Microsatellite instability in European hepatocellular carcinoma - Nature

Oncogene (1999) 18, 181 ± 187 ã 1999 Stockton Press All rights reserved 0950 ± 9232/99 $12.00 http://www.stockton-press.co.uk/onc

Microsatellite instability in European hepatocellular carcinoma Marina Salvucci1, Antoinette Lemoine1, RaphaeÈl Saroy1, Daniel Azoulay2, BeÂneÂdicte LepeÁre1, SteÂphanie Gaillard1, Henri Bismuth2, Michel ReyneÁs3 and Brigitte Debuire1,4 1

Department of Biochemistry, 2Hepatobiliary Surgery and Liver Transplant Research Unit and 3Department of Pathology, Paul Brousse Hospital - 14 avenue Paul Vaillant Couturier - 94804 Villejuif CeÂdex and Faculty of Medicine Paris-Sud, 4UMR 217 CNRS-CEA - Fontenay aux Roses - France

Genetic instability has been detected in many types of cancers but poorly investigated in hepatocellular carcinoma (HCC). We have studied the incidence of microsatellite instability (MI) at eight highly polymorphic microsatellite markers and the poly A tract BAT26 and tested for mutations at two sites of repetitive sequence (poly-A nucleotides 709 ± 718 and GT repeatnucleotides 1931 ± 1936) in the Transforming Growth Factor b (TGFb) type II receptor (RII) gene, in a group of 46 European HCCs and the surrounding nontumour tissue. This analysis showed that 63% of HCCs exhibit MI in at least one chromosome locus and 41% in two or more loci. No mutations of the TGFbRII gene were found in the MI positive tumours. No correlation was found with clinicopathological characteristics of the tumours such as cirrhosis, etiology, number of nodules, Edmondson's grade and vascular invasion. However, in patients who had a rearranged D16S402 microsatellite in their tumour, the recurrent disease and the number of nodules were signi®cantly higher than in the others (P50.005 and P50.02, respectively). We propose to consider D16S402 rearrangement in HCC as a prognostic factor to identify patients presenting a higher risk of recurrence. Keywords: transforming growth factor (TGFb) type II receptor (RII) gene; cirrhosis; clinicopathological characteristics; recurrent disease; prognostic factor

Introduction Instability at widespread highly polymorphic tandem repeat DNA sequences, known as microsatellites, has been reported in hereditary nonpolyposis colorectal cancer (HNPCC; Thibodeau et al., 1993), as well as in a number of other familial and sporadic cancers including colon (Ionov et al., 1993), endometrium (Duggan et al., 1994), oesophagus (Meltzer et al., 1994), stomach (Rhyu et al., 1994), pancreas (Brentnall et al., 1995), lung (Shridhar et al., 1994; Merlo et al., 1994), bladder (Gonzalez-Zulueta et al., 1993), kidney (Uchida et al., 1994), breast (Paulson et al., 1996), ovary (Arzimanoglou et al., 1996), brain (Zhu et al., 1996) and haematopoietic system (Wada et al., 1994; Kaneka et al., 1996), and some preneoplastic or in¯ammatory tissues (Brentnall et al., 1995, 1996; Salvucci et al., 1996). In sporadic tumours, the

Correspondence: B Debuire Received 14 January 1998; revised 7 July 1998; accepted 8 July 1998

