Overexpression of Periostin Protein in Non-Small Cell Lung Carcinoma is Not Related with Clinical Prognostic Significance

http://dx.doi.org/10.4046/trd.2012.72.2.132 ISSN: 1738-3536(Print)/2005-6184(Online) Tuberc Respir Dis 2012;72:132-139

CopyrightⒸ2012. The Korean Academy of Tuberculosis and Respiratory Diseases. All rights reserved.

Overexpression of Periostin Protein in Non-Small Cell Lung Carcinoma is Not Related with Clinical Prognostic Significance

Won Young Park, M.D.

, Dong Hoon Shin, M.D.

, Jae Ho Kim, Ph.D.

, Min Ki Lee, M.D.

, Ho Seok Lee, M.D.

, Chang Hun Lee, M.D.

Departments of

Pathology,

Physiology,

Internal Medicine,

Thoracic Surgery, and

Medical Research Institute, Pusan National University Hospital, Pusan National University School of Medicine, Busan, Korea

Background: Periostin is preferentially expressed in periosteum, indicating a potential role in bone formation.

Recently, there have been emerging controversies about its role in invasion and metastasis of human malignancies.

We attempted to determine the clinicopathological significance of periostin expression in non-small cell lung carcinoma (NSCLC).

Methods: Immunohistochemical staining of periostin protein from 91 cases of NSCLCs was performed using tissue microarray blocks. The results were correlated with clinicopathological parameters.

Results: Positive reaction to periostin was predominantly noted in the tumor stroma. The strongest reaction presented as a band-like pattern just around the tumor nests. Non-neoplastic lung tissue and most in-situ carcinomas did not show a positive reaction in their stroma. With respect to tumor differentiation, moderate to poor differentiated tumors (47/77) revealed even higher periostin expression than the well-differentiated ones (4/14) (p=0.024). High periostin expression was positively correlated with E-cadherin and p53 expression, but was not related with patient age, sex, tumor type, PCNA index, b-catenin, cyclin D1, pTNM-T, pTNM-N, stage, and patient survival (p＞0.05).

Conclusion: These results suggest that periostin might play a role during the biological progression of NSCLC, but may not be related to the clinical prognostic parameters.

Key Words: Carcinoma, Non-Small-Cell Lung; Cadherins; Genes, p53

Address for correspondence: Chang Hun Lee, M.D.

Department of Pathology, Pusan National University Hospital, 1-10, Ami-dong 1-ga, Seo-gu, Busan 602-738, Korea Phone: 82-51-240-7718, Fax: 82-51-510-8040 E-mail: cnlee@pusan.ac.kr

Co-correspondence: Min Ki Lee, M.D.

Department of Internal Medicine, Pusan National University Hospital, 1-10, Ami-dong 1-ga, Seo-gu, Busan 602-738, Korea

Phone: 82-51-240-7845, Fax: 82-51-240-7422 E-mail: leemk@pusan.ac.kr

Received: July. 11, 2011 Revised: July. 25, 2011 Accepted: Oct. 13, 2011

Introduction

　Periostin, homologous with an insect cell adhesion molecule fasciclin I, is a recently identified gene that is preferentially expressed in periosteum, indicating a potential role in bone formation and maintenance of

bone structure

. It has also been shown to be involved in the development of heart valves and other tissues

, and to interact with other extracellular matrix proteins such as fibronectin, tenascin-C, collagen V, and peri- ostin itself

. Functionally, periostin interacts with in- tegrins to support cell adhesion and the spreading of chondrocytes, fibroblasts, and cancer cells

. In addition, recent studies have reported that periostin is frequently found to be highly expressed in various types of human cancer cell lines in vitro and human cancer tissues in vivo

. In tumorigenesis, however, there have been controversies about its role for invasion and metastasis of human malignancies. Kudo et al.

introduced that periostin was highly expressed in head and neck squ- amous cell carcinomas (HNSCC) in comparison with normal tissues, and the level of periostin expression was well correlated with the invasiveness of HNSCC cases.

