C V 2010 The American Laryngological,

The Laryngoscope

Rhinological and Otological Society, Inc.

Increased Amphiregulin Expression as a Biomarker of Cholesteatoma Activity

MiMi P. Macias, PhD; Richard D. Gerkin, MD, MS; John D. Macias, MD

Objectives/Hypothesis: The purpose of this study was to evaluate human surgical specimens for cholesteatoma-associated changes in amphiregulin expression and determine potential relations to clinical disease variables. Amphiregulin, an epidermal growth factor receptor ligand, has functions in normal epithelial proliferation and aberrant neoplastic cell growth and is proinflammatory (e.g., rheumatoid arthritis, fibrosis) and active in hyperproliferative cutaneous conditions including psoriasis and wound healing. These known amphiregulin activities and the characteristic epithelial expansion and bone erosion of cholesteatoma pathophysiology prompted testing of the hypothesis that amphiregulin expression levels are altered in cholesteatoma and correlate to the disease state. Study Design: Prospective experimental study, cross-sectional analysis. Methods: Relative changes in amphiregulin gene expression were quantitated by real-time reverse-transcription polymerase chain reaction analyses of cholesteatoma epithelium compared to uninvolved control tissues from patients postauricular and external auditory canal regions. Western immunoblot assays were performed for qualitative evaluation of amphiregulin protein expression. The t test and Fisher exact test were used for analysis. Results: A statistically significant increase in amphiregulin gene expression was associated with cholesteatoma specimens compared to uninvolved postauricular skin (PAS) and external auditory canal (EAC) skin, P .004 and P .002, respectively. From comparisons of 60 sets of skin pairs, the mean ratio of amphiregulin RNA expression for cholesteatoma/
From the Biomedical Research Program, The EAR Foundation of Arizona (M.P.M.); Scientific Services, Banner Health Research Institute (R.D.G.); and Macias Otology, P.C. (J.D.M.), Phoenix, Arizona, U.S.A. Editors Note: This Manuscript was accepted for publication June 29, 2010. This research was supported by Banner Good Samaritan Medical Center, The Nathan Cummings Foundation with the support of Sheila and Michael Zuieback, The Arizona Community Foundation, The National Organization for Hearing Research Foundation, and The EAR Foundation of Arizona. The authors have no other funding, financial relationships, or conflicts of interest to disclose. Send correspondence to MiMi P. Macias, PhD, Biomedical Research Program, The EAR Foundation of Arizona, 668 North 44th St, Ste 300, Phoenix, AZ 85008. E-mail: mpmacias@earfoundationaz.com DOI: 10.1002/lary.21142

PAS is 4.94 (standard error of the mean [SEM] 1.53, n 30) and for cholesteatoma/EAC is 7.70 (SEM 1.57, n 30). Conclusions: Amphiregulin is overexpressed in epithelial tissues of human cholesteatoma. Significant relationships were identified between increased amphiregulin expression levels and the extent of cholesteatoma migration and bone erosion. Our study results indicate amphiregulin is a potential biomarker of early cholesteatoma disease processes. Key Words: Cholesteatoma, hyperproliferative, migration, biomarker, epidermal growth factor receptor, amphiregulin, real-time reverse-transcription polymerase chain reaction. Level of Evidence: 1b. Laryngoscope, 120:22582263, 2010

INTRODUCTION
Cholesteatoma (CHLST) is a common, progressive disease of the temporal bone that affects both children and adults. The invasive properties of the CHLST sac (which is composed of hyperproliferative desquamating, keratinizing squamous epithelium), frequent bone erosion of the ossicular chain, and high rate of recurrence following surgical removal are responsible for associated hearing loss and other serious morbidity.1 However, clinical progression varies greatly among individuals, with some CHLSTs showing rapid growth and bone erosion and others remaining indolent for years. Surgery represents the only interventional option, and multiple operations are often required for eradication.2 Identifying stimulatory signals that functionally drive the expansion of human CHLST epithelium will help distinguish potential molecular targets for developing preventive and nonsurgical intervention strategies. Amphiregulin (AR) is a major autocrine growth factor for normal proliferation of human keratinocytes and fibroblasts in many tissues including skin, colon, spleen, breast, ovary, and T lymphocytes.3 However, AR also increases the proliferation and invasiveness of various neoplastic cell types4 and is upregulated in inflammatory and hyperproliferative skin lesions such as psoriasis, actinic keratoses, and verrucae.5,6 The experimental overexpression of AR in keratinocytes of transgenic mice is sufficient to induce epidermal hyperproliferation and Macias et al.: Amphiregulin Expression in Cholesteatoma

