Abstract
In a series of 111 patients with squamous cell carcinoma (SCC), we used immunohistochemistry to examine the expression levels of four epidermal growth factor receptor (EGFR) family members (EGFR, HER-2/neu, HER-3, and HER-4). Expression of the EGFR members was not significantly associated with tumor size. However, their expressions (except for HER-4) were significantly associated with the presence of lymph node metastasis, and all of them were significantly associated with distant metastasis. We further examined the association between the expression levels of the EGFR members and the survival rates in 47 oral SCC patients whose detailed clinical follow-ups were available. The expression of all EGFR members was significantly associated with shortened patient survival, and the association was strongest for HER-2/neu. Furthermore, the combination of HER-2, HER-3, and EGFR but not HER-4 significantly improved the predicting power. The expression level of HER-2/neu was significantly correlated with that of EGFR or HER-3. Similar coexpression patterns were also observed in three oral SCC cell lines studied, but not in four other head and neck SCC cell lines. Taken together, these results indicated that expression levels of EGFR, HER-2/neu, and HER-3 may help predict the outcome of patients with oral SCC.
INTRODUCTION
The molecular carcinogenesis of head and neck SCC4 is not yet clear. Several reports have shown the high frequency of p53 gene alterations ,(1, 2, 3) . Amplification of cellular and nuclear oncogenes such as EGFR, c-myc, and bcl-1 have also been reported (4, 5, 6) . A better understanding of the molecular mechanisms and identification of potential oncogenes in the oral cancer may provide more accurate and useful prognostic markers and eventually more effective treatment. Several studies and ours have shown the possible role of some of the EGFR family members in the oral SCC. The EGFR family includes EGFR, HER-2/neu (c-erbB-2), HER-3 (c-erbB-3), and HER-4 (c-erbB-4; Refs. 7, 8, 9 ). The first member of the family is EGFR, which is a transmembrane tyrosine kinase receptor. EGFR is expressed throughout development and in a variety of cell types. The overexpression of EGFR in head and neck cancer was found to significantly correlate with tumor size and stage (10) . Several ligands, such as TGF-α and amphiregulin, can bind to the 170-kDa cell-surface EGFR, resulting in activation of its intrinsic kinase activity (11 , 12) . EGFR can also interact with other members of the family by heterodimerization (13) .
The HER-2/neu gene encodes a transmembrane protein of 185 kDa. HER-2/neu has an extensive sequence homology to EGFR (14 , 15) . Like EGFR, HER-2/neu has intrinsic tyrosine kinase activity (16, 17, 18, 19) and can interact with many different cellular proteins, such as Shc, PLC-γ, and GTPase-activating protein, thereby mediating the signal transduction pathway (20 , 21) . It can also form heterodimers with the other members of the EGFR family. The overexpression of HER-2/neu was shown to increase metastatic potential by promoting the multiple adhesion and invasion steps of the metastatic cascade (22) , suggesting that this gene may play an important role in carcinogenesis. The overexpression of HER-2/neu has been found to correlate with poor survival of patients with breast, ovarian, lung, stomach cancers, and oral SCC (2 , 23, 24, 25, 26, 27) .
The HER-3 gene is the third member of the EGFR family (12 , 28 , 29) . Its gene product is a transmembrane glycoprotein (Mr 180) with a close structural similarity to EGFR and HER-2/neu. The HER-3 protein has been found in normal adult and fetal tissues (29) and in a number of tumor cell lines and primary tumors (30, 31, 32) . One group has demonstrated that high HER-3 protein expression is associated with shorter survival of patients with non-small cell lung carcinoma (32) . These results suggest that HER-3 may play an important role in human neoplasia. HER-3 does not have an intrinsic tyrosine kinase activity. It has been shown that HER-3 can be transphosphorylated by EGFR, resulting in PI3 kinase recruitment (33) . The EGFR/HER-3 chimeric receptor, however, did not couple to PLCγ or GTPase-activating protein, thereby distinguishing the HER-3 signaling pathway from that of other members of the EGFR family (33) . HER-2/neu and HER-3 proteins form heterodimers, and the coexpression of these two proteins in NIH3T3 cells significantly increased the transforming activity (11) . This suggests that the coexpression of these two proteins may be important for human neoplasia and that the detection of both HER-2/neu and HER-3 may have more clinical and prognostic significance than the detection of either protein alone.