Original Article

On the other hand, Kim et al.

reported that the ex- pression of periostin mRNA is significantly repressed in human bladder cancer tissues, and that periostin plays a role as a suppressive factor for invasion and meta- stasis. Collectively, the role of periostin in carcino- genesis is much variable so far. In human lung cancer, Sasaki et al.

demonstrated the non-small cell lung car- cinoma (NSCLC) patients with high periostin expression had significantly poorer survival than the patients with- out periostin expression by reverse transcriptase (RT)- PCR analysis. So, they suggested that periostin could be used as prognostic marker of NSCLC.

　In this study, we analyzed the expression of periostin protein in both tumor and intratumoral stroma in a co- hort group of NSCLC (n=91) by immunohistochemistry and correlated the findings with clinicopathologic pa- rameters such as demography (age, gender), TNM stage, and histologic factors (type, differentiation, β-catenin, E-cadherin, p53 and PCNA index).

Materials and Methods 1. Patients and clinicopathological data

All patients with NSCLC who underwent lobectomy or pneumonectomy at Pusan National University Hospital, Busan, Korea, between June 2000 to October 2008 were considered for inclusion in this study. After exclusion of cases in which there was insufficient patho- logical material remaining for further study, a total of 91 cases were selected for this study (80 men and 11 women). The mean patient age was 65.2 years (range, 40.0∼85.0 years; median, 67.0 years). Tumors were staged according to the TNM classification of the International Union Against Cancer (UICC) staging sys- tem

after reviewing of the clinical, radiological, and pathological data. Other clinical information was ex- tracted from medical records.

The surgically resected specimens were immediately fixed in 10% buffered formalin (pH 7.0). All sections containing both tumor tissue and surrounding lung tis- sue were embedded in paraffin. Pathologic diagnosis was based on the third edition of the World Health

Organization classification

. There were 46 cases of squamous cell carcinoma (SqCC) and 45 cases of ad- enocarcinoma (AdC).

2. Tissue microarray construction

The representative tumor areas to be sampled for tis- sue microarray (TMA) were carefully selected and marked on the H&E slides. TMAs were assembled using a tis- sue-arraying instrument consisting of thin-walled stain- less biopsy needles and stylets used to empty and trans- fer the needle contents. The assembly was held in an X-Y position guide that was manually adjusted by micrometers. Briefly, the instrument was used to create holes in a recipient paraffin block and to acquire tissue cores from the donor block using a thin-walled needle.

The cylindrical sample was retrieved from the selected region in the donor block and extruded directly into the recipient block with defined array coordinates. A solid stylet, closely fitted in the needle, was used to transfer the tissue cores into the recipient block. Taking tumor heterogeneity into account, we used a large-diameter stylet (2.0 mm). The study specimens were routinely oversampled with two replicate core tumor region sam- ples (different areas) from each donor block. Forty-eight cores from 24 tumors were included in each tissue array block. Multiple 4-μm sections were cut with a microtome.

3. Immunohistochemistry

Sections from TMA blocks were transferred to poly-L- lysine-coated glass slides and air-dried overnight at 37

C. They were then dewaxed in xylene (three changes), rehydrated in a graded series of decreasing ethanol con- centrations, and rinsed in pH 7.4 Tris-buffered saline (TBS). Endogenous peroxidase activity was inactivated with 5% hydrogen peroxide in methanol for 15 min at 37

C.

Antigen retrieval, except for anti-proliferating cell nu-

clear antigen (PCNA), was performed using 5-minute

microwave treatment (×2) in TBS. The following pri-

mary antibodies were used: periostin, β-catenin, E-cad-

herin, p53 and PCNA. Each antibody was incubated

with tissue sections at room temperature. Immunohisto-

Table 1. Information of antibodies used in this study

Antigen Source Clone Company Dilution Ab incubation

Periostin MmAb* 345613 R&D Systems 1:100 RT

, 60 min

P53 RmAb

SP5 Lab Vision, USA 1:100 RT, 60 min

PCNA MmAb PC10 SIGMA, USA 1:2,000 Overnight

b-catenin MmAb CAT-5H10 ZYMED, USA 1:500 RT, 60 min

E-cadherein MmAb 4A2C7 ZYMED, USA 1:200 RT, 60 min

*Mouse monoclonal antibody.

