Portal de Periódicos da CAPES

TFF3 Is a Normal Breast Epithelial Protein and Is Associated with Differentiated Phenotype in Early Breast Cancer but Predisposes to Invasion and Metastasis in Advanced Disease

2012; Elsevier BV; Volume: 180; Issue: 3 Linguagem: Inglês

10.1016/j.ajpath.2011.11.022

1525-2191

Ahmed Hamed, Andrew Griffiths, Michael T Tilby, Bruce R. Westley, Felicity E. B. May,

Cancer-related Molecular Pathways

The trefoil protein TFF3 stimulates invasion and angiogenesis in vitro. To determine whether it has a role in breast tumor metastasis and angiogenesis, its levels were measured by immunohistochemistry in breast tissue with a specific monoclonal antibody raised against human TFF3. TFF3 is expressed in normal breast lobules and ducts, at higher levels in areas of fibrocystic change and papillomas, in all benign breast disease lesions, and in 89% of in situ and in 83% of invasive carcinomas. In well-differentiated tumor cells, TFF3 is concentrated at the luminal edge, whereas in poorly differentiated cells polarity is inverted and expression is directed toward the stroma. Expression was high in well-differentiated tumors and was associated significantly with low histological grade and with estrogen and progesterone receptor expression, accordant with induction of TFF3 mRNA by estrogen in breast cancer cells. Paradoxically, TFF3 expression was associated with muscle, neural, and lymphovascular invasion and the presence and number of involved lymph nodes, and it was an independent predictive marker of lymphovascular invasion and lymph node involvement. Consistent with an angiogenic function, TFF3 expression correlated strongly with microvessel density evaluated with CD31 and CD34. In conclusion, TFF3 is expressed in both the normal and diseased breast. Although associated with features of good prognosis, its profile of expression in invasive cancer is consistent with a role in breast tumor progression and tumor cell dissemination. The trefoil protein TFF3 stimulates invasion and angiogenesis in vitro. To determine whether it has a role in breast tumor metastasis and angiogenesis, its levels were measured by immunohistochemistry in breast tissue with a specific monoclonal antibody raised against human TFF3. TFF3 is expressed in normal breast lobules and ducts, at higher levels in areas of fibrocystic change and papillomas, in all benign breast disease lesions, and in 89% of in situ and in 83% of invasive carcinomas. In well-differentiated tumor cells, TFF3 is concentrated at the luminal edge, whereas in poorly differentiated cells polarity is inverted and expression is directed toward the stroma. Expression was high in well-differentiated tumors and was associated significantly with low histological grade and with estrogen and progesterone receptor expression, accordant with induction of TFF3 mRNA by estrogen in breast cancer cells. Paradoxically, TFF3 expression was associated with muscle, neural, and lymphovascular invasion and the presence and number of involved lymph nodes, and it was an independent predictive marker of lymphovascular invasion and lymph node involvement. Consistent with an angiogenic function, TFF3 expression correlated strongly with microvessel density evaluated with CD31 and CD34. In conclusion, TFF3 is expressed in both the normal and diseased breast. Although associated with features of good prognosis, its profile of expression in invasive cancer is consistent with a role in breast tumor progression and tumor cell dissemination. Improvements in the early diagnosis and clinical management of breast cancer have led to an impressive reduction in mortality notwithstanding which >500,000 people die every year as a direct result of disseminated disease. During carcinogenesis, neoplastic cells acquire the ability to migrate, invade, and metastasize either directly into neighboring tissues or via the neural network, lymphatic system, or vasculature to more distant sites. In breast cancer, malignant cells can invade locally to the underlying muscle or overlying skin or can escape through the neural sheath or via the lymph or blood vessels. The presence of tumor metastases in the axillary lymph nodes is an important part of clinical staging and prediction of prognosis. It is accepted that the existence of metastases in regional lymph