HER-4, the fourth member of the family, encodes a 180-kDa transmembrane tyrosine kinase (34) . Its extracellular domain is very similar to that of HER-3, whereas its cytoplasmic domain is highly homologous to EGFR and HER-2/neu. Various human tissues and some human mammary carcinoma cell lines have HER-4 transcripts (8) , and a recent study has shown that the HER-4 protein is overexpressed in papillary thyroid carcinoma (35) . However, the clinical significance of HER-4 overexpression is not yet clear.
In our previous report (27) , we showed that there is a strong correlation between HER-2/neu expression and shortened survival in patients with oral SCC. The complex signaling interactions among EGFR members has been implicated in the transformation of cells in experimental models. It is therefore important to characterize the expression pattern of the EGFR family members in cancer. It will provide further insight to the possible molecular mechanisms of tumorigenesis, and it will also provide a better therapeutic target. In the present study, we characterized the expression of the EGFR family members in oral tumor specimens as well as in seven head and neck SCC cell lines. We then determined the relationship between protein expression and survival in patients with the oral SCC. For the latter, we also investigated the association between overexpression of the EGFR family members and the clinicopathological characteristics of the tumors. The information will provide us with a better understanding of the molecular derangement in oral SCC and the potential prognostic significance of EGFR family members.
MATERIALS AND METHODS
Specimens.
Specimens of primary oral SCC from 111 patients were obtained from the Department of Oral Pathology, Ninth People’s Hospital, Shanghai Second Medical University, Shanghai, People’s Republic of China. These patients (74 male and 37 female) underwent surgical treatment in the hospital between 1989 and 1991. Their ages ranged from 13 to 76 years (median, 51 years). The primary tumors were graded by the WHO classification of histological differentiation. Eighty-two cases were defined as grade I, 24 as grade II, and 5 as grade III. The tissues involved by tumor included tongue (63 cases), gum (16 cases), cheek (10 cases), floor of the mouth (9 cases), palate (5 cases), mucosa of lip (2 cases), mandible (2 cases), mucosa of mandibular molar areas (3 cases), and maxilla (1 case). The specimens were routinely fixed in 10% formalin buffer, embedded in paraffin, sectioned serially every 5 μm, and stained with H&E. Data from all 111 patients were analyzed for association among age, sex, tumor histological grade, tumor size, nodal stage, metastasis, and expression of EGFR family member (EGFR, HER-2/neu, HER-3, HER-4). We have used the modified TNM staging system of the International Union against Cancer (36) . Accurate and detailed clinical information on 47 of 111 patients was available and was used to analyze the association between overall survival rate and the expression of EGFR family members. The patients received either chemotherapy or radiotherapy or both after their surgical treatment. Since the surgical treatment, 10 patients have died of the primary tumor, and 1 has died of metastasis of the primary tumor to the lung.
Immunohistochemical Staining.
The immunoperoxidase staining method was modified from the avidin-biotin complex technique described by Hsu and Raine (37) and Hsu et al. (38) . Briefly, tissue sections were deparaffinized and dehydrated in a graded series of alcohol. They were then digested in 0.05% trypsin for 15 min, blocked in 0.3% hydrogen peroxide in methanol for 15 min, and then treated with 1% (v/v) normal horse serum and normal goat serum for 30 min. The slides were incubated for 3 h at room temperature with EGFR (1005) SC-003 polyclonal antibody (diluted 1:200; Santa Cruz Biotechnology, Inc., Santa Cruz, California), c-neu (Ab3) monoclonal antibody (1:1600; Oncogene Research Products, Cambridge, MA), c-erb-B3 (RTJ.2) monoclonal antibody (1:40), or c-erb-B4 (c-18) polyclonal antibody (1:400; Santa Cruz Biotechnology, Inc.). For detecting the specificity of the EGFR antibody, we also used a specific EGFR-blocking peptide. We observed that the EGFR immunostaining could be totally blocked by the peptide (data not shown). Also for HER-2/neu, other HER-2/neu-specific antibody was used (NCL-CB 11, Vector Laboratories, Inc.), and we also observed for the same pattern of immunostaining (data not shown). For HER-3 and HER-4, other papers have described the use of the antibodies (39) . After extensive washing with PBS, the slides were incubated for another 30 min at room temperature with biotinylated horse antimouse IgG or biotinylated goat antirabbit IgG antibody diluted 