In melanoma, mutations in KIT are most frequent in acral and mucosal subtypes and rarely reported in cutaneous melanomas particularly those associated with intermittent UV exposure. Conversely melanomas arising within chronic sun damaged skin are considered to harbour KIT mutations at higher rates. To characterize the frequency of KIT mutations in a representative melanoma population, 261 patients from two Australian melanoma centres were prospectively screened for mutations in exons 11, 13 and 17 of the KIT gene. A total of 257 patients had cutaneous melanoma arising from non-acral sites and four were acral melanomas. No mucosal or ocular melanomas were analysed. KIT mutations were identified in five tumours (2% of the entire cohort) including two acral melanomas. Two of the three non-acral melanomas with KIT mutations were associated with markers of chronic sun damage as assessed by the degree of skin elastosis. In the remaining cohort, 43% had chronically sun damaged skin. This report confirms that within an Australian population, KIT mutations are infrequent in cutaneous melanomas associated with both intermittent and chronic sun exposed skin.
The majority of melanomas arise from the skin where UV exposure is the single most important causative factor. In rarer melanoma subtypes (mucosal and acral), activating mutations in KIT are more frequent and associated with clinical response to KIT inhibition. KIT mutations are less frequently described in cutaneous melanomas. However in those that have been reported, there appears to be an association with chronic sun damaged skin. It is therefore of clinical importance to establish the frequency of KIT mutations in this most common subtype of melanoma and to gauge the significance of the severity of sun exposure as a factor in discerning which patients are most likely to benefit from KIT directed therapies.
Melanoma represents the second most prevalent cancer in Australia, a country characterized by high environmental UV exposure and a large number of fair skinned inhabitants. The Australian melanoma population is therefore of particular epidemiological interest given these features and a low proportion of acral and mucosal subtypes which make up <5% of all melanoma diagnoses.
Over the past three decades, there has been little improvement in outcome for patients with advanced or metastatic melanoma irrespective of subtype. As standard therapies generally yield low response rates, there is a clear need to provide other avenues for therapeutic intervention. The investigation of molecular therapies in the management of melanoma has been driven by the identification of alterations in oncogenes within different biological pathways that have been linked to specific melanoma subtypes. For example, it has been demonstrated that melanomas associated with intermittent sun exposure have frequent mutations in BRAF (Curtin et al., 2005; Maldonado et al., 2003; Thomas, 2006) and specific BRAF inhibitors are now in therapeutic development for this and other malignancies demonstrating activation of BRAF. KIT is another oncogene that has been investigated as an alternative mechanism of cellular activation in melanoma. KIT is a plausible therapeutic target because it plays an important role in melanocyte development (Grabbe et al., 1994; Nishikawa et al., 1991; Taylor and Metcalfe, 2000) and because signalling through the KIT receptor tyrosine kinase activates downstream targets of the mitogen-activated protein kinase (MAPK) and phosphoinositide 3-kinase/AKT (PI3K-AKT) pathways that mediate the processes of cell survival, proliferation and invasion (Cohen et al., 2002; Dhawan et al., 2002).
A number of recent reports have demonstrated an increased frequency of KIT mutations and/or increased copy number in specific clinical subtypes of melanoma [acral, mucosal and melanomas associated with chronic sun damage (CSD)] and no such aberrations in melanomas arising from skin without CSD (Beadling et al., 2008; Curtin et al., 2006). Additionally, there have been several reports describing in vitro responses to imatinib and other KIT inhibitors in melanomas containing activating KIT mutations (Antonescu et al., 2007; Ashida et al., 2009; Jiang et al., 2008), and clinical responses have been predominantly reported in mucosal subtypes (Hodi et al., 2008, Lutzky et al., 2008). The relationship between sun damage and KIT mutations however remains unclear, making recommendations for KIT mutation detection in cutaneous melanomas arising from non-acral sites uncertain and worthy of further investigation.
This report describes the frequency of KIT mutations in a large and representative group of melanoma patients consecutively referred to two tertiary melanoma centres in Australia, where 98% of cases were cutaneous melanomas arising from non-acral sites. This cohort derives from an area of high solar irradiance by world standards where elastosis scores may be relatively high and is therefore of particular relevance in discerning the relationship between sun exposure and KIT mutation status in melanoma.
