Mutant Epidermal Growth Factor Receptor (EGFRvIII) Contributes to Head and Neck Cancer Growth and Resistance to EGFR Targeting

Every year in the United States, 37,000 new cases of head and neck squamous cell carcinoma (HNSCC) are diagnosed, making it the most common malignancy of the upper aerodigestive tract (1). Although advancements in organ preservation have helped reduce the morbidity associated with treatment of HNSCC (2), 30% of HNSCC patients die every year from this disease (1). New and more efficacious therapeutic modalities are needed to improve HNSCC survival.

Analysis of genetic and epigenetic alterations in human tumors has revealed potential molecular targets for cancer therapy. The epidermal growth factor receptor (EGFR) is a well-characterized proto-oncogene that is present in multiple cancers where it has been shown to promote tumor progression. EGFR is ubiquitously distributed on normal epithelial tissues and is overexpressed in several cancers, such as those of the breast, prostate, lung, and glioma (3). We reported previously that >95% of HNSCCs express elevated EGFR levels compared with levels in normal mucosa from patients without cancer (4). Further investigation showed that elevated EGFR expression levels in HNSCC serve as an independent indicator of poor prognosis and decreased overall survival (5, 6).

Strategies that target EGFR are actively under investigation for the treatment of HNSCC. EGFR-targeted therapies include monoclonal antibodies (mAb) that block the extracellular ligand-binding domain and tyrosine kinase inhibitors (TKI) that prevent activation of the cytoplasmic tyrosine kinase of EGFR. EGFR-specific TKIs and mAbs have shown great promise in cancer cell lines and animal models (7, 8). When combined with high-dose radiation in patients with locoregionally advanced HNSCC, the addition of C225 (EGFR-specific mAb) showed a statistically significant prolongation in overall survival (9). However, limited efficacy has been reported when C225 is administered to HNSCC patients as a single agent (10). Similarly, clinical response to TKIs has also failed to correlate with the promising antitumor effects seen in preclinical studies (11, 12), implicating persistent growth pathways despite blockade of wild-type EGFR (EGFRwt).

The presence of naturally occurring mutations of the EGFR gene in tumors may account for the limited clinical response to EGFR-targeted therapies. Various mutations of the EGFR gene have been described. However, the presence of mutant EGFR (EGFRvIII) and/or EGFRwt has not been systematically evaluated in HNSCC tumors before treatment with EGFR-targeted therapy. Recently, somatic mutations in the tyrosine kinase domain of the EGFR gene have been described in non–small cell lung carcinomas that are associated with increased sensitivity to EGFR-specific TKIs (13, 14). However, in HNSCC, the incidence of these mutations is low and varies according to ethnic origin (1% of Caucasians versus 7% of Asians with HNSCC; refs. 15, 16). A commonly described EGFR mutation is a truncation mutation, EGFR variant III (EGFRvIII). In gliomas, where it has been most extensively studied, EGFRvIII expression correlates with increased tumorigenicity in mouse models (17) and poor prognosis in the clinical setting (18). Moreover, the expression of EGFRvIII is unique to cancer. EGFRvIII has not been observed in normal tissue, but it has been detected in other malignancies, such as non–small cell lung carcinoma, breast cancer, and ovarian carcinoma (19–22). To date, the presence of EGFRvIII has not been investigated in HNSCC.

EGFRvIII harbors an in-frame deletion mutation of exons 2 to 7 spanning the extracellular ligand-binding domain. This deletion produces a truncated 150-kDa protein that is weakly constitutively phosphorylated in a ligand-independent manner (23–25). Ligand-independent activation of EGFRvIII may explain the relative inability of blocking mAbs to down-regulate this receptor.

