p27Kip1 expression is reduced in pancreatic carcinoma but has limited prognostic value

p27Kip1 Expression Is Reduced in Pancreatic Carcinoma but Has Limited Prognostic Value ROGER M. FEAKINS, MB, FRCPI, MRCPATH, AND ADEEL H. GHAFFAR, BSC p27Kip1 is a cyclin-dependent kinase inhibitor whose loss in malignant tumors is associated with disease progression and an unfavorable clinical outcome. There is limited information about its expression in pancreatic carcinomas. In a previous Japanese study, p27Kip1 loss was a powerful negative prognostic factor. In the present study, we assessed the expression of p27Kip1 in resected pancreatic ductal adenocarcinomas from 46 European patients and in associated lymph node metastases from 13 patients, using a standard avidin-biotin peroxidase complex immunohistochemical method. We also analyzed the relationships among p27Kip1 expression, pathologic features, and clinical outcome. The extent of p27Kip1 expression (ie, the percentage of cells expressing p27Kip1) was lower in carcinomas than in nonneoplastic ductal epithelia. Carcinomas with <5% p27Kip1 expression were more likely to be poorly or moderately differentiated than well differentiated (P ⴝ .022). p27Kip1 expression did not correlate with patient gender, tumor maximum dimension, T classification, lymph node metastasis, or International Union Against Cancer stage. No significant difference was seen between the extent of p27Kip1 expression in lymph node metastases and their correspond-

ing primary tumors. Univariate survival analysis showed that an increased risk of death was associated with 2 established prognostic factors: tumor size >5 cm (P ⴝ .011) and incomplete surgical excision (P ⴝ .016). Trends toward worse survival for patients whose primary tumors had <4% p27Kip1 expression (P ⴝ .060) and for patients whose lymph node metastases had <5% p27Kip1 expression (P ⴝ .054) were seen, but multivariate analysis suggested that p27Kip1 expression was not independently prognostic. The findings raise the possibility that abnormalities of p27Kip1 play a role in the pathogenesis of pancreatic ductal adenocarcinoma. However, the extent of p27Kip1 expression in lymph node metastases and primary tumors is similar. Also, reduced p27Kip1 expression has limited prognostic value, at least in European patients. HUM PATHOL 34:385-390. © 2003 Elsevier Inc. All rights reserved. Key words: pancreatic adenocarcinoma, p27Kip1 expression, pathogenesis, prognosis. Abbreviations: CDK, cyclin-dependent kinase; PBS, phosphatebuffered saline; TGF-␤, transforming growth factor beta; UICC, International Union Against Cancer.

Pancreatic adenocarcinoma is a common cause of cancer-related death in Europe and North America.1 Advanced stage, the presence of lymph node metastases, poor differentiation, and incomplete excision may predict a worse outcome, although reported results are conflicting.2-8 Possible additional prognostic factors include p53,9 cyclin D1,10 matrix metalloproteinase 1,11 vascular endothelial growth factor,12 transforming growth factor beta (TGF-␤),13 and DNA ploidy.6 Neoplastic cells progress rapidly through the cell cycle, often ignoring the mechanisms that control cell division. Cyclin-dependent kinases (CDKs) play an important role in cell cycle control. They form cyclin– CDK complexes that facilitate entry into the S phase.14-16 CDKs are bound and inactivated by various proteins, including p27Kip1.14,15 p27Kip1 is an important inhibitor of the cyclin D–CDK4 and cyclin E–CDK2 complexes14,15 and thus reduces cell proliferation. It is up-regulated by cytokines, including TGF-␤.15,17 p27Kip1 is a potential tumor suppressor, and its levels are reduced in many tumor types.18,19 However, p27Kip1 mutations have not been demonstrated in tu-

mors, and reduced expression may possibly reflect enhanced proteasomal degradation.20 p27Kip1 also may have other roles, including promotion of apoptosis, regulation of drug resistance in solid tumors, and mediation of cell differentiation.18 Low p27Kip1 expression may be a poor prognostic factor for various carcinomas, including those of the large intestine, stomach, gall bladder, esophagus, breast, and prostate,19-24 although some subsequent studies have not confirmed its prognostic value.25-27 Studies of p27Kip1 loss in pancreatic adenocarcinoma are, to our knowledge, confined to a Japanese report that demonstrated independent negative prognostic value29 and a Chinese report that did not include survival analysis.30 Furthermore, there appear to be no published reports of p27Kip1 expression in lymph node metastases from pancreatic carcinomas.

