Prognostic Relevance of AGR2 Expression in Breast Cancer

Imaging, Diagnosis, Prognosis

Prognostic Relevance of AGR2 Expression in Breast Cancer Florian Rudolf Fritzsche,1Edgar Dahl,3 Stefan Pahl,1Mick Burkhardt,1Jun Luo,4 Empar Mayordomo,1Tserenchunt Gansukh,1Anja Dankof,1Ruth Knuechel,3 Carsten Denkert,1Klaus-Ju« rgen Winzer,2 Manfred Dietel,1andGlenKristiansen1, 2

Abstract Purpose:We aimed to evaluate the expression of the human anterior gradient-2 (AGR2)inbreast cancer on RNA and protein level and to correlate it with clinicopathologic data, including patient survival. Experimental Design: AGR2 mRNA expression was assessed by reverse transcription-PCR in 25 breast cancer samples and normal tissues. A polyclonal rabbit AGR antiserum was used for immunohistochemistry on155clinicopathologically characterized cases. Statistical analyses were applied to test for prognostic and diagnostic associations. Results: Immunohistochemical detection of AGR2 was statistically significantly associated with positive estrogen receptor status and lower tumor grade. AGR2-positive tumors showed significantly longer overall survival times in univariate analyses. For the subgroup of nodal-negative tumors, an independent prognostic value of AGR2 was found. Conclusions: The expression of AGR2 in breast cancer is strongly associated with markers of tumor differentiation (estrogen receptor positivity, lower tumor grade). A prognostic effect of AGR2 for overall survival could be shown, which became independently significant for the group of nodal-negative tumors.

Despite considerable diagnostic and therapeutic advances in entiation and expression of neural marker genes in a fibroblast the treatment of breast cancer in recent years, there is still an growth factor–dependent way (11). AGR2 has been shown urgent need for further molecular markers to provide the previously to be coexpressed with estrogen receptor (ER) in clinician with useful information concerning patient prognosis breast cancer cell lines (9), suggesting involvement of AGR2 in and possible therapeutic options. Kallikrein 5 (1), urokinase the tumor biology of hormonally responsive breast cancers. plasminogen activator, its inhibitor (2), tissue inhibitor of This is in agreement with the finding that an antiestrogen- metalloproteinases 1 (3), Ep-CAM (4), osteopontin (5), CD24 resistant derivative of the human T47D breast cancer cell line (6), and SFRP1 (7) are just a few examples of a growing list of that has lost ER expression also has a strongly reduced AGR2 potentially useful prognostic markers in breast cancer. We and expression (12). The estrogen-dependent expression of AGR2 others recently described up-regulation of the androgen- in breast cancer may be regulated by the four putative estrogen inducible gene anterior gradient-2 (AGR2) in prostate cancer response elements present in the AGR2 promoter (12). (6, 8). AGR2, which is also known as HAG-2 (9) or Gob-4 Although the Xenopus XAG-2 protein was shown to be secreted (10), is the human orthologue of the secreted Xenopus laevis (11), it is presently not clear whether the human AGR2 protein AGR protein XAG-2. It has a putative role in ectodermal is secreted in normal and malignant breast tissue as well (13). patterning of the frog embryo and is itself regulated by a Persson et al. (14) have shown by Basic Local Alignment number of fundamental embryonic molecules like noggin and Search Tool analysis (15) that AGR2 may represent a novel chordin (11). XAG-2 expression induces cement gland differ- member of the protein disulfide isomerase family. Protein disulfide isomerases are oxidoreductases of the endoplasmic reticulum involved in protein maturation in the endoplasmic reticulum (16). Authors’ Affiliations: 1Institute of Pathology and 2Breast Centre, Charite¤ , 3 In a recent study, Fletcher et al. showed immunohisto- Universita« tsmedizin Berlin, Berlin, Germany; Institute of Pathology, University chemical expression of AGR2 in 83% (n = 58) of breast cancer Hospital of the Rheinisch-Westfa« lische Technische Hochschule Aachen, Aachen, Germany; and 4Department of Urology, Johns Hopkins Medical Institutions, cases using a tissue microarray (17). They also found a Baltimore, Maryland significant correlation with ER expression and an inverse Received 9/20/05; revised 1/6/06; accepted 1/11/06. correlation with epidermal growth factor receptor expression. The costs of publication of this article were defrayed in part by the payment of page However, no significant association between AGR2 expression charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. and tumor grade, patient age, and presence of axillary lymph Note: F. R. Fritzsche, E. Dahl, and S. Pahl contributed equally to this work. node metastasis could be shown. The estrogen responsiveness Requests for reprints: Glen Kristiansen, Institut fu« r Pathologie, Charite¤ , of AGR2 was recently confirmed by Liu et al. (13) as the Universita« tsmedizin Berlin, Schumannstrasse 20-21,10117 Berlin, Germany. Phone: AGR2 mRNA expression in MCF-7 breast cancer cells increased 49-30-450-536145; Fax: 49-30-450-563945; E-mail: glen.kristiansen@ >7-fold in the presence of estrogen. More importantly, their charite.de. F 2006 American Association for Cancer Research. study showed that AGR2 can induce a metastatic pheno- doi:10.1158/1078-0432.CCR-05-2057 type in vivo. AGR2-transfected rat mammary cells (Rama 37)

