differential expression of growth factors in …...differential expression of growth factors in...

Differential Expression of Growth Factors in Squamous CellCarcinoma and Precancerous Lesions of the Lung1

Chandrika J. Piyathilake, Andra R. Frost,Uppender Manne, Heidi Weiss, Walter C. Bell,Douglas C. Heimburger, and William E. GrizzleDepartments Nutrition Sciences [C. J. P., D. C. H.] Pathology[A. R. F., U. M., W. C. B., W. E. G.], Heidi Weiss ComprehensiveCancer Center, University of Alabama at Birmingham, Birmingham,Alabama 35294

ABSTRACTPurpose: This study was conducted to evaluate the clin-

ical significance of the localization of epidermal growthfactor receptor (EGF-r), transforming growth factor(TGF)-�, and erbB-2 in the development, progression andprognosis of squamous cell cancers (SCCs) of the lung.

Experimental Design: The localization of EGF-r,TGF-�, and erbB-2 was evaluated immunohistochemicallyin 60 archival specimens of SCC of the lung and in 60 lungspecimens without cancer. To clarify the patterns of expres-sion of EGF-r in these tumors, the patterns of expression ofEGF-r in cells in culture were monitored after challengingnormal human bronchial epithelial and SCC cell lines witheither EGF or TGF-� at physiological concentrations.

Results and Conclusions: The expression of EGF-r,erbB-2, and TGF-� were significantly higher in SCC andassociated precancerous lesions than in the normal bron-chial epithelium and hyperplastic lesions of noncancer spec-imens. A statistically significant stepwise increase in expres-sion from uninvolved bronchial epithelium to precancerouslesions to SCC was observed with EGF-r and TGF-�. Thelocalization of EGF-r in the cytoplasm (P � 0.04), but not inthe membrane (P � 0.20), of SCCs was significantly associ-ated with poor overall survival of subjects. To demonstratethe biological relevance of cytoplasmic expression of EGF-r,we noted that there was a prompt reduction in the mem-brane expression and a concomitant increase in cytoplasmicexpression of EGF-r after adding either EGF or TGF-� tothe cell culture medium. Overall, the study identified aninvolvement of EGF-r and TGF-� in the development ofSCCs. The prognostic importance of EGF-r expression in

the cytoplasm of lung cancer probably is an indication of theprognostic importance of trafficking of the EGF-r receptorbetween the Golgi apparatus and cell membranes and ofinternalization of EGF-r after an interaction with one of theEGF-r ligands at the cellular membrane surface.

INTRODUCTIONDespite recent advances in the treatments for lung cancer,

improvement in survival has only been modest, indicting thateffective therapeutic approaches are still lacking. Becausegrowth factors and their receptors play a central role in regulat-ing neoplastic processes (1), the expression of various growthfactor receptors and their ligands in human neoplasia has beeninvestigated extensively as potential targets of receptor-directedtherapies. Many studies have identified EGF-r3 and two of itsligands, EGF and TGF-�, and a closely related receptor, erbB-2(p185HER-2), as some of the major growth factors and growthfactor receptors involved in neoplastic transformation (2). How-ever, a great heterogeneity in the incidence of overexpressionand in the prognostic significance of these growth factors andreceptors has been observed both within and between varioustumor types, including cancers of the lung.

Most studies of lung cancer in which growth factors havebeen investigated have grouped together SCC and adenocarci-nomas of the lung as “non-small cell lung cancer (NSCLC),”despite the likely biological differences between the two histo-logical subtypes. EGF-r has been shown to be overexpressed inboth squamous and adenocarcinoma of the lung, but a strongercorrelation between EGF-r overexpression and an adverse out-come was noted for squamous carcinomas than for adenocarci-nomas (3, 4). Therefore, a combination of varying proportionsof SCCs and adenocarcinomas may result in mixed observa-tions. Another major limitation in most studies relates to the lackof data or inconsistency in reporting data on the intensity andlocalization of the receptor in cells of interest and the relation ofgrowth factor expression to survival of patients. For example,some investigators have reported only the cytoplasmic expres-sion of EGF-r (5) and erbB-2 (6), whereas others considered acell positive for EGF-r expression only if membrane stainingwas present (7). However, most investigators have failed toreport the cellular localization of EGF-r expression (8, 9). Aknowledge of the cellular localization of growth factor receptorsand its relation to cancer development and prognosis will aid inunderstanding the future applications of growth factor receptor-directed therapies. Also, in targeted therapy, knowledge of therelative intensity of expression of target antigen in cancer and

Received 3/27/01; revised 12/3/01; accepted 12/17/01.The costs of publication of this article were defrayed in part by thepayment of page charges. This article must therefore be hereby markedadvertisement in accordance with 18 U.S.C. Section 1734 solely toindicate this fact.1 Supported by Preventive Oncology Academic Award K07 CA 70160from the National Cancer Institute.2 To whom requests for reprints should be addressed, at Department ofNutrition Sciences, Division of Nutritional Biochemistry and MolecularBiology, University of Alabama at Birmingham, University Station,Birmingham, AL 35294. Phone: (205) 975-5398; Fax: (205) 934-7049;E-mail: [email protected].

3 The abbreviations used are: EGF-r, epidermal growth factor receptor;TGF, transforming growth factor; SCC, squamous cell carcinoma;CSCS, charcoal-stripped calf serum; PCNA, proliferating cell nuclearantigen.

734 Vol. 8, 734–744, March 2002 Clinical Cancer Research

Research. on July 28, 2020. © 2002 American Association for Cancerclincancerres.aacrjournals.org Downloaded from

http://clincancerres.aacrjournals.org/

associated precancerous lesions and normal tissues is important,but reports of such nature on growth factor and growth factorreceptor expression in the lung are rare.

To overcome these limitations, we examined the cytoplas-mic and/or membrane expression of EGF-r, erbB-2, and TGF-�in archival specimens of SCC, histologically normal bronchialepithelium (uninvolved epithelium), epithelial hyperplasia,metaplasia, and dysplasia associated with SCC and in the nor-mal bronchial epithelium and epithelial hyperplasia in archivallung specimens without cancer (noncancer). We also correlatedthe cytoplasmic and membrane expression of growth factors inSCCs individually and as an average of the two patterns ofexpression with indicators of tumor progression or aggressive-ness (i.e., size of the tumor, nodal status, metastases, stage of thedisease, and differentiation of SCCs) and overall survival fromthe disease. Cell culture studies were conducted to evaluate thepattern of expression of EGF-r in cells after challenging normalhuman bronchial epithelial and SCC cell lines with physiolog-ical concentrations of EGF and TGF-�.