The present study was undertaken to test the hypothesis that EGFRvIII is expressed in HNSCC and contributes to the tumor phenotype. We examined 33 HNSCC tumors for EGFRvIII and EGFRwt overexpression using reverse transcription-PCR (RT-PCR) and immunohistochemistry. We identified EGFRvIII expression in 42% of HNSCC tumors by immunostaining with an EGFRvIII-specific antibody and RT-PCR using primers specific for this mutant receptor. In addition, the expression of EGFRvIII was only detected in the presence of EGFRwt. Because tumors that express EGFRvIII do not retain EGFRvIII expression when grown in long-term tissue cultures, we stably transfected a HNSCC cell line with an EGFRvIII vector. In vitro and in vivo studies using EGFRvIII-expressing HNSCC cells showed a decreased response to the well-characterized EGFR mAb C225/cetuximab/Erbitux when compared with the parental cells, which overexpress only the EGFRwt. Taken together, these data suggest that EGFRvIII is expressed in HNSCC, where this naturally occurring EGFR mutation may contribute to the limited clinical response to EGFR-targeted therapy.

Cell lines, tumors, and reagents. C225/cetuximab/Erbitux was kindly provided by ImClone Systems, Inc. (New York, NY). NR6 (Swiss 3T3 murine fibroblasts) cells expressing EGFRwt (NR6W) was a generous gift from Dr. Alan Wells (University of Pittsburgh School of Medicine; ref. 26). NR6 cells expressing human EGFRvIII (NR6M) were generated as described previously (23). All cells were maintained in DMEM (Mediatech, Inc., Herndon, VA) with 10% heat-inactivated fetal bovine serum (Invitrogen, Carlsbad, CA) and incubated at 37°C in the presence of 5% to 10% CO₂.

Paired primary HNSCC tumor samples were obtained from 33 HNSCC patients undergoing surgical excision with curative intent at the University of Pittsburgh Medical Center under the auspices of an institutional review board–approved protocol (Table 1). Signed informed consent was obtained from each subject.

Generation of EGFRvIII-transfected cells and cloning the vector control. HNSCC cells (UM-22B; 2 × 10⁶) were transfected with 15 μg expression vector containing the EGFRvIII cDNA (pLEARN-L) or the parental vector (PLV) in the presence of liposomes (Mirus Corp., Madison, WI). The transfected cells were selected for resistance to geneticin (800 μg/mL). After 18 days of selection, 24 individual clones were isolated and screened for EGFRvIII expression via RT-PCR. Four stable EGFRvIII-expressing clones (vIII-1 to vIII-4) were selected and expanded for further study. For confirmation of stable EGFRvIII expression, the RT-PCR assay was repeated for each clone after 14 days of colony expansion. To generate the appropriate vector control plasmid (PLV), the EGFRvIII gene was excised from pLEARN-L via enzymatic digestion with SalI, which cuts at both ends of the EGFRvIII gene. The SalI-digested parental vector was purified by agarose gel electrophoresis, eluted, self-ligated with T4 ligase (New England Biolabs, Beverly, MA), and cloned according to established protocols (27).

RT-PCR analysis and cDNA sequencing of EGFRvIII. Total RNA was isolated from 30 mg snap-frozen tumor tissue or from HNSCC cell lines (5 × 10⁶ cells) using RNeasy Mini kit (Qiagen, Valencia, CA) according to the manufacturer's protocol. RNA concentration and integrity was evaluated by measuring absorbance at 260 and 280 nm. Input RNA was diluted to a final concentration of 1 ng in a final reaction volume of 25 μL. To detect the deleted region of EGFRvIII, standard RT-PCR was done using the One-Step RT-PCR kit (Qiagen) with primers (5′-ATGCGACCCTCCGGGACG-3′ and 5′-ATTCCGTTACACACTTTGCGGC-3′) designed to flank the deletion of exons 2 to 7. The primers were diluted to a final concentration of 0.6 μmol/L and incorporated into the reaction mixture. The remaining PCR reagents were diluted according to the manufacturer's protocol. Reverse transcription was done at 50°C for 30 minutes followed by enzyme inactivation and hot-start PCR at 95°C for 15 minutes. Denaturation, annealing, and extension were done at 94°C, 55°C, and 72°C, respectively, for 1 minute each for a total of 35 cycles. The reaction was completed with an extension period at 72°C for 10 minutes. PCR products were visualized on a 1% agarose gel containing ethidium bromide. For confirmatory cDNA sequencing, the agarose-fractionated amplicon corresponding to the EGFR mutant band was excised from agarose and purified using the Qiagen Gel Extraction kit. The DNA product was sequenced by capillary gel electrophoresis using Big Dye Terminator chemistry version 3.1 kit and the ABI Prism Kinetic Analyzer model 3100 by the DNA Core Facility at the University of Pittsburgh School of Medicine.

