Distinctive Expression Pattern of ErbB FamilyReceptors Signifies an Aggressive Variant of Bladder CancerWassim Kassouf, Peter C. Black, Tomasz Tuziak, Jolanta Bondaruk, Sangkyou Lee,Gordon A. Brown, Liana Adam, Caimiao Wei, Keith Baggerly, Menashe Bar-Eli, David McConkey,Bogdan Czerniak and Colin P. Dinney*,†From the Division of Urology, McGill University Health Center (WK), Montreal, Quebec, Canada, and Departments of Urology (PCB,GAB, LA, CPD), Pathology (TT, JB, SL, BC), Biostatistics and Applied Mathematics (CW, KB) and Cancer Biology (MBE, DM),University of Texas M. D. Anderson Cancer Center, Houston, Texas

Purpose: Expression of various members of the ErbB family (epidermal growth factor receptor/ErbB-1, ErbB-2, ErbB-3 andErbB-4) is associated with disease stage and survival in patients with urothelial carcinoma. We examined the correlation ofErbB family receptor expression with the progression of urothelial carcinoma and survival.Materials and Methods: A urothelial carcinoma tissue array was constructed from 248 archival paraffin blocks and qualitycontrol studies were ascertained. The tissue microarray was stained for epidermal growth factor receptor, ErbB-2, ErbB-3and ErbB-4, and analyzed using an automated reader. Patient data included grade, stage, growth pattern, recurrence andsurvival.Results: Kaplan-Meier estimates of 5-year overall and recurrence-free survival were 58% and 27%, respectively. Patientswith high grade, invasive or nonpapillary disease had a worse prognosis than patients with low grade, superficial or papillarydisease (p �0.0001). High epidermal growth factor receptor or low ErbB-4 expression was associated with nonpapillary, highgrade and invasive tumors as well as with significantly shorter recurrence-free and overall survival (p �0.002, 0.028 and0.047, respectively). Levels of ErbB-2 and ErbB-3 expression were not associated with overall or recurrence-free survival.Conclusions: The expression profiles of ErbB-4 and epidermal growth factor receptor are prognostic in urothelial carcinoma.They may help in selecting patients at high risk with bladder cancer for more aggressive therapy.

Key Words: bladder; urothelium; carcinoma; microarray analysis; protein-tyrosine kinases

The initial presentation of bladder cancer can show awide spectrum of biological behavior that may rangefrom less aggressive, low grade, superficial disease to

high grade, muscle invasive cancer with a high risk of dis-ease progression and metastasis. Although conventionalclinicopathological parameters, such as tumor high gradeand stage, are known to be predictors of recurrence, progres-sion and survival, an improved understanding of some bio-logical aspects of bladder cancer has provided novel prog-nostic and therapeutic approaches. Profiling of molecularpathways has gained increasing use to improve prognosti-cation in patients with bladder cancer in an effort to provideoptimal therapy before the development of progressionand/or distant metastasis.

Submitted for publication April 23, 2006.Study received approval from the University of Texas M. D.

Anderson Cancer Center institutional review board.Supported by Cancer Center Core Grant CA16672 from the Na-

tional Cancer Institute, GU Bladder SPORE CA91846 and a T32Training Grant.

* Correspondence: Department of Urology, Unit 1373, Universityof Texas M. D. Anderson Cancer Center, 1515 Holcombe Blvd.,Houston, Texas 77030 (telephone: 713-792-3250; FAX: 713-794-4824; e-mail: cdinney@mdanderson.org).

† Financial interest and/or other relationship with AstraZeneca,

GlaxoSmithKline, Canji/Schering-Plough, Abbott/Vysis, Physicians’Education Resource, Novartis and SciMed.