Laryngoscope 120: November 2010

2258

promote inflammatory skin responses.7 AR is a ligand of the epidermal growth factor receptor (EGFR), a transmembrane signal transducing receptor that is a crucial regulator of cellular proliferative, migratory, and differentiation responses. Previous reports have shown human CHLSTs express elevated levels of EGFR and family member ErbB28 in the basal and suprabasal layers of the diseased epithelium.9,10 The striking functional capacity of AR to stimulate proliferation and invasiveness in other cutaneous systems, along with reports that the EGFR receptors for AR11 are expressed in human CHLST, led us to investigate whether AR expression is dysregulated in CHLST epithelial lesions and potentially contributes to the etiology and pathologic changes in the temporal bone.

MATERIALS AND METHODS Patients

Thirty-three patients undergoing surgery for CHLST removal between May 2003 and September 2008 were prospectively enrolled in this study after providing informed consent for participation. Simple retraction pockets were excluded from the study. Protocols were approved by Banner Health Institutional Review Board. For the study, a skin specimen of approximately 15 mm3 was collected intraoperatively from the patients CHLST and the remainder was sent to pathology for routine examination. In addition, two sources of normal epithelia, postauricular skin (PAS) and external auditory canal (EAC) from the vascular strip, were collected from the subject to serve as uninvolved skin controls. Upon surgical removal, coded specimen sets were immediately transferred to separate 1.5-mL tubes containing stabilizing solution (RNAlater; Ambion, Austin, TX). Before processing, specimens were weighed and documented by digital imaging using a KODAK EDAS 290 System (Eastman Kodak Co., Rochester, NY). Prospectively, individual clinical histories were compiled and coded for subsequent correlative analysis with experimental gene expression data. Catalogued data included patient age, sex, family history of skin conditions, CHLST etiology (congenital, acquired, initial, recurrent), migration landmarks, extent of bone erosion, and presence or absence of infection.

RNA Preparation and Real-time Quantitative Polymerase Chain Reaction

Total RNA from each surgical specimen was isolated and DNase-treated by using the Absolutely RNA Miniprep Kit (Stratagene, La Jolla, CA). Each set of patient CHLST and control tissue specimens was batch processed to maximize consistency in handling. Tissues were homogenized on ice by using a rotor/stator 7-mm rotating-blade homogenizer; concentration and purity of isolated total RNA preparations were determined by spectrophotometry (BioMate3; Thermo Spectronic, Rochester, NY) measurements at 260 nm and 280 nm. Quantitation and integrity were visually confirmed when possible by 0.7% agarose/1X 3-(N-morpholino) propanesulfonic acid gel electrophoretic separation and ethidium bromide or SYBR Gold (Molecular Probes, Laryngoscope 120: November 2010