1:200 in PBS. The slides were subsequently incubated for 60 min at room temperature with the avidin-biotin-peroxidase complex diluted 1:100 in PBS. The peroxidase-catalyzed product was visualized with 0.125% aminoethylcarbazole chromogen stock solution (Sigma Chemical Co.). Between steps, the slides were rinsed for 2 min in PBS (pH 7.6) three times. After light counterstaining with Mayer’s modified hematoxylin (Sigma Chemical Co.), the slides were mounted. Negative controls in which PBS was used instead of the primary antibody were run with each batch of staining. A previously identified strongly staining tumor tissue section was used as a positive control. Interassay and intra-assay consistency was maintained by including these positive and negative controls with each batch of slides stained. The prepared slides were examined by light microscopy. Those tumor cells that were immunostained with red granules in the cytoplasm and cellular membrane were considered positive, and those cells without any immunostaining were considered negative. All slides were reviewed independently by two pathologists. Discrepancies in staining analysis were about 5–8%. These cases were reexamined, and a consensus was reached. Both cytoplasmic and membrane stainings could be observed. The immunoreactivity of EGFR family members was ranked into four groups according to the percentage and staining intensity of positively stained tumor cells, including cytoplasmic and membrane staining: (a) high (+++; ≥50% cells stained), (b) intermediate (++; 20–49% cells stained), (c) low (+; up to 20% cells stained), and (d) none (no staining).
Cell Lines.
We have previously described these seven head and neck SCC cell lines (27) .
Western Blot Analyses and Coimmunoprecipitation.
Western blot analyses were previously described (27 , 40 , 41) . A431 cells, which express a high level of EGFR protein, were used as a positive control for EGFR detection. MDA-MB-453 cells, which express a high amount of HER-3, were used as a positive control for HER-3 Western blot analysis. For coimmunoprecipitation, cells were grown to 90% confluence. The cell lysate preparation and immunoprecipitation were described previously (27 , 40 , 41) . For Western blot analyses, EGFR polyclonal antibody (Upstate Biotech Inc.) and HER-3 (RTJ.2) monoclonal antibody (Neomarker) were used (39) .
Statistical Analyses.
The two-sided Fisher’s exact test was used for analysis of contingency tables, with P < 0.05 as the criterion for statistical significance. Survival curves were calculated by the method of Kaplan and Meier, and differences between curves were analyzed by the log-rank test. Cox multivariate regression analysis was also used to examine the significance of the EGFR family members in predicting the outcome of the disease while including other factors (lymph node, grade, and metastasis). The partial P is calculated based on the fact that two times the log likelihood ratio approximates χ2 value. For example, the combination of HER-2 and HER-3: χ2 ∼ 2× {(log likelihood ratio of HER-2 + log likelihood ratio of HER-3)- log likelihood ratio of HER-2}. From the χ2 value, we can obtain the corresponding partial P. Correlations of coexpression between EGFR family members were analyzed by Spearman rank correlation.
RESULTS
Correlation of EGFR, HER-2/neu, and HER-3 Expression and Poor Survival in 47 Patients with Oral SCC.
Immunostaining of the 111 primary oral SCCs showed wide variation in the levels of EGFR expression in the specimens (Table 1)<$REFLINK> . Representative examples of the staining categories for each of the four EGFR family members in cancer tissue and normal oral tissues are shown in Figs. 1<$REFLINK> and 2<$REFLINK> .
Representative immunohistochemical staining for the expression of four EGFR family members in oral SCC (−, no expression; +, low expression; ++, intermediate expression; +++, high expression).
Representative immunohistochemical staining for the expression of four EGFR family members in normal oral tissues as compared with the carcinoma tissues.
Immunoreactivity of EGFR family members in 111 oral SCCs
There are no significant associations between the expression levels of the EGFR family members and the age, sex, and histological grade (P > 0.05) except that EGFR expression level is significantly associated with age (P < 0.001). The relationship between TNM staging and expression of EGFR family members is shown in Table 2<$REFLINK> . The expression levels of EGFR, HER-2/neu, HER-3, and HER-4 were significantly associated with distant metastasis. All EGFR family members except HER-4 showed a significant association with nodal stage, with P > 0.06 for HER-4. However, there was no association between the expression levels of any EGFR member and tumor size.