Immunohistochemistry of HNSCC tumor sections. Immunohistochemical analysis was done on paraffin-embedded HNSCC tumor tissue with mAb L8A4 (28) specific for the junction of the fusion of exons 1 to 8 found in EGFRvIII and a polyvalent rabbit antiserum raised against the protein product of EGFR exons 2 to 7 (EGFR-1, specific for EGFRwt and unreactive for EGFRvIII; refs. 29, 30). Positive controls for antibody activity were verified by analysis of formalin-fixed cytospins of NR6M (EGFRvIII-expressing cells) and NR6W (EGFRwt-expressing cells). The slides were coded and the intensity and extent of EGFRwt and EGFRvIII expression were independently assessed by a board-certified pathologist and an immunochemist. The intensity of staining was graded on a scale of 0 to 4, with 4 being highest possible intensity. In addition, the extent of tumor staining ranged from a factor of 1 (0-25% of tumor positive), 2 (25-49% of tumor positive), 3 (50-74% of tumor positive), to 4 (>75% of tumor positive).

In vitro growth of EGFRvIII-expressing cells and sensitivity to cisplatin or EGFR inhibition. To determine the growth kinetics of EGFRvIII-expressing HNSCC cells, transfected UM-22B cells (6 × 10³) were seeded onto 35-mm plates. Each cell population was then harvested in duplicate, transferred to a hemocytometer, and counted daily over a period of 6 days. This experiment was done twice. To determine if the sensitivity of HNSCC cells to EGFR inhibition was affected by EGFRvIII expression, vector-transfected and EGFRvIII-expressing HNSCC cells were treated with an EGFR-specific antibody C225 and the viability of each cell population was assessed. HNSCC cells were plated in 24-well plates at a density of 20,000/mL in DMEM containing 10% charcoal-stripped fetal bovine serum. After 24 hours, cells were treated with 200 nmol/L C225. Following a 72-hour treatment with C225, a tetrazolium [3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide] assay for metabolically active cells was carried out according to the manufacturer's protocol. For the in vitro apoptosis studies, HNSCC cells transfected with EGFRvIII cDNA or vector control were treated with 20 μmol/L cisplatin in DMEM containing 10% charcoal-stripped fetal bovine serum. Following treatment with cisplatin, the cells were detached by trypsinization, counted, and pelleted at 1,000 rpm for 5 minutes. Cell pellets were washed once with PBS (pH 7.4) and resuspended in 100 μL Annexin V binding buffer [10 mmol/L HEPES (pH 7.4), 140 mmol/L NaCl, 2.5 mmol/L CaCl₂] as described previously (31). Cells (5 × 10⁵) were incubated with 5 μL Annexin V-Cy3 (BioVision Research Products, Mountain View, CA) at room temperature and in the absence of light for 15 minutes. The stained cell suspension was placed onto a hemocytometer and analyzed under fluorescence microscopy. The ratio (percentage) of apoptotic to total cells (apoptotic + nonapoptotic cells) was calculated for each high-power field. For each treatment, 5 to 10 high-power fields of view were quantitated on each section.