0022-5347/08/1791-0353/0THE JOURNAL OF UROLOGY®

Copyright © 2008 by AMERICAN UROLOGICAL ASSOCIATION

353

Among important mediators of oncogenesis the proteintyrosine kinases have been shown to be involved in regula-tory processes, including proliferation, migration and cellu-lar transformation. Members of the ErbB family receptors ofprotein tyrosine kinases share structural homologies, in-cluding EGFR, ErbB-2, ErbB-3 and ErbB-4. Over expressionof ErbB receptors has prognostic and therapeutic signifi-cance in various malignancies, including breast cancer.1

However, the prognostic significance of ErbB expression inpatients with UC remains controversial since several con-tradictory reports have been published.1–7

A purpose of quantifying ErbB status in patients withbladder UC is to identify potential targets for combinedtherapeutic interventions. Another purpose is to identifyhigh risk features that might direct therapy. Since receptordimerization and cross-phosphorylation of ErbB receptorsare evident in vitro, the implications of co-expression pat-terns in bladder cancer in the clinical setting need furtherclarification. We evaluated the expression of all 4 membersof the ErbB family on a tissue microarray to determine thecorrelations between tumor expression, and patient overalland recurrence-free survival.

MATERIALS AND METHODS

Patient PopulationThe study population, based on patients treated from 1986

to 2002, was 80% male and 20% female, and the median age

Vol. 179, 353-358, January 2008Printed in U.S.A.

DOI:10.1016/j.juro.2007.08.087

DISTINCTIVE EXPRESSION PATTERN OF ERBB FAMILY RECEPTORS354

was 67 years (range 43 to 86). Of the patients included 15%were black and 85% were white. Information was collectedon pathological tumor characteristics, the date of diagnosisand last followup visit, disease recurrence and patient sta-tus (alive or dead).

Tumor samples and followup data were collected fromtransurethral resection or cystectomy and archived accord-ing to the laboratory protocol approved by the InstitutionalReview Board of The University of Texas M. D. AndersonCancer Center. All patients had primary UC of the bladderand none of the patients whose tumor samples were used inthis study received prior chemotherapy or radiotherapy.Since some sections were detached from the glass slide dur-ing tissue processing for immunohistochemistry, not all pro-teins could be measured from all cores. Consequently thenumbers of samples used in the various analyses are slightlydifferent, as the tables show. Furthermore, in a small num-ber of cases survival and/or recurrence-free survival infor-mation was not available. However, when assessing all con-trasts, all potentially informative samples were used.

Tumor identification, preparation and tissue microarrayconstruction were performed as previously described.8 Briefly,histological slides from 248 bladder cancers were reviewedand the most representative, well preserved areas of tumortissue were selected and marked. Tumors were classifiedaccording to the 3-tiered WHO histological grading systemand growth pattern (papillary vs nonpapillary).9 Invasiondepth was recorded according to the TNM staging system.10

Stage T1 (lamina propria invasion) was divided intoT1a—no muscularis mucosae invasion and T1b—muscularismucosae invasion. As in our previous publications, tumors

FIG. 1. Expression of EGFR tested in tissue microarray containingsamples of 248 bladder tumors. A and B, baseline expression innormal urothelium. A, reduced from �8. B, reduced from �200. C,low power view of representative tissue cores corresponding to highgrade (grade 3) invasive (T1b-T4) transitional cell carcinoma withlow (left column) and high (right column) levels of EGFR. Reducedfrom �18. D, higher magnification of tissue core (C) with low EGFRlevels. Reduced from �400. E, higher magnification of tissue core

(C) with high EGFR levels. Low EGFR expression was similar tobaseline expression in normal urothelium. Reduced from �400.

were dichotomized into superficial (Ta-T1a) and invasive(T1b and higher) groups.11

Donor paraffin blocks were punched in areas of interestusing a microarray instrument (Beecher Instruments, SunPrairie, Wisconsin) and 0.6 mm cores of tumor tissue weretransferred to a recipient block. The recipient microarray blockcontained 85 noninvasive low grade (grade 1–2) papillary car-cinomas, 25 noninvasive high grade (grade 3) papillary carci-nomas, 18 invasive high grade (grade 3) papillary carcinomasand 120 invasive high grade (grade 3) nonpapillary carcino-mas. Overall the microarray contained 110 superficial (Ta-T1a) and 138 invasive (T1b and higher) bladder UCs. Thereproducibility of staining in multiple tissue cores from thesame tumor was assessed on a separate microarray contain-ing 5 cores each from 4 randomly selected low grade super-ficial and 4 high grade invasive UCs of the bladder.