Eugene, OR) staining. cDNA was reverse transcribed from 2 lg aliquots of total RNA in a 20 lL reaction volume at 42 C for 90 minutes by using SuperScript II (Invitrogen, Carlsbad, CA) according to manufacturers conditions, 500 lM dNTP mix (Eppendorf, Hamburg, Germany), 1 U/lL RNase inhibitor (Invitrogen, Carlsbad, CA), and 1 lM modified oligo-dT primer, dT-ACP1 primer: 50 -CTGTGAATGCTGCGACTACGATXXXXX(T)1830 (Seegene, Rockville, MD). Real-time quantitative reverse-transcriptionpolymerase chain reaction (QRT-PCR) was run for each set of patient cDNA samples PAS, EAC, and CHLST, concurrently and in triplicate in a Stratagene Mx3000P with analysis software MxPro Version 4.0.1 (Stratagene, La Jolla, CA), for both AR and the internal normalizer control, glyceraldehyde-3-phosphate dehydrogenase (GAPDH). GAPDH demonstrated the most stable expression in these tissues when tested against alternative control gene options including beta-actin, 18S RNA, and cyclophilin (data not shown). Each 25 lL reaction included 30 ng cDNA, Brilliant SYBR Green QPCR Master Mix (Stratagene, La Jolla, CA), 30 nM ROX dye (United States Biochemical, Cleveland, OH), and the AR or GAPDH primer pair at 150 nM each primer: upstream 50 -GCTTAGAA GACAATACGTCAGG-30 , downstream 50 -GGGTCCATTG TCTTATGATCCAC-30 (AR); or upstream 50 -TCTGACTTCAACAGCGACAC-30 , downstream 50 -TGTTGCTGTAGC CAAATTCG-30 (GAPDH). Primers were designed using Amplify software (W. Engels, Madison, WI) to anneal to exon sequences and span genomic intron sequences. Relative amplification efficiencies of the primer pairs were optimized to be similar, and dissociation curves and no template control reactions confirmed specificity of the predicted QRT-PCR products and absence of DNA contamination. An additional level of correction was introduced by including passive fluorescent reference dye, 30 nM ROX, to normalize potential well-to-well differences. Cycling conditions were as follows: 94 C, 10 minutes; (94 C, 30 seconds; 60 C, 1 minute; 72 C, 1 minute) 40 cycles. GAPDH-normalized AR expression is presented as a fold-change by the ratios CHLST/PAS or CHLST/EAC, and calibrator AR ratio values for PAS or EAC are defined as 1.0. Error bars are derived from threshold cycles, fluorescence measurements at each cycle, and software-based estimates of imprecision of the normalizer. Twenty-seven patients had values for CHLST AR expression relative to both PAS and EAC; three had values for only CHLST/PAS, and three had values for only CHLST/EAC.

Protein Preparation and Anti-AR Western Blots

When both RNA and protein were isolated from a set of surgical specimens, each biopsy was first divided into two portions, the RNA portion extracted as described previously and the protein portion processed by using the PARIS lysate preparation system (Ambion, Austin, TX). Concentrations were determined by using Bradford reagent (Sigma-Aldrich, St. Louis, MO), 595 nm absorbance readings and comparison to a standard Macias et al.: Amphiregulin Expression in Cholesteatoma

2259

curve prepared in parallel. Protein samples (15lg/lane) were loaded in duplicate and electrophoretically separated through discontinuous Laemmli 4%/12% sodium dodecyl sulfate polyacrylamide gels for transfer to Whatman Protran BA85, 0.45 lm, nitrocellulose membrane (Sigma-Aldrich, St. Louis, MO). Endogenous cellular AR is synthesized as a 252 amino acid transmembrane precursor protein that is glycosylated and cleaved by matrix metalloproteinase to release a mature 18 kDa polypeptide.6,12 Positive control sample lanes of recombinant human AR (rhAR) contained 2.5 ng/lane of a 98 amino acid, mature form AR (R&D Systems, Minneapolis, MN), and a 50-fold excess (125 ng/lane) of carrier protein, bovine serum albumin (BSA). Two different protein molecular weight marker mixes were used for more accurate sizing: Pre-stained Blue Ranger (Pierce/Thermo Scientific, Rockford, IL) and Sigma Wide (Sigma-Aldrich, St. Louis, MO). Protein gels were electrotransferred by using the Genie Blotter (Idea Scientific, Minneapolis, MN) at 6 V for 3 hours with transfer buffer 25 mM Tris/ 192 mM Glycine/20% methanol (pH 8.3). Duplicate blots were stained with MemCode Reversible (MCR) (Pierce/ Thermo Scientific, Rockford, IL) to assess transfer efficiency and were documented by using the KODAK EDAS 290 System and were then destained before Western blotting steps. Blocking and antibody incubation buffer was Starting Block T20 (Pierce/Thermo Scientific, Rockford, IL). Incubations and washes were performed at room temperature with constant agitation on a rotating platform (Davis R&D, Carlsbad, NM). Two sources of primary antibodies were pooled for anti-AR Western blots: AF262 goat antihuman AR polyclonal sera (R&D Systems, Minneapolis, MN), 1:1,000 working dilution, and N20 goat antihuman AR polyclonal sera (Santa Cruz Biotechnology, Santa Cruz, CA), 1:200 working dilution. Negative control primary antibody was normal goat serum (R&D Systems, Minneapolis, MN). Secondary antibody-enzyme conjugate was donkey antigoat IgG-alkaline phosphatase, 1:5,000 working dilution (Santa Cruz Biotechnology, Santa Cruz, CA). Color substrate for detection was 1-Step NBT/BCIP (Pierce/ Thermo Scientific, Rockford, IL) with 1 mM levamisole added fresh to eliminate endogenous phosphatase activity. Antibody incubation steps were for 1 hour, and color substrate development was for 20 minutes. Western results were recorded with the KODAK EDAS 290.