Clin Cancer Res 2006;12(6) March 15, 2006 1728 www.aacrjournals.org Downloaded from clincancerres.aacrjournals.org on September 27, 2021. © 2006 American Association for Cancer Research. AGR Expression in Breast Cancer

in separate dots (19). An AGR2-specific cDNA fragment (EcoRI/NotI fragment of IMAGE cDNA clone with accession no. AA625485) was radiolabeled using a Megaprime labeling kit (Amersham Biosciences, Braunschweig, Germany), hybridized overnight at 68jC using Expres- sHyb Hybridization Solution (Clontech), washed, and exposed to Kodak XAR-5 X-ray film with an intensifying screen (Eastman Kodak Co., Rochester, NY). RNA preparation from paraffin-embedded tissue specimens. Archival formalin-fixed, paraffin-embedded tissue from 25 well-characterized

representative (44% pT1, 48% pN0, 40% G1-G2) breast cancer specimens and 16 normal breast tissues, both from the archives of the Institute of Pathology of the Rheinisch-Westfa¨lische Technische Hochschule Aachen, were used. For each formalin-fixed, paraffin- embedded tissue specimen, six 4-Am-thick tissue sections were cut with a microtome (Leica Microsystems, Bensheim, Germany) and transferred to a water bath filled with diethylpyrocarbonate-treated water. Sections were mounted on standard glass slides and dried for 1 hour at 60jC. Sections were deparaffinized and rehydrated as follows: 2 Â 15 minutes in xylole; 2 Â 15 minutes in 100% ethanol; and short rinses in 96%, 70%, and 50% ethanol followed by immersion in distilled water. Tissue material was transferred to a microcentrifuge tube and RNA was extracted according to the Trizol protocol supplied by the manufacturer (Life Technologies, Mannheim, Germany). cDNA was Fig. 1. Western blot analysis of AGR2 expression inhuman prostate cancer cell lines. PC3 cells (A) strongly express AGR2 (positive control), whereas CWR22Rv1cells synthesized according to the protocol supplied with the Clontech RT- (B) do not express detectable amounts of AGR2 (negative control). h-Actin was for-PCR-kit. used as loading control. Quantitative reverse transcription-PCR. AGR2 mRNA expression was analyzed using intron-spanning primers on the LightCycler system (Roche Diagnostics, Mannheim, Germany). Glyceraldehyde-3-phosphate injected in the mammary fat pads of syngenic rats induced a dehydrogenase (GAPDH) mRNA was used as reference to obtain relative high incidence of lung metastases. Because the incidence of expression values. Primers used in this study are presented in Table 1. primary tumors in this rat model was not increased, it can be Real-time reverse transcription-PCR (RT-PCR) was carried out with Fast concluded that AGR2 expression may be associated with Start DNA master hybridization probes (Roche Diagnostics). The metastasis but not with initiation of ER-positive tumors (13). conditions were as follows: initial denaturation in one cycle of 15 The study of Liu et al. (13) also analyzed human breast minutes at 95jC, followed by 40 cycles at 95jC for 20 seconds, 60jC tumors (n = 44) immunohistochemically and found a for 20 seconds, and 72jC for 30 seconds. Gene expression was significant correlation between ERa positivity and AGR2 quantified by the comparative CT method (20). expression; however, no correlation to patient survival was Patients. For immunohistochemistry, our study included 155 patients with breast cancer diagnosed at the Institute of Pathology, analyzed. University Hospital Charite´, Berlin, Germany, between 1991 and In our study, we aimed to investigate the expression of AGR2 1997. Patient age at the time of diagnosis ranged from 30 to 81 years immunohistochemically in a large breast cancer collective with a mean of 59 years. Clinical follow-up data, including overall (n = 155) and to evaluate prognostic properties by correlation survival and disease-free survival, were available for all cases. The with clinicopathologic variables. Furthermore, we have careful- median observation time for overall survival was 75 months for ly analyzed the AGR2 mRNA expression by real-time PCR in patients still alive at the time of analysis, and ranged from 1 to 162 ERa-positive and ERa-negative breast cancer. Our data indicate months. Thirty patients (19.4%) died during follow-up and 59 that AGR2 expression is associated with a positive ER status and patients (38.1%) experienced disease progression, defined by either also a favorable prognosis in patients with Unio Internationale metastatic disease or local recurrence. For statistical analysis of the Contra Cancrum stage (UICC) I disease. effect of adjuvant therapy, we arranged the patients into two groups: the first group (n = 69) had received either no adjuvant therapy, radiotherapy, or systemic therapy, excluding tamoxifen. The second Materials and Methods group had received tamoxifen with or without an additional systemic or local therapy (n = 80). For six patients, no data on adjuvant therapy Antibody specificity: Western blot of AGR2 in human prostate cancer were available. cell lines. Cultured PC-3 cells were subjected to standard Western blot analysis as described (18). A polyclonal rabbit antibody (1:1,000 dilution) raised against an AGR2-specific peptide was used to detect AGR2 protein expression in human prostate cancer cell line PC-3 Table 1. AGR2 and GAPDH primers used for real-time (positive control) and CWR22Rv1 (negative control). Protein levels RT-PCR of h-actin were examined by reprobing the same blot and served as loading controls. Only one weak nonspecific band was detected Gene Primer sequence Product (Fig. 1). size (bp) Expression analysis using a matched tumor/normal tissue array. The matched tumor/normal expression array from Clontech (Heidelberg, AGR2 5V-ATTGGCAGAGCAGTTTGTCC-3V 179 Germany) consists of 68 cDNAs, synthesized from human tumorigenic 5V-GAGCTGTATCTGCAGGTTCGT-3V and corresponding normal tissue (http://www.clontech.com/techinfo/ GAPDH 5V-TGGTCACCAGGGCTGCTT-3V 151 manuals/pdf/pt3424-1.pdf). Each pair was independently normalized 5V-AGCTTCCCGTTCTCAGCCTT-3V based on the expression of three housekeeping genes and immobilized