MATERIALS AND METHODSPatients and Tissue Specimens. Formalin-fixed, paraf-

fin-embedded sections of the lung were obtained from thearchival collections of the Veterans Affairs Medical Center andUniversity Hospital in Birmingham, AL. Sixty noncancer spec-imens of the lung were selected at random from a list of lungsurgeries performed between 1988 and 1998 for lung diseases orconditions other than lung cancer. From a list of patients withSCC of the lung who had undergone surgery between 1989 and1996, sixty cases were selected at random. We obtained thesurgical pathology report, demographic information, clinicalfollow-up, and cause of death from the respective hospitalmedical records or hospital tumor registries. The patients werefollowed-up by the respective tumor registries until their date ofdeath or the date of last documented contact.

One tissue block from each noncancer patient was selectedto provide a section that contained adequate normal and unin-flamed bronchial epithelium. One or more tissue blocks fromeach patient with SCC were selected to provide sections thatcontained samples of uninvolved bronchial epithelium, epithe-lial hyperplasia, metaplasia, and dysplasia, when present, inaddition to the invasive carcinomas.

Histological Criteria and Classification of Lesions.Bronchial epithelium was classified histologically as normal (innoncancer specimens), uninvolved (SCC associated, but his-topathologically normal), hyperplastic, metaplastic, or dysplas-tic by the pathologists involved in the study (W. C. B. andA. R. F.). The normal and uninvolved bronchial epithelium wasdefined as pseudostratified ciliated columnar epithelium to cil-iated cuboidal columnar epithelium, depending on the size of thebronchus assessed. The bronchial epithelium was consideredhyperplastic if the number of layers of cells was increased in theepithelium (i.e., more than four cell layers present without anobvious tangential cut). When the normal ciliated columnarepithelium was completely replaced by squamous epitheliumwithout significant cytological atypia, the epithelium was con-sidered to be metaplastic. If the metaplastic squamous epithe-lium demonstrated nuclear atypia, it was deemed dysplastic. All

SCCs of the lung exhibited squamous differentiation with var-ious degrees of keratinization and were classified as well, mod-erately, or poorly differentiated. No mixed cases with well-defined components of adenocarcinoma or small cell carcinomawere included. SCC of the lung was classified into well-differ-entiated, moderately differentiated, and poorly differentiatedsubtypes, depending on the degree of squamous differentiationpresent (10). Tumor staging was evaluated by using the Tumor-Node-Metastasis system (11).

Immunohistochemical Analysis. Our methods of im-munohistochemical analysis have been reported previously (12).Briefly, 5-�m sections were deparaffinized and treated with3.0% H2O2 for 4 min to quench endogenous peroxidase activity.Sections were incubated with preimmune goat serum (1%) for20 min at room temperature to suppress nonspecific staining andthen subsequently incubated with monoclonal antibodies toEGF-r (Zymed Laboratories, Inc., San Francisco, CA), erbB-2,and TGF-� (Oncogene Research Products, Cambridge, MA) ata concentration of 1.0, 0.25, and 0.5 �g/ml, respectively, for 1 hat room temperature. Companion matching slides stained withno primary antibody served as controls (deletes). The remainderof the staining procedure was performed using a biotin-strepta-vidin detection system (BioGenex, San Ramon, CA). The sub-strate diaminobenzidine tetrahydrochloride was used for visual-ization of the antigen-antibody complex (a brown reactionproduct), and sections were counterstained lightly with hema-toxylin.

Assessment of Expression of EGF-r, erbB-2, andTGF-�. The expression of EGF-r, erbB-2, and TGF-� wasscored by three observers (C. J. P., A. R. F., U. M.). Because theintensity of expression of EGF-r and erbB-2 was often differentin basal and luminal cells, a separate score was given for basaland luminal cells in the normal, uninvolved, and hyperplasticbronchial epithelium. Also, EGF-r and erbB-2 exhibited bothcytoplasmic and membrane staining in both basal and luminalcells. However, it was difficult to separate membrane stainingfrom cytoplasmic staining in the basal layer of normal, unin-volved, and hyperplastic epithelium, and thus the basal layerwas assigned a single score that combined the membrane andcytoplasmic immunostaining. In the luminal cells, we were ableto separate membrane staining from cytoplasmic staining, andthus luminal cells were given separate scores for membrane andcytoplasmic staining of EGF-r and erbB-2. TGF-� expressionwas primarily cytoplasmic, and only cytoplasmic staining wasscored. Because the intensity of expression of TGF-� was oftendifferent in basal and luminal cells, a separate score was givenfor basal and luminal cells in the normal, uninvolved bronchialepithelium and epithelial hyperplasia. The basal and luminalscores were averaged in specific analyses. Membrane and cy-toplasmic staining for EGF-r and erbB-2 was scored separatelyin metaplasia, dysplasia, and SCC, and these scores were aver-aged in specific analyses. Each observer graded the intensity ofimmunostaining based on a visual assessment of the intensity ofbrown reaction product within the cell cytoplasm or on the cellmembrane on a scale of 0 (no staining) to 4� (intense staining)in each of the epithelial components. In addition, the percentageof cells at each intensity was estimated (by microscopic exam-ination of the entire tissue section) and multiplied by the appro-priate intensity score to obtain a weighted average of the inten-

735Clinical Cancer Research



sity score (12). The final score reported is the average of thethree observers.

Cell Culture Experiment. The NCI-H226 cell line (de-rived from SCC of the lung) were grown on glass coverslips inmedium with 10% calf serum (13). One day before testing, thismedium was replaced with either medium plus 10% charcoalstripped calf serum (CSCS), or 10% calf serum. Similarly, animmortalized human bronchial epithelial cell line from Ameri-can Type Culture Collection (BEAS-2B) was grown in specifiedmedium plus 10% calf serum, and 1 day before testing, this wasreplaced by medium plus either 10% CSCS or 10% CS. On thetest day, either vehicle, EGF at 10 ng/ml, or TGF-� at 10 ng/mlwas added to the medium, and the effects of the addition of theligands to EGF-r on the pattern of EGF-r expression in cells wasmonitored at 0, 0.5, 1, and 24 h. Specifically, after exposure tothe ligands for the specified times, the cells were fixed in neutralbuffered 10% formalin and stained for EGF-r by immunohisto-chemistry using the EGF-r antibody at 3 �g/ml final dilution.

