- Email: [email protected]
Strong intracellular and negative peripheral expression of fibronectin in tumor cells contribute to invasion and metastasis in papillary thyroid carcinoma
Cancer Letters 146 (1999) 103±109 www.elsevier.com/locate/canlet
Strong intracellular and negative peripheral expression of ®bronectin in tumor cells contribute to invasion and metastasis in papillary thyroid carcinoma Shinso Ryu a,b,*, Shiro Jimi a, Youichi Eura b, Toshihiko Kato b, Shigeo Takebayashi a a
The Second Department of Pathology, School of Medicine, Fukuoka University, 7-45-1 Nanakuma, Jonanku, Fukuoka 814-0180, Japan b Department of Otorhinolaryngology, School of Medicine, Fukuoka University, 7-45-1 Nanakuma, Jonanku, Fukuoka 814-0180, Japan Received 4 March 1999; received in revised form 20 May 1999; accepted 24 June 1999
Abstract The intracellular and stromal expression of ®bronectin (FN) in invading and central parts of papillary thyroid carcinomas with/without lymph node (LN) metastasis (54 metastasizing cases, 52 non-metastasizing cases) were examined immunohistochemically. The intracellular expression of FN in tumor cells in invading parts was stronger than that in central parts in most cases (79/106 cases, 74.5%). In invading parts, negative stromal FN was frequently found at the periphery of the tumor in cases with extracapsular soft tissue invasion (26/37 cases). Tumor cells in invading parts in metastatic cases were signi®cantly more likely to be negative for stromal FN at the periphery of the tumor than those in non-metastatic cases (P , 0:0001). The strong intracellular and negative stromal FN at the periphery of the tumor in invading parts were associated with invasion and metastasis in papillary thyroid carcinoma. These results suggest that these distinctive characteristics of FN may be useful for understanding invasion and metastasis in vivo. q 1999 Elsevier Science Ireland Ltd. All rights reserved. Keywords: Fibronectin; Thyroid carcinoma; Invasion; Metastasis; Immunohistochemistry
1. Introduction Fibronectin (FN), an extracellular matrix protein, is produced by a wide variety of cells such as ®broblasts [1], endothelial cells [2], epithelial cells [3], macrophages [4] and some carcinoma cells [5±8]. FN plays several important roles in cell adhesion, morphology, migration, oncogenic transformation, neovascularization [1,9], tumor invasion and metastasis [10]. Previous reports have demonstrated a relationship between stromal FN, tumor invasion and metastasis. * Corresponding author. Tel.: 181-92-801-1011 ext. 3285; fax: 181-92-863-8383. E-mail address: [email protected] (S. Ryu)
Tumor cells are generally characterized by decreased adhesiveness due to failure to deposit stromal FN, followed by rapid proliferation and migration [11]. Additionally, positive stromal FN detected by immunohistochemical staining is associated with a low metastatic potential [12]. To date, there have been few reports of the relationship between intracellular FN, tumor invasion and metastasis, as very few carcinomas express intracellular FN [5±8]. Our previous study [13] demonstrated that thyroid carcinoma distinctively expresses intracellular FN in vivo after malignant transformation of thyroid tissue. In this study we immunohistochemically examined the stromal and intracellular FN distribution in invading and central parts of thyroid carcinomas, and in
0304-3835/99/$ - see front matter q 1999 Elsevier Science Ireland Ltd. All rights reserved. PII: S 0304-383 5(99)00259-1
104
S. Ryu et al. / Cancer Letters 146 (1999) 103±109
both peripheral and central parts of metastasizing LNs to determine the interaction between FN, invasion and metastasis. 2. Materials and methods 2.1. Patients and samples The specimens used in this study were obtained from surgical samples taken from March 1985 to July 1996 at Fukuoka University Hospital, Fukuoka, Japan. The samples were from 106 patients (19 males and 87 females), mean age 49 years (range 33±77 years). All the specimens were papillary thyroid carcinoma (well-differentiated 98 cases, poorly-differentiated, 8 cases) of which 54 cases had metastasis to LNs alone. The remaining 52 cases had no metastasis to any other organs, including LNs. The clinical stage in all patients was determined according to the International Union Against Cancer (IUCC) [14], and histological diagnosis was made according to the classi®cation established by the World Health Organization [15]. All samples were ®xed in 10% buffered formalin and embedded in paraf®n. 2.2. Immunohistochemical staining for FN Deparaf®nized sections (4 mm thick) were washed in TBS (pH 7.4) and treated with protease (DAKO, Denmark). Immunohistochemical staining was performed by the APAAP method. In brief, the primary antibody against human plasma FN (DAKO) was diluted 1:600 and incubated overnight at 48C. After washing in TBS, the sections were incubated with alkaline phosphatase-conjugated second antibody against rabbit immunoglobulin (DAKO) for FN for 60 min at room temperature. The sections were then incubated after a second wash with alkaline phosphatase-conjugated third antibody against alkaline phosphatase (DAKO) for FN, for 60 min at room temperature. Vehicle alone or an identical concentration of nonimmune animal IgG was used as a negative control. To check the speci®city of the ®rst antibody, an excess amount of antigen, FN (Sigma, MO), was mixed with the ®rst-antibody solution and incubated for 1 h at 378C. This solution was used as a ®rst-antibody solution for immunohistochemistry. Fibroblasts in human skin tissue (which
produce abundant FN under normal physiological conditions in vivo and vitro) were used to test antibody immunoreactivity as a positive control. The sections were then colored by 0.01% new fuchsin (Merck, Germany), 0.01% NaNO2, 10 mg naphthol AS-BI phosphate (Sigma), 0.1 ml N,Ndimethyl formamide (Wako, Osaka) and 10 mg levamisol (Sigma) in 40 ml of 0.2 M Tris±HCl buffer (pH 8.2). Specimens were observed under a microscope. All the stained material in specimens treated with the antibody contained FN. 2.3. Examination of parts and evaluation of staining results The thyroid carcinoma in each case was divided into two parts, an invading part and a central part. Since multicentric tumor nests are formed in papillary thyroid carcinoma, the following de®nition was emphasized: the invading part was de®ned as the area where carcinoma cells invaded thyroid tissue and/or other tissue, such as muscular tissue, fatty tissue, blood vessel or tumor capsule. The central part was de®ned as the center region of the tumor nests. In metastasizing LN, the tumor was also divided into two parts, a peripheral part and a central part. The peripheral part was de®ned as the region where carcinoma cells contacted or were located near lymphoid tissue. The central part was de®ned as the center region of metastasizing tumor nests. Immunohistological evaluation was performed using the following criteria to classify tissue as positive: (1) in carcinoma cells/normal cells adjacent to the carcinoma, when one or more carcinoma cells/ normal cells were intracellulary positive and (2) in stroma, when the stroma completely surrounded carcinoma cells/normal cells adjacent to the carcinoma. FN accumulation in tissue was evaluated semiquantitatively using the following parameters: intracellular FN in normal or carcinoma cells, negative (no cells), focal positive (from one cell up to 74% of the total cells), diffuse positive (75 % or more) and stromal FN (the periphery of epithelial cells): negative (lack of stromal FN), focal positive (weakly uniform stromal FN) and diffuse positive (strongly uniform stromal FN). A statistical analysis was carried out using Student's t-test and x 2 test for independence.