www.aacrjournals.org 172 9 Clin Cancer Res 2006;12(6) March 15, 2006 Downloaded from clincancerres.aacrjournals.org on September 27, 2021. © 2006 American Association for Cancer Research. Imaging, Diagnosis, Prognosis

Evaluation of the immunohistochemical stainings. The immunos- Ta b l e 2 . Clinicopathologic variables and association tains were examined by three pathologists who were blinded to patient with AGR2 expression of the tumor set outcome. We aimed to keep our scoring system of the AGR2 stainings simple to minimize interobserver variability and to enhance the Variable No. patients (%) P reproducibility of our findings in future studies. An immunoreactive score was established, as described by Remmele and Stegner (25). The Patients AGR2 AGR2 staining intensity was classified into four categories: negative (0), negative positive weakly positive (1), moderately positive (2), and strongly positive (3). The percentage of tumor cells staining positively was grouped Patient age (y) accordingly: 0% (0), 1% to 10% (1), 10% to 50% (2), 51% to 80% <60 78 12 (15.4) 66 (84.6) 0.473 (3), and 81% to 100% (4). The immunoreactive score was computed as z 60 77 8 (10.4) 69 (89.6) the product of the categorized staining intensity and percentage of Histology positive cells. Ductal 141 18 (12.8) 123 (87.2) 1.000 Statistical analysis. Statistical analysis was done using SPSS, version Lobular 14 2 (14.3) 12 (85.7) 13.0. Fisher’s exact and m2 tests were applied to assess the statistical pTstatus significance of the associations between expression of AGR2 and various clinicopathologic variables. Wilcoxon test and Mann-Whitney pT1 97 12 (12.4) 85 (87.6) 0.541* U test were used to compare expression levels. Correlations were pT 2 45 8 (17.8) 37 (82.2) calculated according to Spearman. Univariate survival analysis was pT 70(0)7(100) 3 carried out according to Kaplan-Meier, whereas differences in survival pT 60(0)6(100) 4 curves were assessed with the log-rank test. Cox regression analysis was pN status used for multivariate survival analyses. P values <0.05 were considered pN0 73 11 (15.1) 62 (84.9) 0.396* significant. pN1 40 5 (12.5) 35 (87.5) pN 20 2(10.0) 18(90.0) 2 Results pN3 22 2 (9.1) 20 (90.9) Histologic grade AGR2 expression analysis on the RNA level. To analyze the G1 38 2 (5.3) 36 (94.7) 0.001* expression of AGR2 in various solid tumors, including those of G2 76 6 (7.9) 70 (92.1) the breast, an AGR2-specific probe was hybridized onto an G3 41 12 (29.3) 29 (70.7) array containing 68 cDNA pair samples derived from multiple ER human tumors and corresponding normal tissues from Ne ga ti ve 4 5 1 4 (3 1.1) 31 (6 8 . 9) <0.001 individual patients (matched tumor/normal array; Clontech). Positive 98 5(5.1) 93(94.9) As shown in Fig. 2, AGR2 is clearly up-regulated in 6 of 9 C-erbB2 expression tumors derived from the breast but also in 3 of 7 endometrial 0, 1+ 93 13 (14.0) 80 (86.0) 0.786 tumors and 2 of 14 tumors derived from the kidney when 2+, 3+ 37 6 (16.2) 31 (83.8) compared with the corresponding normal cDNA. Furthermore, AGR2 was found to be expressed in A594 cells (see Fig. 2, *m2 test for trends. cell line 6), a human alveolar type II epithelium-like lung carcinoma cell line. AGR2 expression was further analyzed by real-time RT-PCR The selection of cases for this study was based on availability of tissue in a set of 25 primary breast cancer specimens and 16 normal and these were not stratified for any known preoperative or pathologic breast tissues. The result of the LightCycler expression analysis prognostic factors. Patients with systemic disease (pM1) at the time is diagrammatically presented in Fig. 3. Consistent with the of diagnosis were excluded. Histologic typing of tumors was carried data obtained by the matched tumor/normal array, AGR2 out according to the criteria of WHO (21). Tumor stage was determined expression in breast tumors was considerably up-regulated according to the guidelines of the UICC (22). Tumors were graded compared with its expression in normal breast tissue (median according to Bloom and Richardson in the modification of Elston and expression level, 10.9 versus 1.0) demonstrating an f10-fold Ellis (23). Data regarding the ER status, the expression of Her-2/neu up-regulation in the tumor. The statistical significance of this (c-erbB2), and the proliferative fraction (Mib-1) were gathered from the up-regulation was P = 0.006 according to the Mann-Whitney U archival pathology reports. The clinicopathologic data of the cases are y described in Table 2. test for independent variables. Defining a real-time PCR CT Immunohistochemistry. Immunohistochemical analysis was carried value of V8 (i.e., at least the 1/256 expression of the reference out on formalin-fixed, paraffin-embedded archival tissue blocks. These gene GAPDH) as the cutoff level for undoubtful AGR2 were cut, mounted on slides, deparaffinized with xylene, and gradually expression in a tissue, we could show AGR2 mRNA in 6 of rehydrated. Antigen retrieval was achieved by pressure cooking in 0.01 16 normal breast tissues (38%) and 22 of 25 breast tumor mol/L citrate buffer for 5 minutes. For AGR2/Gob-4, we generated samples analyzed (88%). polyclonal rabbit antisera against peptides derived from the AGR2 AGR2 immunostaining in normal breast tissue, and intraductal protein sequence (dilution 1:250; ref. 24). The primary antibody was and invasive carcinoma. The immunohistochemical staining incubated at room temperature for 1 hour. As a negative control, four for AGR2 was highly restricted to the secretory epithelial cells slides were processed without primary antibody. Detection took place by the conventional labeled streptavidin-biotin method (DAKO, Hamburg, of normal breast glands, and cells of intraductal carcinoma Germany) with alkaline phosphatase as the reporting enzyme according and invasive carcinoma, respectively. Blood vessels, connec- to the instructions of the manufacturer. Fast-Red (Sigma-Aldrich, tive tissue cells, and inflammatory cells were negative. The Munich, Germany) served as chromogen. Afterward, the slides were immunohistochemical staining showed a slightly granular briefly counterstained with hematoxylin and aqueously mounted. cytoplasmic appearance. In normal breast tissue adjacent to