Statistical Analysis. Descriptive statistics such as mean,median, and SE were calculated for the expression of EGF-r,erbB-2, and TGF-� for each type of tissue. The difference inexpression was calculated between types of epithelial tissuesand either the Wilcoxon rank sum or Wilcoxon sign rank testwas used to determine whether there is differential expressionbetween different types of tissues. The expression of EGF-r,erbB-2, and TGF-� in SCC was compared according to differenthistopathological characteristics (nodal status, tumor size, stage,and grade of differentiation). The Wilcoxon rank sum was usedto evaluate significant differences in expression of EGF-r,erbB-2, and TGF-� for each histopathological parameter. Fi-nally, the relationship between patient survival and the expres-sion of EGF-r, erbB-2, and TGF-� in the SCCs, either separatelyor in combinations, was evaluated using the log-rank test.

RESULTSSixty noncancer subjects included in the study were oper-

ated upon secondary to the diagnosis of emphysema and/orfibrosis of the lung. All specimens from these subjects containedadequate normal bronchial epithelium for the evaluation ofexpression of EGF-r, erbB-2, and TGF-�. Fifty-eight of the 60specimens examined contained SCC. The majority of subjectshad early stage SCCs (79%, stages 1 and 2; 21%, stages 3 and4), smaller tumors (81%, T1 and T2; 19%, T3 and T4); 69% ofsubjects had no nodal involvement (N0), and 31% of subjectswere N1 and N2.

Fifty-eight specimens contained adequate uninvolved bron-chial epithelium, and 46 specimens also contained adequateepithelial hyperplasia for evaluation. We identified eight dys-plasias and four metaplasias in 60 cancer specimens. Because ofthe small number of dysplastic and metaplastic tissues identi-fied, the immunohistochemical staining scores of EGF-r,erbB-2, and TGF-� in these lesions were not statistically sig-nificant compared with EGF-r, erbB-2, and TGF-� scores in theother types of tissues. Because the average expression of EGF-r,erbB-2, and TGF-� in normal bronchial epithelium from spec-imens with diagnoses of emphysema or fibrosis was not signif-icantly different, the two groups were combined.

The expression of EGF-r, erbB-2 (a combination of cyto-

plasmic and membrane expression in basal cells and cytoplas-mic and membrane expression separately in luminal cells), andTGF-� (cytoplasmic expression in basal and luminal cells) inhistologically normal bronchial epithelial cells and in epithelialhyperplasia associated with SCC was significantly higher com-pared with similar lesions in noncancer specimens (P � 0.05 forall comparisons). In histologically normal bronchial epitheliumand epithelial hyperplasia of noncancer and cancer-associatedlesions, higher expression of EGF-r in the basal cells than inluminal cells was a consistent observation. Similarly, lowermembrane expression of EGF-r in the luminal cells of theselesions was a consistent finding. However, we noted that thedifference between cytoplasmic and luminal expression is largerin normal and hyperplastic lesions of noncancer specimenscompared with similar lesions associated with cancer. Metaplas-tic lesions displayed a similar pattern of membrane:cytoplasmicstaining ratio with that of hyperplastic lesions associated withinvasive cancer. Interestingly, in dysplastic lesions and in inva-sive cancer, expression of EGF-r in the membrane was higherthan its expression in the cytoplasm, suggesting that this switchfrom greater cytoplasmic to greater membrane expression mayoccur at the stage of dysplasia. The expression of erbB-2 wasconsistently lower in the basal cells compared with cytoplasmicexpression of erbB-2 in the luminal cells but was higher com-pared with its membrane expression in the luminal cells ofnormal and hyperplasia of noncancer and cancer specimens. Thecytoplasmic expression of erbB-2 was higher than its membraneexpression in the luminal cells of normal and hyperplasticepithelium, similar to EGF-r. In contrast to EGF-r, this patterndid not change in dysplastic lesions or invasive cancer. Theexpression of TGF-� was mainly cytoplasmic, and the degree ofstaining was higher in the luminal cells compared with basalcells in histologically normal bronchial epithelium and epithelialhyperplasia of noncancer and cancer.

As shown in Fig. 1, when the expression of EGF-r (Fig. 1A)and TGF-� (Fig. 1B) in the cytoplasm and membrane of basaland luminal cells was averaged, there was a statistically signif-icant stepwise increase in their expression from uninvolvedbronchial mucosa to epithelial hyperplasia to cancer, suggestingtheir involvement in lung carcinogenesis. In contrast, a statisti-cally significant stepwise increase was not observed witherbB-2, suggesting that this growth factor may not be involvedin lung carcinogenesis (Fig. 1C). Because of the small numberof samples, metaplastic and dysplastic lesions were not includedin this analysis. The average expression of EGF-r, TGF-�, andalso erbB-2, however, was significantly higher in uninvolvedbronchial mucosa and epithelial hyperplasia associated withSCC compared with their expression in the normal bronchialepithelium and hyperplastic lesions of noncancer specimens,suggesting their early expression in lung carcinogenesis (Fig. 2,A–C). Figs. 2–4 shows the immunohistochemical staining forEGF-r (Fig. 2, A–D), erbB-2 (Fig. 3, A–D), and TGF-� (Fig. 4,A–D) in histologically normal bronchial epithelium in noncan-cer specimens (A) uninvolved bronchial epithelium associatedwith SCC (B), epithelial hyperplasia associated with SCC (C)and SCC (D) of the lung, respectively.