In vivo growth of HNSCC cells expressing EGFRvIII and sensitivity to C225. HNSCC cell line UM-22B-expressing EGFRvIII (vIII-1) or empty vector-transfected parental cells (PLV-1) were cultured in DMEM containing 10% fetal bovine serum and G418 (1.0 mg/mL). Cells were trypsinized and washed thrice with HBSS (Life Technologies, Carlsbad, CA). Cell number and viability of the cells were determined using trypan blue dye exclusion using a hemocytometer. A suspension of 2 × 10⁶ HNSCC cells in 100 μL HBSS was injected s.c. on the right and left flanks of nu/nu athymic nude mice (n = 10; Harlan Sprague-Dawley, Indianapolis, IN). The left flank was injected with vector-transfected cells and the right flank was injected with EGFRvIII-expressing cells. Tumor volumes were measured over 24 days in two dimensions with Vernier calipers. Tumor volumes were calculated using the formula: (length × width²) × 0.5. After 24 days, the mice were sacrificed by cervical dislocation under anesthesia. The tumors were surgically excised and divided into three sections; the first section was fixed in 10% buffered formalin and embedded in paraffin for immunohistochemical analysis with L8A4 or vascular endothelial growth factor (VEGF; Santa Cruz Biotechnology, Santa Cruz, CA), the second section was snap frozen in a dry ice-ethanol bath for gene expression analysis by RT-PCR, and the third section was subjected to cellular lysis for Western blot analysis (below). To investigate the sensitivity to C225 in vivo, xenograft-inoculated animals (n = 4) were treated with i.p. injections of 100 mg/kg C225 thrice weekly for 3 weeks and the tumors were measured biweekly with a Vernier caliper in two dimensions. This experiment was done twice; in the second experiment, the number of vector control cells inoculated was doubled to 4 × 10⁶ cells to ensure equal tumor volumes at the start of treatment. Animal use and care was in strict compliance with institutional guidelines established by the Institutional Animal Care and Use Committee at the University of Pittsburgh.

Western analysis of HNSCC xenografts. Tumors were harvested and cell lysates were prepared and transferred to an Eppendorf tube and centrifuged for 30 minutes at 14,000 rpm. The supernatant was transferred to a clean tube, and protein quantitation was done on the supernatant using Protein Assay Reagent (Bio-Rad Laboratories, Hercules, CA). Proteins (40 μg) were loaded on a 10% SDS-polyacrylamide gel and electrophoresed along with 10 μL prestained broad-range protein marker (Cell Signaling Technology, Beverly, MA). After electrophoresis, proteins were transferred to nitrocellulose filters (Protran, Schleicher & Schuell, Inc., Florham Park, NJ) in a semidry transfer apparatus (Bio-Rad Laboratories). The filters were blocked in 1× PBS with 0.2% Tween 20 and 5% nonfat milk for 1 hour in room temperature. The filters were incubated with primary antibody and subsequently were washed with Blotto solution [50 mmol/L Tris (pH 7.4), dry milk powder, 0.9% NaCl, 0.5% Tween 20] thrice for 15 minutes. The filters were then incubated with secondary antibody for 1 hour and washed with Blotto solution thrice for 15 minutes. The filters were quickly rinsed with rinsing solution, and the blot was developed with Luminol Reagent (Santa Cruz Biotechnology) by autoradiography. Antibodies used for blotting included VEGF and β-actin (Oncogene Research Products, Boston, MA) to show equal loading.

Statistics. Fisher's exact test was used to show correlation between EGFRvIII expression and EGFRwt in HNSCC tumor tissues. Wilcoxon signed rank test was used for the analysis of growth rates in HNSCC xenografts, comparing EGFRvIII and vector controls. Wilcoxon-Mann-Whitney test was used for the analysis of VEGF expression in HNSCC xenografts, comparing EGFRvIII and vector controls. All tests were exact and two-tailed. Statistical analysis was done on the StatExact v.6.1 software (Cytel Software Corp., Cambridge, MA).

EGFRvIII is expressed in HNSCC tumors. Cumulative evidence suggests that EGFR signaling contributes to the pathogenesis of HNSCC. However, the role of EGFRvIII has not been examined previously in HNSCC tissues or cell line models. We first sought to determine if EGFRvIII, a mutation correlated with increased tumorigenicity in glioblastoma, is present in HNSCC tumors. Thirty-three HNSCC tumors were stained using a EGFRwt-specific antibody (anti-exons 2-7) and an EGFRvIII-specific antibody (L8A4) to determine prevalence of the EGFRvIII and EGFRwt expression. Both antibodies were mutually exclusive, as they have been shown not to cross-react with their competing target proteins (29, 30). This was confirmed by immunocytochemical analysis on the control cells, NR6W and NR6M (data not shown). Patient demographic information used in this study and their corresponding clinicopathologic characteristics are summarized in Table 1. EGFRvIII expression was shown in 14 of 33 (42%) of HNSCC tumors (Table 2). Intriguingly, EGFRvIII was shown exclusively in tumors that also stained positive for EGFRwt expression (14 of 33; P = 0.012). Twelve of 33 (36%) tumors stained positively only for the EGFRwt. EGFRvIII was only expressed by tumor cells and not by submucosa or normal adjacent mucosa.