In this group of 261 patients, five patients with KIT mutations were identified representing a frequency of 2%. Of these five patients, two had a primary diagnosis of acral melanoma and three had cutaneous (superficial spreading subtype) melanomas (Table 1). Two samples (one acral and one non-acral) contained the L576P mutation in exon 11 (juxtamembrane region) which is reported to be sensitive to KIT inhibition (Antonescu et al., 2007). One cutaneous sample contained a D816V mutation in exon 17, previously identified as an activating mutation in mastocytosis and acute myeloid leukemia (Longley et al., 1999) and the other patient with acral melanoma had a D820Y mutation in exon 17. Both these regions within exon 17 encode the kinase domain of the receptor and although mutations within this area are considered to be resistant to imatinib therapy (Ma et al., 2002), some of these mutations have been shown to be sensitive to second generation inhibitors such as Sorafenib (Guo et al., 2007). An additional non-acral melanoma contained a novel duplication of 8 amino acids resulting from a 24 base pair duplication at codon 1755 in exon 11. Histological assessment for solar elastosis was performed on 223/261 (85%) of the cohort including all the melanomas containing mutations in KIT to assess the degree of sun damage to the surrounding skin. Grading for elastosis was performed according to a scoring system devised by Bastian (Landi et al., 2006). A multipoint scale from 0 to 3+ was used, where melanomas are classified as arising from non-CSD if they showed only minor signs of elastosis (elastosis score 0 to 2−) and associated with CSD if there was more pronounced solar elastosis (elastosis scores 2 to 3+). In the non-acral samples, where KIT mutations were identified, the elastosis scores varied. Two melanoma samples, both originating from the upper limb, had elastosis scores consistent with CSD skin whereas the melanoma sample originating for the skin on the lower limb had a low elastosis score (Figure S1). Grading for solar elastosis in all the other evaluable cases without KIT mutations revealed that 96 (43%) had a grading of 2 or greater and 127 (57%) were graded between 0 and 2− (Table 2). The distribution of CSD melanomas as indicated by marked elastosis was higher in sites such as the head and neck and upper extremity where 80 and 45% of cases respectively were associated with CSD skin. This is in contrast to the lower limb and lower back where only 26 and 16% of melanomas respectively were considered to have arisen in CSD skin (Figure 1).
Table 1. Clinico-pathological details of melanoma patients with KIT mutations
Table 2. Elastosis score according to anatomical sites where a score between 2 and 3+ is representative of chronic sun damaged (CSD) skin and a score less than 2 indicative of non-CSD skin
()* indicates the number of KIT mutations detected at these melanoma sites.
Head and neck
Details of associated pathological data on all samples with KIT mutations are shown in Table 1. None of the samples had co-existing BRAF or NRAS mutations. Immunohistochemistry (IHC) staining for CD117 (KIT) was only performed on the KIT mutation positive melanomas. The IHC status for the remaining cohort where no KIT mutation was detected was not investigated. Of the samples with KIT mutations, both acral samples stained strongly positive for CD117. Only one non-acral specimen showed positive staining for CD117. This was a thick melanoma with a high mitotic count arising from the upper arm carrying the L576P mutation (Figure S2). Most samples were associated with two or more poor prognostic factors as defined by the standard pathological criteria of a mitotic count of >1 mm², thickness >2.0 mm and presence of ulceration. Follow-up data indicate that the two individuals with the acral melanomas died prior to analysis, one patient remains free of disease at the time of publication and the two other patients have since developed metastatic disease.
In this study, patients referred to two tertiary melanoma services in Australia were evaluated for mutations in KIT to determine the frequency of such KIT aberrations within a representative melanoma population. In this cohort of 261 patients the mutation frequency for KIT was 2%. Conversely within the same cohort, BRAF mutations were found in 45% of primary melanomas (Liu et al., 2007) and NRAS in 15% (manuscript in preparation). This is not surprising given that the majority of this group (98%) comprised of cutaneous melanomas arising from non-acral sites, reflecting the prevalence of this subtype in the average melanoma population in Australia. The remaining 2% comprised of acral melanomas, half (2/4) of which harboured KIT mutations. Mutations in KIT and increased KIT copy number have also been previously reported at higher frequencies in acral melanomas (Beadling et al., 2008; Curtin et al., 2006).
