© 2005 Foundation for Promotion of Cancer Research
Expression of Tumor Suppressor and Tumor-related Proteins in Differentiated Carcinoma, Undifferentiated Carcinoma with Tubular Component and Pure Undifferentiated Carcinoma of the Stomach
1 Department of Pathology and 2 Department of Gastroenterology, Yamagata University School of Medicine, Yamagata, 3 Division of Pathology, Tsubame Rosai Hospital, Tsubame, Niigata, Japan and 4 Molecular Therapeutics Program, National Cancer Institute, National Institutes of Health, MD, USA
For reprints and all correspondence: Gen Tamura, 2-2-2 Iida-nishi, Yamagata 990-9585, Japan. E-mail: gtamura{at}med.id.yamagata-u.ac.jp
Received July 4, 2005; accepted August 21, 2005
 |
Abstract |
|---|
 TOP Abstract INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
Background: The recent development of tissue microarray (TMA) technology allows high-throughput protein expression profiling of cancer tissues by immunohistochemistry. We attempted to clarify the derivation of undifferentiated-type gastric carcinoma with tubular component by using TMA.
Methods: We constructed a TMA system composed of six paraffin blocks in which 274 samples of formalin-fixed gastric carcinoma tissue from 274 patients were embedded. Using this system, we performed immunohistochemical stains for five tumor suppressor and tumor-related proteins, i.e. p53, p16, hMLH1, c-erbB-2 and carcinoembryonic antigen (CEA). The 274 gastric carcinomas were histopathologically divided into the following three groups according to the degree of differentiation: differentiated-type (D-type), undifferentiated-type with tubular component (UT-type) and pure undifferentiated-type (UP-type). Immunohistochemical results were then compared with histological types.
Results: The percentages of abnormal expression of each protein in D-type, UT-type and UP-type carcinomas were as follows: 27% (38/143), 17% (17/98) and 15% (5/33) for p53; 27% (39/143), 19% (19/98) and 18% (6/33) for p16; 38% (54/143), 44% (43/98) and 24% (8/33) for hMLH1; 15% (22/143), 5% (5/98) and 0% (0/33) for c-erbB-2; and 22% (31/143), 35% (34/98) and 70% (23/33) for CEA. UP-type carcinomas exhibited the lowest frequencies of abnormal expression for p53, p16, hMLH1 and c-erbB-2, but the highest frequencies for CEA. UT-type carcinomas generally showed intermediate frequencies between those of D-type and UP-type carcinomas. Differences between D-type and UP-type for c-erbB-2 (P < 0.05) and CEA (P < 0.001) were significant, as were differences between D-type and UT-type for c-erbB-2 (P < 0.05) and CEA (P < 0.05), and differences between UT-type and UP-type for hMLH1 (P < 0.05) and CEA (P < 0.001).
Conclusions: These findings reveal that gastric carcinomas have distinct expression profiles for tumor suppressor and tumor-related proteins depending on histological types, and support the hypothesis that UT-type carcinomas are derived not only from D-type but also from UP-type carcinomas. We also found significant differences between abnormal protein expression and other clinicopathological parameters such as gender, age and status of tumor and nodes.
Key Words: gastric cancer • tissue microarray • immunohistochemistry
|
INTRODUCTION |
|---|
|
TOPAbstractINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
Recent progress in molecular biology has shown that the phenotypic diversity of tumors is associated with corresponding diversity in gene expression (1–3). Techniques involving cDNA microarrays have been applied to study gene expression patterns in different tumor types, providing new insights into development and classification of these cancers. Recently, Kononen et al. (4) developed tissue microarray (TMA) techniques that simultaneously produce comprehensive protein expression profiles for numerous tumors. TMA appears to be particularly useful for immunohistochemical characterization of tumors (5–7). This technology has been used extensively and is the subject of multiple reviews (6–10).