RESULTS
Of the 33 patients enrolled in the study, 20 (60.6%) were males and 13 (39.4%) were females. The mean age was 39.1 years, and ages ranged from 7 to 72 years. At the time of surgery, 16 of the 33 patients (48.5%) were presenting with recurrent disease. Five of these 16 had also undergone contralateral procedures. Of the other 17 patients with initial, acquired CHLST, three had previously undergone surgical removal of CHLST in their contralateral ear. Nineteen of the 33 enrolled patients had infections at the time of surgery. Thirty of the 33 patients had PAS measured, and 27 of these 30 as well as the remaining three subjects had EAC measured. CHLST/PAS comparisons showed 22 of 30 CHLST samples (73.3%) had AR ratios greater than 1.0 (P .004), and the mean AR expression ratio for CHLST/PAS was 4.94 (SEM 1.53, n 30, median 2.87). Twentyfour of the 30 paired CHLST/EAC samples (80.0%) had AR ratios greater than 1.0 (P .002), and the mean AR expression ratio was 7.70 (SEM 1.57, n 30, median 5.08). Interassay variation and reproducibility of technical replicates was measured for all 60 sample pairs, and the mean coefficient of variation was 6.94%. These data are displayed in the comparative quantity charts in Figure 1. Protein lysates were prepared from patients PAS, EAC, and CHLST specimens for anti-AR Western immunoblots to evaluate endogenous AR expression at the protein level. Representative duplicate membranes prepared in parallel are shown with lysate patterns visualized by transient, nonspecific stain (left images in Fig. 2A and 2B). The membranes were then destained and incubated with primary antibodies, either goat antiARspecific antibodies or normal goat serumnegative control. Colorimetric detection reveals AR-specific antibody staining of the 18 kDa AR product in CHLST lysate only (Fig. 2A, right image). Of the six patients tested with Western blots, four patients specimens exhibited AR-specific protein staining identical to the pattern shown in Fig. 2A. Two patients tested did not show endogenous AR staining of any tissue. The positive control lane containing 2.5 ng rhAR protein was as expected, only stained by anti-AR specific polyclonal sera (Fig. 2A) and not normal goat serum (Fig. 2B). Negative control carrier protein BSA (66 kDa), in 50-fold excess of rhAR, was weakly visible by MCR staining but was not recognized by antibodies on either Western blot (Fig. 2A or 2B), further confirming anti-AR antibody specificity. The negative control normal goat serum appeared to react weakly with a few unknown human products, one of approximately 25 kDa in PAS and EAC lysates (Fig. 2B) that were not consistently seen in other patient lysates and one comigrating with BSA at 66 kDa, which may be reactivity or passive reagent trapping (see staining intensity at 66 kDa in all lysate lanes, left images in Fig. 2A and 2B). Classification of the clinical severity of patient pathology was based on evaluation of both the extent of CHLST epithelia migration and the extent of CHLSTassociated bone erosion. For both classifications, Macias et al.: Amphiregulin Expression in Cholesteatoma

Statistical Analysis
Descriptive statistics were used. Continuous variables were reported as means and the standard error of the mean (SEM). Medians were also reported. Categorical variables were reported as percentages. Continuous variables were analyzed using t tests, or Mann-Whitney tests if the data were not normal. The Fisher exact test was used to analyze categorical variables. All specimens were run in triplicate. Standard deviations (SD) and coefficients of variation (SD/mean) were calculated for each sample. A two-tailed P < .05 was considered significant. Laryngoscope 120: November 2010

2260

AR overexpression is clearly associated with diseased CHLST epithelia. Additional tests were performed to examine relationships between levels of AR gene expression and patients clinical information. Analyses demonstrated an inverse relationship in which higher levels of AR expression were found early in CHLST development (category 1 CHLSTs). The CHLST/EAC AR ratio had a statistically significant, inverse relationship to both extent of CHLST migration (P .007) and degree of CHLST-associated bone erosion (P .011) (Table I). The CHLST/PAS AR ratio also demonstrated significant, inverse relation to CHLST-associated bone erosion (P .032) but not to migration, as mean CHLST/PAS AR ratios were elevated similarly in both migration category 1 and 2 lesions (Table I). Neither CHLST/PAS nor CHLST/EAC ratio was related to infection or recurrent disease. Recurrence of disease correlated significantly with the presence of infection (P .006), but not with the extent of migration or bone erosion. Recurrence was also related to age, with younger patients more likely to experience recurrence. The mean age in those with recurrence was 31.3 years, and the mean age in those without recurrence was 45.8 years (P .034). Gender was significantly associated with migration in 18 of 20 males in our study showing more aggressive disease (P .009) as compared with seven of 13 females.