The five samples containing KIT mutations and a further 218 patients without mutations were examined for the degree of elastosis in the surrounding skin as an indicator of CSD. The presence of elastosis relates to patient age and anatomical site and may therefore represent a good surrogate for the cumulative dose of absorbed UV radiation. Sites such as the head and neck tend to be most affected, while in contrast, acral sites show almost no elastosis throughout life (Vollmer, 2007). In the non-acral samples arising from the upper limb that harboured KIT mutations, the degree of elastosis in the surrounding skin was consistent with CSD. A further melanoma containing a novel KIT mutation/duplication arising within non-CSD skin however was also identified. More striking is the observation that almost half of the cohort showed histological evidence of CSD, including over 80% of those originating from the head and neck, however few CSD melanomas with mutations in KIT were identified. This contrasts to what has been previously observed in other studies, where KIT mutations were reported at higher frequencies typically in anatomical sites that are most characteristically associated with chronic sun exposure such as the head and neck region (Curtin et al., 2006). Although the higher frequency of KIT aberrations described in the original reports of samples collected predominantly from patients within Europe and North America may reflect an anatomic bias, they may also reflect the impact of geographic differences in UV exposure and more specifically the interaction between the examined population and geographical location. Alternatively the differences may not be explained by the subtype of melanoma, and factors unrelated to UV exposure might exist to explain the presence of KIT mutations in these cutaneous melanomas. This is feasible given the prevalence of KIT mutations in anatomical sites (acral and mucosal) where UV exposure is limited. Another point of consideration is that KIT amplification has not been tested for in this study. It is possible that KIT amplification is present in a proportion of those samples where CSD is evident. Presently, however, the pathogenic role of KIT amplification in melanoma remains unclear.
Responses to inhibition of the KIT receptor vary according to the location of the mutation within the protein and whether the activation of the corresponding signalling pathway represents a dominant mechanism in the pathogenesis or progression of disease. Mutations affecting the juxtamembrane region of the receptor are of particular significance given that mutations here, e.g. V560G mutations in gastrointestinal stromal tumours (GISTs) are known to result in constitutive activation of the receptor and are highly responsive to imatinib (Heinrich et al., 2003). Two melanomas within the cohort contained an exon 11 (L576P) juxtamembrane mutation and therefore were predicted to be sensitive to KIT kinase inhibition. Additionally, both these samples stained strongly for KIT on IHC. Efforts, however to select melanoma patients for therapy with KIT inhibitors solely based on KIT expression by IHC alone have been largely unsuccessful (Alexis et al., 2005; Kim et al., 2008; Ugurel et al., 2005; Wyman et al., 2006). This is distinct from GIST whereby expression of KIT via IHC correlates highly with activating mutations in the gene (Corless et al., 2004). One possible explanation to account for this is that some constitutively active KIT melanoma mutants, e.g. D816V may signal and be turned over quickly without being transported to the cell surface. In one study, this is supported by the presence of smaller bands on the phosphotyrosine blots, which may represent degraded KIT because of high rate of continuous turnover of constitutively active forms (Frost et al., 2002). In this report, three out of the five cases with KIT mutations (two L576P mutations and one D820Y mutation) stained strongly positive for KIT. The 24 base pair duplication mutation in exon 11 however was negative for KIT staining as was the D816V mutation. Taken together, these data suggest that KIT IHC should not necessarily be relied upon to select for KIT mutation screening in melanoma.
This study demonstrates a low overall frequency of KIT mutations in a cohort of Australian melanoma patients where the majority of patients had cutaneous melanomas arising from non-acral sites. Based on the findings within this representative population of melanoma patients it is difficult to make recommendations for KIT mutation testing in cutaneous melanomas, particularly for an Australian population, based exclusively on the degree of sun exposure and CSD in the affected skin. If low cost methodologies can be found to enable broader screening; however, there would be tangible benefits to those patients whose tumours did carry KIT mutations based on the clinical responses already seen with KIT inhibitors.
Materials and methods
The cohort consisted of 261 patients presenting consecutively to two tertiary melanoma referral centres in Melbourne (Peter MacCallum Cancer Centre and The Alfred Hospital Melanoma Service) between May 2003 and September 2004. These patients consented to their melanoma samples being screened for genetic mutations including BRAF and NRAS (BRAF data published elsewhere (Liu et al., 2007) and NRAS manuscript in preparation). The same samples were subsequently screened for mutations in KIT involving exons 11, 13 and 17. Two hundred and fifty-seven (98%) patients had a diagnosis of primary cutaneous melanoma. The remaining four cases (2%) were from patients with acral melanoma. No mucosal or ocular melanomas were included in the cohort. Clinical and pathological data was collected prospectively for most patients. For those samples in which KIT mutations were identified, IHC for KIT expression (CD117) and scoring for elastosis (as a histological marker of CSD) was performed. A further 218 cases were evaluated for solar elastosis.
Tumour rich areas were micro dissected from unstained sections [derived from fresh frozen paraffin embedded (FFPE) samples] and tumour poor areas left behind (using H&E slide for comparison). Genomic DNA was extracted using a Qiagen DNeasy extraction kit (Qiagen Ltd, Hilden, Germany) as per manufacturer’s instructions except additional proteinase K digestion was added at 24 and 48 h.