nodes is not a precursor of but a manifestation of more distant spread. Distant metastases occur most commonly in bone, followed by lung, liver, and brain. Understanding of the underlying biology of breast cancer and breast cancer metastasis remains obscure yet is increasingly relevant clinically with the move toward stratified medicine and therapeutic monitoring. Systemic therapies are important in the treatment of both early stage and advanced breast cancers. Endocrine therapies, which have been the mainstay of breast cancer therapy, exploit the dependence of tumor cells on estrogens and include partial and pure antiestrogens and type I and type II aromatase inhibitors. Tumors that are likely to respond to endocrine therapies are identified by the presence of the estrogen receptor α, which is a requisite for estrogen action, and the progesterone receptor, which is induced by estrogen. Trefoil proteins are small, secreted proteins that are co-expressed with mucins by epithelia and are thought to be important in normal mucosal protection and repair.1May F.E.B. Westley B.R. Trefoil proteins: their role in normal and malignant cells.J Pathol. 1997; 183: 4-7Crossref PubMed Scopus (92) Google Scholar, 2Thim L. May F.E.B. Structure of mammalian trefoil factors and functional insights.Cell Mol Life Sci. 2005; 62: 2956-2973Crossref PubMed Scopus (121) Google Scholar The three human members of the trefoil factor family, TFF1, TFF2, and TFF3, are characterized by the presence of a conserved 41- to 42-amino acid residue trefoil domain with distinctive structure. TFF1 and TFF3 contain one trefoil domain, whereas TFF2 contains two. TFF3 has an extra trefoil domain cysteine residue that forms intermolecular disulfide bonds.3May F.E.B. Church S.T. Major S. Westley B.R. The closely related estrogen-regulated trefoil proteins TFF1 and TFF3 have markedly different hydrodynamic properties, overall charge, and distribution of surface charge.Biochemistry. 2003; 42: 8250-8259Crossref PubMed Scopus (27) Google Scholar The structure of dimeric TFF3 is considerably more compact than that of dimeric TFF1.3May F.E.B. Church S.T. Major S. Westley B.R. The closely related estrogen-regulated trefoil proteins TFF1 and TFF3 have markedly different hydrodynamic properties, overall charge, and distribution of surface charge.Biochemistry. 2003; 42: 8250-8259Crossref PubMed Scopus (27) Google Scholar, 4Muskett F.W. May F.E.B. Westley B.R. Feeney J. Solution structure of the disulfide-linked dimer of human intestinal trefoil factor (TFF3): the intermolecular orientation and interactions are markedly different from those of other dimeric trefoil proteins.Biochemistry. 2003; 42: 15139-15147Crossref PubMed Scopus (45) Google Scholar, 5Williams M.A. Westley B.R. May F.E.B. Feeney J. The solution structure of the disulphide-linked homodimer of the human trefoil protein TFF1.FEBS Lett. 2001; 493: 70-74Crossref PubMed Scopus (43) Google Scholar The normal functions of trefoil proteins are thought to center on their role in mucosal protection, namely interactions with mucins and stimulation of cell motility.1May F.E.B. Westley B.R. Trefoil proteins: their role in normal and malignant cells.J Pathol. 1997; 183: 4-7Crossref PubMed Scopus (92) Google Scholar, 2Thim L. May F.E.B. Structure of mammalian trefoil factors and functional insights.Cell Mol Life Sci. 2005; 62: 2956-2973Crossref PubMed Scopus (121) Google Scholar TFF3 has been found to stimulate cellular invasion in vitro and to have angiogenic activity.6Emami S. Le Floch N. Bruyneel E. Thim L. May F. Westley B. Rio M. Mareel M. Gespach C. Induction of scattering and cellular invasion by trefoil peptides in src- and RhoA-transformed kidney and colonic epithelial cells.FASEB J. 2001; 15: 351-361Crossref PubMed Scopus (106) Google Scholar, 7Rodrigues S. Van Aken E. Van Bocxlaer S. Attoub S. Nguyen Q.D. Bruyneel E. Westley B.R. May F.E.B. Thim L. Mareel M. Gespach C. Emami S. Trefoil peptides as proangiogenic factors in vivo and in vitro: implication of cyclooxygenase-2 and EGF receptor signalling.FASEB J. 2003; 17: 7-16Crossref PubMed Scopus (125) Google Scholar The role of trefoil proteins in cancer is controversial. TFF1 is thought to be a tumor suppressor in gastric cancer,8Lefebvre O. Chenard M.P. Masson R. Linares J. Dierich A. LeMeur M. Wendling C. Tomasetto C. Chambon P. Rio M.C. Gastric mucosa abnormalities and tumorigenesis in mice lacking the pS2 trefoil protein.Science. 