J Clin Pathol:Mol Pathol 1998;51:21–25 21 Imprinted H19 oncofetal RNA is a candidate tumour marker for

I Ariel, H-Q Miao, X-R Ji, T Schneider, D Roll, N de Groot, A Hochberg, S Ayesh

Abstract This observation was confirmed recently by Li Aims/Background—To study the expres- et al, who found parallel expression of H19 and sion of the H19 in hepatocellular car- insulin-like growth factor 2 (IGF II).15 cinoma. H19 is an imprinted, maternally H19 is one of the few that functions as expressed gene, which is tightly linked, RNA molecules,16 similar to Xist,17 the im- both physically and functionally, to the printed gene in the Prader-Willi syndrome paternally expressed insulin-like growth region18 and the 3' untranslated region of á factor 2 (IGF II). IGF II is known to be tropomyosin.19 In these cases, a protein prod- involved in liver . H19 was uct does not exist, and can, first discovered in the fetal mouse liver to therefore, be determined only by the presence be under the same regulatory genes as á of RNA transcripts in the tissue. We performed fetoprotein (áFP), a widely used tumour radioactive and non-radioactive in situ hybridi- marker for hepatocellular carcinoma. sation (ISH) for H19 RNA on paraYnwax Methods—Using both radioactive and sections of 18 hepatocellular carcinomas and non-radioactive in situ hybridisation, the examined its expression in both the tumour expression of the H19 gene was compared and adjacent liver tissue (when present in the with the presence of áFP,as demonstrated section). by immunohistochemistry, in 18 cases of Alpha fetoprotein (áFP) is a widely used hepatocellular carcinoma. tumour marker in hepatocellular carcinoma.20 21 Results—H19 expression was present in 13 The expression of áFP mRNA in mice is Department of of 18 cases, whereas staining for áFP was determined by at least two transacting genes, Pathology, Perinatal positive in only nine of 18 cases. Concord- raf and Rif. The former determines the basal Pathology and ance was found in 12 of 18 tumours levels of áFP, while the latter induces its Quantitative Molecular 14 Pathology Unit, (66.7%). In general, the staining pattern expression during regeneration. The H19 Hadassah University for H19 was more diVuse than the immu- gene was first characterised as a unlinked Hospital Mount nohistochemical staining for áFP. to áFP, which is also regulated by raf and Rif in Scopus, PO Box 24035, Conclusions—The addition of a non- the liver, but not in other tissues.14 il-91240 Jerusalem, radioactive in situ hybridisation assay for We examined the presence of áFP by immu- Israel H19 RNA to the panel of tumour markers nohistochemical staining, and correlated our I Ariel D Roll used for the histopathological and cyto- findings with the expression of H19 in these logical diagnosis of hepatocellular carci- samples. We found that H19 expression is con- Department of noma might be useful. cordant with the presence or absence of áFP in Biological Chemistry, (J Clin Pathol:Mol Pathol 1998;51:21–25) the tumour in two thirds of the cases, and the Institute for Life ISH for H19 was found to be positive in more Sciences, Hebrew Keywords: hepatocellular carcinoma; H19; cases, although the diVerence was not found to University of á fetoprotein Jerusalem, Jerusalem, be significant in this small series. We suggest Israel that ISH for H19 RNA should be examined T Schneider further, as this molecule might be a useful N de Groot Imprinted genes—genes that are expressed additional tumour marker for the histopatho- A Hochberg from only one , depending on the parental logical and cytological diagnosis of hepatocel- S Ayesh origin—are implicated in playing an important lular carcinoma. role in the tumorigenesis of certain human Department of neoplasms.1–4 To date, the H19 gene is one of Oncology, Tumor Methods and