Rabbit monoclonal antibody.

Room temperature.

Figure 1. Periostin expression in tumors and intratumoral stroma (immunostain, ×200). (A) Non-neoplastic lung stroma shows negative (right) immunostaining in contrast to neoplastic stroma (left). (B) Intratumoral stoma around invasive ad- enocarcinoma shows positive immunoreaction. (C) The stroma of bronchioloalveolar carcinoma is negative for periostin.

(D) Occasionally, the cytoplasm of adenocarcinoma cells shows positive reaction to periostin. (E) The strongest reaction in the positive stroma of squamous cell carcinoma shows a band-like pattern. (F) Intratumoral stroma of well to moder- ately differentiated squamous cell carcinoma shows negative staining to periostin.

chemical procedures were performed using two-step Envision Plus kit (Zymed Co.) with chromogen DAB.

Detailed information concerning the antibodies used is summarized in Table 1.

4. Interpretation of staining

For periostin stain, when the positive reaction to the

antibody was found in less than 25% of tumor area stro-

ma, it was scored semiquantitatively 'low expression'.

Table 2. Relationship between periostin expression in peri- tumoral stroma and clinicopathological parameters in non-small cell lung carcinomas

Parameters Total no.

of cases

Periostin expression

p-value Low

(n=40)

High (n=51) Histological type

SCC* 46 19 27 0.606

AC

45 21 24

Differentiation

WD

14 10 4 0.024

MD-PD

45 30 47

Age (year)

≤65 41 21 20 0.206

＞65 50 19 31

Sex

Female 11 5 6 0.915

Male 80 35 45

Stage

I 42 18 24 0.845

II∼III 49 22 27

TNM-pT

T1 28 11 17 0.550

T2∼4 63 29 34

TNM-pN

N0 47 22 25 0.571

N1∼2 44 18 26

*Squamous cell carcinoma.

Adenocarcinoma.

Well differen- tiated.

Moderately differentiated-poorly differentiated.

Mean age group to create groups.

With more than 25%, it was considered 'high'. For PCNA, the labeling index was determined by the per- centage of positive nuclear staining after counting ap- proximately 1,000 tumor cells from tumor areas at high power field (×400). In case of other antibodies (p53, β-catenin, E-cadherin), if less than 5% of tumor cells were positive, it was considered 'negative'. When 5 to 25% of tumor cells are positive for antibodies, it was '1＋'. When more than 25% of tumor cells are positive, it was considered '2＋'.

5. Statistical analysis

Pearson χ

and Student's t tests were performed in the statistical analyses. For univariate survival analysis, the Kaplan-Meier method was applied. Clinicopathologic variables were further analyzed by multivariate Cox pro- portional hazards regression models to test for in- dependent prognostic value. All analyses were carried out using the SPSS 15.0 software package (SPSS Inc., Chicago, IL, USA). p-values less than 0.05 were re- garded as statistically significant.

Results

The stromas from non-neoplastic lung and most of in situ carcinomas, including bronchioloalveolar carci- noma, were immunonegative to extracellular matrix protein periostin (Figure 1A, C). Positive periostin im- munoreaction was frequently accentuated toward intra- tumoral stroma around invasive carcinoma cells. The strongest reaction in the positive stroma was present as band-like pattern just around tumor nests (Figure 1B, E). Occasionally, tumor cells themselves showed cyto- plasmic expression of periostin in 4 of 46 SqCC cases (8.7%) and in 2 of 45 AdC (4.4%) (Figure 1D). The dis- tribution of tumor cells with cytoplasmic expression was chiefly in the periphery of tumor nests in SqCC, whereas the cells with cytoplasmic expressin were randomly scattered in AdC. None of the tumors with intracellular periostin expression morphologically exhibited spin- dle-cell or mesenchymal morphologic change.