Gastric cancers are classified either as differentiated-type (D-type) carcinoma, which forms tubular or papillary structures roughly corresponding with intestinal type, or undifferentiated-type (U-type) carcinoma, in which such structures are inconspicuous (roughly corresponding with diffuse type) (11,12). D-type carcinomas are thought to evolve from intestinal metaplastic mucosa and U-type tumors from native gastric mucosa (13). Although a significant proportion of D-type carcinomas are thought to progress to U-type (13), there have been few studies which focused on this transformation. For the present study, we subdivided undifferentiated carcinomas into U-type carcinoma with tubular component (UT-type) and pure undifferentiated carcinoma (UP-type) based upon the presence or absence of tubular components. Because different genetic pathways have been proposed for D-type and U-type carcinomas (13,14), they should exhibit different protein expression profiles. Five tumor suppressor and tumor-related proteins, p53, p16, hMLH1, c-erbB-2 and carcinoembryonic antigen (CEA), which frequently show abnormal expression in gastric cancer were selected for this study. We then used TMA to create expression profiles for those proteins in gastric carcinomas according to the histological types.
|
MATERIALS AND METHODS |
|---|
|
TOPAbstractINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
CLINICAL SAMPLES
A total of 274 surgically resected gastric cancers were examined. The patients were 188 males and 86 females (mean age, 68 years; range, 32–88 years). The tumors consisted of 64 early (depth of invasion limited to the mucosa or submucosa) and 210 advanced (depth of invasion reached the muscularis propria) carcinomas. Nodal metastasis was present in 144 patients, and distant metastasis in 7 patients. All specimens were fixed in 10% formalin and embedded in paraffin. The tumors were classified according to tumor-node-metastasis (TNM) staging (8) at the Department of Pathology, Yamagata University School of Medicine, between 1996 and 2001. Histologically, gastric cancers were classified as either of D-type, which forms tubular or papillary structures, or U-type carcinoma, in which such structures are inconspicuous. U-type carcinomas were subdivided into U-type carcinoma with tubular component (UT-type) and pure undifferentiated carcinoma (UP-type) which included poorly differentiated adenocarcinoma without tubular component and signet-ring cell carcinoma.
TMA CONSTRUCTIONS
A map of receiver block was prepared with coordinates for each sample to correctly identify the tumor. Under a microscope, areas of interest that were non-necrotic and rich in tumoral glands were marked with an indelible pen on the whole hematoxylin and eosin stained (H&E) section of each donor block. TMAs were constructed using a manual tissue arrayer (KIN-1; Azumaya, Tokyo, Japan). The system consists of thin-walled stainless-steel needles with an inner diameter of 
2 mm and a stiletto for transferring and emptying needle contents. Specimens are retrieved from selected regions of donor tissue and precisely arrayed in a new recipient block. One core per donor block for each case sampling 2 mm in diameter and using 0.8 mm edge-to-edge spacing was arrayed in six recipient blocks. TMA blocks were incubated at 60°C for 30 min.
IMMUNOHISTOCHEMISTRY
Serial 4 µm thick tissue sections were cut from array blocks, with initial sections H&E stained to verify histology. Immunostaining for p53 (Pab1801; Novocastra, Newcastle, UK) (1 : 40 dilution), p16 (F-12; Santa Cruz Biotechnology, Santa Cruz, CA, USA) (1 : 100 dilution), hMLH1 (G168-728; PharMingen, San Diego, CA, USA) (1 : 50 dilution), c-erbB-2 (DakoCytomation, Glostrup, Denmark) (1 : 100 dilution) and CEA (1B2; IBL, Tokyo, Japan) (1 : 100 dilution) was performed using a standard labeled streptavidin–biotin system (Nichirei, Tokyo, Japan). Positivity was >50% nuclear staining for p53, p16 and hMLH1, membranous staining for c-erbB-2 and cytoplasmic staining for CEA. Immunohistochemical reactivity was independently evaluated by two investigators (T.H., G.T.), with consensus reached after discussion.