High resolution melting analysis (HRM) was used to identify mutations followed by direct sequencing for those samples with aberrant melt profiles (Figure S3). HRM and sequencing assays had been previously optimized against a panel of samples known to harbour mutations in KIT exons 11, 13 and 17. Primers for HRM analysis were designed to span exons 11, 13 and 17 of the KIT gene (NM_000222). Primers for exon 11 were 5′-ATCTATTTTTCCCTTTCTCCCCACAG-3′ (forward) and 5′-GGAAAGCCCTGTTTCATACTGAC-3′ (reverse). Primers for exon 13 were tagged with M13 tails (in italics) and were 5′-TGTAAAACGACGGCCAGTTGCGCTTGACATCAGTTTGCCAG-3′ (forward) and 5′-CAGGAAACAGCTATGACCTAAAAGGCAGCTTGGACACGGCT-3′ (reverse). Primers for exon 17 were 5′-TCCTTACTCATGGTCGGATCACA-3′ (forward) and 5′-TGGGTACTCACGTTTCCTTTAACCA-3′ (reverse). For exon 11, the reaction mixture was made up using HotStarTaq (Qiagen) and consisted of 5 ng of genomic DNA, 1× PCR buffer, 2.5 mM MgCl2 total, 200 and 400 nM of forward and reverse primer respectively, 200 μM of dNTPs, 5 μM of SYTO 9 (Invitrogen, Carlsbad, CA, USA), 0.5 U of HotStarTaq polymerase and PCR grade water in a volume of 20 μl. For exons 13 and 17, reaction mixtures were prepared as for exon 11, but with 400 nM of forward and reverse primers. All samples were tested in duplicate and repeated again in triplicate (total n = 5) for those samples with aberrant melting profiles.
PCR cycling and HRM analysis were performed on the Rotor-Gene (Qiagen) under the following conditions; one cycle of 95°C for 15 min; 55 cycles of 95°C for 10 s, 55°C for 10 s, 72°C for 30 s; one cycle of 97°C for 1 min, and a melt from 70 to 95°C rising at 0.2°C per second. The primer annealing step was preceded by 10 cycles of touch-down annealing decreasing by 1°C per cycle. DNA samples from patients who are known to harbour a mutation in KIT were used as positive controls. Negative controls included cell lines and FFPE samples from patients without mutations in KIT. For sequencing, the initial PCR amplification was performed with conditions as described above, except that Syto9 dye was not included in the PCR mix. Single-stranded excess primers and unincorporated dNTPs were removed from the PCR products using ExoSAP-IT (USB Corporation, Cleveland, OH, USA) as per manufacturer’s instructions. The PCR products were then sequenced using the BigDye v.3.1 Cycle Sequencing kit (Applied Biosystems, Foster City, CA, USA) as per manufacturer’s instructions and using the same primers for each amplicon described above, except that for exon 13, M13 sequencing primers were used. The sequencing products were ethanol precipitated before running on a 3100 Genetic Analyser (Applied Biosystems). The sequencing data was analysed using Sequencher 4.6 (Gene Codes Corporation, Ann Arbor, MI, USA).
Immunohistochemistry staining for KIT (CD117) was performed only on those melanoma samples with KIT mutations. Retrieval was performed using Dako PT link tank, with Dako Envision Flex high pH retrieval solution. Slides were stained on a Dako Autostainer Plus instrument using Dako Envision Flex kit Reagents. The KIT (CD117) primary antibody (Dako polyclonal rabbit) diluted to 1/500 was used. The results were evaluated by light microscopy. Mast cell staining for CD117 was used as an internal positive control. Samples were designated either positive or negative. Only those samples with strong diffuse staining for CD117 were considered positive.
Grading solar elastosis
Wide local excision specimens were stained with H&E and examined under light microscopy for assessment of the degree of solar elastosis in the dermis using the scoring criteria devised by Bastian (Landi et al., 2006). A grading of 2 or greater was considered to be significantly associated with CSD. In addition to the wider excision specimens from the cutaneous melanomas with KIT mutations, a further 218 without mutations in KIT were scored for solar elastosis for comparison. The remaining 43 cases within the cohort were unavailable for evaluation. A proportion of samples were graded by a second person with a inter observer agreement weighted co-efficient of 0.705.
We would like to thank Victoria Beshay for genotyping assistance, Nicholas Jene for immunohistochemistry preparation, Boris Bastian and Glenn Lynch for discussions and Jill Magee, Chris Angel and Graham Mason for their pathology input. Also thanks to Professor Cook for assistance with elastosis scoring and Maurice Loughrey for critical assessment of the manuscript.
This work was supported by grants from the NHMRC and a Sir Edward Dunlop Clinical Research Fellowship from the Cancer Council of Victoria (G.A. McArthur), and an Australian Postgraduate Award from the University of Melbourne (D. Handolias).