Immunohistochemical Staining of UC SpecimensImmunohistochemical studies were performed on formalinfixed, paraffin embedded tissue samples from the tissue mi-croarray using the avidin-biotin peroxidase complex method.Results were related to the corresponding clinicopathologicaldata. Tumor tissue was fixed in formalin and embedded inparaffin according to standard procedures. Sections (2 �m)of formalin fixed, paraffin embedded tissue specimens weredeparaffinized in xylene, followed by treatment with agraded series of alcohol (100%, 95% and 80% ethanol/doubledistilled H2O volume per volume), rehydrated in phosphatebuffered saline (pH 7.5) and pretreated with tris buffer

FIG. 2. Expression of ErbB-4 tested in tissue microarray containingsamples of 248 bladder tumors. A and B, baseline expression innormal urothelium. A, reduced from �8. B, reduced from �200. C,low power view of representative tissue cores corresponding to highgrade (grade 3) invasive (T1b-T4) transitional cell carcinoma withretention (left column) and loss (right column) of ErbB-4 expression.Reduced from �18. D, higher magnification of tissue core (C) withretained ErbB-4 expression. Reduced from �400. E, higher magni-fication of tissue core (C) with loss of ErbB-4 expression. Tumorswith retained expression showed ErbB-4 levels similar to baselineexpression in normal urothelium. Reduced from �400.

(pH 8) at 99C for 45 minutes. Endogenous peroxidase was

DISTINCTIVE EXPRESSION PATTERN OF ERBB FAMILY RECEPTORS 355

blocked using 0.3% hydrogen peroxide in 90% methanol for5 minutes and stained with antibodies against EGFR extra-cellular domain, clone 31G7 (Zymed Laboratories, San Fran-cisco, California), ErbB-2 cytoplasmic domain (Lab VisionCorp., Fremont, California), ErbB-3 C-terminus (Santa CruzBiotechnology, Santa Cruz, California) and ErbB-4 aa 1249-1264 (Lab Vision) overnight at 4C. Bound primary antibodieswere visualized by avidin-biotin complex assay with 3,3=-dia-minobenzidine as the chromogen (Dako™) and hematoxylin asthe counterstain.

Sections of 3 normal ureters from patients undergoingnephrectomy for renal cell carcinoma with no evidence ofurothelial neoplasm involving the pelvis and ureter wereincluded in the microarray and they served as internal pos-itive controls. As a negative control for secondary antibodiesand the color detection system, slides of bladder tumor mi-croarrays including normal ureters were processed withoutthe primary antibodies.

The specificity of staining was additionally assessed onmicroarray sections exposed to primary antibody blockedwith the corresponding immunizing peptide at a 1:10 molarratio. For this purpose we used recombinant full-lengthEGFR (Invitrogen™), the specific ErbB-3 peptide sc-7390P(Santa Cruz Biotechnology) and ErbB-4 immunizing pep-tide, which was custom synthesized elsewhere. No immuniz-ing peptide was available for ErbB-2.

Quantification of Staining in UC SpecimensThe staining intensity of each antigen in the tissue microar-ray was measured using a GenoMx™ automated scanneraccording to manufacturer recommendations. The tumorarea was outlined in each tissue core and its staining inten-

FIG. 3. A, EGFR expression pattern with primary antibody. B, EReduced from �200. C, example of control studies shows reproduc

samples each from 4 randomly selected low grade superficial (1-4) and 4staining with (gray bars) and without (black bars) pre-incubation with i

sity was compared to the intensity of staining in normalurothelium for each protein. For each tumor sample the RELof each protein was calculated using the equation, REL �P � (M � H), where P represents the area of the tumor withpositive staining in percent, M represents the area of the tumorwith medium staining intensity and H represents the area ofthe tumor with high staining intensity. Tumors were dividedinto 2 groups, that is those with REL scores above and belowthe median value for each protein.