From the Department of Histopathology and Morbid Anatomy, Barts and the London, Queen Mary’s School of Medicine and Dentistry, London, UK. Accepted for publication November 19, 2002. Poster presentation at the Pathological Society of Great Britain and Ireland Winter Meeting, London, UK, January 18-21, 2000. Address correspondence and reprint requests to Roger M. Feakins, Department of Histopathology and Morbid Anatomy, Royal London Hospital, Whitechapel, London E1 1BB, UK. © 2003 Elsevier Inc. All rights reserved. 0046-8177/03/3404-0015$30.00/0 doi:10.1053/hupa.2003.23

MATERIALS AND METHODS Patients and Tissue Samples Hematoxylin and eosin (H&E)–stained slides and paraffin wax blocks of pancreatic carcinomas resected from 46 patients (26 male, 20 female; median age, 61 years) between 1965 and 1999 were retrieved from the histopathology archives. A histopathologist (R.M.F.) reviewed the pathologic features of each tumor and selected those with the appearances of primary pancreatic ductal adenocarcinoma. The grade was recorded as either well, moderately, or poorly differentiated. Each tumor was assessed using the TNM classification system and allocated accordingly to an International Union Against Cancer (UICC) stage31 after review of histologic material, pathology reports, and clinical notes. Slides and blocks of lymph node metastases were also re-

385

HUMAN PATHOLOGY

Volume 34, No. 4 (April 2003)

FIGURE 1. (A) A well-differentiated pancreatic ductal adenocarcinoma showing a high level of p27Kip1 expression. Staining is predominantly nuclear. (B) A moderately differentiated pancreatic carcinoma showing a lower level of p27Kip1 expression than the tumor in (A).

trieved. Information about clinical outcome was obtained from clinical notes and a hospital computer database. Ethical approval was obtained from the regional ethics committee.

Immunocytochemistry for p27Kip1 The block containing the most poorly differentiated area of each primary tumor and, in the relevant cases, the block containing the largest lymph node deposit were selected for immunocytochemistry. Sections (3 to 5 ␮ thick) were dewaxed in xylene, rehydrated through graded alcohol, and microwaved for 10 minutes in citrate buffer (pH 6.0). The sections were then blocked with 3% hydrogen peroxide in methanol for 15 minutes, followed by normal horse serum diluted 1:20 in .1 M phosphate-buffered saline (PBS) at pH 6.0, and incubated overnight with a monoclonal p27Kip1 antibody (K25020; Transduction Laboratories, Lexington, KY) diluted 1:200 in PBS. A standard avidin-biotin peroxidase complex method was then used as described previously.19-21 The specificity of the antibody was established previously20 and was confirmed by a blocking reaction. For negative controls, the primary antibody was omitted. Lymphocytes in each section served as positive internal controls.

Statistics Pearson’s ␹2 test and Fisher’s exact test were used to test associations between p27Kip1 expression and pathologic features. The Wilcoxon signed-rank test and the Student t test for independent samples were used to compare p27Kip1 levels in primary tumors with those in their lymph node metastases. The Kaplan-Meier method and the log rank test were used for univariate analysis of the relationship between clinicopathologic features and survival. Features associated with outcome by univariate analysis were entered into a Cox multivariate regression analysis model with forced entry, to determine their independent prognostic value. SPSS version 9.0 (SPSS, Chicago, IL) for Windows (Microsoft, Redmond, WA) was used throughout. A P value ⬍.05 was regarded as significant.

RESULTS p27Kip1 Expression in Primary Carcinomas and in Lymph Node Metastases The majority (⬎90%) of peritumoral and intratumoral lymphocytes showed strong immunocyto-

Analysis of Immunostaining Cells showing nuclear p27Kip1 expression were deemed immunopositive, regardless of the intensity of staining. In each tumor, at least 20 high-power fields (magnification ⫻400) and a minimum of 1000 neoplastic cells were examined. A microscope eyepiece grid was used to facilitate assessment. The percentage extent of p27Kip1 expression was determined by dividing the number of positively staining neoplastic cells in all fields by the total number of neoplastic cells in all fields and multiplying the result by 100. In large heterogeneous tumors, the most poorly differentiated areas were assessed as previously described.19,25 Expression in nonneoplastic acinar, ductal, and islet cells and in lymph node metastases were determined using the same approach. Each slide was examined by the 2 authors independently. Discrepancies of ⬎5% were resolved by reexamination and, if necessary, simultaneous examination of the slide by both authors using a multiheaded microscope.