DISCUSSION
CHLST has long been recognized as a serious condition of the temporal bone, with hearing loss the frequent outcome of its propensity for rapid growth and bone destruction. Better understanding of molecular factors regulating disease progression and the biomarkers indicative of CHLST activity can help identify potential targets for the long-term objectives of developing nonsurgical and preventive treatment strategies for CHLST. Our study is the first to report the statistically significant upregulation of AR in human middle ear CHLST and the correlation to disease state. Several technical and biological limitations were identified and overcome to perform this study. Challenges to studies of human CHLST include 1) limited quantity of clinical biopsy material available for experimental manipulations and 2) inherent biological heterogeneity of human clinical samples. To address logistical constraints, we first optimized protocols for collecting and processing patient specimen sets. Working at this time with only one otologic surgeon (j.d.m.) ensured crucial technical confidence that specimen sets were collected and stabilized in a well-defined manner. Next, sensitive processing methods combined with real-time PCR technology maximized RNA and protein purification yields and the quantitative information obtained from limited starting materials. Intraindividual controls were used for experimental comparisons in our study design. Specimens collected from each patients nondiseased skin regions served as personalized normal controls to minimize interindividual biological differences that potentially complicate data interpretation of heterogeneous clinical isolates. Macias et al.: Amphiregulin Expression in Cholesteatoma

Fig. 1. Comparative quantitation charts. Relative amphiregulin (AR) values with upper and lower error bars are shown for each pair of patient cholesteatoma (CHLST) and postauricular skin (PAS) or external auditory canal (EAC) calibrator reference specimens. Quantity of AR mRNA in PAS or EAC was defined as 1.0 (solid black bars). CHLST AR values (striped bars) are expressed as glyceraldehyde-3-phosphate dehydrogenase (GAPDH)normalized relative ratios of CHLST/PAS or CHLST/EAC. In both graphs, offscale CHLST AR ratio values are indicated next to the truncated striped bar(s). (A) CHLST/PAS AR graph. The mean ratio of AR RNA expressed in CHLST relative to PAS is 4.94 (standard error of the mean [SEM] 1.53, n 30). (B) CHLST/EAC AR graph. The mean ratio of AR mRNA expressed in CHLST relative to EAC is 7.70 (SEM 1.57, n 30). Specimen 22 was measured using endpoint real-time quantitative reverse-transcription polymerase chain reaction and KODAK EDAS 290 digital image analysis system; CHLST/PAS AR ratio 3.8 (standard deviation [SD] 0.2), CHLST/EAC AR ratio 10.6 (SD 0.1).

category 2 CHLSTs are defined to be more pathologically aggressive than category 1. Migration of eight of the 33 CHLSTs was classified as category 1 and limited to the middle ear and/or attic. Migration of the other 25 specimens was classified as category 2 and extended into the antrum and/or mastoid. For bone erosion, eight of 33 CHLSTs were classified as category 1 and exhibited no bone destruction or erosion of one or more ossicles, whereas the 25 category 2 CHLSTs were classified by erosion of the scutum and/or mastoid. Laryngoscope 120: November 2010

2261

Fig. 2. Representative Western blot detection of amphiregulin (AR) protein in a patients set of epithelial lysates (15 lg/lane). Duplicate blots (left images in panels A and B) are transiently stained with MemCode Reversible to visualize protein transfer and pattern and then destained for immunoblotting. (A) Anti-AR Western immunoblot. Human AR protein (18 kDa) detected with anti-AR specific antibodies is indicated with the AR-labeled arrow. Positions of recombinant human AR (rhAR) positive () control at 2.5 ng, bovine serum albumin (BSA) negative (-) control at 125 ng per lane, and Sigma Wide molecular weight standards are indicated. (B) Negative control Western blot assay using normal goat serum. Prestained Blue Ranger molecular weight standards and sizes are indicated to the left. PAS postauricular skin; EAC external auditory canal; CHLST cholesteatoma; con control.

