Analysis of c-KIT expression and KIT gene mutation in human mucosal melanomas

Recent data suggested an increased frequency of KIT aberrations in mucosal melanomas, whereas c-KIT in most types of cutaneous melanomas does not appear to be of pathogenetic importance. However, studies investigating the status of the KIT gene in larger, well-characterised groups of patients with mucosal melanomas are lacking. We analysed 44 archival specimens of 39 well-characterised patients with mucosal melanomas of different locations. c-KIT protein expression was determined by immunhistochemistry, KIT gene mutations were analysed by PCR amplification and DNA sequencing of exons 9, 11, 13, 17 and 18. c-KIT protein expression could be shown in 40 out of 44 (91%) tumours in at least 10% of tumour cells. DNA sequence analysis of the KIT was successfully performed in 37 patients. In 6 out of 37 patients (16%) KIT mutations were found, five in exon 11 and one in exon 18. The presence of mutations in exon 11 correlated with a significant stronger immunohistochemical expression of c-KIT protein (P=0.015). Among the six patients with mutations, in two patients the primary tumour was located in the head/neck region, in three patients in the genitourinary tract and in one patient in the anal/rectal area. In conclusion, KIT mutations can be found in a subset of patients with mucosal melanomas irrespective of the location of the primary tumour. Our data encourage therapeutic attempts with tyrosine kinase inhibitors blocking c-KIT in these patients.


Mutational analysis of KIT
Tumour cells were isolated from paraffin-embedded tissue (either primary tumour, lymph node metastases, skin metastases or local recurrences), if necessary by micrographic dissection using the PALM Laser-MicroBeam System (PALM Wolfratshausen, Germany). DNA extraction was performed with the DNA extraction kit from Qiagen (Hilden, Germany) following the instructions of the manufacturer. Exons 9, 11, 13, 17, 18 of KIT were amplified by LightCycler PCR using specific primers as described in the literature (Tarn et al, 2005;Curtin et al, 2006). Polymerase chain reaction products were DNA sequenced using an ABI Prism 3700 DNA Analyzer (SeqLab, Göttingen, Germany).

Mutational analysis of BRAF
To detect the BRAF V600E mutation a LightCycler fluorescence resonance energy transfer (FRET) assay with two fluorescent hybridisation probes was performed as described earlier (Hay et al, 2007). Real-time PCR was performed by using LightCycler FastStart DNA Master HybProbe (Roche Diagnostics GmbH, Mannheim, Germany). Post amplification fluorescent melting curve analysis was performed by gradual heating of the samples at a rate of 0.21C per second from 45 to 951C. Fluorescent melting peaks were determined by plotting of the negative derivative of fluorescence with respect to temperature.
All PCR products that showed deviation from the Wt (wild-type) genomic DNA melting peak as well as from the positive control samples were confirmed by direct sequencing of exon 15 of the BRAF gene (SequiServe, Vaterstetten, Germany).

Statistical analyses
The software SPSS 13.0 was used for statistical analyses. Kaplan -Meier tests and unpaired t-tests were performed.
Alterations in the KIT gene were observed in 6 out of 37 (16%) patients, five in exon 11 and one in exon 18 ( Figure 1B). Two patients suffered from mucosal melanomas of the head/neck region, three patients from mucosal melanomas located in the genitourinary tract and one patient from mucosal melanoma located in the anal/rectal tract. In one patient (case 27) the KIT mutation could be detected both in lymph node metastases and in skin metastases. Among the five tumours with KIT gene mutation of exon 11, four (80%) tumours showed strong ( þ þ þ þ ) and one showed (20%) high ( þ þ þ ) c-KIT protein expression ( Figure 1A). In contrast, tumours without mutation in exon 11 had significantly lower c-KIT expression (3 out of 32 negative, 7 out of 32 ( þ ), 7 out of 32 ( þ þ ), 8 out of 32 ( þ þ þ ), 7 out of 32 ( þ þ þ þ ), P ¼ 0.015).