FIGURE 2. Metastatic pancreatic carcinoma in a lymph node. The tumor shows ⬍1% p27Kip1 expression (right), contrasting with strong expression by lymphocytes (left).

386

p27Kip1 EXPRESSION IN PANCREATIC CARCINOMA (Feakins and Ghaffar)

TABLE 1. p27Kip1 Expression in Lymph Node Metastases and Corresponding Primary Tumors

Case

Primary tumor p27Kip1 expression (%)

Lymph node metastases p27Kip1 expression (%)

1 2 3 4 5 6 7 8 9 10 11 12 13

.2 .3 1.0 1.6 2.6 3.5 6.9 14.0 18.6 28.6 34.2 61.7 68.5

.1 0 58.3 3.9 5.0 8.4 .4 14.9 70.1 19.4 73.3 53.6 86.6

Median (interquartile range)

6.9 (1.3–31.4)

14.9 (2.2–64.2)

TABLE 3. Univariate Analysis of the Relationship Between Clinicopathologic Features and Clinical Outcome Number of events/ P number of patients (%) value*

NOTE. P ⫽ .221 for the difference between primary tumors and lymph node metastases (Wilcoxon signed rank test).

chemical staining for p27Kip1. Pancreatic islet cells showed more extensive staining (mean extent, 67.6%) than ductal epithelium (mean extent, 27.6%; range, 12.6% to 44.3%), whereas acinar cells showed less extensive staining (mean extent, 13.5%). Within samples, variations in the extent of staining between ducts were seen, but few ducts showed ⬍5% staining. Forty-four of the 46 carcinomas expressed p27Kip1, at least focally (mean extent, 14.4%; median, 6.6%; range, 0 to 83.9%) (Fig 1). The extent of p27Kip1 expression was ⬍1% in 15% of tumors, ⬍5% in 41%, ⬍10% in 56%, and ⬍50% in 89%. The extent of p27Kip1 expression in 13 lymph node metastases (mean, 30.1%; median, 14.9%) (Fig 2) tended to be TABLE 2. Associations Between p27Kip1 Expression and Pathologic Features of Carcinomas No metastases Differentiation Well Moderate Poor Tumor size ⬍ 5 cm ⬎ 5 cm UICC stage 1-2 3-4 Nodal status No metastases Metastases Ki67 index§ ⬍ 23.7% ⬎/⫽ 23.7%

Number with ⬍5% p27Kip1 expression/total (%)

1/11 (9%) 12/20 (60%) 6/15 (40%)

P value*

.022†

12/32 (39%) 7/14 (50%)

.522

10/21 (48%) 9/25 (36%)

.550

10/22 (45%) 9/24 (38%)

.765

9/21 (43%) 8/17 (47%)

1.000

*Fisher’s exact test (unless otherwise specified). †Pearson’s ␹2 test. §Data for some cases are not available.

Gender Male Female Tumor size ⬍ 5 cm ⬎ 5 cm UICC stage 1-2 3-4 Lymph node metastases Not present Present Differentiation Well or moderate Poor Excision† Complete Incomplete Ki67 labelling index† ⬍ 23.7% ⬎ 23.7% p27Kip1 (primary tumor) ⬍ 4% ⬎ 4% p27Kip1 (primary tumor) ⬍ 5% ⬎ 5% p27Kip1 (lymph node metastases) ⬍ 5% ⬎ 5%

13/26 (50%) 14/20 (70%)

.303

14/32 (44%) 13/14 (93%)

.011

10/21 (48%) 17/25 (68%)

.069

11/22 (50%) 16/24 (67%)

.141

18/31 (58%) 9/15 (60%)

.683

15/29 (52%) 12/16 (75%)

.016

13/21 (62%) 10/17 (59%)

.489

12/16 (75%) 15/30 (50%)

.060

14/19 (74%) 13/27 (48%)

.100

5/5 (100%) 3/8 (38%)

.054

*Log rank test. †Data are not available for some cases.