materials Biology Research Unit, the few genes proved to be imprinted in SAMPLES 56 Hadassah University humans. Although it has been suggested by ParaYn wax blocks of 18 patients with hepato- Hospital, Kiryat other investigators that H19 acts as a tumour cellular carcinoma from the department of Hadassah, Jerusalem, suppressor gene,7 we have shown that it exhib- pathology, Qingdao University School of Israel its oncofetal characteristics, and is re-expressed H-Q Miao Medicine in Qingdao, China were submitted in tumours arising from tissues that express the for the study. The medical charts were 8–11 Department of gene in fetal life. One of the organs that searched for clinical and laboratory data Pathology, Qingdao expresses H19 abundantly from the early including tumour size, serum áFP, and hepati- University School of stages of embryogenesis is the liver.12 13 Expres- tis B serum antigen (HBsAg). Medicine, Qingdao, sion is confined to hepatocytes and mesenchy- We performed radioactive and non- China mal tissues, and is undetectable in haemapoi- X-R Ji radioactive ISH studies for H19 RNA and an etic tissue, one of the main tissue components immunohistochemical study for áFP according 10 13 Correspondence to: in the developing liver. H19 expression is to the following protocols. Dr Ariel. downregulated in most adult tissues, including email: [email protected] the liver, but may reappear in the liver during PREPARATION OF H19 VECTOR 14 Accepted for publication regeneration. We were the first to document Human H19 gene (800 base pairs) was 4 December 1997 H19 expression in hepatocellular carcinoma.10 subcloned at EcoRI site into Bluescript II KS 22 Ariel, Miao, Ji, et al

Table 1 Clinical data of 18 patients with hepatocellular carcinoma and the results of the RADIOACTIVE IN SITU HYBRIDISATION in situ hybridisation for H19 RNA and immunohistochemistry for á fetoprotein The procedure was performed on paraYnwax sections as described in detail previously.10 The Case Sex Age Tumour size HBsAg sáFP áFP H19 slides were examined and photographed using 1 M 47 Multicentric + + + f + d an Axioplan 2 (Carl Zeiss Jena GmbH, Göttin- 2 NANANA NANA+f+d 3 M 49 Biopsy − − + d + d gen, Germany) microscope under bright and 4 M 63 Multicentric + + + f + d dark field illumination. 5 F 44 Multicentric + − − + f The negative controls in this study included 6F644×4×4cm + − − +f 7 NANANA NANA+f+d hybridisation with sense RNA probe and 8 M 45 Multicentric + + − − RNase prehybridisation treatment. Negligible 9 M 59 10 × 10 × 10 cm + + − − signal was observed in all the controls. Positive 10 F 48 Multicentric + + − + f 11 F 49 10 × 10 × 10 cm + + + d − controls consisted of sections of fetal renal and 12 M 55 Multicentric − − − + f liver tissue. 13 M 41 Multicentric + + + d + d 14 M 67 Multicentric − + − + f × × 15 M 40 9 8 6cm + + +d +d PREPARATION OF RIBOPROBE FOR 16 M 63 7 × 7 × 7cm + + − − 17 M 52 22 × 14 × 8cm − − +d +d NON-RADIOACTIVE IN SITU HYBRIDISATION 18 M 49 7 × 7 × 7cm + + − − Digoxigenin labelled in vitro RNA transcripts were produced by labelling with DIG-11-UTP HBsAg, hepatitis B surface antigen; sáFP, serum á fetoprotein; f, focal; d, diVuse; NA, data not available. by SP6, T3, or T7 RNA polymerase in an in vitro transcription reaction (Boehringer Mann- (Stratagene) behind the T7 and T3 heim). RNA polymerase binding sites. In vitro RNA transcription with T7 RNA polymerase was used to produce antisense H19 cRNA from NON-RADIOACTIVE IN SITU HYBRIDISATION linearised plasmid DNA. Sense H19 RNA pre- Sections were prepared as for radioactive ISH, pared with T3 polymerase was used for and were rehydrated through a series of control. alcohols, followed by a wash in 0.9% NaCl and then a wash in phosphate buVered saline (PBS). Basic proteins were removed by incuba- PREPARATION OF RIBOPROBE FOR