To determine the subcellular localization of EGFRvIII, formalin-fixed tissues stained with specific antibodies to either EGFRvIII (L8A4) or EGFRwt (EGFR-1) were examined. Figure 1A shows serial sections from a representative HNSCC tumor sample. Figure 1A (I) was stained with L8A4 and Fig. 1A (II) was stained with EGFR-1 (anti-exons 2-7, the ligand-binding segment of EGFR known to be absent in the EGFRvIII gene). Staining for both receptors was specific to the HNSCC tumor nests and was not reactive to the adjacent normal tissue. At ×66 magnification, EGFRvIII staining was homogeneously diffuse with mildly higher intensity in the peripheral regions of the tumor nests (Fig. 1A, I), whereas EGFRwt staining showed areas of focal intensity in addition to homogeneous distribution (Fig. 1A, II). At ×132 magnification (Fig. 1A, III), EGFRvIII expression showed diffuse cytoplasmic staining in addition to strong membranous staining. Negative primary reagent controls were provided by murine IgG at the same concentration as L8A4 and purified normal rabbit IgG at the same concentration as anti-exons 2-7 (Fig. 1A, IV). There was no evidence of EGFRvIII expression in the adjacent normal mucosa or in the submucosa.

To confirm the protein expression data, RT-PCR was done using primers that flank exons 2 to 7 designed to synthesize two discrete amplicons distinguishing the expression of EGFRvIII from the larger EGFRwt transcript. Figure 1B shows mRNA expression of the mutant receptor in one of three representative tumor samples where coexpression with EGFRwt is evident, consistent with the protein expression data. Moreover, DNA sequencing of the PCR products confirmed the identity of the 5′ regions of EGFRvIII and EGFRwt (data not shown). These results showed that EGFRvIII is present in nearly half of HNSCC tumors.

EGFRvIII contributes to increased HNSCC cell proliferation and tumor growth. A universal feature among cells undergoing malignant transformation is the increase in cellular growth rates. Overexpression of EGFR has been associated with increased tumor growth and metastasis (32, 33). Therefore, a constitutively active form of this receptor, EGFRvIII, could contribute to increased growth potential. Human tumors that express EGFRvIII do not retain expression of the mutant receptor when grown in vitro. To investigate the contribution of EGFRvIII in HNSCC proliferation, we established an in vitro model by generating a stable cell line that constitutively expresses this mutant receptor. Twenty-four clones of a total of 2 × 10⁶ transfected cells (HNSCC cell line, UM-22B) were selected, all of which initially scored positive for EGFRvIII expression by RT-PCR (data not shown). Four of the most stable clones constitutively expressing EGFRvIII were selected for further study. RT-PCR analysis of two representative clones confirmed stable expression of EGFRvIII as well as the absence of EGFRvIII expression in the vector-transfected control cells (Fig. 2A). When these paired cells were grown in culture, the population doubling time was shortened in EGFRvIII-expressing cells and was immediately evident by day 3. By day 5, the number of EGFRvIII-expressing HNSCC cells was greater than doubled when compared with the vector-transfected control cell population (Fig. 2B).