Relationship between TNM staging and EGFR families expressing in 111 cases of oral SCCS
The significance of the expression of the EGFR family members in human oral SCC was studied retrospectively in 47 patients that had been followed up since their surgical treatment. Table 3<$REFLINK> provides clinical data for these patients. The tumor specimens were observed under light microscopy analysis. Oncoprotein expression levels were ranked according to the percentage and staining intensity of positively stained tumor cells. They were grouped into three categories: low or no expression, moderate expression, and high expression. The survival curves with respect to different EGFR members were calculated by the Kaplan-Meier method, and the significance between curves was analyzed by log-rank tests. The protein expression levels of EGFR, HER-2/neu, HER-3, and HER-4 and the survival rates in these 47 oral SCC cases were significantly associated as shown in Fig. 3<$REFLINK> (log-rank tests of all four EGFR family members, P < 0.001).
Survival curves for 47 patients expressing EGFR (A), HER-2/neu (B), HER-3 (C), and HER-4 (D) in 47 patients with oral SCC. The log-rank tests of all four EGFR family members, P < 0.0001.
Characteristics of 47 oral SCCs
HER-2/neu and Outcome of Oral SCC.
To examine the significance of a combination of EGFR family members in predicting the outcome of the disease, we performed Cox multivariate analyses on the data from 47 oral SCC patients. We ranked the relative significance of each member based on their log likelihood ratio. As shown in Table 4<$REFLINK> , HER-2/neu was ranked first in predicting the outcome of these 47 cases (log likelihood ratio = −19.20; P < 0.001), followed by HER-4 (log likelihood ratio = −19.54; P < 0.001), HER-3 (log likelihood ratio = −20.72; P < 0.001), and EGFR (log likelihood ratio = −25.75; P < 0.001). We then examined whether the coexpression of these EGFR family members would improve the predicting power as compared with the most significant single factor, HER-2/neu. We first examined the effect of combining HER-4, HER-3, or EGFR with HER-2/neu on predicting the outcome of the patients. The combination of HER-2/neu and HER-3 was the most significant pair among those of HER-2/neu and HER-4 or HER-2/neu and EGFR (log likelihood ratios are −13.94, −16.84, and −18.35, respectively). Furthermore, their combination is better than the expression of either HER-2/neu or HER-3 alone (log likelihood ratio = −13.94; P < 0.005; Table 3<$REFLINK> ). When EGFR was combined with HER-2/neu and HER-3, the predicting power was significantly improved (log likelihood ratio = −11.99; P < 0.05), whereas the addition of HER-4 did not improve the predicting power (log likelihood ratio = −13.82; P > 0.05). Further addition of HER-4 with HER-2/neu, HER-3, and EGFR does not significantly improve the predicting power (log likelihood ratio = −11.66; P > 0.05). The results indicate that the combination of EGFR, HER-2/neu, and HER-3 can significantly improve the predicting power better than any individual members can.
Results of univariate and multivariate analyses using the Cox regression analysis model
Coexpression of EGFR, HER-2/neu, and HER-3 in Oral SCCs.
Because the combination of EGFR, HER-2/neu, and HER-3, but not HER-4 shows significant improvement in predicting the survival of oral SCC patients, we therefore focused on investigating the relationship between the expression levels of HER-2/neu and those of EGFR or HER-3 in the 111 cases of oral SCC. The expression level of HER-2/neu and that of EGFR (r = 0.62, P < 0.05) or HER-3 (r = 0.57, P < 0.05) were significantly correlated, with the EGFR or HER-3 expression level increasing with a higher HER-2/neu expression level (Fig. 4, A and B)<$REFLINK> . The similar correlation is also observed between EGFR and HER-3 (Fig. 4C<$REFLINK> ; r = 0.58; P < 0.05). These findings suggest that the coexpression of EGFR, HER-2/neu, and HER-3 may be a characteristic of oral SCC.
Correlation of coexpression between HER-2/neu and EGFR (A; r = 0.62; P < 0.05), between HER-2/neu and HER-3 (B; r = 0.59; P < 0.05), and between EGFR and HER-3 (C; r = 0.58; P < 0.05) in 111 patients with oral SCC.
Expression of EGFR, HER-2/neu, and HER-3 in Seven Head and Neck SCC Cell Lines.