1996; 274: 259-262Crossref PubMed Scopus (462) Google Scholar whereas in other tumors, including breast, prostate, and bladder, trefoil factor expression has been considered to be ectopic.1May F.E.B. Westley B.R. Trefoil proteins: their role in normal and malignant cells.J Pathol. 1997; 183: 4-7Crossref PubMed Scopus (92) Google Scholar Their expression has been reported to be predictive both of better survival9Foekens J.A. Rio M.C. Seguin P. van Putten W.L. Fauque J. Nap M. Klijn J.G.M. Chambon P. Prediction of relapse and survival in breast cancer patients by pS2 protein status.Cancer Res. 1990; 50: 3832-3837PubMed Google Scholar, 10Thompson A.M. Elton R.A. Hawkins R.A. Chetty U. Steel C.M. PS2 mRNA expression adds prognostic information to node status for 6-year survival in breast cancer.Br J Cancer. 1998; 77: 492-496Crossref PubMed Scopus (15) Google Scholar and of a more aggressive phenotype.11Yamachika T. Werther J.L. Bodian C. Babyatsky M. Tatematsu M. Yamamura Y. Chen A. Itzkowitz S. Intestinal trefoil factor: a marker of poor prognosis in gastric carcinoma.Clin Cancer Res. 2002; 8: 1092-1099PubMed Google Scholar, 12Dhar D.K. Wang T.C. Tabara H. Tonomoto Y. Maruyama R. Tachibana M. Kubota H. Nagasue N. Expression of trefoil factor family members correlates with patient prognosis and neoangiogenesis.Clin Cancer Res. 2005; 11: 6472-6478Crossref PubMed Scopus (69) Google Scholar, 13Yio X. Zhang J.Y. Babyatsky M. Chen A. Lin J. Fan Q.X. Werther J.L. Itzkowitz S. Trefoil factor family-3 is associated with aggressive behavior of colon cancer cells.Clin Exp Metastasis. 2005; 22: 157-165Crossref PubMed Scopus (33) Google Scholar, 14Babyatsky M. Lin J. Yio X. Chen A. Zhang J.Y. Zheng Y. Twyman C. Bao X. Schwartz M. Thung S. Lawrence Werther J. Itzkowitz S. Trefoil factor-3 expression in human colon cancer liver metastasis.Clin Exp Metastasis. 2009; 26: 143-151Crossref PubMed Scopus (34) Google Scholar TFF1 was discovered first as an abundant estrogen-regulated molecule in breast cancer cells, whereas TFF2 is expressed only at very low levels in breast cancer cells.15Masiakowski P. Breathnach R. Bloch J. Gannon F. Krust A. Chambon P. Cloning of cDNA sequences of hormone-regulated genes from the MCF-7 human breast cancer cell line.Nucleic Acids Res. 1982; 10: 7895-7903Crossref PubMed Scopus (702) Google Scholar, 16May F.E.B. Westley B.R. Cloning of estrogen-regulated messenger RNA sequences from human breast cancer cells.Cancer Res. 1986; 46: 6034-6040PubMed Google Scholar, 17May F.E.B. Westley B.R. Identification and characterization of estrogen-regulated RNAs in human breast cancer cells.J Biol Chem. 1988; 263: 12901-12908Abstract Full Text PDF PubMed Google Scholar, 18May F.E.B. Semple J.I. Prest S.J. Westley B.R. Expression and motogenic activity of TFF2 in human breast cancer cells.Peptides. 2004; 25: 865-872Crossref PubMed Scopus (22) Google Scholar In 1997, TFF3 mRNA was detected by in situ hybridization in normal breast, in benign and in situ lesions, and in 31 invasive carcinomas19Poulsom R. Hanby A.M. Lalani E.N. Hauser F. Hoffmann W. Stamp G.W.H. Intestinal trefoil factor (TFF 3) and pS2 (TFF 1), but not spasmolytic polypeptide (TFF 2) mRNAs are co-expressed in normal, hyperplastic, and neoplastic human breast epithelium.J Pathol. 1997; 183: 30-38Crossref PubMed Scopus (99) Google Scholar and was shown to be induced by estrogen and inhibited by antiestrogens in estrogen-responsive breast cancer cells.20May F.E.B. Westley B.R. Expression of human intestinal trefoil factor in malignant cells and its regulation by oestrogen in breast cancer cells.J Pathol. 