To investigate the mitogenic effect of EGFRvIII expression in vivo, athymic nude mice were inoculated with HNSCC cells expressing EGFRvIII on the right flank and vector-transfected control cells on the left flank over a period of 24 days (Fig. 2C). The tumor volumes were measured in 10 athymic nude mice. By day 24, the median tumor volume from HNSCC xenografts expressing EGFRvIII measured >7-fold greater in size in comparison with the contralateral HNSCC xenografts derived from vector-transfected control cells (P < 0.004 for days 5, 9, 16, 20, and 24 and P < 0.032 for day 12). The median tumor size difference between vIII-1 and PLV-1 was calculated from individual mice and was increased to 751 mm³ by day 24 (Fig. 2D). The tumors were harvested at the conclusion of the experiment and stained for expression of EGFRvIII by immunohistochemistry. Figure 2E depicts a representative slide showing the expression of the EGFRvIII protein in vivo. This was further validated at the gene expression level by RT-PCR, showing the coexpression of EGFRwt (also seen in the control xenografts) in addition to EGFRvIII mRNA expression (Fig. 2F).

HNSCC tumors expressing EGFRvIII appeared more erythematous by gross observation, characteristic in appearance for tumor angiogenesis (data not shown). Previous studies in astrocytoma cells expressing EGFRvIII have shown increases in VEGF secretion through the activation of the Ras oncogene (34). To investigate the possible role of VEGF expression in EGFRvIII-mediated tumorigenesis in head and neck cancer, xenografts derived from EGFRvIII-transfected HNSCC cells were stained for VEGF (Fig. 3A). Quantitation of VEGF expression by densitometry revealed that the level of VEGF was nearly doubled in HNSCC xenografts expressing the EGFRvIII (Fig. 3B; P = 0.04). This observation was confirmed by Western blot analysis showing increased VEGF expression in EGFRvIII-transfected xenografts from three representative nude mice (Fig. 3C). These results suggest that the expression of EGFRvIII in HNSCC contributes to increased growth in vitro and in vivo and is associated with elevated expression of VEGF.

EGFRvIII abrogates antitumor responses to chemotherapy and EGFR targeting. Chemotherapy (generally in combination with radiation therapy) is the standard approach for the treatment of patients with recurrent and/or metastatic HNSCC and, more recently, for patients with advanced locoregional disease. The benefits of cisplatin were shown ∼20 years ago (35) and this remains a common agent either alone or in combination with other chemotherapy drugs. EGFRvIII has been reported to confer resistance to cisplatin in glioma cells (36). To investigate the role of EGFRvIII in response to cisplatin, HNSCC cells expressing EGFRvIII- or vector-transfected controls were treated with cisplatin followed by staining for apoptotic cells with Annexin 5-Cy3 flourophor. Cisplatin treatment induced apoptosis in 43% of the vector-transfected controls. However, the percentage of cells undergoing apoptosis in HNSCC cells expressing EGFRvIII was reduced by 2-fold (19.6%; P < 0.001; Fig. 4). EGFRvIII-expressing cells are, therefore, more resistant to apoptosis when treated with a potent cytotoxic agent, such as cisplatin.

Several strategies designed to block EGFR activity are currently under clinical investigation or have been approved by the Food and Drug Administration for use in selected cancer patients. These include receptor-blocking α-EGFR antibodies, TKIs, EGFR ligand-conjugated toxins, and EGFR antisense gene therapy (37). Among these agents, the monoclonal α-EGFR antibodies (C225/cetuximab/Erbitux) and the small-molecule TKIs (ZD1839/gefitinib/Iresssa or OSI774/erlotinib/Tarceva) are the most extensively studied. Clinical trials of these agents administered as monotherapy showed limited antitumor effects (38). We hypothesized, therefore, that the aberrant expression of EGFRvIII may contribute to the limited response of HNSCC to EGFR targeting. To investigate the contribution of EGFRvIII to EGFR-blocking strategies, HNSCC cells expressing EGFRvIII- or vector-transfected controls were treated with the chimeric human-murine mAb to EGFR (C225) followed by a tetrazolium-based [3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide] viability assay to monitor cell death. The percentage of killing from C225 treatment was reduced in EGFRvIII cells (33%) compared with untransfected parental cells (55%) or vector-transfected controls (54%; Fig. 5A). In contrast to the relative resistance to treatment with an α-EGFR antibody, EGFRvIII-expressing cells and vector-transfected controls were equally susceptible to EGFR TKI treatment (data not shown). This is consistent with the fact that both EGFRvIII and EGFRwt contain an intact tyrosine kinase domain.