S. Ryu et al. / Cancer Letters 146 (1999) 103±109
3. Results 3.1. Immunohistochemical distribution of FN in carcinomas and metastasizing LNs The invading parts of thyroid carcinomas were diffusely positive for intracellular FN in most cases (85/106 cases, 80.2%) while approximately half of the cases were negative for stromal FN (50/106 cases, 47.2%) (Table 1). Thirty-seven cases had extracapsular soft tissue invasion, including 2 cases that also had capsular and blood vascular invasion. The remaining 69 cases had thyroid tissue invasion. Negative stromal FN at the periphery of the tumor in extracapsular soft tissue invasion was signi®cantly more common than positive stromal FN (26 negative cases versus 11 positive cases, P , 0:05). However, the more frequent occurance of negative stromal FN at the tumor periphery was not found in intra-thyroid tissue invasion. The central parts of thyroid carcinomas were focally positive for intracellular FN and stromal FN in almost all cases (90/106 cases, 84.9% and 84/106 cases, 79.2%) (Table 1). A pattern in which the invading part of the carcinoma was diffusely positive for intracellular FN and negative for stromal FN at the periphery of the tumor (Fig. 1A) was found in 47/106 cases (44.3%). A second pattern in which the invading part of the carcinoma was diffusely positive for intracellular FN and focally positive for stromal FN at the periphery of the tumor (Fig. 1B) was found in 31/106 cases (29.2%). A third pattern was found in which the central part of the carcinoma was focally positive for both intracellular FN and stromal FN (Fig. 1C) in 78/106 cases (73.4%). Thyroid tissue was negative for intracellular FN in all cases and focally positive for stromal FN in almost all cases (95/106 cases, 89.6 %). The distribution of FN
105
in metastasizing LNs was similar to that of the original carcinomas (Table 2). 3.2. Comparison of the FN distribution in metastatic and non-metastatic cases With regard to the distribution of FN in thyroid carcinomas (Table 1), the intracellular expression of FN in most cases (79/106, 74.5%) was stronger in invading parts was stronger than in central parts. The invading parts were negative for stromal FN at the periphery of the tumor. However, this pattern was not associated with metastasis (P 0:8711) (Table 3). Tumor cells in invading parts of metastatic cases were signi®cantly more likely to be negative for stromal FN at the periphery of the tumor than those of non-metastatic cases (P , 0:0001) (Table 3). Other parameters such as age, sex and differentiation were also examined but found to have no signi®cant relationship to metastasis (Table 4). 4. Discussion Our data indicated that in thyroid carcinoma, negative stromal FN at the periphery of tumor cells in invading parts was signi®cantly associated with metastasis, and that the intracellular expression of FN in invading parts was stronger than that in central parts but not associated with metastasis (Table 3). Subsequently, negative stromal FN at the periphery of the tumor was frequently found in extracapsular soft tissue invasion. To our knowledge, this is the ®rst report of a relationship between FN distribution and invasion or metastasis of thyroid carcinoma. Previous studies [12,16] have demonstrated that negative staining for stromal FN has signi®cant association
Table 1 The immunohistochemical distribution of FN in invading and central parts of papillary thyroid carcinomas Cases/total cases (%) Invading part
Positive Diffusely Focally Negative
Central part
Tumor cell
Stroma
Tumor cell
Stroma
85/106 (80.2) 21/106 (19.8) 0
16/106 (15.1) 40/106 (37.7) 50/106 (47.2)
6/106 (5.7) 90/106 (84.9) 10/106 (9.4)
20/106 (18.9) 84/106 (79.2) 2/106 (1.9)
106
S. Ryu et al. / Cancer Letters 146 (1999) 103±109
with high metastatic potential in invasive breast carcinoma. Although our materials are different (thyroid carcinoma versus breast carcinoma) the results are very similar. Previous studies [6,7,17,18] have demonstrated that FN is expressed in tumor cells of sarcoma and in some carcinomas, such as hepatocellular carcinoma and breast carcinoma. However it appears there was no comparison of a tumor's invading and central parts. Our previous study [13] indicated that thyroid carcinoma distinctively expresses intracellular FN in vivo in comparison with other carcinomas (including hepatocellular carcinoma, transitional cell carcinoma and gastric adenocarcinoma). In this study, we compared intracellular FN in invading and central tumor parts and discovered that intracellular expression of FN was stronger in the invading parts. It has been reported that malignant transformation in vitro is often accompanied by diminished stromal FN synthesis [5], the point at which cells acquire invasive or metastatic characteristics [11]. This phenomenon is consistent with our result that negative stromal FN was found at the tumors periphery in invading parts, especially in extracapsular soft tissue invasion. This correlated well with metastasis. However, tumor invasion and metastasis are highly complex multistage processes. The interaction of tumor cells with matrix (such as FN, collagen and laminin) also plays a key role in tumor invasion and metastasis [19,20] including not only diminished matrix synthesis in tumors but also degradation of matrix in invading regions [21] and increased production of matrix by host cells in response to the presence of a tumor [22]. However, the histological heterogeneosity of tumors is great enough such that these phenomena may take place separately or simultaFig. 1. FN localization in cases of (A,C) papillary thyroid carcinoma with metastasis (65 year old, female), (B) papillary thyroid carcinoma without metastasis (52 year old, female). (A) In an invading part of the carcinoma, tumor cells were diffusely positive for intracellular FN (right), but negative for stromal FN at the periphery of the tumor. Thyroid tissue was negative for intracellular and stromal FN (left). (B) In an invading part of the carcinoma, tumor cells were diffusely positive for intracellular FN (arrow), and focally positive for stromal FN at the periphery of the tumor. Thyroid tissue was negative for intracellular and stromal FN. (C) In a central part of the carcinoma, tumor cells were focally positive for intracellular FN, and stroma was focally positive for stromal FN. (original magni®cation of all sections, £ 200).