We used Western blot analysis to examine head and neck SCC cell lines for the expression of EGFR, HER-2/neu, and HER-3. All seven cell lines expressed moderate to high levels of EGFR, but all of the three oral SCC cell lines (Tu 138, 686/LN-1, and Tu 167) expressed moderate levels of HER-3 (Fig. 5)<$REFLINK> , and the other four cell lines that were established from other head and neck regions showed either low or no expression of HER-3 except for the cell line 1483. We previously showed that these three oral cell lines also expressed relatively high levels of HER-2/neu compared with the other four head and neck SCC cell lines studied (27) . These results suggest that the expression of EGFR is a characteristic of head and neck SCC, but that HER-2/neu and HER-3 are more common characteristics of oral SCC. Their coexpression pattern (Fig. 4)<$REFLINK> and the stronger predicting power by their combination suggest that HER-2/neu and HER-3 may cooperate in tumorigenesis. We therefore examined whether HER-2/neu can also form heterodimers with HER-3 in the oral SCC. We performed coimmunoprecipitation with the anti-HER-2/neu antibody and then immunoblotted with either anti-HER-2/neu or HER-3. Heterodimerization of HER-2/neu and HER-3 was observed in Tu 138 cells, which coexpress HER-2/neu and HER-3, but not in Tu 177 cells, which express low or no HER-2/neu and HER-3 (Fig. 6)<$REFLINK> . These data indicate that heterodimerization between HER-2/neu and HER-3 can also occur in oral SCC.
Western blot analyses of EGFR and HER-3 expression in seven head and neck SCC cell lines. A, A431 cells, which express a high level of EGFR, were used as a positive control for EGFR protein detection in the cell lines. B, MDA-MB-453 human breast cancer cells, which express a high level of HER-3 expression, were used as a positive control for HER-3 detection in the cell lines.
Heterodimer formation in an oral cancer cell line coexpressing HER-2/neu and HER-3. Coimmunoprecipitation was performed as described in “Materials and Methods.” Tu 138 cells, which coexpress HER-2/neu and HER-3, were used to detect heterodimer formation. Tu 177 cells, which express low or no HER-2/neu and HER-3, were used as a control.
DISCUSSION
The present study examined the expression of four EGFR family members in oral SCCs and their relationship with TNM staging, patient survival, and other EGFR family members. We demonstrated that all four EGFR members can be expressed in oral SCC patient samples. We further showed that there was a significant association of the expression levels of EGFR, HER-2/neu, HER-3, and HER-4 with shorter survival. By the Cox regression analyses (Table 3)<$REFLINK> , HER-2/neu was found to be the most significant single factor in predicting disease outcome followed by HER-4, HER-3, and then EGFR in our model. The combination of EGFR, HER-2/neu, and HER-3 but not HER-4 was more significant than any individual member in predicting the outcome.
In our previous report on oral SCC, we have discussed the findings of cytoplasmic and membrane staining of HER-2/neu (27) . The validity of cytoplasmic staining of HER-2/neu was questioned. Some argue that HER-2/neu cytoplasmic staining could be an artifact or a nonspecific finding (42, 43, 44) . However, reports have shown that the HER-2/neu cytoplasmic staining in breast tumors correlates with HER-2 amplification (45) . Apart from membrane staining, EGFR and HER-2/neu can also be detected in the nucleus (46, 47, 48) . We used another HER-2/neu specific antibody, which demonstrated a similar cytoplasmic and membrane staining. Taken together, cytoplasmic staining is as important as membrane staining and is a specific finding as well. In the study reported here, we observed both cytoplasmic and membrane staining of other EGFR family members in our tumor specimens as well. For HER-3, several other groups have also reported cytoplasmic staining in ovarian, breast, and papillary thyroid carcinomas (35 , 49 , 50) . One report has shown that the HER-3 gene encodes secreted as well as transmembrane receptor forms (51) . HER-3 may have other unknown functions in addition to being a receptor tyrosine kinase. For HER-4, both cytoplasm and membrane immunostaining were reported in papillary thyroid carcinoma (35) . The significance of cytoplasmic staining, however, is not clear at the present time.
In the present study, the levels of immunoreactivity of HER-2/neu were significantly associated with those of EGFR and of HER-3. The concomitant high to moderate expression of these three members of the EGFR family was also observed in the three oral SCC cell lines. A similar frequent coexpression of different EGFR family members was reported in papillary thyroid carcinoma (35) . These authors suggested that there may be some common mechanism regulating the expression of the EGFR family members. The common coexpression of EGFR family members in oral cancer supports the notion.