The effects of C225 on EGFRvIII-expressing HNSCC tumor growth were also evaluated in vivo. Athymic mice were inoculated s.c. with HNSCC cells expressing EGFRvIII- and vector-transfected control cells and subsequently treated with C225 (100 mg/kg, three injections weekly). The tumor volumes were measured in four athymic nude mice over a period of 3 weeks. By day 22, the mean tumor volume from HNSCC xenografts expressing EGFRvIII nearly doubled in size when compared with the vector control tumors (235 versus 128 mm³, respectively; P < 0.04; data not shown). Due to the faster growth rates observed in the EGFRvIII-expressing xenografts, this experiment was done twice. In the second experiment, athymic nude mice were inoculated with twice the number of vector control cells (4 × 10⁶ versus 2 × 10⁶ cells) to ensure equal tumor volumes at the start of the experiment. By day 24, the mean tumor volume from the EGFRvIII-expressing xenografts more than tripled in size when compared with the vector control tumors (54 versus 171 mm³, respectively; P < 0.008). The median tumor size difference between vIII-1 and PLV-1 was also calculated from each individual mice and was increased to 117 mm³ by day 24 (Fig. 5B). Taking together, these results suggest that the limited efficacy of chemotherapy and specific agents that inhibit the ligand-binding activity of EGFRwt in a subset of HNSCC patients may be due to the contribution of an aberrant, ligand-independent EGFRvIII expression.

The EGFR has long been associated with increased tumorigenesis in cancer models where expression levels in human tumors correlate with poor prognosis in the clinical setting. As a growth-promoting molecule, overexpression of EGFR leads to up-regulation of downstream signaling pathways leading to increased cell proliferation, invasion, and metastasis (39). EGFR is up-regulated in >90% of HNSCC tumors where overexpression of this receptor as well as its autocrine ligand, transforming growth factor-α, are independent indicators of poor prognosis (4).

EGFR-driven carcinogenesis has been investigated in several tumor systems. Genomic analysis of EGFR amplification in glioblastomas led to the discovery of tumor-specific EGFR mutations (25, 40). The most common mutant found is EGFRvIII, a ligand-independent, constitutively active receptor that was subsequently shown to correlate with poor prognosis in glioblastoma patients (41). A novel truncation mutation of EGFR was detected in malignant oral keratinocytes as well as oral SCC (42, 43). The presence of EGFRvIII has not been reported in HNSCC.

In this study, we show EGFRvIII expression in 42% of HNSCC by immunohistochemistry. EGFRvIII was also detected at the mRNA level by RT-PCR. Sequencing of the RT-PCR product confirmed the characteristic mRNA deletion of exons 2 to 7 corresponding to the EGFRvIII. This frequency of EGFRvIII expression by immunohistochemistry is similar to that seen in other tumors types, including glioblastomas, anaplastic astrocytomas, non–small cell lung carcinoma, breast cancers, and ovarian carcinomas (19–22, 44). More recent mutational analysis of EGFR in lung cancers has shown that EGFRvIII is uncommon. In this study, EGFRvIII mutation was present in 5% (3 of 56) of lung SCC but was not present in lung adenocarcinomas (0 of 123; ref. 45). To our knowledge, this study is the first report of EGFRvIII expression in HNSCC.

We find the extensive and homogeneous expression of EGFRvIII in head and neck carcinoma similar to that found in glioblastomas. Demonstration of EGFRvIII in HNSCC is not entirely unexpected given its presence in other cancers of ectodermal cell origin. In contrast to glioblastomas, where 40% of the tumors overexpress EGFR at the genomic level by gene amplification (25), increased expression of EGFR in HNSCC is accompanied by gene amplification in ∼10% of cases (46). HNSCC cells expressing elevated levels of EGFR have been shown to proliferate, invade, and metastasize to a greater extent than those expressing lower EGFR levels. We hypothesize that EGFR overexpression leads to rapid cell proliferation resulting in late-stage mutations, such as EGFRvIII. In support of this hypothesis, we found that EGFRvIII expression in HNSCC is only evident in the presence of EGFRwt overexpression. In glioblastomas, EGFR mutations are generally not detected in tumors without EGFRwt amplification (47). It is also possible that, given our relatively small cohort of 33 HNSCC patients, analysis of additional cases may reveal a tumor uniquely expressing EGFRvIII in the absence of EGFRwt.