S. Ryu et al. / Cancer Letters 146 (1999) 103±109
107
Table 2 The immunohistochemical distribution of FN in metastasizing LNs of papillary thyroid carcinomas Cases/total cases (%) Peri-tumor part
Positive Diffusely Focally Negative
Central part
Tumor cell
Stroma
Tumor cell
Stroma
45/54 (83.3) 9/54 (16.7) 0
3/54 (5.6) 25/54 (46.3) 26/54 (48.1)
1/54 (1.9) 51/54 (94.4) 2/54 (3.7)
2/54 (3.7) 38/54 (70.4) 14/54 (25.9)
neously in different regions of the same tumors [23]. These phenomena may be explained by negative stromal FN at the tumor's periphery in extracapsular soft tissue invasion, but not in intra-thyroid tissue invasion. Our data and those of previous studies [12,16] may have resulted from these complicated interactions. A previous in vitro study [24] demonstrated that invasion by squamous cell carcinoma is highly correlated with cell-matrix adhesion by FN. It is well known that RGD (Arg-Gly-Asp), a sequence in the middle of FN, is recognized by the cell surface receptor of integrins, such as the FN receptor (a 5b 1 integrin). This results in the induction of cell attachment and adhesion [25,26]. Tumor cells that express a high level of a5b1 integrin adhere ®rmly to FN matrix and consequently exhibit less migratory activity [27]. In this study, almost all cases (104/106 cases, 98.1%) were positive for stromal FN in the central parts of thyroid tumors, an area where migratory activity should have been less than the invading parts. In
general, tumor cells, as compared to normal cells, diminish their own extracellular matrix proteins which causes greater proliferative and migratory activity [11]. We found that relatively few thyroid tumors were positive for stromal FN in invading parts, (56/106 cases, 52.8 %) and that cases with negative expression of stromal FN at the periphery of the tumor were signi®cantly associated with metastasis (Table 3). This is consistent with previous studies [12,16]. Almost all cases in this study exhibited positive expression of intracellular FN in thyroid tumors in invading and central parts (Table 1). However, there was a large difference in the rate of positive expression of stromal FN in the two parts (52.8% for invading parts versus 98.1% for central parts). One possible explanation for this ®nding is that thyroid carcinoma cells of invading parts secrete less extracellular stromal FN than central parts of tumor nests. Another alternative is that tumor cells in invading parts express more intracellular FN than those in central parts.
Table 3 Comparison of FN distribution in metastatic and non-metastatic cases of papillary thyroid carcinoma a Cases/total cases (%) Total
Metastasis
Non-metastasis
Intracellular FN Invading part . central part Invading part central part
79/106 (74.5) 27/106 (25.5)
40/54 (74.1)* 14/54 (25.9)*
39/52 (75.0)* 13/52 (25.0)*
Stromal FN in invading part Positive Negative
56/106 (52.8) 50/106 (47.2)
11/54 (20.4)** 43/54 (79.6)**
45/52 (86.5)** 7/52 (13.5)**
a
*P 0:8711, **P , 0:0001 by chi-square test for independence.
108
S. Ryu et al. / Cancer Letters 146 (1999) 103±109
Table 4 Comparison of clinical and histological ®ndings of thyroid carcinoma in metastatic and non-metastatic cases
Cases Age (years) M/F Differentiation Poorly/well a b
the Society for the Promotion of International OtoRhino-Laryngology (SPIO).
