated from the dark homogeneous streaks pattern group revealed KIT mutations in the catalytic loop of the distal kinase domains H790Q, A794V, and I798M (Figure 4) (eTable 3 [Supplement]). All 3 of these point mutations were novel. The dark homogeneous streaks pattern group was wild type for BRAF, NRAS, HRAS, and GNAQ.
The location of the KIT gene on chromosome 4. Exon 17 includes the second tyrosine kinase domain. The highlighted bases are those included in the catalytic domain. All 3 mutations identified are in or directly next to the catalytic domain and are shown in the sequences in the panel to the right of each tumor image.
Dermoscopy allows for visualization of melanoma growth patterns. Using a dermoscopy-based, pattern-matching approach, we have identified 3 melanoma growth patterns that lack mutations in BRAF (exon 15), NRAS (exons 2 and 3), HRAS (exons 2 and 3), GNAQ (exon 5), and KIT (exons 11, 13, 17, and 18). Despite the high prevalence of BRAF and NRAS mutations in melanomas as a whole, all of these tested patterns were wild type. This may merely be due to a selection bias, but it is also possible that BRAF and NRAS are overrepresented in published data sets owing to increased growth, tumor bulk and availability, and lethality. Reports have noted that fast-growing and vertical growth phase melanomas are more frequently mutated for BRAF and NRAS.8 Ellerhorst12 et al have shown that the median Breslow thicknesses of BRAF- and NRAS-mutated melanomas were 1.28 mm and 1.4 mm, respectively, whereas for wild-type melanomas the median was 0.93 mm. Thus, it is possible that thinner and low-risk tumors will be found to have fewer BRAF and NRAS mutations than previously thought. Since the rhomboid, negative, and polygonal dermoscopic patterns of melanoma did not demonstrate common driving mutations, this pattern of tumors could be used to identify as-yet unknown melanoma growth–promoting gene mutations. They may also uncover a set of mutations that occur prior to the development of BRAF and NRAS mutations if BRAF and NRAS mutations occur later in progression.
The dark homogeneous streaks pattern was found to have a specific KIT gene mutation. KIT mutations have been previously reported to be present in 4.3% of the cutaneous melanomas1,4; of them, 4.4% were from exon 17, which suggests that only 0.19% of cutaneous melanomas will have a mutation in exon 17. Thus, the chance that 3 out of 3 melanomas would have mutations in KIT exon 17 is 6.8 in 1 billion; therefore, it is rational to speculate that this specific mutation gives rise to this specific dermoscopic growth pattern. Although all the other cases of melanomas studied were from unique patients, it is also of interest that 2 of the dark homogeneous streaks pattern were from different primary tumors on the same patient. While KIT was mutated in both primary tumors, the specific mutated sequence was different, confirming the different origins of the 2 tumors. Patients often have similar mole patterns, and it has also been noted that patients with multiple melanomas often have tumors with similar patterns.13 Thus, it is possible that patients are predisposed to developing melanomas with mutations in specific pathways.
Our KIT-mutated tumors were dark, and this is consistent with the findings of Wu et al,14 who previously reported that KIT is mutated and/or overexpressed in darkly pigmented cutaneous melanomas. KIT mutations have been previously related to a histologic lentiginous pattern present in chronically sun-damaged areas.1 Two of the KIT-positive melanomas presented herein were described histologically as lentiginous, but only 1 was in a chronically sun-damaged area.
The 3 mutations noted herein are novel and are all located in the second kinase domain, within or next to the catalytic loop. This domain is thought to maintain the enzyme in its active form.15 Given that KIT also has mutations in other domains and interacts with a number of different pathways, the different KIT mutations may eventually be found to give rise to different subsets of melanoma growth patterns.
In summary, one of the dermoscopic patterns, the dark homogeneous streaks pattern, seems to correlate with KIT mutations in or around the second kinase domain, suggesting the possibility that specific growth patterns are caused by specific driving mutations. Two of the similarly patterned primary tumors occurred in the same patient, and, while having different specific nucleotide changes, the mutations occurred in the same domain of the same gene, suggesting the possibility that patients may have a predisposition to develop and/or maintain certain mutations. Three of the 4 patterns tested lacked common melanoma-driving mutations suggesting the possibility that these patterns could be used to identify other novel driving genes and/or identify genes that are mutated in early melanoma prior to the development of the common driving mutations. Further study is required, but it is possible that dermoscopic growth patterns will lead to improved molecular classification of melanocytic neoplasias.
Accepted for Publication: September 7, 2013.
Corresponding Author: James M Grichnik, MD, PhD, Department of Dermatology and Cutaneous Surgery, Miller School of Medicine, University of Miami, Room 912, BRB, 1501 NW 10th Ave, Miami, FL 33136 (firstname.lastname@example.org).
Published Online: April 2, 2014. doi:10.1001/jamadermatol.2013.8442.
Author Contributions: Drs Sanchez and Grichnik had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Sanchez, Grichnik.
Acquisition of data: Sanchez, Rabinovitz, Oliviero, Elgart, Perez, Puig, Malvehy.
Analysis and interpretation of data: Sanchez, Grichnik.
Drafting of the manuscript: Sanchez, Grichnik.
Critical revision of the manuscript for important intellectual content: All authors.
Obtained funding: Grichnik.
Administrative, technical, and material support: Sanchez, Perez, Grichnik.
Study supervision: Grichnik.
Conflict of Interest Disclosures: Dr Rabinovitz has served as a consultant to 3Gen, and Ms Oliviero has been a speaker for 3Gen. Drs Rabinovitz and Grichnik have served as consultants for and received meeting support and equipment from Caliber ID Inc. Dr Rabinovitz has also served as a consultant to MELA Sciences, DermaTech, and SciBase Inc. Grichnik is also a major shareholder in DigitalDerm Inc. No other disclosures are reported.
Funding/Support: This study was supported in part by the Frankel Family Division of Melanocytic Tumors, Department of Dermatology and Cutaneous Surgery, the Anna Fund Melanoma Program at Sylvester Comprehensive Cancer Center, University of Miami. The research at the Melanoma Unit in Barcelona, Spain, is partially funded by grants from Fondo de Investigaciones Sanitarias PI 09/01393 and PI 12/00840; by the CIBER de Enfermedades Raras of the Instituto de Salud Carlos III; grant AGAUR 2009 SGR 1337 of the Catalan Government, Spain; by the European Commission under the Sixth Framework Programme, contract No. LSHC-CT-2006-018702 (GenoMEL).
Role of the Sponsor: The funding sources had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
Additional Contributions: We are indebted to the Anna Fund Melanoma Program at Sylvester Comprehensive Cancer Center, the Frankel Family Division of Melanocytic tumors, Department of Dermatology and Cutaneous Surgery, Department of Dermatology, and our many other benefactors, especially William Rubin and his family and friends. They did not receive compensation for their assistance.
eFigure 1. Rhomboidal pattern
eFigure 2. Polygonal pattern
eTable 1. Clinical and histopathologic characteristics of melanomas genetically analyzed
eTable 2. Sequence of polymerase chain reaction primers used in the study
eTable 3. Results of genetically analyzed cutaneous melanomas
Sanchez MI, Rabinovitz HS, Oliviero MC, Elgart GW, Perez C, Puig S, Malvehy J, Grichnik JM. Dark Homogeneous Streak Dermoscopic Pattern Correlating With Specific KIT Mutations in Melanoma. JAMA Dermatol. 2014;150(6):633-639. doi:10.1001/jamadermatol.2013.8442