RADIOACTIVE IN tion in 0.1 N HCl at room temperature for 15 SITU HYBRIDISATION minutes. After a further wash in distilled water, 35S-labelled in vitro RNA transcripts (107 cpm/ sections were treated with 10 µg/ml proteinase µg) were produced using the Amersham RPN K (Sigma, Poole, Dorset, UK) in 50 mM Tris, 2006 Kit and RNA polymerases from Boe- 5 mM EDTA for 30 minutes at 37°C, rinsed in hringer Mannhein (Mannheim, Germany). PBS, and then refixed for five minutes in 4% Linearised were prepared by diges- paraformaldehyde/PBS. Thereafter, the sec- tion with HindIII (antisense) and EcoRI tions were rinsed in PBS and acetylated for 10 (sense). The fragments were separated from minutes in fresh acetic anhydride diluted 1/400 unincorporated nucleotidases by ethanolic in 0.1 M triethanolamine (Sigma) at pH 8.0. precipitation. The slides were then rinsed in PBS for five

Figure 1 (A) Immunohistochemical staining for á fetoprotein in hepatocellular carcinoma. Positive staining is present in single tumour cells (arrows). (B) H19 expression in the same tumour, as demonstrated by digoxigenin labelled in situ hybridisation. Expression is evident in most tumour cells. H19 in hepatocellular carcinoma 23

Figure 2 H19 expression in hepatocellular carcinoma. (A) Bright and (B) dark field. The expression is most prominent at the boundary of the tumour nodule. (In situ hybridisation of 35S-labelled antisense of H19 with haematoxylin and eosin as a counterstain.) minutes and in 0.9% NaCl for another five probe; and hybridisation with hybridisation minutes, dehydrated, and air dried. buVer without the probe. A section of fetal kid- The hybridisation buVer contained 50% ney, which expresses H19 in the metanephric deionised formamide, 0.3 M NaCl, 20 mM blastema, was used as a positive control. Tris-HCl (pH 7.4), 5 mM EDTA, 10 mM × NaH2PO4 (pH 8.0), 10% dextran sulphate, 1 Denhardt’s solution, and 0.5 µg/ml yeast IMMUNOHISTOCHEMISTRY A polyclonal rabbit primary antibody was used tRNA. Each section was covered with 30 µl of for histochemical detection of FP on paraYn the hybridisation solution containing 30– á wax sections (DPC ImmunoStain, Los Ange- 100 ng of DIG labelled RNA probe. The les, California, USA). Visualisation of FP in sections were covered with siliconised cover- á the sections was made by the streptavidin– slips. Hybridisation was performed at 48–52°C biotin immunoperoxidase technique using His- for at least 16 hours in a humid chamber. After tostain SP Kit (Zymed Lab Inc, San Francisco, hybridisation, coverslips were gently floated off California, USA) according to the manufactur- in 5× saline sodium citrate (SSC) at 50°C for er’s directions. 30 minutes. Subsequently, tissues were sub- jected to a stringent wash at 60°Cin50% deionised formamide, 2× SSC for 20 minutes. Results After a double rinse for 10 minutes in wash- The clinical and laboratory data are presented ing buVer (0.4 M NaCl, 10 mM Tris-HCl, and in table 1. 5 mM EDTA), treatment with RNase A Expression of H19 was detected by ISH in (20 µg/ml) was performed for 30 minutes at 13 of 18 samples with hepatocellular carci- 37°C. The sections were rinsed again for five noma. Expression was found to be prominent minutes in washing buVer, then washed in 2× and diVuse in eight of 13 positive samples (fig SSC, 0.5× SSC, and 0.1× SSC at 37°C. The eight), and weak or prominent and focal (in less sections were equilibrated in buVer 1 (100 mM than one third of the tumour) in the remaining Tris-HCl and 150 mM NaCl, pH 7.5) at room five samples (table 1). In six cases, we noted a temperature. The sections were then incubated gradual increase in the amount of expression in in whole sheep’s serum diluted 1/50 for 30 the malignant hepatocytes at the boundary of minutes (to avoid non-specific crossreactions the tumour nodules (fig 2). The results of the of the primary antibody). Incubation with the radioactive and the newly developed non- antidigoxigenin antibody (Boehringer) diluted radioactive studies were found to be essentially 1/1000 in buVer 1 was performed for two hours identical. at room temperature. After two washes in Immunohistochemical staining for áFP was buVer 1, the sections were equilibrated in found to be positive in nine of 18 samples. Of buVer 2 (100 mM Tris-HCl, 100 mM NaCl, those, it was present only focally in single cells