We observed a statistically significant positive Spearmancorrelation coefficient between cytoplasmic expression ofEGF-r and TGF-� (r � 0.27, P � 0.04) and a nonsignificant

736 Growth Factors in Lung Cancer Development and Prognosis



positive correlation between membrane expression of EGF-rand cytoplasmic expression of TGF-� (r � 0.17, P � 0.22) inSCCs. erbB-2 expression, either cytoplasmic (r � 0.22, P �0.09) or membrane (r � 0.09, P � 0.47), was not significantlyassociated with expression of TGF-�. There was also no statis-tically significant association between expression of EGF-r anderbB-2, either cytoplasmic or membrane (r � 0.26, P � 0.05;r � 0.22, P � 0.10, respectively). The correlation between thedegree of cell proliferation as measured by immunohistochem-ical assessment of PCNA, and expression of EGF-r in thecytoplasm and membrane was positive and statistically signifi-

cant (r � 0.40, P � 0.002 for both comparisons). The correla-tion between PCNA and TGF-� was also positive and signifi-cant (r � 0.41, P � 0.002). In contrast, there was no statisticallysignificant correlation between PCNA and expression of erbB-2,cytoplasmic or membrane (r � 0.02, P � 0.90; r � 0.12, P �0.12, respectively).

The relationship between expression of EGF-r, erbB-2, andTGF-� and the pathological features of SCCs is shown in Table1. Although the overall results were suggestive of higher ex-pression of EGF-r and TGF-� with advancing disease, as indi-cated by nodal involvement, larger tumors, and higher stage,

Fig. 1 Average expression*(average of each growth factorexpression, cytoplasmic andmembrane in basal and luminalcells) of EGF-r (A), TGF-� (B),and erbB-2 (C) in differenttissue types of noncancer andcancer.




these differences were statistically nonsignificant. A higher ex-pression of both cytoplasmic and membrane EGF-r in tumorswith nodal involvement compared with tumors with an absenceof nodal involvement reached statistical significance. Whencytoplasmic and membrane expression of EGF-r was averaged,the results remained the same (data not shown). The degree ofexpression of erbB-2, cytoplasmic or membrane, did not seem tobe associated with pathological features of the tumors. Grade ofdifferentiation of the tumors (well, moderate, and poor) was notassociated with the degree of expression of EGF-r, erbB-2, orTGF-� (data not shown).

A higher level (�median) of cytoplasmic expression ofEGF-r in the cancer was associated with a statistically signifi-cant poor survival of subjects (P � 0.02; Fig. 5A), whereas its

membrane expression was not (P � 0.20; Fig. 5B). When theexpression of EGF-r in the cytoplasm and membrane was aver-aged, it was not significantly associated with survival (P �0.26). The association between higher expression of TGF-� andpoor patient survival approached statistical significance (P �0.08; Fig. 5C). The expression of erbB-2, cytoplasmic (Fig. 5D),membrane, or an average of the two, was not significantlyassociated with survival of subjects.

Cancers expressing higher levels of TGF-� and EGF-r,either cytoplasmic or membrane, demonstrated statistically sig-nificant poor survival (P � 0.03, Fig. 6A; P � 0.04, Fig. 6B,respectively). A higher expression of TGF-� along with a higherexpression of an average of cytoplasmic and membrane expres-sion of EGF-r also was associated with a significantly poorsurvival of subjects (P � 0.04). The expression of erbB-2 incombination with EGF-r or TGF-� was not associated withsurvival of patients (P � 0.05 for all comparisons). The analysisof survival data using receiver operating characteristic curves(14) to define cutoff levels provided very similar results as whenthe median values were used in the analysis.

The pattern of EGF-r expression when NCI-H226 andBEAS-2B cells were grown in medium with CSCS versus intactcalf serum is similar; in both cases, EGF-r is expressed stronglyon cell membranes (Figs. 7 and 8, A, solid, notched arrows) andto a lesser extent in the cytoplasm. After addition of 10 ng/ml ofeither EGF or TGF-� to the medium, there is a prompt reductionin the membrane expression of EGF-r within 30 min (Figs. 7 and8, B, solid, notched arrows) and a marked accentuation ofperinuclear expression of EGF-r (open arrows). The loss ofEGF-r expression on cellular membranes is almost completeafter 1 h of exposure to EGF; however, the loss of EGF-rexpression on cellular membranes required 4 h of exposure toTGF-� in the NCI H226 cells. In contrast to our observations inDU145 cells (prostate cancer; Ref. 15), EGF-r expression wasnot down-regulated at 24 h, where in both cell lines there wasthe return of some membrane staining as well as the presence of

Fig. 2 The immunohistochemical staining of EGF-r in histologicallynormal bronchial epithelium in noncancer specimens (A), SCC associ-ated uninvolved bronchial epithelium (B), epithelial hyperplasia (C), andSCC (D).

Fig. 3 The immunohistochemical staining of erbB-2 in histologicallynormal bronchial epithelium in noncancer specimens (A), SCC associ-ated uninvolved bronchial epithelium (B), epithelial hyperplasia (C), andSCC (D).

Fig. 4 The immunohistochemical staining of TGF-� in histologicallynormal bronchial epithelium in noncancer specimens (A), SCC associ-ated uninvolved bronchial epithelium (B), epithelial hyperplasia (C), andSCC (D).




perinuclear staining (Figs. 7 and 8, D). When EGF or TGF-�was withdrawn after 4 h and replaced with CSCS, the pattern ofEGF-r expression returned to the original pattern of expressionafter 20 h (data not shown).

DISCUSSIONUnderstanding the role of growth factor biochemical sig-

naling networks in cancer is pivotal to improvements not only in

diagnosis and prognosis of this disease but also in the develop-ment of targeted therapies. It is likely that knowledge of therelationship between differential expression and cellular local-ization of growth factor receptors and their ligands in preneo-plastic and neoplastic lesions and patient survival will prove tobe beneficial to the development of novel pharmacologicalapproaches to therapy. However, studies that address theseissues are scarce. We have examined the localization of immu-

Table 1 Mean � SE and median immunostaining scores of EGF-r, erbB-2, and TGF-� in SCCs by pathological diagnosis: N0 (absence of nodalinvolvement), N1 � N2 (presence of nodal involvement), tumor size (T1 � T2 and T3 � T4), and stage (stages 1 � 2 and stages 3 � 4)

Growth factor

Mean � SE and median of immunostaining score in SCCs

N0

n � 40N1 � N2

n � 18 PT1 � T2

n � 47T3 � T4

n � 11 PStages 1 � 2

n � 46Stages 3 � 4

n � 12 P

EGF-r cytoplasmic 1.3 � 0.1 1.5 � 0.1 0.09 1.3 � 0.1 1.5 � 0.1 0.50 1.3 � 0.1 1.5 � 0.1 0.301.4 1.7 1.4 1.6 1.4 1.6