higher than in the 13 corresponding primary tumors (mean, 18.4%; median, 6.9%), but the differences were not significant (Wilcoxon rank sum test) (Table 1), and neither the mean nor the median values differed (t test for independent samples). Association of Clinicopathologic Features and p27Kip1 Expression Tumors in which the extent of p27Kip1 expression was ⬍5% were more likely to be moderately or poorly differentiated than well differentiated (P ⫽ .022). p27Kip1 expression was not related to patient gender, tumor maximum dimension, T classification, the presence of lymph node metastases, or UICC stage (Table 2). Features Associated With Clinical Outcome Of the 46 patients (mean follow-up time, 20.1 months; median, 9 months; range, 1 to 152 months), 27 died during the follow-up period of the study (mean time to death, 8.7 months; median, 6 months; range, 1 to 65 months). Univariate survival analysis showed a trend toward worse survival in patients whose tumors had ⬍4% p27Kip1 expression (P ⫽ .060) (Fig 3A and Table 3). When alternative thresh-

387

HUMAN PATHOLOGY

Volume 34, No. 4 (April 2003)

FIGURE 3. (A) p27Kip1 expression (primary tumors) and survival. (B) p27Kip1 expression (lymph node metastases) and survival. (C) Completeness of excision and survival. (D) Tumor size and survival.

olds for defining p27Kip1 positivity, including the median value (6.6% expression), 1%, 2%, 3%, 5%, and 10% expression, were used, the associations with clinical outcome were weaker. There was a trend toward a worse outcome for patients whose lymph node metastases had ⬍5% p27Kip1 expression (P ⫽ .054) (Fig 3B). Death was associated with incomplete tumor excision (P ⫽ .016) (Fig 3C) and tumor

maximum dimension ⬎5 cm (P ⫽ .011) (Fig 3D and Table 3). When p27Kip1 expression, tumor size, nodal metastasis, and completeness of excision were entered into multivariate survival analysis models, completeness of excision was the only feature that consistently showed an association with clinical outcome. An example is given in Table 4.

TABLE 4. Multivariate Analysis of the Association Between Clinicopathologic Features and Clinical Outcome (Cox Regression With Forced Entry of Variables)

p27Kip1 expression Tumor size Completeness of excision Lymph node metastases

Beta

Standard error

Relative risk (90% confidence interval)

P value

.781 .113 .901 .577

.420 .548 .422 .422

2.18 (1.09–4.35) .89 (.36–2.19) 2.46 (1.22–4.93) .56 (.28–1.12)

.063 .836 .033 .171

NOTE. In this model, all features were entered as categorical (dichotomous) variables.

388

p27Kip1 EXPRESSION IN PANCREATIC CARCINOMA (Feakins and Ghaffar)

DISCUSSION A recent Japanese study showed ⬍1% p27Kip1 expression in 56% of pancreatic ductal adenocarcinomas,29 and a Chinese study showed ⬍5% expression in 66% of adenocarcinomas.30 In the present study, there was ⬍1% p27Kip1 expression in 15% of tumors and ⬍5% expression in 41% of tumors. This suggests that reduced p27Kip1 expression is less prevalent in tumors from European patients than in those from Asian patients. Methodologic differences might be partly responsible for the discrepancies, although Lu et al’s selection of the most strongly staining areas for analysis would be expected to decrease the prevalence of apparent p27Kip1 loss.29 p27Kip1 expression may correlate with the stage and grade of carcinomas.19,21,23,25,32 In pancreatic adenocarcinomas, p27Kip1 loss correlated with high stage in 2 previous studies29,30 but not in the present study. It was associated with poor differentiation in 1 of 2 previous studies29,30 and in the present study. The inconsistencies suggest that associations between p27Kip1 expression and pathologic features are weak. Low p27Kip1 expression was a strong independent negative prognostic factor for pancreatic adenocarcinoma in a previous study29 but was a weak prognostic factor in the present study, despite the use of the same antibody and a similar immunocytochemical procedure. This discrepancy has several possible explanations. Different approaches to the analysis of staining may be partly responsible. Also, the present study included fewer variables in the multivariate model, although this does not explain the striking difference between the P values obtained by univariate analysis (P ⫽ .060 and P ⫽ .004).29 Furthermore, the small size of both studies (n ⫽ 46 and n ⫽ 34) may have caused statistical errors. Nevertheless, it is conceivable that the role of p27Kip1 in the pathogenesis of gastric and pancreatic adenocarcinomas differs between Eastern and Western populations. Indeed, p27Kip1 expression was a prognostic factor for gastric adenocarcinomas in 3 Japanese studies21,22,33 but not in 2 European studies.25,26 p27Kip1 may influence intercellular adhesion. Accordingly, reduced p27Kip1 expression may favor loosening of cells from one another and from the extracellular matrix, in turn increasing the likelihood of metastasis.19 Indeed, p27Kip1 expression in lymph node metastases was lower than in the corresponding primary carcinomas of the large bowel,34 prostate,35 and stomach33,36 in previous studies, but not in carcinomas of the stomach in another study25 or in pancreatic carcinomas in the present study. Interestingly, low p27Kip1 expression in lymph node metastases tended to be associated with a worse outcome in the present study. Although this finding needs to be confirmed in a larger series, it raises the possibility that p27Kip1 loss in metastatic deposits reduces cell adhesion, in turn facilitating further metastasis and hastening death. The findings suggest that p27Kip1 loss plays a role in the pathogenesis of pancreatic ductal adenocarcinoma but is not a major determinant of clinical out-