Correlation between histologic types of NSCLC and

periostin expression was not shown. Regarding tumor differentiation, moderately to poorly differentiated tu- mors (47/77) revealed even higher periostin expression than well-differentiated ones (4/14) (p=0.024, Table 2).

Otherwise, high periostin expression was positively cor- related with E-cadherin and p53 expression (Table 3).

However, it was not related with PCNA index, β-cat- enin, and cyclin D1 (respectively, p＞0.05). Among other clinicopathologic parameters, patient age, sex, TNM-pT, TNM-pN and stage were not correlated with the degree of periostin expression (respectively, p＞

0.05). Occasional expression of periostin in tumor cell

cytoplasm also did not reveal any specific relation with

clinicopathologic parameters, although the ratio was

very small (4.4∼8.7%).

Table 3. Relationship between periostin expression in peritumoral stroma and clinicopathological parameters in non-small cell lung carcinomas

Parameters Grade Total no.

of cases

Periostin expression

p-value Low (n=40) High (n=51)

β-catenin neg 39 13 26 0.886

1＋ 26 12 14

2＋ 36 15 21

E-cadherin neg 17 10 7 0.048

1＋ 18 11 7

2＋ 56 19 37

p53 neg 49 28 21 0.017

1＋ 32 8 24

2＋ 10 4 6

Cyclin D1 neg 39 22 31 0.481

1＋ 26 17 20

2＋ 36 1 0

PCNA, %* 91 42.0±22.2 39.7±22.3 0.638

*Expressed as mean±SD.

SD: standard deviation.

Figure 2. Overall survival curves for patients with non- small cell lung carcinoma showing low expression (n=40) and high expression (n=51) of periostin.

Table 4. Factor related to the prognosis of non-small cell lung carcinoma in this study

Factor HR (95% CI) p-value

Nodal metastasis

No 1

Yes 4.16 (1.55∼11.24) 0.005

HR: hazard ratio; CI: confidence interval.

With regard to the degree of periostin expression, univariate survival analysis of NSCLC patients was per- formed by Kaplan-Meier method. Thirty-three of 40 pa- tients with low expression were censored and 7 were died (mean and median survival times, 81.0 and 42.5 months, respectively). Thirty-five of 51 patients with high expression were censored and 16 died (mean and median survival times, 79.1 and 34.0 months, re- spectively). The overall survival was not related with

the degree of periostin expression (p=0.126 by log-rank test, Figure 2). Additionally, multivariate Cox propor- tional hazards regression analysis showed independent prognostic value only in TNM-N stage (Table 4), but other factors including periostin failed to reveal sig- nificant prognostic value.

Discussion

There has been a growing awareness that the cellular

microenvironment throughout tumorigenesis plays a

critical role in providing supportive microsurroundings

for cancer cell proliferation, survival, migration and in-

vasion

. Cancer cells launch various signaling path-

ways to respond to the altered external microenviron-

ment

.

Periostin is a secreted protein that was originally iso- lated from MC3T3-E1 osteoblast cells to modulate cell adhesion as an extracellular matrix (ECM)-associated molecule

. Though structurally unrelated, it could di- rectly interact with integrin as ECM-associated molecules.

Baril et al.

demonstrated that periostin promotes the survival of pancreas cancer cells by inducing Akt phos- phorylation through binding to ß4 integrin and activa- tion of PI3 kinase pathway. Furthermore, periostin-in- tegrin interaction may affect the ECM-integrin inter- action and trigger the intracellular signaling and activa- tion of certain genes that are involved in tumor pro- gression

.

In the present study, the extracellular matrix protein periostin regularly showed an expression gradient in the stroma as band-like fashion immediately surrounding in- vasive nests of cancer cells. In NSCLC, periostin was shown to be up-regulated on mRNA level at the tumor periphery, predominantly in the stromal cells just around the carcinoma cells, with very little expression in the tumor cells themselves

. The present study also showed similar results, although it was just performed by immunohistochemical stain. The significance of this spatial distribution of periostin expression remains un- clear, although the question arises if a peritumoral band-like expression of periostin in the tumor stroma facilitate invasion or rather mediate protective functions against the invading tumor. When interpreting im- munohistochemical result of periostin, we originally had used four-tiered system based on the extent of the stain- ing, as 0, 1＋, and 2＋, and 3＋. But, we finally de- cided to change the system into two-tiered one to de- crease possible interobserver discrepancies and improve statistical significance. Because of the small diameter of each specimen and relatively unfamiliar peritumoral im- unostaining pattern, relatively concurred results com- pared with others were not made in this study.