Previous publications investigating pathophysiology of human CHLSTs generally used either PAS or EAC as appropriate normal skin controls, PAS being a distal ear-associated epithelium and EAC a proximal, migratory epithelial reference. Incorporating both in this current study allowed simultaneous comparisons of CHLST to two reference skin sources for better understanding of gene expression patterns in these pertinent yet distinct epithelial regions of the ear. Despite differences in the magnitude of AR upregulation, mean expression levels of AR in CHLST are increased as compared with both PAS and EAC reference controls. AR gene expression levels correspond inversely to the clinical stage of disease based on migration and temporal bone erosion. Elevated AR expression in early stage clinical specimens was independent of relation to age, recurrence, or infection and suggests AR stimulation may contribute to progressive growth during CHLST development. These findings are even more striking when considering the clinically diverse backgrounds under which the samples were collected. If AR plays an important functional role during a specific point in CHLST

development, reporting averaged values from clinical specimens at different stages of progression potentially underestimates the relative magnitude of AR overexpression. Understandably, CHLST is often identified later in disease when epithelial expansion has caused notable sequelae such as hearing loss or infection. And at the time of surgery, the metabolic status of specimens is varied and generally unknown. Specimens collected during a metabolically active phase of disease (e.g., response to proinflammatory signaling molecules) or an indolent or latent period (e.g., following antibiotic resolution of a secondary infection) may present correspondingly up- and down-regulated levels of AR. To address the relationship of metabolic status to AR expression, we used QRT-PCR assays to measure the expression of proliferating cell nuclear antigen (PCNA) in several of our paired specimen sets. Our results are consistent with those of previous reports identifying CHLST as hyperproliferative tissue and having upregulated proliferation markers PCNA13 and Ki-67 nuclear antigen.14 Our CHLST study specimens with higher AR expression also had higher PCNA expression (data not shown).

TABLE I. Cholesteatoma Amphiregulin Expression Levels and Relation to Clinical Variables.

Migration Category 1(n 8) Migration Category 2(n 25) P Value* Erosion Category 1(n 8) Erosion Category 2(n 25) P Value*

CHLST/PAS AR mean (SEM) CHLST/EAC AR mean (SEM)

5.50 (3.22) 15.08 (2.89)

4.77 (1.78) 5.45 (1.59)

.845 .007

10.30 (2.77) 14.73 (2.93)

2.99 (1.67) 5.56 (1.62)

.032 .011

Migration category 1 extension to middle ear and/or attic; migration category 2 extension to antrum and/or mastoid; erosion category 1 none or erosion of one or more ossicles; erosion category 2 erosion of scutum and/or mastoid. *Statistical significance is indicated by P  .05. CHLST cholesteatoma; PAS postauricular skin; AR amphiregulin; SEM standard error of the mean; EAC external auditory canal.

Laryngoscope 120: November 2010

Macias et al.: Amphiregulin Expression in Cholesteatoma

2262

This study demonstrates AR is overexpressed in CHLST; in the presence of its cell-surface receptor EGFR, AR has the potential to play an important role in driving dynamic processes of the disease state. Future studies of the function of AR in human CHLST disease progression will likely require synchronized, manipulatable model systems such as the gerbil animal model, which has been carefully defined,15 or development of a short-term human skin organ culture system to test effects of growth factor and signaling inhibitors on molecular responses in human CHLST tissues. In both normal and tumor cells, activation of EGFR-dependent signaling cascades by binding of specific ligand such as AR induces an intrinsic tyrosine kinase and mediates complex proliferative, invasive, angiogenic, and antiapoptotic responses. Recent advances in U.S. Food and Drug Administrationapproved inhibitors of EGFR and overexpression of components of the EGFR signaling pathway in CHLST make this molecular network a compelling target for future studies of treatment options for CHLST. Development of topical delivery systems for anti-EGFR therapeutics16 (e.g., monoclonal antibodies, tyrosine kinase inhibitors, ligand-toxin conjugates, or antisense oligonucleotides) may be efficacious for inhibiting CHLST progression in nonsurgical candidates or preventing recurrence of disease.