EGF-r membrane 1.6 � 0.1 1.9 � 0.1 0.07 1.6 � 0.1 2.0 � 0.1 0.30 1.6 � 0.1 2.0 � 0.1 0.101.7 2.1 1.8 2.1 1.8 2.1

erbB-2 cytoplasmic 1.3 � 0.1 1.3 � 0.1 0.70 1.3 � 0.1 1.3 � 0.1 0.70 1.3 � 0.1 1.2 � 0.1 0.901.3 1.2 1.2 1.3 1.2 1.3

erbB-2 membrane 1.0 � 0.1 0.9 � 0.1 0.40 1.0 � 0.1 1.0 � 0.2 0.80 1.0 � 0.1 1.0 � 0.2 0.800.9 0.7 0.9 0.6 0.9 0.7

TGF-� cytoplasmic 1.2 � 0.1 1.3 � 0.1 0.60 1.2 � 0.1 1.4 � 0.1 0.20 1.2 � 0.1 1.4 � 0.1 0.101.2 1.4 1.3 1.5 1.3 1.5

Fig. 5 Overall survival after resection of SCCs in two groups defined by expression of EGF-r, TGF-�, and erbB-2, �median and �median ofimmunostaining score.




nohistochemically demonstrable EGR-r, erbB-2, and TGF-�expression in histologically normal bronchial epithelial cells andhyperplastic epithelial lesions of smokers who have not devel-oped lung cancer and in SCCs and in histologically normalbronchial epithelium, epithelial hyperplasia, metaplasia, anddysplasia associated with SCC.

The noncancer subjects identified in our study are likely tohave a higher risk of developing smoking-related lung cancerbecause of their exposures to carcinogens from cigarette smokesimilar to SCC subjects. This is supported by their reportedlong-term smoking habits and by the presence of emphysema-tous changes of the lung, which are associated with cigarettesmoking through the activation of elastase leading to elastictissue damage (16). The differential localization of all threemarkers was significantly higher in histologically normal bron-chial mucosa and epithelial hyperplasia of subjects who haddeveloped lung cancer compared with those who had not, iden-tifying these as susceptibility markers for the development ofSCCs. To our knowledge, this is the first report to include asubset of smokers who have been resistant to the developmentof lung cancer enabling us to document such an observation.

The expression of EGF-r was higher in basal cells com-pared with luminal cells of normal and hyperplastic lesions,suggesting its association with cell proliferation. In contrast, theexpression of erbB-2 and TGF-� was lower in basal cellscompared with luminal cells in the same lesions, suggesting alack of association with cell proliferation. To our knowledge,

the clinical significance of cellular localization (membrane ver-sus cytoplasmic) of growth factor receptors in the lung has notbeen addressed previously. Receptor localization may signifyadaptations that permit growth factor responsiveness in a milieuof available ligand. The down-regulation of growth factor re-ceptors in the cytoplasm may reflect receptor-ligand internal-ization, a rapid process that occurs after ligand binding (17).Although internalized receptors in most cell culture systemshave been shown to be degraded rapidly (17), a proportion ofinternalized receptor in EGF-treated hepatocytes was thought torecycle to the cell surface (18). Intracellular localization ofreceptor also may represent newly synthesized molecules withinthe Golgi that have yet to be processed and inserted into themembranes. Lack of detection of the receptor on the membraneby immunohistochemical means, however, cannot be interpretedas its total absence but may reflect a limited number of receptorson the surface.

We observed that the expression of EGF-r in the cytoplasmis higher than its membrane expression in early lesions of lungcancer, including metaplastic lesions. In dysplastic lesions andSCCs, however, the membrane expression is higher comparedwith cytoplasmic expression. Such a pattern was not observedwith erbB-2. Because it is thought that for the EGF-r to bindligand it must be expressed at the cell surface, the observedresults may suggest an important role for EGF-r in lung carci-nogenesis. We also observed a stepwise increase in averageEGF-r and TGF-� expression from SCC-associated uninvolved

Fig. 6 Overall survival after re-section of SCCs in two groupsdefined by expression of �and � median immunostainingscores of both EGF-r cytoplasmicand TGF-� (A) and EGF-r mem-brane and TGF-� (B).




bronchial mucosa to epithelial hyperplasia to SCC, suggestingtheir involvement in lung carcinogenesis. The absence of such astepwise increase with expression of erbB-2 may suggest itslack of importance in lung carcinogenesis. We emphasize thatthis stepwise alteration was present with an averaged expres-sion, which combines marker expression in basal cells andmembrane and cytoplasmic expression in luminal cells. Thesame pattern is not observed with some markers if the results aresummarized differently, e.g., the degree of expression of TGF-�in the luminal cells of epithelial hyperplasia is similar to that ofinvasive cancer. These observations indicate the importance ofstating precisely the manner in which the evaluation of markerswas performed so that results can be compared among studies.

Although the overexpression of EGF-r and its ligand,TGF-�, is shown to be frequent in resectable non-small cell lungcancer (19), its relation to tumor progression is inconsistent;some studies report no relationship to tumor progression (19),whereas others report a relationship between the expression ofEGF-r and TGF-� and aggressiveness of SCCs of the lung (20).The study of Pavelic et al. (20) is also suggestive of an associ-ation between expression of EGF-r and progression of adeno-carcinomas of the lung. The expression of TGF-� was notidentified as an independent tumor maker in several studies ofnon-small cell lung cancer, which includes both squamous andadenocarcinomas (21). Our results are suggestive of an associ-ation between higher expression of EGF-r, both cytoplasmic andmembrane, and nodal involvement, but not the other features oftumor aggressiveness. Similar associations have been reportedpreviously by Gorgoulis et al. (9). The overexpression of EGF-rin breast cancer has been correlated with both a poor prognosis

(22), a lack of response to endocrine therapy (23), and increasedmetastatic potential (24). An increased expression of EGF-r hasalso been associated with adverse outcome of patients withcancers of the cervix (25), ovary (26), and endometrium (27).Also, EGF-r overexpression in primary laryngeal cancer wasshown to be an independent prognostic factor of neck noderelapse (28). Increased EGF-r expression was shown to be abiomarker of bronchial metaplasia of smokers but not a biomar-ker of retinoid response in lung cancer chemoprevention trials(29). These studies, however, have not distinguished cytoplas-mic expression from membrane expression.