Clin Cancer Res 2006;12(6) March 15, 2006 1730 www.aacrjournals.org Downloaded from clincancerres.aacrjournals.org on September 27, 2021. © 2006 American Association for Cancer Research. AGR Expression in Breast Cancer

significantly correlated with higher pT status, nodal status, lower proliferative fraction (Mib-1), and positive ER status. There was no significant correlation with tumor grade or c-erbB2 status (Table 3). For further statistical analyses, we delineated groups of tumors with no (immunoreactive score = 0) versus positive (immunoreactive score = 1-12) AGR2 expression. In crosstables, we found AGR2 expression significantly associated with positive ER status and lower histologic tumor grade (Table 2). AGR2 expression and survival times. In univariate survival analysis, cumulative survival curves were calculated according to the Kaplan-Meier method. Differences in survival were assessed with the log-rank test. We analyzed the effect of AGR2, patient age, histologic tumor type, pT status, pN status, tumor grade, ER status, cerbB2 status, and type of therapy on overall survival time and disease-free survival time. The conventional prognostic markers pT status, nodal status, histologic grade, and ER status reached significance for overall and disease-free survival. Patients treated with tamoxifen had significantly longer overall survival times. For patients with AGR2-expressing breast carcinomas, significantly longer overall survival times (5-year survival rate 88% versus 73%, P = 0.035) became apparent, whereas disease-free survival times (P = 0.528) remained insignificant (Fig. 5A and B). In Cox multivariate analysis, AGR2 failed significance (data not shown). Survival analysis in patient subgroups. To evaluate the prognostic value of AGR2 expression in selective patient groups, we repeated the univariate and multivariate survival analysis in subgroups stratified according to pT status, pN status, tumor grade, ER status, and cerbB2 status, respectively. A trend toward Fig. 2. Expression analysis of AGR2 using the matched tumor/normal array. longer disease-free survival times was revealed in various A, organization of the cDNA samples derived from tumor and normal tissue of individual patients on the array. N, normal;T, tumor.1to 9, cDNAs derived from the subgroups that failed significance (data not shown). In the following cell lines: 1,HeLa; 2, Daudi; 3, K562; 4, HL60; 5, G361;6, A594; 7, Molt4; subgroups of small tumors (pT1, P = 0.001), nodal-negative 8, SW480; 9, Raji. B, hybridization results obtained with the AGR2 probe.The filter used contains nine breast cancer samples. tumors (pN0, P = 0.001) and tumors of UICC stage I (P = 0.002) significance was reached for overall survival time (Fig. 5). For the subgroup of nodal-negative tumors, the the tumor, weak AGR2 expression was seen in 121 (89.0%) significant prognostic effect of AGR2 (P = 0.044) was of 136 conclusive cases (Fig. 4A). Intraductal carcinoma confirmed by Cox multivariate analysis (Table 4), whereas it revealed a mostly moderate to strong cytoplasmic expression failed significance in the other subgroups (data not shown). in 87 (92.6%, n = 94; Fig. 4B) and invasive carcinoma in 135 (87.1%, n = 155) cases (Fig. 4C). The immunoreactive Discussion score for AGR2 was significantly higher in intraductal carcinoma (median 8) and invasive carcinoma (median 6) AGR2, the human orthologue of XAG2 protein, is a small, compared with adjacent normal tissue (median 4, Wilcoxon possibly secreted molecule (13) of yet weakly defined func- test: P = 0.016). Expression of AGR2 was statistically tions that is widely expressed in human tissues. Whether it acts

Fig. 3. Abundant up-regulation of AGR2 expression in breast cancer as shown by quantitative RT-PCR. Median AGR2 expression was >10-fold up-regulated in breast tumors (samples 1-25) compared with normal breast tissues (samples A-P). Calculation of error bars according to Applied Biosystems user manual.

www.aacrjournals.org 173 1 Clin Cancer Res 2006;12(6) March 15, 2006 Downloaded from clincancerres.aacrjournals.org on September 27, 2021. © 2006 American Association for Cancer Research. Imaging, Diagnosis, Prognosis