and 50 mM MgCl2, pH 9.5) for two minutes throughout the tumour in four cases (fig 1), and incubated with a freshly prepared colour and in a relatively large number of cells in the substrate solution containing nitroblue tetrazo- remaining five positive tumours (table 1). lium salt (340 µg/ml), 5-bromo-4-chloro-3- Non-neoplastic liver tissue was present in 14 indolyl, phosphate toluidine salt (170 µg/ml), of 18 sections and in all but one section this and levamizole (1 mM) in buVer 2. The slides non-neoplastic tissue was negative for H19 were placed in a humid chamber and allowed expression. In the one positive sample, promi- to develop in the dark for 12–14 hours at room nent expression of H19 was noted in the nod- temperature. The reaction was stopped in ules of cirrhotic liver, whereas no expression buVer 3 (10 mM Tris-HCl and 1 mM EDTA, was detected in the tumour itself. It is of inter- pH 7.4) for a few minutes. Finally, the sections est to note that this sample was from an HBsAg were counterstained with 3% Giemsa stain, positive patient. The remaining eight samples quickly dehydrated, and mounted in Enthelan. from HBsAg positive patients that included Controls for the specificity of the ISH non-neoplastic liver tissue in the histological included as follows: RNase A pretreatment of section were negative for H19 in the cirrhotic the sections; hybridisation with a sense RNA liver tissue. 24 Ariel, Miao, Ji, et al

Table 2 Comparison of When the expression of H19 (by ISH) was raised serum áFP concentration is always staining for H19 RNA by compared with immunohistochemical staining present in another liver tumour, hepatoblas- ISH and for áFP by immunohistochemistry for áFP, concordance was found in 12 of 18 toma. High serum áFP is also found frequently cases: in eight samples, both were found to be in germ cell tumours with a yolk sac compo- H19+ H19− positive and in four samples, both were nent. Occasionally, raised concentrations of áFP+ 8 1 negative. In the remaining six cases, the results áFP are seen in other types of , áFP− 5 4 were discordant: in five samples, the H19 gene predominantly tumours originating in the gas- was expressed and áFP was not present, and in trointestinal tract. Adenocarcinomas with one sample, H19 expression was not detected hepatoid diVerentiation, present among pan- but áFP was found to be positive (table 2). creatic, lung, and renal cancer, seem to be associated consistently with raised serum áFP. Discussion It should be mentioned, however, that many Despite extensive research for more than a other gastrointestinal carcinomas present with decade, the biological role of the H19 gene is raised serum áFP as well as positive immuno- still not understood. A clue might lie in its tight histochemical staining of tumour cells.41 linkage with the gene for IGF II. The latter is a The final diagnosis of hepatocellular carci- member of a family of growth factors and their noma is made by histopathological examin- receptors that play an important role in cell ation of biopsy material. The diagnosis is based proliferation and diVerentiation.22 The H19 on morphology and is assisted by a panel of and IGF II genes are reciprocally imprinted special stains and immunohistochemistry.42 and their expression is linked in fetal and adult The diVerential