STATISTICAL ANALYSIS
Correlation between expression of each tumor suppressor and tumor-related protein and tumor differentiation, as well as correlation with other clinicopathological parameters, was compared by Fisher's exact probability test and chi-square for independence test. A P-value <0.05 was considered to be statistically significant.
|
RESULTS |
|---|
|
TOPAbstractINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
EXPRESSION OF TUMOR SUPPRESSOR AND TUMOR-RELATED PROTEINS IN DIFFERENTIATED AND UNDIFFERENTIATED GASTRIC CANCERS (TABLE 1)
UP-type carcinomas exhibited the lowest frequencies of abnormal expression for p53, p16, hMLH1 and c-erbB-2, but the highest frequency for CEA. UT-type carcinomas generally showed intermediate frequencies between those of D-type and UP-type. Differences between D-type and UP-type for c-erbB-2 (P < 0.05) and CEA (P < 0.001), differences between D-type and UT-type for c-erbB-2 (P < 0.05) and CEA (P < 0.05), and differences between UT-type and UP-type for hMLH1 (P < 0.05) and CEA (P < 0.001) were significant. An example of immunohistochemistry is shown in Fig. 1.
|
|
CORRELATION BETWEEN TUMOR SUPPRESSOR AND TUMOR-RELATED PROTEIN EXPRESSION AND OTHER CLINICOPATHOLOGICAL PARAMETERS (TABLE 2)
Overexpression of p53 was observed more frequently in older patients (P < 0.05), with early tumor status (P < 0.05) and with negative nodes (P < 0.05). Loss or reduction of p16 was associated with negative nodes (P < 0.01). Expression of hMLH1 was more frequently reduced or lost in older patients (P < 0.01) and with advanced tumor status (P < 0.05). Expression of c-erbB-2 was more common with early tumor status (P < 0.05). Cytoplasmic CEA expression was frequent in tumors from female (P < 0.05) and younger (P < 0.05) patients, and with positive nodes (P < 0.05).
|
|
DISCUSSION |
|---|
|
TOPAbstractINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
TMA allows simultaneous analysis of large numbers of specimens, resulting in high-throughput data acquisition. Because all tissue specimens analysed are arrayed on one identical TMA, antigen retrieval, reagent concentrations, incubation times with primary antibodies, temperatures and wash conditions are identical for each core (15,16), resulting in an unprecedented level of standardization over and above that available with standard histopathological techniques. In addition, because small quantities of reagent and fewer personnel are required, this method is extremely efficient and cost-effective. Histopathological benefits include minimal destruction of original tissue blocks, which are often considered vital resources, and the presence of internal positive and negative controls on the TMA. Because such small specimens are sampled, many rare cell lines can be assessed with ease (17). One potential limitation of TMA is that very small cores may not be representative of whole tumors, particularly in heterogeneous cancers such as prostate adenocarcinoma. However, excellent correlation has been reported between data obtained from TMA and conventional blocks in a variety of tumor types such as breast, prostate, bladder and human fibroblastic tumors (15). Furthermore, in the most extensive study investigating validation in prostate cancer, Rubin et al. (18) found that only 3–4 optimal cores of 0.6 mm diameter were necessary to predict outcome after radical prostatectomy for local prostate cancer. Similarly, other groups have found that sampling with optimal cores was sufficient to accurately detect clinicopathological correlations (15,19). Interestingly, these studies have shown that increasing the number of cores to compensate for heterogeneity and inevitable losses confers only a slightly higher rate of validity while producing the significant disadvantages of additional labor in arraying a larger number of arrays and lack of efficiency in processing tissues. In the present study, we used a single core after microscopic identification of representative histopathology with reference to a serial H&E stained section for each patient. Our large core size, 2 mm in diameter, allowed easy recognition of cell types with positive and negative immunoreactivity.
In the present study, we have demonstrated that UP-type carcinomas exhibited the lowest frequencies of abnormal expression for p53, p16, hMLH1 and c-erbB-2, but the highest frequencies for CEA. UT-type carcinomas generally showed intermediate frequencies between those of D-type and UP-type carcinomas. Differences between D-type and UT-type for c-erbB-2 and CEA were statistically significant, as were differences between UT-type and UP-type for hMLH1 and CEA. Thus, UT-type carcinoma have aspects of both D-type and UP-type at protein level suggesting that UT-type carcinomas might be derived not only from D-type but also from UP-type carcinomas. Recently, Peng et al. (20) have shown that poorly differentiated gastric carcinoma with a tubular component, which corresponds to UT-type in the present study, usually develops from tubular adenocarcinomas (D-type) and sometimes from signet-ring cell carcinoma (UP-type) by comparative genomic hybridization (CGH) analyses.