Fig. 4. AGR2 immunohistochemistry. A, whereas secretory epithelia (bold arrows) show a homogenous cytoplasmic staining forAGR2, myoepithelial cells (small arrows)areAGR2 negative. B, strong cytoplasmic staining of intraductal carcinoma with intraluminal cellular debris. C, strong cytoplasmic staining of invasive ductal carcinoma (small arrows) surrounding AGR2-negative normal parenchymal tissue (white arrows). D, strong cytoplasmic staining of invasive lobular carcinoma (short arrows)with adjacent lobular carcinoma in situ (long arrows). Original magnification, Â200 (A and D), Â400 (B and C).

as a protein disulfide isomerase, as has been suggested by clinicohistopathologic data. The analysis of the matched Persson et al. (14), has to await functional studies. Analyzing a tumor/normal array showed that AGR2 up-regulation in breast dot blot array with cDNAs from 11 different human tumor cancer is very significant compared with that in other tumor entities and corresponding normal tissues we detected a very entities because six of nine matching tumor/normal pairs strong AGR2 mRNA expression in normal human colon, exhibited a strong up-regulation in the tumor. Additionally, stomach, rectum, prostate, and breast. These organs secrete a a significant up-regulation of AGR2 was found in endometrial variety of small molecules necessary to maintain tissue and renal tumors, which awaits further study. The up- homeostasis and integrity. Our data are in agreement with regulation of AGR2 in breast cancer could be further previous findings that AGR2 is predominately expressed in confirmed by quantitative real-time RT-PCR in a set of 25 tissue that contain mucus-secreting cells (10) and/or function primary invasive ductal carcinomas that were compared with as endocrine organs (26, 27). Currently, nothing is known normal breast tissue. Liu et al. (13) recently published a about the function of XAG-2 in adult organs of Xenopus. similar quantification of AGR2 mRNA in breast cancer using However, in the Xenopus embryo, XAG-2 is responsible for semiquantitative RT-PCR with only 25 cycles. They found patterning the cement gland, which is a mucus-secreting tissue AGR2 expression in three of nine (33%) normal breast tissues as well. The cement gland is involved in the attachment of the compared with 6 of 16 (38%) normal breast tissues in this Xenopus embryo to a solid support before swimming and study. In the tumor tissue, Liu et al. (13) detected AGR2 feeding. In contrast to the situation in frogs, the biological expression in 44 of 56 cases (79%) compared with 22 of 25 function of the AGR2 protein in human cancer is still largely (88%) tumor samples in our study. When Liu et al. quantified unknown. AGR2 expression according to the ER status, they detected a In this study, we carefully and quantitatively analyzed the considerable lower expression of AGR2 mRNA in ER-negative expression of AGR2 in human breast cancer both on the RNA tumors (13 of 22, 59%) compared with ER-positive tumors and protein level and correlated these expression data to (31 of 34, 91%). In analogy, we found AGR2 expression on

Ta b l e 3 . Correlation of AGR2 expression in breast cancer with conventional clinical or tumor variables

AGR2 pTstatus pN status Grading Mib-1 ER status c-erbB2 status

Correlation coefficient 0.281 0.360 À0.059 À0.271 0.246 À0.094 Significance (two sided) <0.001 <0.001 0.466 0.007 0.003 0.286 n 155 155 155 99 143 130

Clin Cancer Res 2006;12(6) March 15, 2006 1732 www.aacrjournals.org Downloaded from clincancerres.aacrjournals.org on September 27, 2021. © 2006 American Association for Cancer Research. AGR Expression in Breast Cancer

Fig. 5. Kaplan-Meier curves with univariate analyses (log rank) for patients without AGR2 expression (dotted line) versus AGR2-expressing tumors (bold line). A and B, overall survival time (A)and disease-free survival time (B) analyses showing a significantly longer overall survival times of tumors with AGR2 expression. C and D, overall survival time analyses in the subgroups of small (pT1) tumors (C) and nodal-negative patients (D) showing significantly longer survival times forAGR2-expressing tumors.