Statistical AnalysisDifferences in the relative expression levels of markers be-tween clinical groups were compared using Van der WaerdenScore tests. The Kaplan-Meier product limit method was usedto determine the overall survival probability for subgroupsbased on the staining score. Median receptor staining intensitywas used as the cutoff point to dichotomize patients into highand low score groups. Survival curves were compared using thelog rank test with 2-sided p � 0.05 considered significant for alltests. Univariate and multivariate Cox proportional hazardsmodels were fit to assess the relationships between immuno-staining scores and survival in the absence and presence ofother clinical covariates. Statistical analyses were performedusing the R, version 1.9.0 (http://www.r-project.org/) and S-Plus® statistical packages.

RESULTS

Superficial bladder cancer was diagnosed in 34% of patients,including Ta in 18% and T1 in 16%, while 66% had moreadvanced stage tumors, including T2-4N0 in 42% and TxN�in 24%, on final pathological analysis. Of the tumors 73%

expression pattern after inhibition with its immunizing peptide.y of EGFR expression measurement (calculation of REL) in 5 core

GFRibilit

high grade invasive (5-8) bladder UCs. Results are presented formmunizing peptide.

DISTINCTIVE EXPRESSION PATTERN OF ERBB FAMILY RECEPTORS356

were high grade and 50% were characterized by nonpapil-lary growth patterns. Patients with high grade disease hada worse prognosis than patients with low grade disease witha median overall survival of 25 vs 84.6 months, patients withinvasive disease had a worse prognosis than those withoutinvasion with a median overall survival of 20.9 vs 75.2months and similarly patients with nonpapillary tumorshad a worse prognosis than those without papillary tumorswith a median overall survival of 23.5 vs 67.4 months (eachp �0.0001). Kaplan-Meier estimates of 5-year overall andrecurrence-free survival for all patients were 58% and 27%,respectively.

Staining of tumors by all the ErbB family members waspredominantly membranous with occasional detectableimmunoreactivity in cytoplasm. Normal urothelium stainedpositively for all ErbB family members and staining waspredominantly membranous with a minimal diffuse cyto-plasmic reaction. Figures 1 and 2 show examples of EGFRand ErbB-4 to demonstrate this pattern of predominantmembrane staining. Normal urothelium and tumor samplesthat were not incubated with primary antibody did not showany positive reaction. The specificity of staining for EGFR,ErbB-3 and ErbB-4 was verified by pre-incubation of theprimary antibody with its respective immunizing peptide,which showed a greater than 80% decrease in staining in-tensity. The reproducibility of staining patterns was testedon a separate microarray containing 4 low grade superficial(Ta, T1a) and 4 high grade invasive (T1b and higher) tu-

TABLE 1. Cox proportional hazards analysis of variablespredicting overall and recurrence-free survival

No. Pts

Survival p Value

Overall Recurrence-free

EGFR 182 NS 0.028HER-2 184 NS NSHER-3 128 NS NSHER-4 124 0.0037 NSCombined EGFR � HER-4: 0.075 NS

High EGFR/low HER-4 32Low EGFR/high HER-4 35

Growth pattern: �0.0001 NSPapillary 128Nonpapillary 120

Grade: �0.0001 NSLow 84High 164

Invasiveness: �0.0001 NSSuperficial 110Invasive 138

FIG. 4. Kaplan-Meier estimate of overall survival stratified by ErbB-4 extumors.

mors. Figure 3 shows an example of these control studies forEGFR.