p53 expression was most frequent in D-type (26.6%) and infrequent (15.2%) in UP-type tumors, although this difference was not significant. The relatively low rate of p53 immunoreactivity (21%, 60/274) might be owing to the cut-off value of 50%; however, comparison of single-strand conformation polymorphism/sequencing and immunohistochemistry of p53 alterations revealed the propriety of using a cut point of 
50%. Infrequent p53 mutation (3.8%, 1/26) in undifferentiated gastric carcinoma has been noted (21). In addition, p53 overexpression was seen more frequently in tumors with early tumor status (Tis and T1) than with advanced status (T2a–T4), a finding consistent with a previous report that found p53 mutation was more frequent in early carcinomas than in advanced ones (22). The significantly lower rate of immunoreactivity against p53 in younger patients was also consistent with a previous report on p53 mutations (23). Correlation between p53 expression and nodal status is controversial (23,24). We found less frequent p53 expression in node-positive cases, probably because early carcinomas commonly expressed p53.
Reduced expression of p16 owing to promoter hypermethylation results in higher cyclin D1-dependent protein kinase activity and aberrant phosphorylation of Rb that play roles in the pathogenesis of gastric cancer (25–27). No significant differences between p16 expression and histological differentiation were observed, probably because p16 was commonly affected by promoter hypermethylation in both differentiated and undifferentiated gastric cancers (28). In the present study, reduced p16 protein expression was more frequent in node-negative tumors, although no significant association was noted in previous studies (25,26,29).
Since inactivation of hMLH1 in association with DNA methylation was first reported in colorectal cancer (30), similar epigenetic alterations have been described in gastric cancer (13,31). DNA methylation of hMLH1 promoter region CpG islands is tightly associated with loss of hMLH1 expression in gastric cancers that exhibit microsatellite instability (MSI) (31,32). Loss of hMLH1 protein was less frequent in UP-type carcinoma, which is in agreement with the finding that MSI-H (high frequency MSI) is absent in this type of carcinoma (21) and loss of hMLH1 expression owing to promoter hypermethylation plays a critical role in D-type carcinoma (32,33). The frequencies of loss of hMLH1 expression increased with patient age and advanced tumor stage, both of which are consistent with previous reports (28,34).
Overexpression of c-erbB-2 has been detected in 10–50% of gastric carcinomas and is correlated with a poor prognosis (35–37). The great majority of gastric carcinomas with c-erbB-2 overexpression are histologically intestinal type (35–37), which corresponds with D-type carcinoma in our study. The correlation between c-erbB-2 expression and status of tumor and nodes has been controversial (35–37), although c-erbB-2 overexpression was more frequent in early carcinomas in the present study.
CEA is one of the most useful tumor markers of adenocarcinoma (38). All signet-ring cell carcinomas are intensively positive for CEA (38). In the current study, cytoplasmic CEA expression was significantly associated with undifferentiated histology, especially UP-type, which includes signet-ring cell carcinoma. In addition, CEA expression was associated with female gender, younger age and positive nodes, confirming the significance of this protein expression as a biological marker.
Thus, gastric carcinomas show distinct expression profiles of tumor suppressor and tumor-related proteins depending on tumor differentiation. This supports the hypothesis that UT-type carcinomas are derived not only from D-type but also from UP-type carcinomas. Furthermore, abnormal protein expression has a characteristic association with clinicopathological parameters. TMA is a useful technique for rapid identification of protein expression profiles using minimal samples from archived tissues.
|
References |
|---|
|
TOPAbstractINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
1 Golub RT, Slonim KD, Tamayo P, Huard C, Gaasenbeek M, Mesirov PJ, et al. Molecular classification of cancer: class discovery and class prediction by gene expression monitoring. Science 1999;286:531–53.
2 Perou CM, Sorlie T, Elsen MB, van de Rijn M, Jeffrey SS, Rees CA, et al. Molecular portraits of human breast tumors. Nature 2000;406:747–52.[CrossRef][Medline]
3 Ramaswamy S, Tamayo P, Rifkin R, Mukherjee S, Yeang HC, Angelo M, et al. Multiclass cancer diagnosis using tumor gene expression signatures. Proc Natl Acad Sci USA 2001;98:15149–54.