protein level significantly correlated to a positive ER status. status and positive nodal status that elude a convincing expla- Still, in the group of ER-negative tumors, 31 of 45 (68.9%) nation thus far. showed AGR2 expression, which suggests that AGR2 expres- Importantly, we found that AGR2 might be a new sion in human tumors is only partially dependent on the prognostic marker in breast cancer for the subgroups of nodal presence of a functional ER. negative, pT1, and Unio Internationale Contra Cancrum stage I Our immunohistochemical study on AGR2 protein expres- tumors. For the group of nodal-negative breast tumors, we sion in breast cancer specimens (n = 155) is the largest analysis could show an independent prognostic value for AGR2 described thus far. We could confirm the findings of Fletcher expression being associated with favorable overall survival et al. (9), who found a clear correlation between AGR2 (P = 0.044). Further studies are clearly needed to verify these expression and ER status. However, in our larger data set, we findings to establish AGR2 as a prognostic marker in breast found additionally significant associations with the tumor cancer and to clarify its role in carcinogenesis by functional grade and the proliferation rate. The coexpression of AGR2 and analysis. ER, the higher expression rates seen in tumors with a lower proliferative fraction, and the association with lower tumor grades indicate that AGR2 is a marker of differentiation, Acknowledgments although further functional studies are clearly needed to under- We thank Britta Beyer (Charite¤ Berlin) and Inge Losen (Rheinisch-Westfa« lische score these findings. We also found the somehow contradictory Technische Hochschule Aachen) for excellent technical assistance and Ilka Olson significant correlations of higher AGR2 expression with pT for discussions.

Ta b l e 4 . Cox multivariate analysis for overall survival in the subgroup of nodal-negative tumors

Variable Relative risk (95% confidence interval) P AGR2 (positive vs negative) 15.063 (1.082-209.771) 0.044 pTstatus (1vs 2/3/4) 0.000 0.994 Histologic grade (1vs 2 vs 3) 0.969 (0.170-5.531) 0.972 ER status (positive vs negative) 13.115 (0.687-250.273) 0.087 c-erbB2 status (positive vs negative) 0.826 (0.097-7.018) 0.861

www.aacrjournals.org 173 3 Clin Cancer Res 2006;12(6) March 15, 2006 Downloaded from clincancerres.aacrjournals.org on September 27, 2021. © 2006 American Association for Cancer Research. Imaging, Diagnosis, Prognosis