1997; 182: 404-413Crossref PubMed Scopus (85) Google Scholar Subsequently, large expression microarray studies identified TFF3 as one of the genes whose expression is most closely associated with expression of estrogen receptor α.21Gruvberger S. Ringnér M. Chen Y. Panavally S. Saal L.H. Borg A. Fernö M. Peterson C. Meltzer P.S. Estrogen receptor status in breast cancer is associated with remarkably distinct gene expression patterns.Cancer Res. 2001; 61: 5979-5984PubMed Google Scholar, 22West M. Blanchette C. Dressman H. Huang E. Ishida S. Spang R. Zuzan H. Olson JA J.R. Marks J.R. Nevins J.R. Predicting the clinical status of human breast cancer by using gene expression profiles.Proc Natl Acad Sci U S A. 2001; 98: 11462-11467Crossref PubMed Scopus (1100) Google Scholar The expression and role of TFF3 protein in human breast tissue have not been studied. We report that TFF3 protein is expressed in the normal breast, benign breast disease, and in situ carcinoma and that its expression is lost in some invasive breast carcinomas. We provide evidence that, although TFF3 protein expression is higher in well-differentiated tumors and is associated with markers of estrogen responsiveness, TFF3 has a more sinister role in breast cancer invasion and metastasis. Ethical permission for the studies was obtained from the Joint Newcastle Health Hospitals and University of Newcastle on Tyne Ethical Committee. Samples were collected from archived material of patients who presented sequentially between 2002 and 2004. Tissue was selected, and 1-mm cores of tissue were embedded in tissue microarrays. Duplicate cores of tissue were taken from different blocks and arrayed in separate paraffin blocks. In total, 34 samples of normal breast, 86 benign breast lesions, 116 in situ carcinomas, 266 invasive carcinomas, and 76 involved lymph nodes were included in the study. In addition, whole sections of normal breast tissue from 40 women between the ages of 18 and 39 were analyzed. Sections of formalin-fixed, paraffin-embedded breast tissue were deparaffinized and rehydrated. They were incubated in 0.5% hydrogen peroxide for 10 minutes to block endogenous peroxide activity. Antigen retrieval was with trypsin for 10 minutes at room temperature, in a 10 mmol/L citrate buffer, pH 6.0, in a pressure cooker for 0.5 to 3 minutes or in a microwave for 2 × 5 minutes. After antigen retrieval, the tissue was washed in Tris-HCl, pH 7.6, buffered saline and 1% Tween 20. The tissues were incubated with a specific anti-TFF3 monoclonal antibody (manuscript in preparation) raised against correctly folded human TFF3 protein,3May F.E.B. Church S.T. Major S. Westley B.R. The closely related estrogen-regulated trefoil proteins TFF1 and TFF3 have markedly different hydrodynamic properties, overall charge, and distribution of surface charge.Biochemistry. 2003; 42: 8250-8259Crossref PubMed Scopus (27) Google Scholar, 4Muskett F.W. May F.E.B. Westley B.R. Feeney J. Solution structure of the disulfide-linked dimer of human intestinal trefoil factor (TFF3): the intermolecular orientation and interactions are markedly different from those of other dimeric trefoil proteins.Biochemistry. 2003; 42: 15139-15147Crossref PubMed Scopus (45) Google Scholar a specific anti-TFF1 monoclonal antibody raised against correctly folded human TFF1 protein,23Westley B.R. Griffin S.M. May F.E.B. Interaction between TFF1, a gastric tumor suppressor trefoil protein, and TFIZ1, a Brichos domain-containing protein with homology to SP-C.Biochemistry. 2005; 44: 7967-7975Crossref PubMed Scopus (91) Google Scholar anti-CD31 (JC70A; Dako UK Ltd., Ely, Cambridgeshire, UK), anti-D34 (QBend-10; Dako UK Ltd.), or anti-Ki-67 (MIB-1; Dako UK Ltd.) diluted in Tris-buffered saline and 1% Tween 20 at room temperature and then with the Universal Probe [X-Cell Plus Polymer HRP (horseradish peroxidase) detection kit; A. Menarini Diagnostics, Firenze, Italy] at room temperature and finally with HRP-Polymer (A. Menarini Diagnostics). The brown colorimetric reaction was developed by incubation with 3,3′ diaminobenzidine, and the sections were counterstained with hematoxylin. The immunohistochemistry results were analyzed to generate a histoscore that combines the intensity of the immunoreaction with the number of positive cells. One thousand cells were scored individually as negative, weakly positive, moderately positive, or strongly positive. These four categories are weighted as 0, 0.1, 0.3, and 1.0, respectively, in recognition of the range of intensities of the immunoreaction. The histoscores varied between 0 and 1000. A sample of normal human colonic mucosa was included in all immunohistochemical experiments to serve as a reference of the intensity of the immunoreaction. The final histoscore is the mean of scores from the two sections for each sample. The concordance between the two values was strong and significant (Spearman's ρ correlation, P < 0.001). The results were analyzed statistically with the SPSS-PC package version 17 (SPSS Inc., Chicago, IL). Probabilities of <0.05 were considered statistically significant. The statistical significance of differences observed was analyzed by Kruskal-Wallis II or Mann-Whitney U tests and of correlations by Pearson's correlation or Spearman's ρ correlation. To adjust for potential confounding variables and to single out independent predictors, a multivariate analysis was undertaken. The different histoscores were scaled, and the predictive ability of the continuous variables was evaluated by binary logistic regression and refined by stepwise selection. The expression of TFF3 protein was characterized in normal breast tissue from 34 premenopausal and perimenopausal women. There was strong TFF3 immunoreaction in the cytoplasmic compartment of normal breast lobular epithelial cells which showed frequently clear cellular polarity (Figure 1, A and D). TFF3 protein was localized toward the luminal borders of the acini and was often absent at the basal membrane, which is consistent with TFF3 being secreted from the acini into the lumina of the lobules. No immunoreaction was observed in myoepithelial or stromal cells. TFF3 expression was detected in most of the normal breast lobules. TFF3 expression was detected also in epithelial cells of the intralobular and interlobular ducts (Figure 1B) and in the main lactiferous duct (Figure 1C). TFF3 was concentrated toward the lumina of the polarized epithelial cell monolayers. TFF3 has been measured in human milk,24Vestergaard E.M. Nexo E. Wendt A. Guthmann F. Trefoil factors in human milk.Early Hum Dev. 2008; 84: 631-635Crossref PubMed Scopus (29) Google Scholar but the breast lobules in which TFF3 expression was detected in the present study were not from lactating women. These results show that TFF3 is synthesized in the normal nonlactating breast and are consistent with it being secreted from the breast epithelial cells and having a role in maintenance of the epithelial integrity of the lobular and ductal lumina. We investigated the possibility that expression of TFF3 in the normal breast lobule might depend on cyclical changes that occur during the menstrual cycle. The morphological features of the breast which change during the menstrual cycle were assessed by histologic parameters.25Vogel P.M. Georgiade N.G. Fetter B.F. Vogel F.S. McCarty Jr, K.S. The correlation of histologic changes in the human breast with the menstrual cycle.Am J Pathol. 1981; 104: 23-34PubMed Google Scholar, 26Ramakrishnan R. Khan S.A. Badve S. Morphological changes in breast tissue with menstrual cycle.Mod Pathol. 2002; 15: 1348-1356Crossref PubMed Scopus (104) Google Scholar The breast lobules in the two cases in which TFF3 expression was not detected were in the early follicular phase that is characterized by low circulating levels of estrogen and progesterone. Late follicular phase coincides with an increase in circulating estrogens and is characterized by clear acinar formation with distinct luminal cells. After ovulation and resultant increased circulating progesterone, the breast enters the early luteal phase, and lumina become enlarged compared with previous phases. The secretory phase is characterized by apocrine secretion from the luminal epithelial cells. The fifth, menstrual phase is associated with estrogen and progesterone withdrawal, and apocrine secretion abates.25Vogel P.M. Georgiade N.G. Fetter B.F. Vogel F.S. McCarty Jr, K.S. The correlation of histologic changes in the human breast with the menstrual cycle.Am J Pathol. 1981; 104: 23-34PubMed Google Scholar A trend toward elevated TFF3 expression was observed in the late follicular phase compared with the early follicular phase, but the difference was not statistically significant. The potential association between TFF3 expression and morphological changes that occur in breast lobular epithelial cells during the menstrual cycle was examined further in breast lobules of 40 younger women, aged between 18 and 39 years. Cytoplasmic expression of TFF3 in breast luminal epithelial cells was higher during late follicular and early luteal phases of the menstrual cycle (median, 560 and 430) when estrogen levels are higher than during the early follicular or late luteal phases (median, 25 and 110) when estrogen levels are low (Kruskal-Wallis II test, P < 0.01). TFF3 immunoreaction was strong within the apocrine secretions of luminal cells in the secretory phase. During the menstrual phase, TFF3 immunoreaction was restricted mainly to the apical edge of the cells and residual secretory material within the lumina. The results indicate that there is an association between TFF3 expression in breast lobular epithelial cells and the stage of menstrual cycle. TFF3 was expressed in all 86 benign breast lesions examined (Figure 1, E–H). As with normal breast tissue, TFF3 protein was expressed only by the epithelial cells. Even in cases of sclerosing adenosis, in which the increased number of involved lobules are compressed and obliterated by stromal fibrosis (Figure 1E), the myoepithelial cells did not express TFF3. The intensity of the immunoreaction varied considerably among the 14 cases of sclerosing adenosis studied from almost undetectable to relatively strong. In the 10 fibroadenomas studied, TFF3 expression was detected always, but the intensity of the immunoreaction varied markedly between cells (Figure 1F). Fibroadenomas in which the glandular component was dominant tended to express more TFF3 than those in which the stromal component was dominant. In the six intraductal papillomas (Figure 1G), a higher proportion of cells contained detectable levels of TFF3 than in fibroadenomas or cases of sclerosing adenosis. In fibrocystic disease (Figure 1H), all epithelial cells present in areas of fibrocystic change expressed detectable TFF3, and the immunoreaction was strong in most. TFF3 was detected always in the cytoplasm, and the immunoreaction was sometimes granular in appearance. When acini were visible, TFF3 was concentrated toward the lumina, and the immunoreaction was strong in areas in which cellular release of large secretory granules was apparent. The luminal release of TFF3 was particularly evident in apocrine metaplasia. TFF3 expression was especially high in cells that exhibit apocrine metaplasia within areas of fibrocystic change but was absent in squamous metaplasia. TFF3 expression in the normal breast tissues and in benign breast lesions was quantified to generate a histoscore (Figure 2A). Four cases each of sclerosing adenosis (29%), fibroadenoma (40%), and papilloma (67%) and 53 cases of fibrocystic change (95%) had high histoscores. TFF3 expression is significantly higher in papilloma of the breast and in areas of fibrocystic change than in normal breast tissue (Mann-Whitney U-test, P < 0.01 and P < 0.001, respectively). TFF3 expression in papilloma and in fibrocystic change is higher also than in sclerosing adenosis and fibroadenoma (Kruskal-Wallis II test, P < 0.001). TFF3 was detected in 93 of the 106 cases (88%) of ductal carcinoma in situ studied, which comprised 50 of mixed histologic type, 33 solid, 17 comedo, 4 cribriform, and 2 papillary tumors (Figure 1, I–K). TFF3 immunoreaction was generally strong and, where cellular polarity was evident, was concentrated at the luminal edge but was distributed evenly throughout the cytoplasm in nonpolarized malignant epithelial cells. No statistically significant differences were observed in TFF3 expression between subtypes of carcinoma in situ (Figure 2B) or between different grades; 81 lesions were of high grade, 13 of intermediate grade, and 12 of low grade. TFF3 immunoreaction was detected in all cases of lobular carcinoma in situ (Figure 1L). TFF3 was detected in 215 of 266 invasive tumors (Figure 1). The immunoreaction was cytoplasmic with a tendency toward perinuclear condensation in some cells, consistent with the concentration of the TFF3 protein in the Golgi apparatus before its secretion. Most of the immunoreactive tumor cells were moderately to strongly positive. TFF3 was not detected in the surrounding stromal, endothelial, inflammatory, or muscle cells. Malignant cells in tumors with strong TFF3 expression were scattered diffusely throughout the stroma (Figure 1O), whereas those in tumors with weaker expression had a tendency to cluster or aggregate together within the stroma (Figure 1, M and N). Within these clusters, malignant cells with higher TFF3 expression were concentrated on the edges of the cell aggregates and in some instances had polarized TFF3 