4 Kononen J, Bubendorf L, Kallioniemi A, Barlund M, Schrami P, Leighton S, et al. Tissue microarray for high-throughput molecular profiling of tumor specimens. Nat Med 1998;4:844–7.[CrossRef][Web of Science][Medline]
5 Henshall S. Tissue microarrays. J Mammary Gland Biol Neoplasia 2003;8:347–58.[CrossRef][Web of Science][Medline]
6 Shergill IS, Shergill NK, Arya M, Patel HRH. Tissue microarrays: a current medical research tool. Curr Med Res Opin 2004;20:707–12.[CrossRef][Medline]
7 van de Rijn M, Gilks CB. Applications of microarrays to histopathology. Histopathology 2004;44:97–108.[CrossRef][Web of Science][Medline]
8 Sobin LH, Wittekind C. TNM Classification of Malignant Tumours. 6th edn. New York: Wiley-Liss 2002.
9 Tatematsu M, Hasegawa R, Ogawa K, Kato T, Ichinose M, Miki K, Ito N. Histogenesis of human stomach cancers based on assessment of differentiation. J Clin Gastroenterol 1992;14:S1–7.[CrossRef][Web of Science][Medline]
10 Williamson M, Naaby-Hansen S, Masters JRW. 21st century molecular biology in urology. BJU Int 2001;88:451–7.[Medline]
11 Lauren P. The two histological main types of gastric carcinoma: diffuse and so-called intestinal type carcinoma. An attempt at a histological classification. Acta Pathol Microbiol Scand 1965;64:31–49.[Web of Science][Medline]
12 Nakamura K, Sugano H, Takagi K. Carcinoma of the stomach in incipient phase: its histogenesis and histological appearances. Gann 1968;59:251–8.[Web of Science][Medline]
13 Tamura G. Promoter methylation status of tumor suppressor and tumor-related genes in neoplastic and nonneoplastic gastric epithelia. Histol Histopathol 2004;19:221–8.[Web of Science][Medline]
14 Tahara E. Genetic alterations in human gastrointestinal cancers. The application to molecular diagnosis. Cancer 1995;75:1410–7.[CrossRef][Medline]
15 Rimm DL, Camp RL, Charette LA, Olsen DA, Provost E. Amplification of tissue by construction of tissue microarrays. Exp Mol Pathol 2001;70:255–64.[CrossRef][Medline]
16 Mousses S, Kallioniemi A, Kauraniemi P, Elkahloun A, Kallioniemi OP. Clinical and functional target validation using tissue and cell microarrays. Curr Opin Chem Biol 2001;6:97–101.
17 Brandt B, Roetger A, Bidart JM, Packeisen J, Schier K, Mikesch JH, et al. Early placenta insulin-like growth factor (proEPIL) is overexpressed and secreted by c-erbB-2-positive cells with high invasion potential. Cancer Res 2002;62:1020–4.
18 Rubin MA, Dunn R, Strawderman M, Pienta KJ. Tissue microarray sampling strategy for prostate cancer biomarker analysis. Am J Surg Pathol 2002;26:313–9.
19 Camp RL, Charette LA, Rimm DL. Validation of tissue microarray technology in breast carcinoma. Lab Invest 2000;80:1943–9.[Web of Science][Medline]
20 Peng DF, Sugihara H, Mukaisho K, Ling ZQ, Hattori T. Genetic lineage of poorly differentiated gastric carcinoma with a tubular component analysed by comparative genomic hybridization. J Pathol 2004;203:884–95.[Medline]
21 Tamura G, Sato K, Akiyama S, Tsuchiya T, Endoh Y, Usuba O, et al. Molecular characterization of undifferentiated-type gastric carcinoma. Lab Invest 2001;81:593–8.[Web of Science][Medline]
22 Tamura G, Sakata K, Maesawa C, Suzuki Y, Terashima M, Satoh K, et al. Microsatellite alterations in adenoma and differentiated adenocarcinoma of the stomach. Cancer Res 1995;55:1933–6.