Metastasis
Non-metastasis
P value
References
54 48.8 ^ 14.8 12/42
52 50.8 ^ 15.9 7/45
0.532 a 0.237 b
[1] K.M. Yamada, J.A. Weston, Isolation of a major cell surface glycoprotein from ®broblasts, Proc. Natl. Acad. Sci. USA 71 (1974) 3492±3496. [2] E.A. Jaffe, D.F. Mosher, Synthesis of ®bronectin by cultured human endothelial cells, J. Exp. Med. 147 (1979) 1779±1791. [3] L.B. Chen, N. Maitland, P.H. Gallimore, J.K. McDougall, Detection of the large external transformation-sensitive protein on some epithelial cells, Exp. Cell Res. 106 (1977) 39±46. [4] B. Villiger, D.G. Kelly, C. Kuhn, J.A. McDonald, Human alveolar macrophage ®bronectin: synthesis and ultrastructural localisation, Clin. Res 28 (1980) 745A. [5] K. Olden, K.M. Yamada, Mechanism of the decrease in the major cell surface protein of chick embryo ®broblasts after transformation, Cell 11 (1997) 957±969. [6] H.S. Smith, J.L. Riggs, N.W. Mosesson, Production of ®bronectin by human epithelial cells in culture, Cancer. Res. 39 (1979) 4138±4144. [7] E. Pearlstein, R.O. Hynes, L.M. Franks, V.J. Hemmings, Surface proteins and ®brinolytic activity of cultured mammalian cell, Cancer Res. 36 (1976) 1475±1479. [8] A. Vaheri, M. Kurkinen, V.P. Lehto, E. Linder, R. Timpl, Codistribution of pericellular matrix protein in cultured ®broblasts and loss in transformation: ®bronectin and procollagen, Proc. Natl. Acad. Sci. USA 75 (1978) 4944±4948. [9] R.O. Hynes, K.M. Yamada, Fibronectin: multifunctional modular glycoproteins. , J. Cell Biol. 95 (1982) 369±377. [10] A.J. D'ardenne, J.O'D. McGee, Fibronectin in disease, J. Pathol. 12 (1984) 235±351. [11] H. Mohri, Fibronectin and integrin interactions, J. Invest. Med. 44 (1996) 429±441. [12] L. Christensen, M. Nielsen, J. Andersen, I. Clemmensen, Stromal ®bronectin staining pattern and metastasizing ability of human breast carcinoma, Cancer Res. 48 (1988) 6227±6233. [13] S. Ryu, S. Jimi, S. Takebayashi, Thyroid carcinoma distinctively expresses intracellular ®bronectin in vivo, Cancer Lett. 121 (1997) 189±193. [14] P. Hermanek, L.H. Sobin, Classi®cation of Malignant Tumours, in: P, Hermanek, L.H. Sobin (Eds.), fourth ed., International Union Against Cancer. Springer: New York, pp. (1987) 33±35. [15] C. Hedinger, E.D. Williams, L.H. Sobin, Histological typing of thyroid tumours, second ed, World Health Organization, Geneva, 1988. [16] H. Takei, Y. Iino, J. Horiguchi, T. Yokoe, Immunohistochemical ®bronectin staining pattern and prognosis in invasive breast carcinoma, Oncology 52 (1995) 106±111. [17] S. Korach, M.F. Poupon, J.A. Du Villard, M. Becker, Differential adhesiveness of rhabdomyosarcoma-derived cloned metastatic cell lines to vascular endothelial monolayers, Cancer Res. 46 (1986) 3624±3629.
5/49
3/49
0.531 b
Student's t-test. x 2 test for independence.
Our results also showed that negative stromal FN at the tumor's periphery in the invading part of a carcinoma was signi®cantly more common in metastatic cases than other parameters such as age, sex and differentiation (Tables 3 and 4). Previous studies [12,16] have reported that negative staining for stromal FN is superior to such parameters as clinical stage or histological type or grade as an index of metastasis and prognosis of breast carcinoma. In this study, we could not examine the prognosis using negative staining for stromal FN since almost all cases (data not shown) had very good prognosis (which is common in differentiated thyroid carcinoma) [28]. In this study we found the distribution of FN in metastasizing LNs was very similar to that which occurred in the original thyroid carcinomas (Table 2). In our previous study [29], we demonstrated that intracellular FN is distinctive to thyroid carcinomas, even when they metastasize to other organs, such as the lung, kidney and LN. In the present study, we found that intracellular and stromal FN retained their distinctive distribution in metastasizing LNs. In conclusion, the intracellular expression of FN in thyroid carcinoma was stronger in invading parts than in central parts and that negative stromal FN at the tumor's periphery in invading parts was signi®cantly correlated with extracapsular soft tissue invasion and metastasis to LNs. These results may be useful for understanding invasion and metastasis in vivo.
Acknowledgements This work was supported by a research grant from
S. Ryu et al. / Cancer Letters 146 (1999) 103±109 [18] S. Matsui, T. Takahashi, Y. Oyanagi, S. Takahashi, S. Boku, K. Takahashi, Expression, localization and alternative splicing pattern of ®bronectin messenger RNA in ®brotic human liver and hepatocellular carcinoma, J. Hepatol. 27 (1997) 843±853. [19] L.A. Liotta, Tumor invasion and metastasis. Role of extracellular matrix, Cancer Res. 46 (1986) 1±7. [20] L.A. Liotta, C.N. Rao, S.H. Barsky, Tumor invasion and the extracellular matrix, Lab. Invest. 49 (1983) 636±649. [21] S.H. Barsky, G. Siegal, F. Jannotta, L.A. Liotta, Loss of basement membrane components by invasive tumors but not by their benign counterparts, Lab. Invest. 49 (1983) 140±147. [22] S.H. Barsky, C.N. Rao, G.R. Grotendorst, L.A. Liotta, Increased content of type V collagen in desmoplasia of human breast carcinoma, Am. J. Pathol. 108 (1982) 276±283. [23] I.J. Fidler, D.M. Gersten, I.R. Hart, The biology of cancer invasion and metastasis, Adv. Cancer Res. 28 (1978) 149±157. [24] E. Kawahara, Y. Okada, I. Nakanishi, K. Iwata, S. Kojima, S. Kumagai, The expression of invasive behavior of differen-
[25] [26] [27] [28]
[29]
109
tiated squamous carcinoma cell line evaluated by an in vitro invasion model, Jpn. J. Cancer Res. 84 (1993) 409±418. M.D. Pierschbacher, E. Ruoslahti, Cell attachment activity of ®bronectin can be duplicated by small synthetic fragments of the molecule, Nature 309 (1984) 30±33. E. Ruoslahti, M.D. Pierschbacher, Arg-Gly-Asp: a versatile cell recognition signal, Cell 44 (1986) 517±518. F.G. Giancotti, E. Ruoslahti, Elevated levels of the a5b1 ®bronectin receptor suppress the transformed phenotype of Chinese hamster ovary cells, Cell 60 (1990) 849±859. M.L. Carcangiu, G. Zampi, A. Pupi, A. Castagnoli, J. Rosai, Papillary carcinoma of the thyroid. A clinicopathologic study of 241 cases treated at the University of Florence, Italy, Cancer 55 (1985) 508±528. S. Ryu, S. Jimi, Y. Eura, T. Kato, S. Takebayashi, Retention of intracellular ®bronectin expression in original and metastatic thyroid carcinoma: an immunohistochemical study, Cancer Lett. 133 (1998) 215±222.