extension and the type of microsatellite alterations is generally less pronounced than in HNPCC patients and often appears as one or few additional alleles at only one or few loci (Wooster et al., 1994). Recent studies have suggested intriguing correlations between microsatellite instability and clinicopathological parameters of some tumour types (De Marchis et al., 1997; Chung et al., 1996; Lothe et al., 1993) as well as an incidence of microsatellite instability on clinical outcome and patient survival (Thibodeau et al., 1993; Paulson et al., 1996; Lothe et al., 1993; Honchel et al., 1994). If mechanisms resulting in genetic instability lead to accelerated tumorigenesis and poor disease outcome in aected individuals, it is surprising that microsatellite instability has been associated with a better prognosis and increased survival in patients with HNPCC (Thibodeau et al., 1993; Lothe et al., 1993; Honchel et al., 1994). If dierent mechanisms are involved in microsatellite instability observed in sporadic cancers or if mutated target genes are dierent from one tissue to another, the ultimate eects on disease progression and patient survival may be dierent to that observed in colon cancer (Lothe et al., 1993). Therefore, it is important to clearly determine the consequences of microsatellite instability observed in dierent types of sporadic tumours, which constitute the majority of human cancers. We recently reported a frequent microsatellite instability in cirrhosis, especially in cirrhosis consecutive to hepatitis B viral infection (Salvucci et al., 1996). Cirrhosis is the most frequent underlying liver disease in hepatocellular carcinoma and very little is known about genomic instability in this tumour type (Hann et al., 1993). Treatment of HCC is rather disappointing and liver resection or transplantation can only be oered with a good chance of success to a small number of patients. Indeed, the results are much worse in patients presenting with large and multiple tumours, with a median survival as poor as 10 months (Akashi et al., 1991; Bismuth et al., 1993). However 30% of these latter patients have no reccurent disease within 3 years following surgery. Therefore, it would be of interest to have a molecular marker to help recognize patients at higher risk of recurrence. In the present study, we have analysed a series of both hepatocellular carcinoma (HCC) and the surrounding nontumour liver tissue DNA at dinucleotide repeats located on chromosomes 2p, 3p, 4q, 5q, 9p, 13q, 16q and 17p respectively. We also examined the poly(A) tract BAT 26, the repeat of 26 deoxyadenosines localized in an intron of the DNA mismatch repair gene hMSH2 on chromosome 2, known to provide remarkable results as an indicator of the replication error phenotype (RER) in colorectal

Microsatellite instability in hepatocellular carcinoma M Salvucci et al

182

cancers and cell lines (Hoang et al., 1996). Finally, we tested for mutations in two stretches of repetitive sequences in the Transforming Growth Factor b (TGFb) type II receptor (RII) gene. TGFb is a potent inhibitor of epithelial cell growth (Polyak et al., 1996). Its growth inhibitor eect is mediated by two distantly related transmembrane serine/threonine kinases called receptors I and II (RI and RII). The RII gene contains two stretches of repetitive sequence, poly A (nucleotides 709 ± 718) and GT repeat (nucleotides 1931 ± 1936) (Lin et al., 1992; Togo et al., 1996), which are thought to be the target of mutation in HNPCC and some sporadic gastrointestinal cancers (Zborowska et al., 1995; Lu et al., 1995; Parsons et al., 1995). Overall, our purpose was to determine the incidence of microsatellite instability and TGF-bRII gene mutation in HCC and to investigate the relationship between these alterations and clinicopathological variables and patient survival.

Results Microsatellite instability incidence in HCC Eight polymorphic microsatellite markers and one mononucleotide repeat, from eight chromosomal regions were ampli®ed from both the tumour and nontumour liver DNAs and the corresponding blood or non liver normal DNA from 46 HCC patients. Each PCR ampli®cation was repeated at least twice. As a preliminary experiment, comparison of a series of ten normal liver DNAs extracted from formalin-®xed paran-embedded blocks to the matched blood DNAs showed no dierences. The appearance of fragments of altered length in tumour and/or nontumour liver DNA indicated an alteration in microsatellite size. Samples showing changes in microsatellite sequences were veri®ed by isolating DNA from a new paran section and subjecting it to a second, independent PCR analysis. When samples were informative (78% of cases) it was also possible to detect a loss of heterozygosity (LOH). Results presented in Table 1A show that microsatellite alterations were detected in at least two chromosome loci in either the tumour and/or nontumour liver DNA in 19 of the 46 patients tested (41%). In one patient, the alterations found in the tumour DNA were not present in the nontumour liver DNA. Conversely, one patient had microsatellite alterations in the nontumour liver DNA that were not found in the tumour DNA. Ten additional patients exhibited microsatellite instability at only one locus (22%) (Table 1B), of these, two patients did not have alterations in the nontumour DNA. When samples were informative, most of the time, only one allele was aected. However, patient 27 exhibited both a shift of one allele and loss of the other allele in the peritumourous tissue (Figure 1). Of the eight chromosomes investigated, loci on chromosome 5 (8/17, 17%), chromosome 9 (9/46 patients, 20%) and chromosome 16 (19/46 patients, 41%) were preferentially altered. Figure 1 shows some examples of the alterations we observed. In 11/29 cases (38%), the rearrangements seen in the nontumour liver DNA were not identical to that observed in the corresponding tumour DNA and in