Functional characterization studies of EGFRvIII have shown that several downstream modulators confer increased tumorigenicity to EGFRvIII-expressing cells. In vitro studies suggest that EGFRvIII-expressing cells may be more chemotherapy and radiotherapy resistant as well as display a more aggressive phenotype (36, 48–50). In contrast to EGFRwt, EGFRvIII does not activate the Ras-Raf-MEK pathway and instead seems to preferentially activate the phosphatidylinositol 3-kinase pathway (51). It has recently been shown that astrocytic cells expressing EGFRvIII may be radioresistant in part due to increased phosphatidylinositol 3-kinase signaling (52). Other studies have shown that EGFRvIII induces metalloproteinases and extracellular matrix components known to be involved in tumor invasion (50). Nagane et al. showed down-regulation of Bcl-X_L as well as decreased apoptosis in response to cisplatin treatment (36, 53). Moreover, Holland et al. showed that all of the tumors generated in their in vivo gene transfer model for gliomagenesis required EGFRvIII expression (54).

Because of the relatively high frequency of EGFR overexpression and its correlation with decreased survival in HNSCC, this receptor has been investigated as a target for therapeutic intervention in this malignancy. EGFR-targeted therapies, such as TKIs, EGFR-specific mAbs, and ligand-linked toxins, have successfully reduced HNSCC tumor burden in vivo (55). However, clinical trials evaluating these reagents as monotherapy have achieved only modest results (56). This may be because patients are not routinely selected for treatment based on EGFR overexpression and/or activation status. Consequently, those patients who do not show increased EGFR expression or activation may not be amenable to EGFR-targeted therapies. Alternatively, the modest clinical response may be due to the presence of EGFRvIII.

Our study shows that the expression of EGFRvIII abrogates antitumor responses to EGFR-specific mAb, C225. C225 binds to the extracellular ligand-binding domain of EGFRwt with an affinity equal to its ligand and blocks its activation (57). Therefore, it is not surprising that EGFRvIII expression could circumvent the effects of C225, given the fact that this truncated mutant receptor is ligand independent and constitutively active. Although there is no conclusive evidence of C225 binding of EGFRvIII, the development of EGFR-targeted C225 immunoliposomes that bind and internalize in tumor cells, which overexpress EGFRvIII, have been described in vitro (58). However, in a follow-up study by the same authors, specific binding of C225 to EGFRvIII in vivo could not be determined because the EGFRvIII-expressing xenografts also expressed EGFRwt (59). The Food and Drug Administration has recently granted priority review for the approval of C225 as monotherapy in HNSCC patients with recurrent and/or metastatic disease where prior platinum-based chemotherapy has failed or in combination with radiation for locally or regionally advanced disease (fda.gov press, released March 1, 2006).

EGFRvIII is an attractive candidate target for therapeutic intervention because, unlike EGFRwt, EGFRvIII is not found in normal tissue. The in-frame recombination of exons 1 and 8 produces a novel glycine epitope for EGFRvIII-specific antibodies. Antibody-linked chemotherapies would target tumor tissue with theoretically little or no toxicity to normal tissue. Several data support this hypothesis. An EGFRvIII peptide vaccine increased survival time in mice with established i.c. tumors (60). Systemic administration of an EGFRvIII-specific mAb (806) to nude mice bearing i.c. glioblastoma xenografts led to tumor shrinkage and increased survival (61).

In conclusion, EGFRvIII is expressed in more than one third of HNSCC tumors. This mutant receptor is a tumor-specific, constitutively active, ligand-independent EGFR mutant that may contribute to the initiation, promotion, or progression to the malignant phenotype of HNSCC. Expression of EGFRvIII, at least in part, abrogates antitumor responses to chemotherapy and EGFR targeting in vitro and in vivo. Functional studies in HNSCC models are required to further elucidate the role of EGFRvIII in HNSCC pathogenesis and its potential as a therapeutic target.