Strong intracellular and negative peripheral expression of ®bronectin in tumor cells contribute to invasion and metastasis in papillary thyroid carcinoma Shinso Ryu a,b,*, Shiro Jimi a, Youichi Eura b, Toshihiko Kato b, Shigeo Takebayashi a a
The Second Department of Pathology, School of Medicine, Fukuoka University, 7-45-1 Nanakuma, Jonanku, Fukuoka 814-0180, Japan b Department of Otorhinolaryngology, School of Medicine, Fukuoka University, 7-45-1 Nanakuma, Jonanku, Fukuoka 814-0180, Japan Received 4 March 1999; received in revised form 20 May 1999; accepted 24 June 1999
Abstract The intracellular and stromal expression of ®bronectin (FN) in invading and central parts of papillary thyroid carcinomas with/without lymph node (LN) metastasis (54 metastasizing cases, 52 non-metastasizing cases) were examined immunohistochemically. The intracellular expression of FN in tumor cells in invading parts was stronger than that in central parts in most cases (79/106 cases, 74.5%). In invading parts, negative stromal FN was frequently found at the periphery of the tumor in cases with extracapsular soft tissue invasion (26/37 cases). Tumor cells in invading parts in metastatic cases were signi®cantly more likely to be negative for stromal FN at the periphery of the tumor than those in non-metastatic cases (P , 0:0001). The strong intracellular and negative stromal FN at the periphery of the tumor in invading parts were associated with invasion and metastasis in papillary thyroid carcinoma. These results suggest that these distinctive characteristics of FN may be useful for understanding invasion and metastasis in vivo. q 1999 Elsevier Science Ireland Ltd. All rights reserved. Keywords: Fibronectin; Thyroid carcinoma; Invasion; Metastasis; Immunohistochemistry
1. Introduction Fibronectin (FN), an extracellular matrix protein, is produced by a wide variety of cells such as ®broblasts [1], endothelial cells [2], epithelial cells [3], macrophages [4] and some carcinoma cells [5±8]. FN plays several important roles in cell adhesion, morphology, migration, oncogenic transformation, neovascularization [1,9], tumor invasion and metastasis [10]. Previous reports have demonstrated a relationship between stromal FN, tumor invasion and metastasis. * Corresponding author. Tel.: 181-92-801-1011 ext. 3285; fax: 181-92-863-8383. E-mail address: [email protected] (S. Ryu)
Tumor cells are generally characterized by decreased adhesiveness due to failure to deposit stromal FN, followed by rapid proliferation and migration [11]. Additionally, positive stromal FN detected by immunohistochemical staining is associated with a low metastatic potential [12]. To date, there have been few reports of the relationship between intracellular FN, tumor invasion and metastasis, as very few carcinomas express intracellular FN [5±8]. Our previous study [13] demonstrated that thyroid carcinoma distinctively expresses intracellular FN in vivo after malignant transformation of thyroid tissue. In this study we immunohistochemically examined the stromal and intracellular FN distribution in invading and central parts of thyroid carcinomas, and in
0304-3835/99/$ - see front matter q 1999 Elsevier Science Ireland Ltd. All rights reserved. PII: S 0304-383 5(99)00259-1
104
S. Ryu et al. / Cancer Letters 146 (1999) 103±109
both peripheral and central parts of metastasizing LNs to determine the interaction between FN, invasion and metastasis. 2. Materials and methods 2.1. Patients and samples The specimens used in this study were obtained from surgical samples taken from March 1985 to July 1996 at Fukuoka University Hospital, Fukuoka, Japan. The samples were from 106 patients (19 males and 87 females), mean age 49 years (range 33±77 years). All the specimens were papillary thyroid carcinoma (well-differentiated 98 cases, poorly-differentiated, 8 cases) of which 54 cases had metastasis to LNs alone. The remaining 52 cases had no metastasis to any other organs, including LNs. The clinical stage in all patients was determined according to the International Union Against Cancer (IUCC) [14], and histological diagnosis was made according to the classi®cation established by the World Health Organization [15]. All samples were ®xed in 10% buffered formalin and embedded in paraf®n. 2.2. Immunohistochemical staining for FN Deparaf®nized sections (4 mm thick) were washed in TBS (pH 7.4) and treated with protease (DAKO, Denmark). Immunohistochemical staining was performed by the APAAP method. In brief, the primary antibody against human plasma FN (DAKO) was diluted 1:600 and incubated overnight at 48C. After washing in TBS, the sections were incubated with alkaline phosphatase-conjugated second antibody against rabbit immunoglobulin (DAKO) for FN for 60 min at room temperature. The sections were then incubated after a second wash with alkaline phosphatase-conjugated third antibody against alkaline phosphatase (DAKO) for FN, for 60 min at room temperature. Vehicle alone or an identical concentration of nonimmune animal IgG was used as a negative control. To check the speci®city of the ®rst antibody, an excess amount of antigen, FN (Sigma, MO), was mixed with the ®rst-antibody solution and incubated for 1 h at 378C. This solution was used as a ®rst-antibody solution for immunohistochemistry. Fibroblasts in human skin tissue (which
produce abundant FN under normal physiological conditions in vivo and vitro) were used to test antibody immunoreactivity as a positive control. The sections were then colored by 0.01% new fuchsin (Merck, Germany), 0.01% NaNO2, 10 mg naphthol AS-BI phosphate (Sigma), 0.1 ml N,Ndimethyl formamide (Wako, Osaka) and 10 mg levamisol (Sigma) in 40 ml of 0.2 M Tris±HCl buffer (pH 8.2). Specimens were observed under a microscope. All the stained material in specimens treated with the antibody contained FN. 2.3. Examination of parts and evaluation of staining results The thyroid carcinoma in each case was divided into two parts, an invading part and a central part. Since multicentric tumor nests are formed in papillary thyroid carcinoma, the following de®nition was emphasized: the invading part was de®ned as the area where carcinoma cells invaded thyroid tissue and/or other tissue, such as muscular tissue, fatty tissue, blood vessel or tumor capsule. The central part was de®ned as the center region of the tumor nests. In metastasizing LN, the tumor was also divided into two parts, a peripheral part and a central part. The peripheral part was de®ned as the region where carcinoma cells contacted or were located near lymphoid tissue. The central part was de®ned as the center region of metastasizing tumor nests. Immunohistological evaluation was performed using the following criteria to classify tissue as positive: (1) in carcinoma cells/normal cells adjacent to the carcinoma, when one or more carcinoma cells/ normal cells were intracellulary positive and (2) in stroma, when the stroma completely surrounded carcinoma cells/normal cells adjacent to the carcinoma. FN accumulation in tissue was evaluated semiquantitatively using the following parameters: intracellular FN in normal or carcinoma cells, negative (no cells), focal positive (from one cell up to 74% of the total cells), diffuse positive (75 % or more) and stromal FN (the periphery of epithelial cells): negative (lack of stromal FN), focal positive (weakly uniform stromal FN) and diffuse positive (strongly uniform stromal FN). A statistical analysis was carried out using Student's t-test and x 2 test for independence.