23 Monig SP, Eidt S, Zirbes TK, Stippel D, Baldus SE, Pichlmaier H. p53 expression in gastric cancer. Dig Dis Sci 1997;42:2463–7.[CrossRef][Web of Science][Medline]
24 Uchino S, Noguchi M, Ochiai A, Saito T, Kobayashi M, Hirohashi S. p53 mutation in gastric cancer: a genetic model for carcinogenesis is common to gastric and colorectal cancer. Int J Cancer 1993;54:759–64.[Web of Science][Medline]
25 Feakins RM, Nickols CD, Bidd H, Walton SJ. Abnormal expression of pRb, p16, and cyclin D1 in gastric adenocarcinoma and its lymph node metastases: relationship with pathological features and survival. Hum Pathol 2003;34:1276–82.[CrossRef][Medline]
26 Osawa T, Chong JM, Sudo M, Sakuma K, Uozaki H, Shibahara J, et al. Reduced expression and promoter methylation of p16 gene in Epstein–Barr virus-associated gastric carcinoma. Jpn J Cancer Res 2002;93:1195–200.
27 Shim YH, Kang GH, Ro JY. Correlation of p16 hypermethylation with p16 protein loss in sporadic gastric carcinomas. Lab Invest 2000;80:689–95.[Web of Science][Medline]
28 Waki T, Tamura G, Tsuchiya T, Sato K, Nishizuka S, Motoyama T. Promoter methylation status of E-cadherin, hMLH1, and p16 genes in nonneoplastic gastric epithelia. Am J Pathol 2002;161:399–403.
29 Tsujimoto H, Hagiwara A, Sugihara H, Hattori T, Yamagishi H. Promoter methylations of p16INK4a and p14ARF genes in early and advanced gastric cancer. Correlations of the modes of their occurrence with histologic type. Pathol Res Pract 2002;198:785–94.[Medline]
30 Herman JG, Umar A, Polyak K, Graff JR, Ahuja N, Issa JP, et al. Incidence and functional consequences of hMLH1 promoter hypermethylation in colorectal carcinoma. Proc Natl Acad Sci USA 1998;95:6870–5.
31 Fleisher AS, Esteller M, Wang S, Tamura G, Suzuki H, Yin J, et al. Hypermethylation of the hMLH1 gene promoter in human gastric cancers with microsatellite instability. Cancer Res 1999;59:1090–5.
32 Guo RJ, Arai H, Kitayama Y, Igarashi H, Hemmi H, Arai T, et al. Microsatellite instability of papillary subtype of human gastric adenocarcinoma and hMLH1 promoter hypermethylation in the surrounding mucosa. Pathol Int 2001;51:240–7.[CrossRef][Medline]
33 Endoh Y, Tamura G, Ajioka Y, Watanabe H, Motoyama T. Frequent hypermethylation of the hMLH1 gene promoter in differentiated-type tumors of the stomach with the gastric foveolar phenotype. Am J Pathol 2000;157:717–22.
34 Nakajima T, Akiyama Y, Shiraishi J, Arai T, Yanagisawa Y, Ara M, et al. Age-related hypermethylation of the hMLH1 promoter in gastric cancers. Int J Cancer 2001;94:208–11.[CrossRef][Web of Science][Medline]
35 Mizutani T, Onda M, Tokunaga A, Yamanaka N, Sugisaki Y. Relationship of c-erbB-2 protein expression and gene amplification to invasion and metastasis in human gastric cancer. Cancer 1993;72:2083–8.[CrossRef][Web of Science][Medline]
36 Polkowski W, van Sandick JW, Offerhaus GJ, ten Kate FJ, Mulder J, Obertop H, van Lanschot JJ. Prognostic value of Lauren classification and c-erbB-2 oncogene overexpression in adenocarcinoma of the esophagus and gastroesophageal junction. Ann Surg Oncol 1999;6:290–7.[CrossRef][Web of Science][Medline]
37 Uchino S, Tsuda H, Maruyama K, Kinoshita T, Sasako M, Saito T, et al. Overexpression of c-erbB-2 protein in gastric cancer. Its correlation with long-term survival of patients. Cancer 1993;72:3179–84.[CrossRef][Web of Science][Medline]
38 Wittekind C, Wachner R, Henke W, von Kleist S. Localization of CEA, HCG, lysozyme, alpha-1-antitrypsin, and alpha-1-antichymotrypsin in gastric cancer and prognosis. Virchows Arch A Pathol Anat Histopathol 1986;409:715–24.[Medline]
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||