diagnosis includes primary tissues.12 23–25 The relation between their allelic cholangiocarcinoma and metastatic carci- status and abundance of expression has been noma. Further diYculty may arise in diVeren- investigated in certain types of human tiating atypical regenerating hepatocytes in cancer.1 3 15 26–30 The two genes reside close to liver cirrhosis from hepatocellular carcinoma in each other on 11p15.5, and it has cores of needle biopsies. been shown in mice that they share a common When immunohistochemical staining for enhancer.31 áFP is performed, a weak staining is often seen, IGF II is one of the growth factors known to although this was not our experience in the be involved in liver carcinogenesis in humans present study. DiYculty in the interpretation of and animal models.4 15 22 32–38 It regulates tissue this weak staining is probably the reason for the specific gene expression in rat and human wide range of positive staining for áFP hepatomas, and this eVect seems to be depend- documented in the literature.42 Moreover, áFP ent on the degree of tumour diVerentiation.22 may also be detected in non-neoplastic hepato- In the experimental model of woodchuck liver cytes by the immunohistochemical method.41 carcinogenesis, IGF II blocks apoptosis in- When comparing the immunohistochemical duced by N-myc.36 37 Loss of heterozygosity in staining of áFP in our series to the expression SV40 Tag induced murine hepatocellular of the H19 gene as demonstrated by ISH, the carcinoma was demonstrated in one third of latter method gave a more diVuse staining pat- the tumours with partial or complete loss of tern (fig 1) and was positive in more cases maternal chromosome 7 (the mouse syntenic (table 2), although the diVerence was not gene of the human ), which statistically significant (p = 0.29, using Fisher’s resulted in overexpression of IGF II and exact test). The use of the ISH technique for suppression of H19 expression.39 Although the diagnosis in routine laboratories of pathology allelic status of H19 was not the subject of our has become possible as a result of the develop- study, it appears from our findings of H19 ment of commercial kits for non-radioactive expression in 13 of 18 tumours, that this is not ISH. The procedure is only slightly more com- the case in human hepatocellular carcinoma. In plicated than immunohistochemistry. We in- fact, a deletion at 11p13–14, not at the region tend to study more cases of hepatocellular car- of IGF II/H19, was found to be induced by cinoma in order to assess the possible value of hepatitis B virus (HBV) in hepatocellular H19 as a tumour marker. carcinoma in humans.40 It was also found that Recently, the diagnosis of hepatocellular car- IGF II expression in human hepatomas is cinoma in fine needle aspirates has been upregulated by HBV at the post-transcriptional introduced.43 44 The cytological findings are level. assisted by various immunohistochemical Another gene known to be co-expressed with stains, including áFP. Staining for áFP was H19 in the liver (in mice) is áFP.14 áFP is a found to be positive in only 44% of the cases glycoprotein synthesised by the fetal liver and and was present focally in single cells in the fine yolk sac. Hepatic synthesis is reduced markedly needle aspirates.43 We have recently developed after birth: newborn serum contains a highly a method for detection of H19 expressing cells variable concentration of between 10 and in cytological specimens, such as urine (un- 200 mg/l but by 8 months of age, concentra- published data). It is possible that ISH for H19 tions have reached the adult level of less than expression, taking into account the more 20 mg/l.41 Raised concentrations of serum áFP diVuse and uniform pattern in tissue sections, are seen both in benign (mainly hepatic) and may prove to be useful for the diagnosis of malignant diseases, and it is widely used as a hepatocellular carcinoma in fine needle aspi- tumour marker for the diagnosis and follow up rates. of liver cancer.20 21 Of the liver tumours, We suggest that further work should be car- hepatocellular carcinoma is often associated ried out to investigate the use of H19 as a can- with a remarkable increase of serum áFP and a didate tumour marker to be applied to H19 in hepatocellular carcinoma 25