References 1. Yousef GM, Scorilas A, Kyriakopoulou LG, et al. breast cancer cell lines. BiochemBiophys Res Com- racemase: a new molecular marker for prostate can- Human kallikrein gene 5 (KLK5) expression by mun 19 9 8 ; 2 51: 111 ^ 6 . cer. Cancer Res 2002;62:2220 ^ 6. quantitative PCR: an independent indicator of poor 10. KomiyaT,TanigawaY,Hirohashi S. Cloningof the gene 19. Zhumabayeva B, Diatchenko L, Chenchik A, et al. prognosis in breast cancer. Clin Chem2002;48: gob-4, which is expressed in intestinal goblet cells Use of SMART-generated cDNA for gene expression 1241 ^ 5 0. inmice. Biochim Biophys Acta1999;1444:434 ^ 8. studies in multiple human tumors. Biotechniques 2. Manders P,Tjan-HeijnenVC, Span PN, et al. Complex 11. Aberger F, Weidinger G, Grunz H, et al. Anterior 2001;30:158 ^ 63. of urokinase-type plasminogen activator with its type specification of embryonic ectoderm: the role of the 20. Fink L, Seeger W, Ermert L, et al. Real-time quanti- 1inhibitor predicts poor outcome in 576 patients with Xenopus cement gland-specific gene XAG-2. Mech tative RT-PCR after laser-assisted cell picking. Nat lymph node-negative breast carcinoma. Cancer 2004; Dev1998;72:115^30. Med 1998;4:1329 ^ 33. 101:48 6 ^ 94. 12. Huber M, Bahr I, Kratzschmar JR, et al. Comparison 21. Tavassoli FA, Devilee P, editors. WHO classification 3. Nakopoulou L, Giannopoulou I, Stefanaki K, et al. of proteomic and genomic analyses of the human of tumours. Pathology and genetics of tumours of the Enhanced mRNA expression of tissue inhibitor of breast cancer cell line T47D and the antiestrogen- breast and female genital organs. Lyon (France): IARC metalloproteinase-1 (TIMP-1) in breast carcinomas is resistant derivative T47D-r. Mol Cell Proteomics Press; 2003. correlated with adverse prognosis. JPathol 2002;197: 2004;3:43 ^ 55. 22. Sobin LH, Fleming ID. TNM classification of malig- 307^13. 13. Liu D, Rudland PS, Sibson DR, et al. Human homo- nant tumors. 5th ed. Union Internationale Contre le 4. Spizzo G, Went P, Dirnhofer S, et al. High Ep-CAM logue of cement gland protein, a novel metastasis in- Cancer and the American Joint Committee on Cancer. expression is associated with poor prognosis in ducer associated with breast carcinomas. Cancer Res Cancer 1997;80:1803 ^ 4. node-positive breast cancer. Breast Cancer Res Treat 2005;65:3796^805. 23. Elston CW, Ellis IO. Pathologic prognostic factors in 2004;86:207 ^ 13. 14. Persson S, Rosenquist M, Knoblach B, et al. Diver- breast cancer. I.The value of histologicalgradeinbreast 5. Rudland PS, Platt-Higgins A, El Tanani M, et al. Prog- sity of the protein disulfide isomerase family: identifi- cancer: experience froma large study with long-term nostic significance of the metastasis-associated pro- cation of breast tumor induced Hag2 and Hag3 as follow-up. Histopathology1991;19:403 ^ 10. tein osteopontin in human breast cancer. Cancer Res novel members of the protein family. Mol Phylogenet 24. Kristiansen G, Pilarsky