immunoreaction that was more intense toward the stromal edge of the tumor cell. TFF3 expression in the 204 invasive ductal carcinomas of no specific type (Figure 1, M–O) ranged from undetectable (not shown) to very high. In tumors with lower TFF3 expression (Figure 1, M and N), the intensity of the immunoreaction varied considerably between individual malignant cells. In tumors with very high TFF3 expression (Figure 1O), most or all of the malignant cells were intensely immunoreactive. Variable levels of TFF3 expression were detected also in 31 cases of invasive lobular carcinoma (Figure 1P). Signet ring cells and those with a leading edge within lobular carcinomas had high TFF3 expression. TFF3 was not detected in four cases of medullary carcinoma (Figure 1Q), and expression was generally low in the eight cases of the other special types of invasive ductal carcinoma. These comprised three micropapillary, two metaplastic (Figure 1R), one apocrine, and one cribriform (Figure 1S) carcinoma and one phyllodes tumor of the breast (Figure 1T). TFF3 expression is significantly higher in tubular (Figure 1U) and mucinous (Figure 1V) carcinomas, which are well differentiated and have a favorable prognosis, than in all other types of invasive breast carcinoma (Figure 2C, Kruskal-Wallis II test, P < 0.01). In mucinous carcinoma, the intracellular mucus and occasional extracellular mucus was immunoreactive for TFF3. We investigated whether TFF3 expression is associated with histological grade of the tumors. All grade 1 tumors express detectable TFF3, but a third of grade 3 tumors contain no TFF3 protein (Figure 3A). The median values for TFF3 expression were 420 for grade 1 tumors, 340 for grade 2 tumors, and only 100 for grade 3 tumors; there is a significant negative association between tumor grade and TFF3 protein expression (Spearman's ρ coefficient, P < 0.01). Thus, TFF3 expression is significantly lower in high-grade breast tumors than in low-grade tumors (Kruskal-Wallis II test, P < 0.001). Greater than 70% of the tumors in which TFF3 expression was not detected was grade 3. To investigate if there was coordinate loss of expression of the paralagous trefoil protein, TFF1 in these tumors, TFF1 expression was analyzed by immunohistochemistry with an antibody raised against TFF1 protein.23Westley B.R. Griffin S.M. May F.E.B. Interaction between TFF1, a gastric tumor suppressor trefoil protein, and TFIZ1, a Brichos domain-containing protein with homology to SP-C.Biochemistry. 2005; 44: 7967-7975Crossref PubMed Scopus (91) Google Scholar TFF1 immunoreaction was detected in only 2 of the 36 grade 3 tumors in which TFF3 expression was absent. These results indicate that there is coincident loss of TFF3 and TFF1 expression in breast cancer, which is consistent with interconnectivity of regulation of the TFF1 and TFF3 genes in breast cancer cells. It is noteworthy that cases with a more anaplastic appearance, including a high nuclear-to-cytoplasmic ratio and absence of tubular differentiation, with cytologic atypia tended to express less TFF3 protein (Figure 1W), whereas TFF3 was expressed at markedly higher levels in invasive breast cancer types that are characterized by good differentiation: tubular and mucinous carcinomas. These observations coupled with the finding that TFF3 is secreted by normal breast acini suggest that the negative association of TFF3 expression with tumor grade is explained in part by its association with good differentiation. Another possible reason for the strong negative association between TFF3 expression and tumor grade is that TFF3 expression is related to the proportion of mitotic cells. Cases with multinucleated cells or with a high mitotic incidence tended to express less TFF3 protein (Figure 1, W and X). Cases with a low mitotic incidence tended to express more TFF3, and TFF3 was not detectable in dividing cells within tumors that do express TFF3 (Figure 1M). These observations of a negative association between TFF3 expression and mitotic incidence are exemplified by the absence of TFF3 protein in medullary carcinomas (Figure 1Q) that have a high mitotic index and by its higher level in tubular carcinomas (Figure 1U) that have a low mitotic index. An association between expression of Ki-67, a nuclear nonhistone protein, and mitotic index has been rep