Patients whose tumors expressed higher EGFR and lowerErbB-4 levels had significantly shorter recurrence-free andoverall survival (p � 0.028 and 0.0037, respectively, table 1).Based on the published literature we predicted that EGFRexpression levels would correlate with outcome but we wereintrigued that ErbB-4 expression levels were prognostic. Inpatients whose tumors retained ErbB-4 expression the5-year overall survival rate was 65% compared with 35%in those whose tumors expressed lower ErbB-4 levels(p � 0.0037). Median overall survival was 10.1 and 2.8years, respectively (fig. 4, A). On multivariate analysis ad-justed for grade, invasiveness and growth pattern a higherEGFR level remained significantly associated with recur-rence-free survival (p �0.04), although the ErbB-4 level wasno longer significantly associated with overall survival. Onsubset analyses of patients with invasive and noninvasivedisease separately the only statistically significant correla-tion was between high EGFR expression and decreased re-currence-free survival in those with noninvasive disease(p � 0.026). There was no such correlation with overallsurvival or in patients with invasive disease. Of note ErbB-2and ErbB-3 levels were not associated with overall or recur-rence-free survival. High EGFR and low ErbB-4 levels weresignificantly associated with nonpapillary, high grade andinvasive tumors (p �0.002, table 2). Although ErbB-2 levelwas associated with growth pattern and grade, it was notsignificantly associated with invasive disease. The ErbB-3level was not associated with growth pattern, grade or stage.

We evaluated whether there was any correlation betweenexpression levels of ErbB family receptors in bladder tu-mors. We found that the strongest statistically significantcorrelation was between EGFR and ErbB-4 (correlation coef-ficient –0.38, p �0.0001) compared with EGFR and ErbB-2(correlation coefficient –0.27, p � 0.001), and with ErbB-3 andErbB-4 (correlation coefficient 0.14, p �0.0001). Since EGFRand ErbB-4 levels were significantly associated with sur-vival, we compared patients whose tumors expressed highEGFR as well as low ErbB-4 with patients whose tumorsexpressed low EGFR and retained ErbB-4 expression. Toevaluate the combination we dichotomized the levels of ex-pression into high vs low with the median serving as thecutoff value, rather than using the level of receptor expres-sion as a continuous variable. Kaplan-Meier curve demon-strated that patients whose tumors expressed a high level ofEGFR as well as a low level of ErbB-4 had shorter overall

pression, and combined EGFR and ErbB-4 expression in urothelial

DISTINCTIVE EXPRESSION PATTERN OF ERBB FAMILY RECEPTORS 357

survival than patients whose tumors expressed a low level ofEGFR but retained ErbB-4 expression. However, the differ-ence attained only borderline statistical significance (p �0.075, fig. 4, B).

DISCUSSION

We observed that high EGFR and low ErbB-4 levels wereassociated with a high grade, invasive phenotype of UC.Similarly high EGFR and low ErbB-4 levels were associatedwith shorter recurrence-free and overall survival in patientswith UC.

EGFR expression was previously shown to predict dis-ease progression to muscle invasive or metastatic UC and itwas found to be an independent prognostic factor for deathon multivariate analysis.12 Our study confirmed these find-ings. We found that patients with higher EGFR levels haddecreased rates of recurrence-free survival compared tothose in patients with low EGFR expression levels. In con-trast to several reports, EGFR expression was also found tobe prognostic on multivariate analysis adjusted for grade,invasiveness and growth pattern. Interestingly tumors thatdid not lose ErbB-4 expression were associated with a morefavorable prognosis in our study. Patients with tumors thathad not lost ErbB-4 expression had improved 5-year overallsurvival rates (65% vs 35%) and they were more likely tohave superficial low grade tumors.