C,Wissmann C, et al. Expres- 2002;62:3417^ 27. Evol 2005;36:734 ^ 40. sion profiling of microdissected matched prostate 6. Kristiansen G,Winzer KJ, Mayordomo E, et al. CD24 15. Altschul SF, MaddenTL, SchafferAA, et al. Gapped cancer samples reveals CD166/MEMD and CD24 as expression is a new prognostic marker in breast can- BLAST and PSI-BLAST: a new generation of protein new prognostic markers for patient survival. J Pathol cer. Clin Cancer Res 2003;9:4906 ^ 13. database search programs. Nucleic Acids Res 1997; 2005;205:359 ^ 76. 7. Klopocki E, Kristiansen G, Wild PJ, et al. Loss of 25:3389 ^ 402. 25. Remmele W, Stegner HE. [Recommendation for SFRP1is associated with breast cancer progression 16. Sevier CS, Kaiser CA. Formation and transfer of uniformdefinition of an immunoreactive score (IRS) and poor prognosis in early stage tumors. Int J Oncol disulphide bonds in living cells. Nat Rev Mol Cell Biol for immunohistochemical estrogen receptor detection 2004;25:641 ^ 9. 2002;3:836 ^ 47. (ER-ICA) in breast cancer tissue]. Pathologe 1987;8: 8. Zhang JS, Gong A, Cheville JC, et al. AGR2, an 17. Fletcher GC, Patel S, Tyson K, et al. hAG-2 and 13 8 ^ 4 0. androgen-inducible secretory protein overexpressed hAG-3, human homologues of genes involved in dif- 26. Sive H, Bradley L. A sticky problem: the Xenopus in prostate cancer. Genes Chromosomes Cancer ferentiation, are associated with oestrogen receptor- cement gland as a paradigm for anteroposterior pat- 2005;43:249 ^ 59. positive breast tumours and interact with metastasis terning. Dev Dyn1996;205:265 ^ 80. 9. Thompson DA,Weigel RJ. hAG-2, the human homo- gene C4.4a and dystroglycan. Br J Cancer 2003;88: 27. Sive HL, Hattori K,Weintraub H. Progressive deter- logue of the Xenopus laevis cement gland gene 579 ^ 85. mination during formation of the anteroposterior axis XAG-2, is coexpressed with estrogen receptor in 18. LuoJ,ZhaS,GageWR,etal.a-Methylacyl-CoA in Xenopus laevis. Cell1989;58:171^80.

Clin Cancer Res 2006;12(6) March 15, 2006 1734 www.aacrjournals.org Downloaded from clincancerres.aacrjournals.org on September 27, 2021. © 2006 American Association for Cancer Research. Prognostic Relevance of AGR2 Expression in Breast Cancer

Florian Rudolf Fritzsche, Edgar Dahl, Stefan Pahl, et al.

Clin Cancer Res 2006;12:1728-1734.

Updated version Access the most recent version of this article at: http://clincancerres.aacrjournals.org/content/12/6/1728

Cited articles This article cites 26 articles, 6 of which you can access for free at: http://clincancerres.aacrjournals.org/content/12/6/1728.full#ref-list-1

Citing articles This article has been cited by 8 HighWire-hosted articles. Access the articles at: http://clincancerres.aacrjournals.org/content/12/6/1728.full#related-urls

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

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://clincancerres.aacrjournals.org/content/12/6/1728. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.