DECREASED EXPRESSION OF p27 PROTEIN IS ASSOCIATED WITH ADVANCED TUMOR STAGE IN HEPATOCELLULAR CARCINOMA Andrea TANNAPFEL 1 * , Dorothee GRUND 1 , Alexander KATALINIC 2 , Dirk UHLMANN 3 , Ferdinand KO ¨ CKERLING 4 , Ulrike HAUGWITZ 5 , Mark WASNER 5 , Johann HAUSS 3 , Kurt ENGELAND 5 and Christian WITTEKIND 1 1 Institute of Pathology, University of Leipzig, Leipzig, Germany 2 Institute of Cancer Epidemiology, University of Lu ¨beck, Lu ¨beck, Germany 3 Department of Surgery, University of Leipzig, Leipzig, Germany 4 Clinic of Surgery, Hannover, Germany 5 Department of Internal Medicine II, University of Leipzig, Leipzig, Germany Reduced expression of the cyclin-dependent kinase inhibi- tor p27 has previously been correlated with fatal clinical outcome in some tumors, including gastric, breast, and pros- tate cancers. For hepatocellular carcinoma, the findings are equivocal. In situ hybridization and immunohistochemistry were performed on a series of 203 curatively (R0) resected hepatocellular carcinomas and in corresponding non-cancer- ous liver tissue to detect p27. Patients receiving liver trans- plantation were excluded. The results were correlated with histopathological stage according to the UICC system, Ed- mondson grade, several other histopathological factors of possible prognostic significance, and finally patient survival. Whereas p27 mRNA was expressed homogeneously in all carcinomas examined, the p27 protein was found in various amounts. The labeling index of p27 protein was significantly lower in advanced stages of the disease (P < 0.001,  2  28.1). We observed decreased p27 protein in higher pT cat- egories (P < 0.001,  2  24.7) and in multiple tumor nodules (P < 0.001,  2  9.3). Multivariate Cox survival analysis identified age, co-existing cirrhosis, and Edmondson grade as independent prognostic factors. We conclude that evaluation of p27 in hepatocellular carcinoma is useful to predict stage of disease and may have clinical significance, e.g., in predict- ing optimal therapeutic regimes. Int. J. Cancer (Pred. Oncol.) 89:350 –355, 2000. © 2000 Wiley-Liss, Inc. Identification of multiple clinical and pathological prognostic factors in hepatocellular carcinomas (HCCs) has permitted a rea- sonable degree of risk stratification (Wittekind, 1995). However, clinical and pathological data fail to predict the actual virulence of a tumor concerning relapse or spread to distant organs in individ- ual cases. Cell-cycle progression is regulated by a family of cyclin-dependent kinases (cdks) and inhibited by cdk inhibitors. Cyclins, cdks, and cdk inhibitors are frequently deregulated in cancer (Nurse, 1997). Members of the kinase inhibitor protein (KIP) family, currently composed of p21 Cip1/Waf1/Sdi1 , p27 Kip1 , and p57 Kip2 , bind and inhibit cyclin E– cdk2 and cyclin A– cdk2 complexes (Ladha et al., 1998). p27 Kip1 is an inhibitor in cells arrested by TGF- and regulated by growth-inhibitory cytokines and by contact inhibition (Santoni-Rugiu et al., 1999). p27 is strongly expressed in non-proliferating cells and plays important roles in the regulation of both quiescence and G 1 progression (Glaise et al., 1998). Levels of p27 decrease as cells re-enter the cell cycle, mostly due to ubiquitin-proteasome– dependent degra- dation (Loda et al., 1997). Although mutations of the p27 Kip1 gene are rare in human tumors (Kawamata et al., 1995; Spirin et al., 1996), decreased p27 protein levels have been found in aggressive breast (Fredersdorf et al., 1997; Tan et al., 1997), lung (Esposito et al., 1997; Catzavelos et al., 1999), gastric (Mori et al., 1997; Ohtani et al., 1999), and colon (Thomas et al., 1998) cancers, suggesting that its loss may both reflect and participate in the process of tumor progression (Tsihlias et al., 1999). The relationship between p27 Kip1 expression and prognosis or histopathological parameters has not been examined in a large series of HCCs. To date, all reported studies have been based on small numbers of patients (Ito et al., 1999), have employed dif- ferent techniques for p27 assessment, and have not assessed pre- dictive value. The latter is an important concept since it is becom- ing clear that identification of specific molecular defects may be used to predict not only prognosis but also optimal treatment regimes. Additionally, potential prognostic indicators have almost exclusively been investigated in Asian (Qin et al., 1998; Ito et al., 1999) rather than Western medical centers. Given the differences in patient populations and etiology of HCC, it is reasonable to question whether the relative importance of various new prognos- tic factors differs as well. We therefore assessed p27 mRNA and protein expression in a series of surgically removed primary HCCs, to establish whether there is a correlation between degree of expression in the primary tumor and histopathological data including tumor grade, stage, and lymph node involvement. MATERIAL AND METHODS Patients and tissue samples Two hundred and three patients with HCC undergoing partial hepatectomy (segmental or lobar resection) between 1979 and 1997 were included in our retrospective study. No patient received pre-operative or adjuvant chemo- or radio- therapy. All patients were operated curatively (R0 resections). Patients who received orthotopic liver transplantation were ex- cluded. Each tumor was re-evaluated with regard to typing, staging, and Edmondson grading. Tumor typing and staging was performed using WHO (Ishak, 1994) and UICC (Sobin and Wittekind, 1997) criteria. Maximum tumor diameter was measured macroscopically in fresh specimens. In addition, every tumor was examined macro- and microscopi- cally for the presence of vascular invasion, capsule formation, satellites, multiplicity, inflammatory reaction, necrosis, bile pro- duction, giant cells, dysplasia within the surrounding liver, and co-existing cirrhosis. Multiplicity includes multiple nodules rep- resenting multiple, independent primary tumors as well as intra- hepatic metastasis from a single primary hepatic carcinoma. Vas- cular invasion includes gross as well as histological involvement. Tumor samples, free of necrosis or hemorrhage, were obtained from all patients. By macroscopic inspection, the periphery of the given tumor was preferred, to obtain 4 pieces of tissue in each case. For all analyses, the material was routinely fixed in 4% formaldehyde solution and embedded in paraffin. Sections (4 m thick) were cut, dewaxed in xylene, and then rehydrated. In all cases, slides from 4 different paraffin blocks, sampled from dif- ferent tumor areas, were examined. For histopathological data see Tables 1 and 2. *Correspondence to: Andrea Tannapfel, Institute of Pathology, Univer- sity of Leipzig, Liebigstrasse 26, 04103 Leipzig, Germany. Fax: +49-341- 9715009. E-mail: tana@medizin.uni-leipzig.de Received 15 November 1999; Revised 17 February 2000 Int. J. Cancer (Pred. Oncol.): 89, 350 –355 (2000) © 2000 Wiley-Liss, Inc. Publication of the International Union Against Cancer