We report that cytoplasmic expression of EGF-r, but not itsmembrane expression, is significantly associated with poor pa-tient survival. Higher expression of EGF-r, either cytoplasmic ormembrane, together with higher expression of TGF-�, however,was significantly associated with poor survival of patients, sug-gesting the significance of coexpression of these markers. Be-cause most previous reports have failed to indicate the cellularlocalization of EGF-r, it is not possible to compare our resultswith others. Higher expression of EGF-r and TGF-� together,however, was shown to predict poor survival rate of subjectsdiagnosed with primary adenocarcinomas of the lung. EGF-rand TGF-� have been shown to be up-regulated in a variety oftumors other than lung, including esophagus (30) and head andneck (31, 32). Grandis et al. (33), Ozanne et al. (34), and Cernyet al. (35) reported that the elevated expression of both EGF-rand TGF-� in primary tumors was the most important prognos-tic factor identified in cancers of the head and neck.

erbB-2 expression in lung adenocarcinomas has beenshown to predict shortened survival in some studies (36, 37),

Fig. 7 A (�600) demonstratesthat cancer cells grown on cov-erslips and exposed for 24 h tomedium with CSCS have astrong expression of EGF-r oncellular membranes (solidnotched arrow); however, thereis some diffuse cytoplasmic ex-pression as well as perinuclearstaining (open arrow). AfterEGF (10 ng/ml) was added tothe medium, the patternchanged within 30 min (B,�600), but the expression ofEGF-r on membranes was stillpresent (solid notched arrows),although there is marked accen-tuation of perinuclear staining(open arrows). The loss of ex-pression of EGF-r on mem-branes with a shift to perinu-clear accumulation of EGF-r isalmost complete at 1 h of expo-sure (C, �600). The stainingremains unchanged even after24 h of exposure (D, �600),except for some slight mem-brane staining (solid notchedarrow) as well as some mem-brane staining in small circularareas of cells (line arrows).




whereas others have reported no significant association withsurvival (6). Our results clearly indicated that erbB-2 expressionis very unlikely to be associated with progression of SCCs or inthe modification of patient survival. Overexpression ofc-erbB-2, however, has been shown to be a frequent event inoral SCC and was also correlated with poor survival (38). Incontrast to our results of SCC of the lung, erbB-2 expressionin oral SCCs has also been shown to correlate significantly withexpression of EGF-r, and coexpression of erbB-2 and EGF-rwas shown to predict a significantly poor survival of thesepatients (39). These observations suggest that the relationshipbetween clinical outcome and expression of erbB-2 in SCCsmay vary with the site of origin of the SCCs. We also noted thatthe expression of EGR-r and TGF-�, but not erbB-2, waspositively associated with PCNA (data not shown), suggestingthat in SCCs of the lung, EGF-r and TGF-� may form anautocrine stimulatory loop and hence be involved in the controlof cell proliferation leading to poor outcome of patients.

The development of novel therapies for neoplastic pro-cesses including immunotherapy, ligand receptor therapy, andgene therapy has been a high priority goal in the 1990s (40). Thenew targets for these therapies include oncogenes, growth fac-tors and their receptors, signal transduction pathways, and celldifferentiation signals. For each of these therapies, knowledgeof the expression of specific biomarkers in preneoplastic lesionsand neoplastic lesions and their localization is important. Al-though the documentation of biomarker expression in cancershas advanced over the past decade, a large majority of studieshave been conducted with a heavy focus on biomarker expres-sion in invasive cancers rather than in corresponding preneo-

plastic lesions. The knowledge of the relative expression ofthese markers in adjacent normal and preneoplastic lesions is ofgreat value, especially when toxins are delivered to cells ex-pressing some of these markers [for example, EGF-diphtheriatoxin (15)]. We observed that the relative distribution of growthfactors and their ligands in preneoplastic tissues may vary withthe specific cellular localization and protocols used to documentthese patterns. Although the effectiveness of targeted therapymay vary with specific cellular localization (membrane versuscytoplasmic) of growth factors and receptors in the tumor, thereare no reports addressing this issue. We documented that theassociation between expression of growth factor receptors andsurvival is dependent on the specific cellular localization of thegrowth factor receptors, suggesting the importance of detailedevaluation of their expression.

The pattern of phenotypic expression of EGF-r varies in alltissues with the strongest membrane and cytoplasmic expressionin squamous epithelia and in tumors and in cell lines derivedfrom squamous epithelia. Because EGF-r is known to be inter-nalized after interaction with its ligands including EGF andTGF-� and to traffic from the Golgi apparatus to cell mem-branes, the localization of EGF-r within cells varies based on theextent of interaction of its ligands with EGF-r and with thesynthesis of EGF-r. The changes in the phenotypic pattern ofEGF-r expression are demonstrated in both cell lines from thelung (cancer and normal). Thus, our finding of the prognosticimportance of EGF-r expression in the cytoplasm of lung cancerprobably is an indication of the prognostic importance of traf-ficking of the EGF-r receptor between the Golgi apparatus and

Fig. 8 A (�600) demonstratesthat normal bronchial epithelialcells grown on coverslips andexposed for 24 h to mediumwith CSCS have a strong ex-pression of EGF-r on cellularmembranes (solid notched ar-row); however, there is somediffuse cytoplasmic expressionas well as perinuclear staining(open arrow). After EGF (10ng/ml) was added to the me-dium, the pattern changedwithin 30 min (B, �600). Theexpression of EGF-r on mem-branes was very weak, andthere is accentuation of perinu-clear staining (open arrows).The loss of expression of EGF-ron membranes with a shift toperinuclear accumulation ofEGF-r is almost complete at 1 hof exposure (C, �600). A reap-pearance of some expression ofEGF-r on membranes stainingwas observed after 24 h of ex-posure (D, solid notched ar-rows, �600).




cell membranes and of internalization of EGF-r after an inter-action with one of the EGF-r ligands.

REFERENCES1. Sporn, M. B., and Todaro, G. J. Autocrine growth factors and cancer.Nature (Lond.)., 313: 745–751, 1985.

2. Goustin, A. S., Leof, E. B., Shipley, G. D., and Moses, H. L. Growthfactors and cancer. Cancer Res., 46: 1015–1029, 1986.