come in European patients. Nonetheless, assessment of p27Kip1 expression in pancreatic carcinoma may become more important in the future if treatments aimed at specific targets are developed. Acknowledgement. The authors are very grateful to Alex Brown and Carole Nickols for their technical work and support. REFERENCES 1. Rosewicz S, Wiedenmann B: Pancreatic carcinoma. Lancet 349:485-489, 1997 2. Geer RJ, Brennan MF: Prognostic indicators for survival after resection of pancreatic adenocarcinoma. Am J Surg 165:6872, 1993 3. Fortner JG, Klimstra DS, Senie RT, et al: Tumor size is the primary prognosticator for pancreatic cancer after regional pancreatectomy. Ann Surg 223:147-153, 1996 4. Magistrelli P, Antinori A, Crucitti A, et al: Prognostic factors after surgical resection for pancreatic carcinoma. J Surg Oncol 74: 36-40, 2000 5. Lundin J, Nordling S, von Boguslawsky K, et al: Prognostic value of Ki-67 expression, ploidy and S-phase fraction in patients with pancreatic cancer. Anticancer Res 15:2659-2668, 1995 6. Allison DC, Piantadosi S, Hruban RH, et al: DNA content and other factors associated with ten-year survival after resection of pancreatic carcinoma. J Surg Oncol 67:151-159, 1998 7. Luettges J, Schemm S, Vogel I, et al: The grade of pancreatic ductal carcinoma is an independent prognostic factor and is superior to the immunohistochemical assessment of proliferation. J Pathol 191:154-161, 2000 8. Takahashi T, Niino N, Ishikura H, et al: Predictive factors for long-term survival in patients with pancreatic carcinoma. Hepatogastroenterology 44:1463-1468, 1997 9. Sato Y, Nio Y, Song MM, et al: p53 protein expression as prognostic factor in human pancreatic cancer. Anticancer Res 17: 2779-2788, 1997 10. Gansauge S, Gansauge F, Ramadani M, et al: Overexpression of cyclin D1 in human pancreatic carcinoma is associated with poor prognosis. Cancer Res 57:1634-1637, 1997 11. Ito T, Ito M, Shiozawa J, et al: Expression of the MMP-1 in human pancreatic carcinoma: Relationship with prognostic factor. Mod Pathol 12:669-674, 1999 12. Seo Y, Baba H, Fukuda T, et al: High expression of vascular endothelial growth factor is associated with liver metastasis and a poor prognosis for patients with ductal pancreatic adenocarcinoma. Cancer 88:2239-2245, 2000 13. Coppola D, Lu L, Fruehauf JP, et al: Analysis of p53, p21WAF1, and TGF-beta 1 in human ductal adenocarcinoma of the pancreas: TGF-beta 1 protein expression predicts longer survival. Am J Clin Pathol 110:16-23, 1998 14. Toyoshima H, Hunter T: p27, a novel inhibitor of G1 cyclinCdk protein kinase activity, is related to p21. Cell 78:67-74, 1994 15. Polyak K, Lee MH, Erdjument-Bromage H, et al: Cloning of p27Kip1, a cyclin-dependent kinase inhibitor and a potential mediator of extracellular antimitogenic signals. Cell 78:59-66, 1994 16. Kamb A: Cell-cycle regulators and cancer. Trends Genet 11:136-40, 1995 17. Ewen ME, Sluss HK, Whitehouse LL, et al: TGF beta inhibition of cdk4 synthesis is linked to cell cycle arrest. Cell 74:1009-1020, 1993 18. Lloyd RV, Erickson LA, Jin L, et al: p27kip1: A multifunctional cyclin-dependent kinase inhibitor with prognostic significance in human cancers. Am J Pathol 154:313-323, 1999 19. Catzavelos C, Bhattacharya N, Ung YC, et al: Decreased levels of the cell-cycle inhibitor p27Kip1 protein: Prognostic implications in primary breast cancer. Nat Med 3:227-230, 1997 20. Loda M, Cukor B, Tam SW, et al: Increased proteasomedependent degradation of the cyclin-dependent kinase inhibitor p27 in aggressive colorectal carcinomas. Nat Med 3:231-234, 1997 21. Ohtani M, Isozaki H, Fujii K, et al: Impact of the expression