In fact, clinicopathological studies revealed that peri- ostin can serve as a prognostic marker that is overex- pressed in many kinds of tumor tissues such as non- small-cell lung carcinomas, ovarian carcinoma, oral can-

cer, and head and neck squamous cell carcinoma

. These results suggest that periostin should play an im- portant role in the tumor cell adhesion of NSCLC, and partly contribute its aggressive behavior. However, the serum level of periostin expression may not be asso- ciated with well-known prognostic indicators of NSCLC such as stage, tumor size and nodal status

. Interestingly, authors found that there was no relation between the expression level of periostin and progression variables such as tumor size, nodal status, stage, etc. Soltermann et al.

reported that patients of SqCC histology, male gender, or lower age had a particularly periostin-rich stroma. On the contrary, our study result was also out of line with theirs.

Herein, high periostin expression was not correlated with advanced stage and periostin expression showed no statistical relationship with histologic type, patient age and sex. According to our results, there was no re- lationship with periostin expression and patient's surviv- al from the Kaplan-Meier method for univariate survival analysis. Additionally, multivariate Cox proportional hazards regression analysis was also applied. From the analysis, only TNM-nodal stage showed independent prognostic value, but other factors including periostin failed to reveal any prognostic value. These results may be related with some limitations of this study. Thus, the demerits of the study could include a relatively small number of cases studied, a large proportion of early stage tumors (stage I in 46.2% out of the total 91 cases), and a relatively short postoperative follow-up period.

In spite of the limitations, however, we presume that periostin may be not a consistent prognostic marker of NSCLC.

On the other hand, periostin expression was related with tumor differentiation. High periostin expression was correlated with moderately to poorly differentiated tumors. The result suggests that periostin expression might be related with aggressiveness of tumor cells.

Further study using a large number of cases with poorly

differentiated tumors is needed to confirm significance

of periostin expression in tumor differentiation and ag-

gressive behavior.

Hong et al.

showed that periostin could promote the proliferation and migration of A549 cells (NSCLC cell line) by inducing vimentin and N-cadherin expression and downregulating E-cadherin expression. On the oth- er hand, Shao et al.

used three tumor cell lines to en- gineer stable cell lines that overexpress periostin. Unexpec- tedly, the proliferation rate of these periostin-producing cells was noticeably slower than that of the control cells in culture, suggesting that periostin does not promote proliferation of tumor cells in vitro. Kudo et al.

also demonstrated that periostin overexpression does not promote cell proliferation, but it dramatically enhances the invasiveness of the head and neck cancer cell lines HSC2 and HSC3. In the present study, immunohisto- chemical expression levels of periostin in NSCLC were not correlated with tumor cell PCNA index (42.0±

22.2% in low expression group vs. 39.7±22.3% in high group), either. Instead, the proliferation index in high expression group was slightly lower than that in low group. The expression of cyclin D1, another indicator of tumor cell proliferation, was also not related with the level of periostin expression.

Recently, it is suggested that secreted periostin inter- acts with αvβ5 integrins, and the intracellular signal- ing activation via cross-talk between integrins, and EGFR promotes the cell to undergo epithelial-mesen- chymal transition (EMT), resulting in tumor invasion and metastasis

. The molecular basis of EMT in cancer cells that is a process of morphologic transdifferentiation in- volves multiple changes in expression, distribution, and/or function of several proteins that include peri- ostin, vimentin, integrins, matrix metalloproteinases, and cadherins

. In the present study, high periostin expression was positively correlated with E-cadherin and p53 expression, but it was not related with β-cate- nin, a member of the wnt signaling. Cadherins are a multifunctional family of Ca

-dependent, transmembrane glycoproteins, and their expression is known to be re- quired for the assembly of cells into solid tissues.