S. Ryu et al. / Cancer Letters 146 (1999) 103±109
3. Results 3.1. Immunohistochemical distribution of FN in carcinomas and metastasizing LNs The invading parts of thyroid carcinomas were diffusely positive for intracellular FN in most cases (85/106 cases, 80.2%) while approximately half of the cases were negative for stromal FN (50/106 cases, 47.2%) (Table 1). Thirty-seven cases had extracapsular soft tissue invasion, including 2 cases that also had capsular and blood vascular invasion. The remaining 69 cases had thyroid tissue invasion. Negative stromal FN at the periphery of the tumor in extracapsular soft tissue invasion was signi®cantly more common than positive stromal FN (26 negative cases versus 11 positive cases, P , 0:05). However, the more frequent occurance of negative stromal FN at the tumor periphery was not found in intra-thyroid tissue invasion. The central parts of thyroid carcinomas were focally positive for intracellular FN and stromal FN in almost all cases (90/106 cases, 84.9% and 84/106 cases, 79.2%) (Table 1). A pattern in which the invading part of the carcinoma was diffusely positive for intracellular FN and negative for stromal FN at the periphery of the tumor (Fig. 1A) was found in 47/106 cases (44.3%). A second pattern in which the invading part of the carcinoma was diffusely positive for intracellular FN and focally positive for stromal FN at the periphery of the tumor (Fig. 1B) was found in 31/106 cases (29.2%). A third pattern was found in which the central part of the carcinoma was focally positive for both intracellular FN and stromal FN (Fig. 1C) in 78/106 cases (73.4%). Thyroid tissue was negative for intracellular FN in all cases and focally positive for stromal FN in almost all cases (95/106 cases, 89.6 %). The distribution of FN
105
in metastasizing LNs was similar to that of the original carcinomas (Table 2). 3.2. Comparison of the FN distribution in metastatic and non-metastatic cases With regard to the distribution of FN in thyroid carcinomas (Table 1), the intracellular expression of FN in most cases (79/106, 74.5%) was stronger in invading parts was stronger than in central parts. The invading parts were negative for stromal FN at the periphery of the tumor. However, this pattern was not associated with metastasis (P 0:8711) (Table 3). Tumor cells in invading parts of metastatic cases were signi®cantly more likely to be negative for stromal FN at the periphery of the tumor than those of non-metastatic cases (P , 0:0001) (Table 3). Other parameters such as age, sex and differentiation were also examined but found to have no signi®cant relationship to metastasis (Table 4). 4. Discussion Our data indicated that in thyroid carcinoma, negative stromal FN at the periphery of tumor cells in invading parts was signi®cantly associated with metastasis, and that the intracellular expression of FN in invading parts was stronger than that in central parts but not associated with metastasis (Table 3). Subsequently, negative stromal FN at the periphery of the tumor was frequently found in extracapsular soft tissue invasion. To our knowledge, this is the ®rst report of a relationship between FN distribution and invasion or metastasis of thyroid carcinoma. Previous studies [12,16] have demonstrated that negative staining for stromal FN has signi®cant association
Table 1 The immunohistochemical distribution of FN in invading and central parts of papillary thyroid carcinomas Cases/total cases (%) Invading part
Positive Diffusely Focally Negative
Central part
Tumor cell
Stroma
Tumor cell
Stroma
85/106 (80.2) 21/106 (19.8) 0
16/106 (15.1) 40/106 (37.7) 50/106 (47.2)
6/106 (5.7) 90/106 (84.9) 10/106 (9.4)
20/106 (18.9) 84/106 (79.2) 2/106 (1.9)
106
S. Ryu et al. / Cancer Letters 146 (1999) 103±109
with high metastatic potential in invasive breast carcinoma. Although our materials are different (thyroid carcinoma versus breast carcinoma) the results are very similar. Previous studies [6,7,17,18] have demonstrated that FN is expressed in tumor cells of sarcoma and in some carcinomas, such as hepatocellular carcinoma and breast carcinoma. However it appears there was no comparison of a tumor's invading and central parts. Our previous study [13] indicated that thyroid carcinoma distinctively expresses intracellular FN in vivo in comparison with other carcinomas (including hepatocellular carcinoma, transitional cell carcinoma and gastric adenocarcinoma). In this study, we compared intracellular FN in invading and central tumor parts and discovered that intracellular expression of FN was stronger in the invading parts. It has been reported that malignant transformation in vitro is often accompanied by diminished stromal FN synthesis [5], the point at which cells acquire invasive or metastatic characteristics [11]. This phenomenon is consistent with our result that negative stromal FN was found at the tumors periphery in invading parts, especially in extracapsular soft tissue invasion. This correlated well with metastasis. However, tumor invasion and metastasis are highly complex multistage processes. The interaction of tumor cells with matrix (such as FN, collagen and laminin) also plays a key role in tumor invasion and metastasis [19,20] including not only diminished matrix synthesis in tumors but also degradation of matrix in invading regions [21] and increased production of matrix by host cells in response to the presence of a tumor [22]. However, the histological heterogeneosity of tumors is great enough such that these phenomena may take place separately or simultaFig. 1. FN localization in cases of (A,C) papillary thyroid carcinoma with metastasis (65 year old, female), (B) papillary thyroid carcinoma without metastasis (52 year old, female). (A) In an invading part of the carcinoma, tumor cells were diffusely positive for intracellular FN (right), but negative for stromal FN at the periphery of the tumor. Thyroid tissue was negative for intracellular and stromal FN (left). (B) In an invading part of the carcinoma, tumor cells were diffusely positive for intracellular FN (arrow), and focally positive for stromal FN at the periphery of the tumor. Thyroid tissue was negative for intracellular and stromal FN. (C) In a central part of the carcinoma, tumor cells were focally positive for intracellular FN, and stroma was focally positive for stromal FN. (original magni®cation of all sections, £ 200).