3. Hendler, F. J., and Ozanne, B. W. Human squamous cell lung cancersexpress increased epidermal growth factor receptors. J. Clin. Investig.,74: 647–651, 1984.4. Veale, D., Ashcroft, T., Marsh, C., Gibson, G. J., and Harris, A. L.Epidermal growth factor receptors in non-small cell lung cancer. Br. J.Cancer, 55: 513–516, 1987.5. Rusch, V., Klimstra, D., Linkov, I., and Dmitrovsky, E. Aberrantexpression of p53 or the epidermal growth factor receptor is frequent inearly bronchial neoplasia and coexpression precedes squamous cellcarcinoma development. Cancer Res., 55: 1365–1372, 1995.6. Harpole, D. H., Marks, J. R., Richards, W. G., Herndon, J. E., andSugarbaker, D. J. Localized adenocarcinoma of the lung: oncogeneexpression of erbB-2 and p53 in 150 patients. Clin. Cancer Res., 1:659–664, 1995.7. Pfeiffer, P., Clausen, P. P., Andersen, K., and Rose, C. Lack ofprognostic significance of epidermal growth factor receptor and theoncoprotein p185HER-2 in patients with systemically untreated non-small-cell lung cancer: an immunohistochemical study on cryosections.Br. J. Cancer, 74: 86–91, 1996.8. Rusch, V., Baselga, J., Cordon-Cardo, C., Orazem, J., Zaman, M.,Hoda, S., McIntosh, J., Kurie, J., and Dmitrovsky, E. Differentialexpression of the epidermal growth factor receptor and its ligands inprimary non-small cell lung cancers and adjacent benign lung. CancerRes., 53: 2379–2385, 1993.9. Gorgoulis, V., Aninos, D., Mikou, P., Kanavaros, P., Karameris, A.,Joardanoglou, J., Rasidakis, A., Veslemes, M., Ozanne, B., and Span-didos, D. A. Expression of EGF, TGF-� and EGFR in squamous celllung carcinomas. Anticancer Res., 12: 1183–1187, 1992.10. Colby, T. V., Koss, M. N., and Travis, W. D. Squamous cellcarcinoma and variants. In: J. Rosai (ed.), Tumors of the Lower Res-piratory Tract. Atlas of Tumor Pathology, Fascicle 13, pp. 157–178.Washington, DC: Armed Forces Institute of Pathology, 1995.11. Rosai, J. Ackerman’s Surgical Pathology, Ed. 8 V2, pp. 2587–2595.St. Louis: Mosby, 1996.12. Grizzle, W. E., Myres, R. B., Manne, U., and Srivastava, S. Immu-nohistochemical evaluation of biomarkers in prostatic and colorectalneoplasia. In: M. Hanausek and Z. Walaszek (eds.), John Walker’sMethods in Molecular Medicine. Tumor Marker Protocols, pp. 143–160.Totowa, NJ: Humana Press, Inc., 1998.13. Myers, R. B., Srivastava, S., Grizzle, W. E., and Lewis, Y. Antigenas detected by the monoclonal antibody BR96 is expressed strongly inprostatic adenocarcinoma. J. Urol., 153: 1572–1574, 1995.14. Choi, B. C. Slopes of a receiver operating characteristic curve andlikelihood ratios for a diagnostic test. Am. J. Epidemiol., 148: 1127–1132, 1998.15. Myers, R. B., Marshall, M. E., Oelschlager, D. K., and Grizzle,W. E. An EGF-diphtheria fusion protein kills the prostate cancer celllines DU145 and LNCaP. Proc. Am. Assoc. Cancer Res., 36: 428, 1995.16. Lipscomb, M. F. The respiratory system. In: V. Kumar, R. S.Cotran, and S. L. Robbins (eds.), Basic Pathology, pp. 392–394. Phila-delphia: W. B. Saunders Company, 1992.17. Beguinot, L., Lyall, R. M., Willingham, M. C., and Pastan, I.Down-regulation of the epidermal growth factor receptor in KB cells isdue to receptor internalization and subsequent degradation in lysosomes.Proc. Natl. Acad. Sci. USA, 81: 2384–2388, 1984.18. Dunn, W. A., and Hubbard, A. L. Receptor-mediated endocytosis ofepidermal growth factor by hepatocytes in the perfused rat liver: ligandand receptor dynamics. J. Cell Biol., 98: 2148–2159, 1984.

19. Rusch, V., Klimstra, D., Venkatraman, E., Pisters, P. W., Langen-feld, J., and Dmitrovsky, E. Overexpression of the epidermal growthfactor receptor and its ligand transforming growth factor � is frequent inresectable non-small cell lung cancer but does not predict tumor pro-gression. Clin. Cancer Res., 3: 515–522, 1997.

20. Pavelic, K., Banjac, Z., Pavelic, J., and Spaventi, S. Evidence for arole of EGF receptor in the progression of human lung carcinoma.Anticancer Res., 13: 1133–1137, 1993.

21. Pecur, L., Kapitanovic, S., Sonicki, Z., Pavicic, F., Spaventi, S.,Gluckman, J. L., Stambrook, P. J., Pavelic, Z. P., Spaventi, R., andPavelic, K. Prognostic significance of transforming growth factor �(TGF-�) in human lung carcinoma: an immunohistochemical study.Anticancer Res., 14: 2839–2843, 1994.

22. Fox, S. B., Smith, K., Hollyer, J., Greenall, M., Hastrich, D., andHarris, A. L. The epidermal growth factor receptor as a prognosticmarker: results of 370 patients and review of 3009 patients. BreastCancer Res. Treat., 29: 41–49, 1994.

23. Nicholson, S., Halcrow, P., Sainsbury, J. R., Angus, B., Chambers,P., Farndon, J. R., and Harris, A. L. Epidermal growth factor receptor(EGFr) status associated with failure of primary endocrine therapy inelderly postmenopausal patients with breast cancer. Br. J. Cancer, 58:810–814, 1988.

24. Toi, M., Osaki, A., Yamada, H., and Toge, T. Epidermal growthfactor receptor expression as a prognostic indicator in breast cancer. Eur.J. Cancer, 27: 977–980, 1991.