Importantly, E-cadherins are involved in a homotypic epithelial adhesion

. Therefore, we assumed that the expression level of periostin, an EMT-related protein,

would be inversely correlated with E-cadherin. Howev- er, the result was the converse. Though the precise rea- son of this discordance remains unclear in this study, it might be related with a rather small sample size of NSCLC, interpretation method for immunohistochemical staining, and any character of monoclonal antibody used. Further studies would be necessary to identify the regulators of periostin in tumorigenesis.

In conclusion, these results suggest that periostin might play a role in the aggressive biologic behavior of NSCLC. However, in contrast with the previous re- sults, their expression may not be associated with well-known prognostic indicators such as stage and tu- mor cell proliferation.

Acknowledgements

This work was supported by a grant from the National R&D Program for Cancer Control, Ministry for Health, Welfare and Family Affairs, Republic of Korea (0920050).

References

1. Horiuchi K, Amizuka N, Takeshita S, Takamatsu H, Katsuura M, Ozawa H, et al. Identification and charac- terization of a novel protein, periostin, with restricted expression to periosteum and periodontal ligament and increased expression by transforming growth factor beta. J Bone Miner Res 1999;14:1239-49.

2. Kühn B, del Monte F, Hajjar RJ, Chang YS, Lebeche D, Arab S, et al. Periostin induces proliferation of dif- ferentiated cardiomyocytes and promotes cardiac repair. Nat Med 2007;13:962-9.

3. Lie-Venema H, Eralp I, Markwald RR, van den Akker NM, Wijffels MC, Kolditz DP, et al. Periostin expression by epicardium-derived cells is involved in the develop- ment of the atrioventricular valves and fibrous heart skeleton. Differentiation 2008;76:809-19.

4. Norris RA, Borg TK, Butcher JT, Baudino TA, Banerjee I, Markwald RR. Neonatal and adult cardiovascular pathophysiological remodeling and repair: devel- opmental role of periostin. Ann N Y Acad Sci 2008;

1123:30-40.

5. Norris RA, Damon B, Mironov V, Kasyanov V, Rama-

murthi A, Moreno-Rodriguez R, et al. Periostin regu-

lates collagen fibrillogenesis and the biomechanical properties of connective tissues. J Cell Biochem 2007;101:695-711.

6. Takayama G, Arima K, Kanaji T, Toda S, Tanaka H, Shoji S, et al. Periostin: a novel component of sub- epithelial fibrosis of bronchial asthma downstream of IL-4 and IL-13 signals. J Allergy Clin Immunol 2006;

118:98-104.

7. Bao S, Ouyang G, Bai X, Huang Z, Ma C, Liu M, et al. Periostin potently promotes metastatic growth of colon cancer by augmenting cell survival via the Akt/PKB pathway. Cancer Cell 2004;5:329-39.

8. Siriwardena BS, Kudo Y, Ogawa I, Kitagawa M, Kitajima S, Hatano H, et al. Periostin is frequently over- expressed and enhances invasion and angiogenesis in oral cancer. Br J Cancer 2006;95:1396-403.

9. Kudo Y, Siriwardena BS, Hatano H, Ogawa I, Takata T. Periostin: novel diagnostic and therapeutic target for cancer. Histol Histopathol 2007;22:1167-74.

10. Puglisi F, Puppin C, Pegolo E, Andreetta C, Pascoletti G, D'Aurizio F, et al. Expression of periostin in human breast cancer. J Clin Pathol 2008;61:494-8.

11. Kudo Y, Ogawa I, Kitajima S, Kitagawa M, Kawai H, Gaffney PM, et al. Periostin promotes invasion and an- chorage-independent growth in the metastatic process of head and neck cancer. Cancer Res 2006;66:6928-35.

12. Kim CJ, Yoshioka N, Tambe Y, Kushima R, Okada Y, Inoue H. Periostin is down-regulated in high grade hu- man bladder cancers and suppresses in vitro cell in- vasiveness and in vivo metastasis of cancer cells. Int J Cancer 2005;117:51-8.