histological and cytological preparations, along 24 Ariel I, Weinstein D, Voutilainen R, et al. Genomic imprint- ing and the endometrial cycle. the expression of the with other special stains and immunohisto- imprinted gene H19 in the human female reproductive chemistry, for the diagnosis of hepatocellular organs. Diagn Mol Pathol 1997;6:17–25. 25 Eversole-Cire P, Fergusson-Smith AC, Surani AM, et al. carcinoma. Coordinate regulation of Igf-2 and H19 in cultured cells. Cell Growth DiV 1995;6:337–45. This work was supported by the Trilateral Grant (Germany- 26 Elkin M, Shevelev A, Schulze E, et al. The expression of the Israel-Palestinian Authority) from the DFG, Germany, Grant imprinted H19 and IGF-2 genes in human bladder No. 1490 by the DKFZ, Heidelberg, Germany and Grant No. carcinoma. FEBS Lett 1995;374:57–61. 95.00163 from the US–Israel Binational Science Foundation. 27 Kondo M, Suzuki H, Ueda R, et al. Frequent loss of imprinting of the H19 gene is often associated with its overexpression in human lung . Oncogene 1995;10: 1 Feinberg AP. and gene activation in 1193–8. cancer. Nat Genet 1993;4:110–13. 28 Taniguchi T, Sullivan MJ, Ogawa O, et al. Epigenetic 2 Hochberg A, Gonik B, Goshen R, et al. A growing relation- changes encompassing the IGF2/H19 locus associated with ship between genomic imprinting and tumorigenesis. Can- relaxation of IGF2 imprinting and silencing of H19 in cer Genet Cytogenet 1994;73:82–3. Wilms’ tumor. Proc Natl Acad Sci USA 1994;92:2159–63. 3 Tycko B. Genomic imprinting: mechanism and role in 29 Van Gurp RJLM, Oosterhuis JW, Kalscheuer V, et al. human pathology. Am J Pathol 1994;144:431–43. Human testicular germ cell tumors show biallelic expres- 4 De Souza AT, Yamada T, Mills JJ, et al. Imprinted genes in sion of the H19 and IGF2 gene. J Natl Cancer Inst 1994;86: liver carcinogenesis. FASEB J 1997;11:60–7. 1070–5. 5 Rachmilewitz J, Goshen R, Ariel I, et al. Parental imprinting 30 Yballe CM, Vu TH, HoVman AR. Imprinting and of the human H19 gene. FEBS Lett 1992;309:25–8. expression of insulin-like growth factor-II and H19 in nor- 6 Zhang Y, Tycko B. Monoallelic expression of the human mal breast tissue and breast tumor. J Clin Endocrinol Metab H19 gene. Nat Genet 1992;1:40–4. 1996;81:1607–12. 7 Hao Y, Crenshaw M, Moulton T, Tumour-suppressor et al. 31 Leighton PA, Ingram RS, Eggenschwiler J, et al. Disruption activity of H19 RNA. Nature 1993;365:764–7. of imprinting caused by deletion of the H19 region in mice. 8 Ariel I, Lustig O, Oyer CE, . Relaxation of imprinting in et al Nature 1995;375:34–9. trophoblastic disease. Gynecol Oncol 1994;53:212–19. 9 Ariel I, Lustig O, Schneider T, The imprinted H19 as a 32 Cariani E, Lassere C, Kemeny F, et al. DiVerential et al. expression of insulin-like growth factor II mRNA in human tumor marker in bladder carcinoma. Urology 1995;45:335– 8. primary liver cancer, benign liver tumors and liver 10 Ariel I, Ayesh S, Perlman EJ, et al. The product of the cirrhosis. Cancer Res 1988;48:6844–9. imprinted H19 gene is an oncofetal RNA. J Clin Pathol:Mol 33 d’Arville CN, Nouri-Aria KT, Johnson P, et al. Regulation of Pathol 1997;50:34–44. insulin-like growth factor II gene expression by hepatitis B 11 Biran H, Ariel I, de Groot N, et al. Human imprinted genes virus in hepatocellular carcinoma. Hepatology 1991;13: as oncodevelopmental markers. Tumor Biol 1994;15:123– 310–15. 