DISCUSSION
A recent report showed a possible role of c-KIT in subsets of melanoma, in particular, mucosal melanomas (21% KIT mutations, 61% c-KIT overexpression), acral cutaneous melanomas (11% KIT mutations, 75% c-KIT overexpression) and cutaneous melanomas on skin with chronic sun damage (17% KIT mutations, 100% c-KIT overexpression) (Curtin et al, 2006). This suggests that c-KIT might be of pathogenetic relevance and therefore a therapeutic target in these subtypes of melanoma. In contrast, KIT mutations are rarely found in the major subtype of cutaneous melanoma originating from skin without chronic sun damage (Curtin et al, 2006) and unselected cutaneous melanomas (2 out of 100 in Willmore-Payne et al (2005); 1 out of 39 in Went et al (2004)). Moreover, therapeutic phase II studies with the c-KIT blocker imatinib in unselected melanoma patients without known KIT mutation status were disappointing (Ugurel et al, 2005;Wyman et al, 2006;Becker et al, 2007).
The aim of this study was to further elucidate c-KIT alterations in our well-characterised group of patients with mucosal melanomas that might support the role of c-KIT as a new therapeutic target in this subgroup of melanoma. The successful genetic analysis of KIT in 37 patients revealed mutations in 6 patients (16%). We could show KIT mutations in 2 out of 12 mucosal melanomas from head/neck, 3 out of 11 from the genitourinary tract and 1 out of 8 from the anal/rectal tract. This is consistent with the findings of Antonescu et al and Rivera et al who detected mutations of the KIT gene in 3 out of 20 (15%) and 4 out of 18 (22%) patients with mucosal melanomas of the anal region and oral cavity, respectively (Antonescu et al, 2007;Rivera et al, 2008). Thus, KIT mutations occur in up to 20% of mucosal melanomas irrespective of the location of the primary tumour.
The majority of KIT mutations in mucosal melanomas (11 out of 16 tumours in Curtin et al (2006), 3 out of 3 in Antonescu et al (2007), 4 out of 4 in Rivera et al (2008) and 5 out of 6 in our study) were detected in the juxtamembrane region of KIT encoded by exons 11 and 13, presumably resulting in the activation of c-KIT.
The L576P und W557R mutations of our patients have already been described both in mucosal melanomas and gastrointestinal stromal tumours (GIST) (Antonescu et al, 2004(Antonescu et al, , 2007Rivera et al, 2008). Deletions covering the 579 position (such as in our patient 24) have also been frequently described in GIST (Tarn et al, 2005). The K550N mutation found in our patient 2 has, to our knowledge, not been described yet. However, other alterations in the proximal part of exon 11 at codons 550 -562 have been reported frequently in GIST (Lasota et al, 1999;Longley et al, 2001).
Thus, the minor subgroup of patients with mucosal melanomas and activating KIT mutations might be susceptible to therapeutic c-KIT blockade. This is supported by findings in GIST, which show KIT mutations in 75 -80%, and respond significantly better to a therapy with the c-KIT inhibitor imatinib than tumours without KIT mutations (Heinrich et al, 2003). Therefore, therapeutic c-KIT blockade could be considered for the treatment of patients with mucosal melanomas and an activating KIT mutation. This is supported by two case reports published very recently of single patients suffering from metastasising anal melanoma that harboured a KIT mutation in exon 11 and exon 13, respectively. These patients were successfully treated with the c-KIT blocker imatinib (Hodi et al, 2008;Lutzky et al, 2008). c-KIT protein expression could be observed in 91% of our primary mucosal melanomas, which is similar to the rates reported in other series of primary mucosal melanomas of the anal/rectal tract (12 out of 16 in Chute et al, 2006) and oral cavity (16 out of 18 in Rivera et al, 2008), respectively. However, Antonescu et al (2007) reported lower number (6 out of 26 (23%)) of c-KIT positivity in mucosal melanomas of the anal/rectal tract. In addition, cutaneous melanomas have been reported to express c-KIT by immunohistochemistry in 22.8% (Potti et al, 2003) up to 84% (Giehl et al, 2007) of the cases. The high range between these studies is being explained with the different qualities of immunohistochemistry. Thus, c-KIT protein expression appears to be in a similar range in mucosal melanomas and cutaneous melanomas. In line with this, Giehl et al could find no differences of the c-KIT expression in melanomas with sun exposure compared with melanomas without sun exposure (Giehl et al, 2007). In contrast to metastatic melanoma from primary cutaneous melanoma, where the c-KIT expression is decreased (Montone et al, 1997;Shen et al, 2003), we could observe c-KIT expression in 9 out of 9 metastases of primary mucosal melanoma. Similarly, Antonescu et al detected c-KIT reactivity in 6 out of 6 metastases from their anal mucosal melanomas (Antonescu et al, 2007). Furthermore, in agreement with earlier studies we could not show a prognostic relevance of the c-KIT protein expression in mucosal melanomas (Chute et al, 2006).
We could observe a significant higher immunohistochemical expression of c-KIT in mucosal melanomas that harbour a potentially activating KIT mutation as compared with tumours A similar correlation has also been found for anal melanomas (Antonescu et al, 2007) and melanomas of the oral cavity (Rivera et al, 2008). Thus, immunohistochemistry might be a useful tool to screen for patients that are subjected to KIT mutation analysis. However, immunohistochemistry might not be sufficient to detect tumours with mutations susceptible for c-KIT blockade, as overexpression of c-KIT can also occur in tumours without mutation. This is illustrated by the report of three patients with metastatic melanoma and strong immunohistochemical c-KIT expression who did not respond to a therapy with the c-KIT blocker imatinib (Alexis et al, 2005).
BRAF mutations were rarely identified in 3 out of 27 (11%) patients in our series, similarly to other studies with patients suffering from mucosal melanomas who reported 1 out of 17 (Cohen et al, 2004) and 0 out of 13 (Edwards et al, 2004), respectively, whereas the majority of cutaneous melanomas on skin without chronic sun-induced damage harbour BRAF V600E mutations (Curtin et al, 2005). It is interesting to note that one of our patients with a BRAF mutation suffered from melanoma located in an UV-exposed area of the conjunctiva. In line with this, Gear et al (2004) reported BRAF mutations in 5 out of 22 patients with conjunctival melanomas.
In conclusion, our study supports the finding that KIT mutations presumably activating the tyrosine kinase activity of c-KIT can be found in a subgroup of patients with mucosal melanomas irrespective of the origin of the primary tumour. This encourages clinical studies with c-KIT blockers in patients with mucosal melanomas and appropriate KIT mutations. Immunohistochemistry for c-KIT expression might be a useful tool to screen for patients who are subjected to mutational analysis but cannot replace genetic analysis.  Table 1) and melting curve peak at 60.61C from BRAF V600E heterozygous mutation (cases 2 and 5, corresponding to cases 16 and 38 in Table 1).
c-KIT in melanoma I Satzger et al