BIBLIOGRAPHY
1. Albino AP, Kimmelman CP, Parisier SC. Cholesteatoma: a molecular and cellular puzzle. Am J Otol 1998;19: 719. 2. Rosenfeld RM, Moura RL, Bluestone CD. Predictors of residual-recurrent cholesteatoma in children. Arch Otolaryngol Head Neck Surg 1992;118:384391. 3. Plowman GD, Green JM, McDonald VL, et al. The amphiregulin gene encodes a novel epidermal growth factorrelated protein with tumor-inhibitory activity. Mol Cell Biol 1990;10:19691981. 4. Willmarth NE, Ethier SP. Autocrine and juxtacrine effects of amphiregulin on the proliferative, invasive, and migratory properties of normal and neoplastic human mammary epithelial cells. J Biol Chem 2006;281: 3772837737. 5. Yamane S, Ishida S, Hanamoto Y, et al. Proinflammatory role of amphiregulin, an epidermal growth factor family member whose expression is augmented in rheumatoid arthritis patients. J Inflamm (Lond) 2008;5:5. 6. Piepkorn M. Overexpression of amphiregulin, a major autocrine growth factor for cultured human keratinocytes, in hyperproliferative skin diseases. Am J Dermatopathol 1996;18:165171. 7. Cook PW, Piepkorn M, Clegg CH, et al. Transgenic expression of the human amphiregulin gene induces a psoriasis-like phenotype. J Clin Invest 1997;100:22862294. 8. Sakamoto T, Kondo K, Yamasoba T, et al. Overexpression of ErbB-2 protein in human middle ear cholesteatomas. Laryngoscope 2004;114:19881991. 9. Bujia J, Kim C, Holly A, et al. Epidermal growth factor receptor (EGF-R) in human middle ear cholesteatoma: an analysis of protein production and gene expression. Am J Otol 1996;17:203206. 10. Ergun S, Zheng X, Carlsoo B. Expression of transforming growth factor-alpha and epidermal growth factor receptor in middle ear cholesteatoma. Am J Otol 1996;17:393396. 11. Klingelhofer J, Moller HD, Sumer EU, et al. Epidermal growth factor receptor ligands as new extracellular targets for the metastasis-promoting S100A4 protein. FEBS J 2009;276:59365948. 12. Shoyab M, McDonald VL, Bradley JG, et al. Amphiregulin: a bifunctional growth-modulating glycoprotein produced by the phorbol 12-myristate 13-acetate-treated human breast adenocarcinoma cell line MCF-7. Proc Natl Acad Sci U S A 1988;85:65286532. 13. Bujia J, Sudhoff H, Holly A, et al. Immunohistochemical detection of proliferating cell nuclear antigen in middle ear cholesteatoma. Eur Arch Otorhinolaryngol 1996;253: 2124. 14. Mallet Y, Nouwen J, Lecomte-Houcke M, et al. Aggressiveness and quantification of epithelial proliferation of middle ear cholesteatoma by MIB1. Laryngoscope 2003;113: 328331. 15. Kim HJ, Tinling SP, Chole RA. Increased proliferation and migration of epithelium in advancing experimental cholesteatomas. Otol Neurotol 2002;23:840844. 16. Robinson R, Bertram JP, Reiter JL, et al. New platform for controlled and sustained delivery of the EGF receptor tyrosine kinase inhibitor AG1478 using poly(lactic-coglycolic acid) microspheres. J Microencapsul 2010;27: 263271.

CONCLUSION
Our study results show the growth factor AR is overexpressed in the epithelium of aural CHLST compared to nondiseased epithelial sites of the ear. Demographic and clinical variables were prospectively evaluated to investigate correlations between AR levels and CHLST and to provide evidence that increased AR expression is a potential biomarker of CHLST activity. Significant inverse relationships were identified between AR gene expression and both the extent of expansion into the temporal bone and the severity of erosion. AR levels were significantly higher when CHLST migration was in the earliest stages of temporal bone invasion, before the development of significant tissue destruction. The significantly increased expression of AR detected at the onset of CHLST pathology suggests a functional role during developmental stages of disease progression and makes AR a promising target for pharmacologic medical intervention.

Acknowledgment
The authors gratefully acknowledge the support of Victor Hsu, Jamie Lee, Tanya Thal, John Pepper, Jerry and Amy Hugo, Lylis Olsen, Linda Mardel, Kim Greenlief, and Shayne Davis.

Laryngoscope 120: November 2010

Macias et al.: Amphiregulin Expression in Cholesteatoma

2263