Another interesting question concerning coexpression of EGFR family members in cancer is whether their coexpression contributes to tumorigenesis. Overexpression of HER-2/neu by itself has been shown to enhance the metastatic potential by increasing the multiple adhesion and invasion steps of the metastatic cascade in an experimental model (22) and to correlate with poor survival of patients with breast, ovarian, lung, stomach, or oral cavity cancer (2 , 23, 24, 25, 26, 27) . Overexpression of EGFR can transform mouse fibroblast cells in the presence of EGFR (52, 53, 54) . EGFR is also believed to influence important steps of tumor invasion and dissemination, including enhancing cell motility, cytoskeletal changes, and production of extracellular matrix degradative enzymes (9) . As mentioned above, HER-2/neu is known to enhance metastatic potential. The roles of HER-3 and HER-4 individually in relation to metastasis are less clear. Further studies will be required to gain a better understanding. However, their ability to form heterodimer and cross-talk among the EGFR family, HER-3, and HER-4 through the stimulation by their ligands can enhance the signaling pathways of EGFR or HER-2/neu. In our analyses, we found that the expression levels of EGFR family members have a significant association with distant metastasis (Table 2)<$REFLINK> . They tend to coexpress in the tumor samples (Fig. 4)<$REFLINK> . It could be possible that coexpression of different receptor tyrosine kinases cooperates in the distant metastasis. Furthermore, among the EGFR family members studied here, HER-2/neu was the most significant factor in predicting the outcome of the patients with oral SCC, which suggests that HER-2/neu may, at least in oral SCC, play the most significant role. Although its ligands have not been found, HER-2/neu tyrosine kinase activity can be indirectly stimulated by the formation of heterodimers with other members of the family. HER-2/neu and HER-3 can work cooperatively, resulting in qualitative as well as quantitative changes in the substrate phosphorylation and enhancing neoplastic transformation (11) . These data strongly suggest that cooperation exists among members of the EGFR family. When we grouped HER-2/neu and EGFR, HER-2/neu and HER-3, or HER-2 and HER-4 immunoreactivities and analyzed for their association with survival, the coexpression of HER-2/neu and HER-3 was more significant than either HER-2/neu or HER-3 individually, the coexpression of HER-2/neu and EGFR, or the coexpression of HER-2 and HER-4 (Table 3)<$REFLINK> . Furthermore, we observed the heterodimerization of HER-2/neu and HER-3 in an oral SCC cell line, suggesting interaction between these two oncoproteins in the oral SCC. With the addition of EGFR to the HER-2/neu and HER-3 combination in the multivariate analysis model, the predicting power was further improved (log likelihood ratio changed from −13.94 to −11.99; P <0.05); however, it was not improved with the addition of HER-4 (log likelihood ratio changed from −13.94 to −13.82; P > 0.5). The findings indicate that the combination of these three EGFR family members can significantly improve the predicting power over that for any individual member. Because EGFR improved the log likelihood ratio only slightly, HER-2/neu together with HER-3 may be sufficient to predict the outcome of the disease. Taken together, the results indicate that the expression of EGFR, HER-2/neu, and HER-3 may help predict the outcome of patients with oral SCC.
Acknowledgments
We thank the Department of Oral Pathology, Ninth People’s Hospital, Shanghai Second Medical University, Shanghai, People’s Republic of China, for providing the patients’ clinical information.
Footnotes
-
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
-
↵1 This work was supported in part by United States Department of Health and Human Services Cancer Center Core Support Grant CA-16672 from the National Cancer Institute, NIH Grants CA58880 and CA60856, M. D. Anderson Faculty Achievement Award (to M-C. H.), and Grant DAMD 17-94-J4315 (to M-C. H.) from the Department of Defense.
-
↵2 The first two authors contributed equally to this work.
-
↵3 To whom requests for reprints should be addressed, at Box 108, Section of Molecular Cell Biology, Department of Cancer Biology, University of Texas M. D. Anderson Cancer Center, 1515 Holcombe Boulevard, Houston, TX 77030. Phone: (713) 792-3668; Fax: (713) 794-0209; E-mail: mchung{at}tumorb01.mdacc.tmc.edu
-
↵4 The abbreviations used are: SCC, squamous cell carcinoma; EGFR, epidermal growth factor receptor; TNM, tumor, lymph node, and distant metastasis.
- Accepted September 23, 1999.
- Received May 13, 1999.
- Revision received September 22, 1999.