25. Hale, R. J., Buckley, C. H., Gullick, W. J., Fox, H., Williams, J.,and Wilcox, F. L. Prognostic value of epidermal growth factor receptorexpression in cervical carcinoma. J. Clin. Pathol., 46: 149–153, 1993.

26. Bartlett, J. M., Langdon, S. P., Simpson, B. J., Stewart, M., Katsa-ros, D., Sismondi, P., Love, S., Scott, W. N., Williams, A. R., Lessells,A. M., Macleod, K. G., Smyth, J. F., and Miller, W. R. The prognosticvalue of epidermal growth factor receptor mRNA expression in primaryovarian cancer. Br. J. Cancer, 73: 301–306, 1996.

27. Niikura, H., Sasano, H., Matsunaga, G., Watanabe, K., Ito, K., Sato,S., and Yajima, A. Prognostic value of epidermal growth factor receptorexpression in endometrioid endometrial carcinoma. Hum. Pathol., 26:892–896, 1995.

28. Almadori, G., Cadoni, G., Galli, J., Ferrandina, G., Scambia, G.,Exarchakos, G., Paludetti, G., and Ottaviani, F. Epidermal growth factorreceptor expression in primary laryngeal cancer: an independent prog-nostic factor of neck node relapse. Int. J. Cancer, 84: 188–191, 1999.

29. Kurie, J. M., Shin, H. J., Lee, J. S., Morice, R. C., Ro, J. Y.,Lippman, S. M., Hittelman, W. N., Yu, R., Lee, J. J., and Hong, W. K.Increased epidermal growth factor receptor expression in metaplasticbronchial epithelium. Clinical Cancer Res., 2: 1787–1793, 1996.

30. Yoshida, K., Kyo, E., Tsuda, T., Tsujino, T., Ito, M., Niimoto, M.,and Tahara, E. EGF and TGF-�, the ligands of hyperproduced EGFR inhuman esophageal carcinoma cells, act as autocrine growth factors. Int.J. Cancer, 45: 131–135, 1990.

31. Grandis, J. R., Melhem, M. F., Barnes, E. L., and Tweardy, D. J.Quantitative immunohistochemical analysis of transforming growth fac-tor-alpha and epidermal growth factor receptor in patients with squa-mous cell carcinoma of the head and neck. Cancer (Phila.), 78: 1284–1292, 1996.

32. Beenken, S., Sellers, M., Huang, P., Peters, G., Krontiras, H.,Dixon, P., Stockard, C., Listinsky, C., and Grizzle, W. E. Transforminggrowth factor � (TGF�), expression in dysplastic oral leukoplakia:modulation by 13-cis retinoic acid. Head Neck, 21: 566–573, 1999.

33. Grandis, J. R., Melhem, M. F., Gooding, W. E., Day, R., Holst,V. A., Wagener, M. M., Drenning, S. D., and Tweardy, D. J. Levels ofTGF-� and EGFR protein in head and neck squamous cell carcinomaand patient survival. J. Natl. Cancer Inst., 90: 824–832, 1998.

34. Ozanne, B., Richards, C. S., Hendler, F., Burns, D., and Gusterson,B. Overexpression of the EGF receptor is a hallmark of squamous cellcarcinomas. J. Pathol., 149: 9–14, 1986.




35. Cerny, T., Barnes, D. M., Hasleton, P., Barber, P. V., Healy, K.,Gullick, W., and Thatcher, N. Expression of epidermal growth factorreceptor (EGF-R) in human lung tumours. Br. J. Cancer, 54: 265–269,1986.36. Kern, J. A., Schwartz, D. A., Nordberg, J. E., Weiner, D. B.,Greene, M. I., Torney, L., and Robinson, R. A. p185neu expression inhuman lung adenocarcinomas predicts shortened survival. Cancer Res.,50: 5184–5187, 1990.37. Tateishi, M., Ishida, T., Kohdono, S., Hamatake, M., Fukuyama, Y.,and Sugimachi, K. Prognostic influence of the co-expression of epider-mal growth factor receptor and c-erbB-2 protein in human lung adeno-carcinoma. Surg. Oncol., 3: 109–113, 1994.

38. Xia, W., Lau, Y. K., Zhang, H. Z., Liu, A. R., Li, L., Kiyokawa, N.,Clayman, G. L., Katz, R. L., and Hung, M. C. Strong correlationbetween c-erbB-2 overexpression and overall survival of patients withoral squamous cell carcinoma. Clin. Cancer Res., 3: 3–9, 1997.39. Xia, W., Lau, Y. K., Zhang, H. Z., Xiao, F. Y., Johnston, D. A., Liu,A. R., Li, L., Katz, R. L., and Hung, M. C. Combination of EGFR,HER-2/neu, and HER-3 is a stronger predictor for the outcome of oralsquamous cell carcinoma than any individual family members. Clin.Cancer Res., 5: 4164–4174, 1999.40. Albelda, S. M., Wiewrodt, R., and Zuckerman, J. B. Genetherapy for lung disease: hype or hope? Ann. Intern. Med., 132:649 – 660, 2000.




2002;8:734-744. Clin Cancer Res Chandrika J. Piyathilake, Andra R. Frost, Uppender Manne, et al. Carcinoma and Precancerous Lesions of the LungDifferential Expression of Growth Factors in Squamous Cell

Updated version

http://clincancerres.aacrjournals.org/content/8/3/734

Access the most recent version of this article at:

Cited articles

http://clincancerres.aacrjournals.org/content/8/3/734.full#ref-list-1

This article cites 34 articles, 12 of which you can access for free at:

Citing articles

http://clincancerres.aacrjournals.org/content/8/3/734.full#related-urls

This article has been cited by 9 HighWire-hosted articles. Access the articles at:

E-mail alerts related to this article or journal.Sign up to receive free email-alerts

Subscriptions

Reprints and

[email protected] at

To order reprints of this article or to subscribe to the journal, contact the AACR Publications

Permissions

Rightslink site. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC)

.http://clincancerres.aacrjournals.org/content/8/3/734To request permission to re-use all or part of this article, use this link



http://clincancerres.aacrjournals.org/content/8/3/734.full#ref-list-1

http://clincancerres.aacrjournals.org/content/8/3/734.full#related-urls

http://clincancerres.aacrjournals.org/cgi/alerts

mailto:[email protected]



differential expression of growth factors in …...differential expression of growth factors in...

Documents