S. Ryu et al. / Cancer Letters 146 (1999) 103±109
107
Table 2 The immunohistochemical distribution of FN in metastasizing LNs of papillary thyroid carcinomas Cases/total cases (%) Peri-tumor part
Positive Diffusely Focally Negative
Central part
Tumor cell
Stroma
Tumor cell
Stroma
45/54 (83.3) 9/54 (16.7) 0
3/54 (5.6) 25/54 (46.3) 26/54 (48.1)
1/54 (1.9) 51/54 (94.4) 2/54 (3.7)
2/54 (3.7) 38/54 (70.4) 14/54 (25.9)
neously in different regions of the same tumors [23]. These phenomena may be explained by negative stromal FN at the tumor's periphery in extracapsular soft tissue invasion, but not in intra-thyroid tissue invasion. Our data and those of previous studies [12,16] may have resulted from these complicated interactions. A previous in vitro study [24] demonstrated that invasion by squamous cell carcinoma is highly correlated with cell-matrix adhesion by FN. It is well known that RGD (Arg-Gly-Asp), a sequence in the middle of FN, is recognized by the cell surface receptor of integrins, such as the FN receptor (a 5b 1 integrin). This results in the induction of cell attachment and adhesion [25,26]. Tumor cells that express a high level of a5b1 integrin adhere ®rmly to FN matrix and consequently exhibit less migratory activity [27]. In this study, almost all cases (104/106 cases, 98.1%) were positive for stromal FN in the central parts of thyroid tumors, an area where migratory activity should have been less than the invading parts. In
general, tumor cells, as compared to normal cells, diminish their own extracellular matrix proteins which causes greater proliferative and migratory activity [11]. We found that relatively few thyroid tumors were positive for stromal FN in invading parts, (56/106 cases, 52.8 %) and that cases with negative expression of stromal FN at the periphery of the tumor were signi®cantly associated with metastasis (Table 3). This is consistent with previous studies [12,16]. Almost all cases in this study exhibited positive expression of intracellular FN in thyroid tumors in invading and central parts (Table 1). However, there was a large difference in the rate of positive expression of stromal FN in the two parts (52.8% for invading parts versus 98.1% for central parts). One possible explanation for this ®nding is that thyroid carcinoma cells of invading parts secrete less extracellular stromal FN than central parts of tumor nests. Another alternative is that tumor cells in invading parts express more intracellular FN than those in central parts.
Table 3 Comparison of FN distribution in metastatic and non-metastatic cases of papillary thyroid carcinoma a Cases/total cases (%) Total
Metastasis
Non-metastasis
Intracellular FN Invading part . central part Invading part central part
79/106 (74.5) 27/106 (25.5)
40/54 (74.1)* 14/54 (25.9)*
39/52 (75.0)* 13/52 (25.0)*
Stromal FN in invading part Positive Negative
56/106 (52.8) 50/106 (47.2)
11/54 (20.4)** 43/54 (79.6)**
45/52 (86.5)** 7/52 (13.5)**
a
*P 0:8711, **P , 0:0001 by chi-square test for independence.
108
S. Ryu et al. / Cancer Letters 146 (1999) 103±109
Table 4 Comparison of clinical and histological ®ndings of thyroid carcinoma in metastatic and non-metastatic cases
Cases Age (years) M/F Differentiation Poorly/well a b
the Society for the Promotion of International OtoRhino-Laryngology (SPIO).