389

HUMAN PATHOLOGY

Volume 34, No. 4 (April 2003)

of cyclin-dependent kinase inhibitor p27Kip1 and apoptosis in tumor cells on the overall survival of patients with non-early stage gastric carcinoma. Cancer 85:1711-1718, 1999 22. Myung N, Kim MR, Chung IP, et al: Loss of p16 and p27 is associated with progression of human gastric cancer. Cancer Lett 153:129-136, 2000 23. Cote RJ, Shi Y, Groshen S, et al: Association of p27Kip1 levels with recurrence and survival in patients with stage C prostate carcinoma. J Natl Cancer Inst 90:916-920, 1998 24. Anayama T, Furihata M, Ishikawa T, et al: Positive correlation between p27Kip1 expression and progression of human esophageal squamous cell carcinoma. Int J Cancer 79:439-443, 1998 25. Feakins RM, Mulcahy HE, Quaglia A, et al: p27(Kip1) loss does not predict survival in patients with advanced gastric carcinoma. Cancer 89:1684-1691, 2000 26. Muller W, Grabsch H, Takeno S, et al: Prognostic value of the cyclin-dependent kinase inhibitor p27Kip1 in gastric cancer. Anticancer Res 20:1787-1792, 2000 27. Itami A, Shimada Y, Watanabe G, et al: Prognostic value of p27(Kip1) and cyclin D1 expression in esophageal cancer. Oncology 57:311-317, 1999 28. Resnick MB, Schacter P, Finkelstein Y, et al: Immunohistochemical analysis of p27/kip1 expression in thyroid carcinoma. Mod Pathol 11:735-739, 1998 29. Lu CD, Morita S, Ishibashi T, et al: Loss of p27kip1 expression independently predicts poor prognosis for patients

with resectable pancreatic adenocarcinoma. Cancer 85:1250-1260, 1999 30. Hu YX, Watanabe H, Li P, et al: An immunohistochemical analysis of p27 expression in human pancreatic carcinomas. Pancreas 21:226-230, 2000 31. Sobin LH, Wittekind C: Digestive system tumors: Pancreas, in Sobin LH, Wittekind C (eds): TNM Classification of Malignant Tumors (ed 5). New York, NY, Wiley-Liss, 1997, pp 87-90 32. Singh SP, Lipman J, Goldman H, et al: Loss or altered subcellular localization of p27 in Barrett’s associated adenocarcinoma. Cancer Res 58:1730-1735, 1998 33. Yasui W, Kudo Y, Semba S, et al: Reduced expression of cyclin-dependent kinase inhibitor p27Kip1 is associated with advanced stage and invasiveness of gastric carcinomas. Jpn J Cancer Res 88:625-629, 1997 34. Thomas GV, Szigeti K, Murphy M, et al: Down-regulation of p27 is associated with development of colorectal adenocarcinoma metastases. Am J Pathol 153:681-687, 1998 35. Fernandez PL, Arce Y, Farre X, et al: Expression of p27/Kip1 is down-regulated in human prostate carcinoma progression. J Pathol 187:563-566, 1999 36. Kim DH, Lee HI, Nam ES, et al: Reduced expression of the cell-cycle inhibitor p27Kip1 is associated with progression and lymph node metastasis of gastric carcinoma. Histopathology 36:245-251, 2000

390