13. Sasaki H, Lo KM, Chen LB, Auclair D, Nakashima Y, Moriyama S, et al. Expression of Periostin, homologous with an insect cell adhesion molecule, as a prognostic marker in non-small cell lung cancers. Jpn J Cancer Res 2001;92:869-73.

14. Sobin LH, Wittekind C. International Union against Cancer. TNM Classification of malignant tumours. 5th ed. New York: Wiley-Liss; 1997.

15. Travis WD, Colby TV, Corrin B, Shimosato Y, Brambilla E, Sobin LH. Histological typing of lung and pleural tumours. International histological classification of tumours.

3rd ed. Berlin: Springer-Verlag; 1999.

16. Hu M, Polyak K. Microenvironmental regulation of can- cer development. Curr Opin Genet Dev 2008;18:27-34.

17. Bissell MJ, Radisky D. Putting tumours in context. Nat Rev Cancer 2001;1:46-54.

18. Radisky D, Hagios C, Bissell MJ. Tumors are unique organs defined by abnormal signaling and context.

Semin Cancer Biol 2001;11:87-95.

19. López-Maury L, Marguerat S, Bähler J. Tuning gene ex-

pression to changing environments: from rapid re- sponses to evolutionary adaptation. Nat Rev Genet 2008;9:583-93.

20. Takeshita S, Kikuno R, Tezuka K, Amann E. Osteoblast- specific factor 2: cloning of a putative bone adhesion protein with homology with the insect protein fasciclin I. Biochem J 1993;294:271-8.

21. Baril P, Gangeswaran R, Mahon PC, Caulee K, Kocher HM, Harada T, et al. Periostin promotes invasiveness and resistance of pancreatic cancer cells to hypoxia-in- duced cell death: role of the beta4 integrin and the PI3k pathway. Oncogene 2007;26:2082-94.

22. Sasaki H, Dai M, Auclair D, Fukai I, Kiriyama M, Yamakawa Y, et al. Serum level of the periostin, a ho- mologue of an insect cell adhesion molecule, as a prognostic marker in nonsmall cell lung carcinomas.

Cancer 2001;92:843-8.

23. Ouyang G, Liu M, Ruan K, Song G, Mao Y, Bao S.

Upregulated expression of periostin by hypoxia in non-small-cell lung cancer cells promotes cell survival via the Akt/PKB pathway. Cancer Lett 2009;281:213-9.

24. Hong L, Sun H, Lv X, Yang D, Zhang J, Shi Y.

Expression of periostin in the serum of NSCLC and its function on proliferation and migration of human lung adenocarcinoma cell line (A549) in vitro. Mol Biol Rep 2010;37:2285-93.

25. Soltermann A, Tischler V, Arbogast S, Braun J, Probst- Hensch N, Weder W, et al. Prognostic significance of epithelial-mesenchymal and mesenchymal-epithelial transition protein expression in non-small cell lung cancer. Clin Cancer Res 2008;14:7430-7.

26. Shao R, Bao S, Bai X, Blanchette C, Anderson RM, Dang T, et al. Acquired expression of periostin by hu- man breast cancers promotes tumor angiogenesis through up-regulation of vascular endothelial growth factor receptor 2 expression. Mol Cell Biol 2004;24:

3992-4003.

27. Yan W, Shao R. Transduction of a mesenchyme- specif- ic gene periostin into 293T cells induces cell invasive activity through epithelial-mesenchymal transforma- tion. J Biol Chem 2006;281:19700-8.

28. Larue L, Bellacosa A. Epithelial-mesenchymal transition in development and cancer: role of phosphatidylinosi- tol 3' kinase/AKT pathways. Oncogene 2005;24:7443-54.

29. Mimeault M, Batra SK. Interplay of distinct growth fac- tors during epithelial mesenchymal transition of cancer progenitor cells and molecular targeting as novel can- cer therapies. Ann Oncol 2007;18:1605-19.

30. Takeichi M. Cadherin cell adhesion receptors as a mor-

phogenetic regulator. Science 1991;251:1451-5.