34. 34 Fiorentino M, Grigioni WF, Baccarini P, et al.DiVerent in 12 Goshen R, Rachmilewitz J, Schneider T, et al. The situ expression of insulin-like growth factor type II in hepa- expression of the human H19 and IGF2 genes during tocellular carcinoma. An in situ hybridization and immu- human embryogenesis and placental development. Mol nohistochemical study. Diagn Mol Pathol 1994;3:59–65. Reprod Dev 1993;34:374–9. 35 Su Q, Liu YF, Zhang JF, et al. Expression of insulin-like 13 Lustig O, Ariel I, Ilan J, et al. Expression of the imprinted growth factor II in hepatitis B, cirrhosis and hepatocellular H19 gene in the human fetus. Mol Reprod Dev 1994;38: carcinoma: its relationship with hepatitis B antigen expres- 239–46. sion. Hepatology 1994;20:788–99. 14 Pachnis V, Belagew A, Tilghman SM. Locus unlinked to 36 Ueda K, Ganem D. Apoptosis is induced by N-myc expres- á-fetoprotein under the control of the murine raf and Rif sion in hepatocytes, a frequent event in hepadnavirus genes. Proc Natl Acad Sci USA 1984;84:5523–7. oncogenesis, and is blocked by insulin-like growth factor II. 15 Li X, Nong Z, Ekstrom C, et al. Disrupted IGF2 promoter J Virol 1996;70:1375–83. control by silencing of promoter P1 in human hepatocellu- 37 Yang D, Faris R, Hixon D, et al. Insulin-like growth factor II lar carcinoma. Cancer Res 1997;57:2048–54. blocks apoptosis of N-myc2-expressing woodchuck liver 16 Brannan CI, Dees EC, Ingram RS, et al. The product of the epithelial cells. J Virol 1996;70:6260–8. H19 gene may function as an RNA. Mol Cell Biol 1990;10: 38 Tabor E. Tumor suppressor genes, growth factor genes, and 28–36. oncogenes in hepatitis B virus-associated hepatocellular 17 Brockdorf N, Ashworth A, Kay GF, et al. The product of the carcinoma. J Med Virol 1994;42:357–65. mouse Xist gene is a 15 kb inactive X specific transcript 39 Casola S, Ungaro P, Padone PV, et al. Loss of heterozygosity containing no conserved ORF and located in the nucleus. of imprinted genes in SV/40 t/T antigen induced Cell 1992;71:515–26. hepatocellular carcinoma. 1995;11:711–21. 18 Wevrick R, Kerns JA, Francke U. Identification of a novel Oncogene paternally expressed gene in the Prader–Willi syndrome 40 Rogler CE, Sherman M, Su CY, et al. Deletion in chromo- some 11p associated with a hepatitis B integration site in region. 1994;3:1877–82. Hum Mol Genet hepatocellular carcinoma. 1985; :319–22. 19 Rastinejad F, Conboy MJ, Rando TA, et al. Tumor suppres- Science 230 sion by RNA from the 3' untranslated region of 41 Christiansen M, Ishiguro T, Hogdall C, et al. Alpha- á-tropomyosin. Cell 1993;75:1107–17. fetoprotein. In: Ballesta AM, et al, eds. Updating on tumor 20 Abelev GI. Alpha-fetoprotein: 25 years of study. Tumor Biol markers in tissues and biological fluids. Torino: Edizioni Min- 1989;10:63–74. erva Medica, 1993:245–69. 21 Tatarinov YS. The past and the future for cancer testing by 42 Hurlimann J, Gardiol D. Immunohistochemistry in the dif- alpha-fetoprotein. A review. J Nucl Med Allied Sci ferential diagnosis of liver carcinomas. Am J Surg Pathol 1989;33(suppl 3):5–11. 1991;15:280–8. 22 Zvibel I, Brill S, Reid LM. Insulin-like growth factor II 43 Guindi M, Yazdi HM, Gilliatt MA. Fine needle aspiration regulation of gene expression in rat and human hepatomas. biopsy of hepatocellular carcinoma. Value of immunohisto- J Cell Physiol 1995;162:36–43. chemical and ultrastructural studies. Acta Cytol 1994;38: 23 Voutilainen R, Ilvesmaki V, Ariel I, et al. Parallel regulation 385–91. of parentally imprinted H19 and insulin-like growth 44 Wee A, Nilsson B, Tan LK, et al. Fine needle aspiration factor-II genes in cultured human fetal adrenal cells. Endo- biopsy of hepatocellular carcinoma. Diagnostic dilemma at crinology 1994;134:2051–6. the ends of the spectrum. Acta Cytol 1994;38:347–54.