Conflicting reports on this topic have been published inbreast cancer.13,14 However, more recently Memon et alfound that ErbB-4 levels were associated with better sur-vival in patients with bladder cancer,15 which is consistentwith the findings in our study. These findings may be aresult of transcriptional regulation of the ErbB-4 gene inpatients with UC.16 We also found that of members of theErbB receptor family the strongest significant correlationwas between the levels of EGFR and ErbB-4 receptors. In-terestingly patients whose tumors expressed high levels ofEGFR and low levels of ErbB-4 had shorter overall survivalthan patients whose tumors expressed low levels of EGFRand retained ErbB-4 expression. Although it was not con-firmed in our study due to small sample size of the subsetanalysis, a combination of these 2 variables may lead tobetter prognostication in patients with bladder cancer thanwhen the variables are used alone.

To our knowledge the importance of ErbB-2 and ErbB-3in bladder cancer progression and survival has yet to beclarified. Several studies performed in various tumorsshowed that higher ErbB-2 levels are associated with poor

TABLE 2. ErbB family receptors and b

EGFR HER-2

Mean � SD p Value Mean � SD

Growth pattern: 0.0007Papillary 33.5 � 24.4 34.3 � 30.1Nonpapillary 46.2 � 30.6 26.4 � 26.7

Grade: 0.002Low 31.4 � 25.2 30.0 � 27.8High 44.0 � 28.9 31.8 � 29.7

Stage: �0.0001Superficial 31.1 � 23.8 35.3 � 25.6Invasive 46.6 � 29.7 26.8 � 27.8

prognosis,7,17,18 whereas other studies showed the con-

trary.6 Similarly reports of the association between ErbB-3levels and prognosis are also conflicting.6,15,19,20 Differencesin the incidence and prognostic significance of ErbB-2 andErbB-3 expression are likely attributable to different meth-odologies, including the assessment of receptor status (iedetection of amplification vs detection of over expression),the method used (ie polymerase chain reaction, fluorescencein situ hybridization or immunohistochemistry) and the def-inition of receptor positivity.

CONCLUSIONS

The expression profile of high EGFR and loss of ErbB-4 mayidentify a distinct phenotype of UC with an adverse prognosisand it may help select patients with bladder cancer for moreaggressive therapy. Approaches to modifying the activity ofEGFR and ErbB-4 may also have therapeutic potential for thetreatment of select patients with UC of the bladder.

ACKNOWLEDGMENTS

ErbB-4 immunizing peptide was synthesized at GenScriptCorp., Piscataway, New Jersey.

Abbreviations and Acronyms

EGFR � epidermal growth factor receptorNS � not significant

REL � relative expression levelUC � urothelial carcinoma

REFERENCES

1. Rajkumar T, Stamp GW, Pandha HS, Waxman J and GullickWJ: Expression of the type 1 tyrosine kinase growth factorreceptors EGF receptor, c-erbB2 and c-erbB3 in bladdercancer. J Pathol 1996; 179: 381.

2. Mellon K, Wright C, Kelly, P. Horne CH and Neal DE: Long-term outcome related to epidermal growth factor receptorstatus in bladder cancer. J Urol 1995; 153: 919.

3. Sriplakich S, Jahnson S and Karlsson MG: Epidermal growthfactor receptor expression: predictive value for the outcomeafter cystectomy for bladder cancer? BJU Int 1999; 83: 498.

4. Liukkonen T, Rajala P, Raitanen M, Rintala E, Kaasinen Eand Lipponen P: Prognostic value of MIB-1 score, p53,EGFr, mitotic index and papillary status in primary super-ficial (stage pTa/T1) bladder cancer: a prospective compar-ative study. The Finnbladder Group. Eur Urol 1999; 36:

er tumor pathological characteristics

HER-3 HER-4

alue Mean � SD p Value Mean � SD p Value

.037 NS �0.000130.6 � 28.0 19.1 � 20.825.7 � 22.5 4.6 � 7.1

S NS �0.000130.4 � 29.0 19.4 � 21.126.8 � 23.5 7.5 � 12.6

.036 NS �0.000131.7 � 28.0 20.6 � 21.025.8 � 23.3 5.6 � 9.6

ladd

p V

0

N

0

393.