Metastasis
Non-metastasis
P value
References
54 48.8 ^ 14.8 12/42
52 50.8 ^ 15.9 7/45
0.532 a 0.237 b
[1] K.M. Yamada, J.A. Weston, Isolation of a major cell surface glycoprotein from ®broblasts, Proc. Natl. Acad. Sci. USA 71 (1974) 3492±3496. [2] E.A. Jaffe, D.F. Mosher, Synthesis of ®bronectin by cultured human endothelial cells, J. Exp. Med. 147 (1979) 1779±1791. [3] L.B. Chen, N. Maitland, P.H. Gallimore, J.K. McDougall, Detection of the large external transformation-sensitive protein on some epithelial cells, Exp. Cell Res. 106 (1977) 39±46. [4] B. Villiger, D.G. Kelly, C. Kuhn, J.A. McDonald, Human alveolar macrophage ®bronectin: synthesis and ultrastructural localisation, Clin. Res 28 (1980) 745A. [5] K. Olden, K.M. Yamada, Mechanism of the decrease in the major cell surface protein of chick embryo ®broblasts after transformation, Cell 11 (1997) 957±969. [6] H.S. Smith, J.L. Riggs, N.W. Mosesson, Production of ®bronectin by human epithelial cells in culture, Cancer. Res. 39 (1979) 4138±4144. [7] E. Pearlstein, R.O. Hynes, L.M. Franks, V.J. Hemmings, Surface proteins and ®brinolytic activity of cultured mammalian cell, Cancer Res. 36 (1976) 1475±1479. [8] A. Vaheri, M. Kurkinen, V.P. Lehto, E. Linder, R. Timpl, Codistribution of pericellular matrix protein in cultured ®broblasts and loss in transformation: ®bronectin and procollagen, Proc. Natl. Acad. Sci. USA 75 (1978) 4944±4948. [9] R.O. Hynes, K.M. Yamada, Fibronectin: multifunctional modular glycoproteins. , J. Cell Biol. 95 (1982) 369±377. [10] A.J. D'ardenne, J.O'D. McGee, Fibronectin in disease, J. Pathol. 12 (1984) 235±351. [11] H. Mohri, Fibronectin and integrin interactions, J. Invest. Med. 44 (1996) 429±441. [12] L. Christensen, M. Nielsen, J. Andersen, I. Clemmensen, Stromal ®bronectin staining pattern and metastasizing ability of human breast carcinoma, Cancer Res. 48 (1988) 6227±6233. [13] S. Ryu, S. Jimi, S. Takebayashi, Thyroid carcinoma distinctively expresses intracellular ®bronectin in vivo, Cancer Lett. 121 (1997) 189±193. [14] P. Hermanek, L.H. Sobin, Classi®cation of Malignant Tumours, in: P, Hermanek, L.H. Sobin (Eds.), fourth ed., International Union Against Cancer. Springer: New York, pp. (1987) 33±35. [15] C. Hedinger, E.D. Williams, L.H. Sobin, Histological typing of thyroid tumours, second ed, World Health Organization, Geneva, 1988. [16] H. Takei, Y. Iino, J. Horiguchi, T. Yokoe, Immunohistochemical ®bronectin staining pattern and prognosis in invasive breast carcinoma, Oncology 52 (1995) 106±111. [17] S. Korach, M.F. Poupon, J.A. Du Villard, M. Becker, Differential adhesiveness of rhabdomyosarcoma-derived cloned metastatic cell lines to vascular endothelial monolayers, Cancer Res. 46 (1986) 3624±3629.
5/49
3/49
0.531 b
Student's t-test. x 2 test for independence.
Our results also showed that negative stromal FN at the tumor's periphery in the invading part of a carcinoma was signi®cantly more common in metastatic cases than other parameters such as age, sex and differentiation (Tables 3 and 4). Previous studies [12,16] have reported that negative staining for stromal FN is superior to such parameters as clinical stage or histological type or grade as an index of metastasis and prognosis of breast carcinoma. In this study, we could not examine the prognosis using negative staining for stromal FN since almost all cases (data not shown) had very good prognosis (which is common in differentiated thyroid carcinoma) [28]. In this study we found the distribution of FN in metastasizing LNs was very similar to that which occurred in the original thyroid carcinomas (Table 2). In our previous study [29], we demonstrated that intracellular FN is distinctive to thyroid carcinomas, even when they metastasize to other organs, such as the lung, kidney and LN. In the present study, we found that intracellular and stromal FN retained their distinctive distribution in metastasizing LNs. In conclusion, the intracellular expression of FN in thyroid carcinoma was stronger in invading parts than in central parts and that negative stromal FN at the tumor's periphery in invading parts was signi®cantly correlated with extracapsular soft tissue invasion and metastasis to LNs. These results may be useful for understanding invasion and metastasis in vivo.
Acknowledgements This work was supported by a research grant from
S. Ryu et al. / Cancer Letters 146 (1999) 103±109 [18] S. Matsui, T. Takahashi, Y. Oyanagi, S. Takahashi, S. Boku, K. Takahashi, Expression, localization and alternative splicing pattern of ®bronectin messenger RNA in ®brotic human liver and hepatocellular carcinoma, J. Hepatol. 27 (1997) 843±853. [19] L.A. Liotta, Tumor invasion and metastasis. Role of extracellular matrix, Cancer Res. 46 (1986) 1±7. [20] L.A. Liotta, C.N. Rao, S.H. Barsky, Tumor invasion and the extracellular matrix, Lab. Invest. 49 (1983) 636±649. [21] S.H. Barsky, G. Siegal, F. Jannotta, L.A. Liotta, Loss of basement membrane components by invasive tumors but not by their benign counterparts, Lab. Invest. 49 (1983) 140±147. [22] S.H. Barsky, C.N. Rao, G.R. Grotendorst, L.A. Liotta, Increased content of type V collagen in desmoplasia of human breast carcinoma, Am. J. Pathol. 108 (1982) 276±283. [23] I.J. Fidler, D.M. Gersten, I.R. Hart, The biology of cancer invasion and metastasis, Adv. Cancer Res. 28 (1978) 149±157. [24] E. Kawahara, Y. Okada, I. Nakanishi, K. Iwata, S. Kojima, S. Kumagai, The expression of invasive behavior of differen-
[25] [26] [27] [28]
[29]
109
tiated squamous carcinoma cell line evaluated by an in vitro invasion model, Jpn. J. Cancer Res. 84 (1993) 409±418. M.D. Pierschbacher, E. Ruoslahti, Cell attachment activity of ®bronectin can be duplicated by small synthetic fragments of the molecule, Nature 309 (1984) 30±33. E. Ruoslahti, M.D. Pierschbacher, Arg-Gly-Asp: a versatile cell recognition signal, Cell 44 (1986) 517±518. F.G. Giancotti, E. Ruoslahti, Elevated levels of the a5b1 ®bronectin receptor suppress the transformed phenotype of Chinese hamster ovary cells, Cell 60 (1990) 849±859. M.L. Carcangiu, G. Zampi, A. Pupi, A. Castagnoli, J. Rosai, Papillary carcinoma of the thyroid. A clinicopathologic study of 241 cases treated at the University of Florence, Italy, Cancer 55 (1985) 508±528. S. Ryu, S. Jimi, Y. Eura, T. Kato, S. Takebayashi, Retention of intracellular ®bronectin expression in original and metastatic thyroid carcinoma: an immunohistochemical study, Cancer Lett. 133 (1998) 215±222.