Somatic mutation of epidermal growth factor receptor in a small subset of cutaneous squamous cell carcinoma

To the Editor,

Non-melanoma skin cancer (NMSC) is the most prevalent cancer in man. For reasons that are unclear NMSC is dramatically increased in organ transplant recipients (OTR) compared to immunocompetent patients (non-OTR) (Euvrard et al, 2003). Increased risk of metastasis and the presence of multiple tumors are a significant source of morbidity in OTR. These patients present with different types of NMSC with squamous cell carcinoma (SCC), and keratoacanthoma (KA) being particularly prevalent. The genetic aberrations driving these cancers in OTRs and non-OTRs are poorly understood.

Receptor tyrosine kinases (RTKs) are frequently activated by somatic mutations and/or amplification in human cancers, particularly in epithelial tumors. To determine whether RTKs are mutated in SCC, we searched the literature to identify RTKs that play a role in epidermal homeostasis and thus could be candidate oncogenes in squamous lesions. We chose to analyze epidermal growth factor receptor (EGFR) that is highly expressed in a small subset of metastatic cutaneous SCCs (Bauknecht et al, 1985; Shimizu et al, 2001; Maubec et al, 2005); FGFR3 that is mutated in familial acanthosis nigricans and Crouzon’s syndrome, a type of craniosynostosis (Berk et al, 2007) and induces acanthosis and benign tumors in transgenic mice (Logie et al, 2005); FGFR2, that is also mutated in Crouzon’s syndrome and in this disease is associated with acanthosis nigricans (Meyers et al, 1995). We included the insulin like growth factor receptor 1(IGF1R); mice lacking this receptor have hypoplastic skin (Liu et al, 1993; De Moerlooze et al, 2000), and MET the receptor for the ligand HGF. Mice over expressing the MET receptor exhibit enhanced numbers of hair follicles and accelerated hair follicle morphogenesis (Lindner et al, 2000), a feature associated with cyclosporine use in OTRs. Finally, we assessed ERBB2 that induces SCCs when targeted to mouse skin (Kiguchi et al, 2000).

We determined the mutation status of the kinase domains of EGFR, IGF1R, MET and ERBB2 and the regions of FGFR2 and FGFR3 that are mutated in Crouzon’s syndrome in a cohort of 95 tumors that consisted of 70 SCCs and 25 KAs from 55 OTR and 40 non-OTR tumors; not every tumor was analyzed for every gene. Genomic DNA was extracted from archival formalin fixed paraffin embedded samples and amplified with M13 sequence-tailed primers (supplementary table 1).

Mutations were found in EGFR, FGFR2 and FGFR3 but not in ERRB2, MET and IGF1R (Table 1). The somatic nature of the mutations was confirmed by sequencing the adjacent normal skin in all three cases in which mutations were found. EGFR was mutated in one of 40 (2.5%) SCCs, a frequency not dissimilar to that detected in head and neck SCCs (7.3%) (Willmore-Payne et al, 2006). The particular Y727H mutation we found in exon 18 of EGFR has been observed in SCC of the lung (Pallis et al, 2007). In addition to mutational activation, amplification of wild-type EGFR can drive tumorigenesis in a variety of cancers and in head and neck SCC cell lines (Weichselbaum et al, 1989). In a dataset of array-based comparative genomic hybridization (CGH) of a larger cohort of SCCs (n=268, 173 OTR tumors and 95 non-OTR tumors) and KAs (n=46, 27 OTR tumors and 19 non-OTR tumors) we found amplifications of the EGFR region on chromosome 7 in three SCCs (1.1%) but not in KAs (Figure 1a) (Ridd et al. manuscript in preparation). Interestingly, all of the tumors harboring an EGFR amplification were from non-OTRs. The amplifications were confirmed by fluorescence in situ hybridization (FISH) (Figure 1b). The tumor with the EGFR mutation had no amplification of EGFR, and the three tumors with amplification did not show any mutations. We also analyzed EGFR protein levels in 275 SCCs (157 OTR tumors and 118 non-OTR tumors) and 69 KAs (28 OTR tumors and 41 non-OTR tumors) using tissue microarrays. EGFR was over-expressed compared to adjacent normal skin in 19 of the 275 SCCs (6.9%) tumors and none of the KAs (Figure 1c and 1d). EGFR was significantly over-expressed in a greater number of non-OTR SCCs (14/118 (11.8%) compared to OTR SCCs (5/157 (3.1%)) (p<0.0001). Amplification of EGFR was associated with high EGFR protein levels by immunohistochemistry and a representative image is shown in Figure 1c. There was not sufficient tissue to determine the EGFR protein level in the single tumor with an EGFR mutation. EGFR overexpression was not associated with an increased pERK expression (data not shown), but pERK levels generally do not correlate with EGFR activation even in cancers with a demonstrated pathogenetic role for EGFR such as non-small cell lung cancer (Vicent et al, 2004; Han et al, 2005; Conde et al, 2006; Sonobe et al, 2007; Hosokawa et al, 2009). This is explained by the extensive feedback mechanisms that exist for MAPK signaling (Citri and Yarden 2006; Amit et al, 2007).

We found a somatic mutation of FGFR2 in one of 41 SCCs (2.4%) and none of 19 KAs. The P303L mutation found in exon 8 is different from the somatic mutations found in primary uterine tumors or endometrial cell lines (Pollock et al, 2007) and the germline mutations found in craniosynostosis syndromes (Kan et al, 2002).

Mutations in FGFR3 were detected in one of 17 (6%) KA but were not found in a series of 41 SCCs. Interestingly, the mutation was found in exon 7 (A265T) and thus differs from the mutations of FGFR3 described in seborrheic keratoses, bladder and cervical cancer (Cappellen et al, 1999; Logie et al, 2005). Our results are concordant with a previous study that did not report mutations in FGFR3 in SCC (Karoui et al, 2001).

No mutations were found in IGF1R, and to date mutations have not been reported in human cancers. Mutations were also absent in MET in our samples, while mutations are found in renal cell carcinomas and in 25% of SCC of the head and neck (Schmidt et al, 1999; Di Renzo et al, 2000). Finally, we did not find mutations in ERBB2 in NMSC consistent with studies on SCC of the head and neck (Willmore-Payne et al, 2006).

Our study identifies genetic activation of EGFR by mutation or amplification in a small subset of SCC. EGFR is a validated therapeutic target in other cancers and approved drugs targeting EGFR exist, so that this finding may provide a therapeutic opportunity for a small subset of patients with advanced SCC refractory to conventional treatment. Future studies will have to confirm, whether EGFR activation is indeed more frequent in non-OTRs as suggested by the mutation pattern in our data set, By contrast, somatic mutations in other RTKs found in other epithelial cancers and hyperproliferative epidermal conditions are not a common finding in cutaneous SCC.