DISTINCTIVE EXPRESSION PATTERN OF ERBB FAMILY RECEPTORS358

5. Miyamoto H, Kubota Y, Noguchi S, Takase K, Matsuzaki J,Moriyama M et al: C-ERBB-2 gene amplification as a prog-nostic marker in human bladder cancer. Urology 2000; 55:679.

6. Chow NH, Chan SH, Tzai TS, Ho CL and Liu HS: Expressionprofiles of ErbB family receptors and prognosis in primarytransitional cell carcinoma of the urinary bladder. ClinCancer Res 2001; 7: 1957.

7. Sato K, Moriyama M, Mori S, Saito M, Watanuki T, Terada Ket al: An immunohistologic evaluation of C-erbB-2 geneproduct in patients with urinary bladder carcinoma. Cancer1992; 70: 2493.

8. Kim JH, Tuziak T, Hu L, Wang Z, Bondaruk J, Kim M et al:Alterations in transcription clusters underlie developmentof bladder cancer along papillary and nonpapillary path-ways. Lab Invest 2005; 85: 532.

9. Mostofi FK: Histological typing of urinary bladder tumours. In:Histological Typing of Urinary Bladder Tumours. Edited byFK Mostofi, CJ Davis and IA Sesterhenn. New York: Springer1999.

10. Sobin LH and Wittekind C: TNM Classification of MalignantTumors. New York: John Wiley 1997.

11. Sen S, Zhou H, Zhang RD, Yoon DS, Vakar-Lopez F, ItoS et al: Amplification/overexpression of a mitotic kinasegene in human bladder cancer. J Natl Cancer Inst 2002;94: 1320.

12. Nguyen PL, Swanson PE, Jaszcz W, Aeppli DM, Zhang G,Singleton TP et al: Expression of epidermal growth factorreceptor in invasive transitional cell carcinoma of the uri-nary bladder. A multivariate survival analysis. Am J Clin

Pathol 1994; 101: 166.

13. Bieche I, Onody P, Tozlu, S. Driouch K, Vidaud M andLidereau R: Prognostic value of ERBB family mRNA ex-pression in breast carcinomas. Int J Cancer 2003; 106: 758.

14. Suo Z, Risberg B, Kalsson MG, Willman K, Tierens A,Skovlund E et al: EGFR family expression in breastcarcinomas. c-erbB-2 and c-erbB-4 receptors have differ-ent effects on survival. J Pathol 2002; 196: 17.

15. Memon AA, Sorensen BS, Meldgaard P, Fokdal L, Thykjaer Tand Nexo E: The relation between survival and expressionof HER1 and HER2 depends on the expression of HER3 andHER4: a study in bladder cancer patients. Br J Cancer2006; 94: 1703.

16. Amsellem-Ouazana D, Bieche I, Tozlu S, Botto H, Debre B andLidereau R: Gene expression profiling of ERBB receptorsand ligands in human transitional cell carcinoma of thebladder. J Urol 2006; 175: 1127.

17. Slamon DJ, Godolphin W, Jones LA, Holt JA, Wong SG, KeithDE et al: Studies of the HER-2/neu proto-oncogene in hu-man breast and ovarian cancer. Science 1989; 244: 707.

18. Gullick WJ: The role of the epidermal growth factor receptorand the c-erbB-2 protein in breast cancer. Int J Cancer,suppl., 1990; 5: 55.

19. Witton CJ, Reeves JR, Going JJ, Cooke TG and Bartlett JM:Expression of the HER1-4 family of receptor tyrosine ki-nases in breast cancer. J Pathol 2003; 200: 290.

20. Pawlowski V, Revillion F, Hebbar M, Hornez L and Peyrat JP:Prognostic value of the type I growth factor receptors in alarge series of human primary breast cancers quantifiedwith a real-time reverse transcription-polymerase chain

reaction assay. Clin Cancer Res 2000; 6: 4217.