nine of the 19 patients (47%) with microsatellite alterations at two or more loci, additional alterations were observed in the tumour DNA comparatively to the adjacent nontumour liver DNA. Mutations of TGF bRII gene Genomic DNA of the 29 HCC specimens with microsatellite instability in at least one locus was screened for mutations by DNA sequence analysis in two nucleotide sequences, positions 625 ± 773 and 1851 ± 2024 containing the (A)10 or (GT)3 repeats of the TGF-bRII gene, respectively. No mutation was detected in any of the samples analysed. Correlations between microsatellite instability and tumour characteristics The potential impact of microsatellite instability on tumour growth and progression was determined by examining the association with a variety of clinical and pathological characteristics of the primary HCCs. Unintentional bias was prevented by coding patient tissue samples so that microsatellite instability determination was done without knowledge of the patient and tumour characteristics. The results are summarized in Table 1. Of the 19 tumours exhibiting microsatellite instability in two or more chromosome loci, 13 had developed on a cirrhotic liver of viral (n=9) or alcoholic origin (n=3); in one patient (patient 7), cirrhosis was of unknown origin. Infection by HBV alone was observed in four cases, HCV alone in ®ve cases, HBV+HCV in two cases and HBV+delta in one case (Table 1A). Ten additional patients had a tumour presenting microsatellite instability in only one locus, of these seven had cirrhosis induced by HCV in two, HBV+HCV in three and alcohol in two (Table 1B). A third group of 17 patients did not exhibit microsatellite instability in their tumour which had developed on a cirrhotic liver in ten, induced by HBV in one, HCV in ®ve and alcohol in two. In one patient, cirrhosis was of unknown origin and in another patient (patient 45) cirrhosis was consecutive to hemochromatosis (Table 1C). Four patients in this group had chronic hepatitis. Overall no correlation was found between the presence or not of cirrhosis and microsatellite instability in the tumour. We tested the samples to see whether HBV virus DNA was present in the tumours, irrespective of the etiology of the disease, using a technique employing PCR ampli®cation of part of the HBX gene, followed by hybridization with a speci®c probe (Bourdon et al., 1995). Results are given in Table 1. HBV imprinting was found in seven tumour DNAs, one of which had a negative corresponding serology. Only two of the seven HBX positive tumour DNAs did not exhibit at least one microsatellite alteration and overall 10/12 (83%) HBV positive patients were positive for microsatellite alterations in their tumour. In 13/46 patients (28%) serological markers of HCV were detected exclusively of other markers of hepatotropic viruses. Seven of them (54%) presented a genomic instability in their tumour. The number of nodules varied from one to more than six and their size from 9 mm to 200 mm. The size was not correlated with the microsatellite instability


Table 1 Clinicopathological characteristics and microsatellite instability analysis of a series of 46 European HCCs Number of HBX nodules Tumor* Portal PCR Surgery (size) grade invasion

Sex/ Patient Age

Cirrhosis

A 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19

F/56 F/56 M/37 M/45 M/48 M/66 M/57 M/54 F/43 F/37 M/36 M/57 M/63 M/50 M/45 M/75 M/62 M/36 F/62

+ HCV + alcohol ± HCV + HBV + HBV+HCV ± alcohol + ± + HBV+HCV ± ± ± HBV + HBV+D + HBV + HCV ± alcohol CH HCV + alcohol + alcohol + HBV + HCV

± ± ± + + ± ± ± ± + ± + ± ± ± ± ± + ±

PH 1 (100) OLT 1 (28) OLT >6 (45) PH 1 (50) OLT >6 (50) PH 1 (120) OLT >6 (70) OLT 1 (35) PH >6 (100) PH 1 (120) OLT 1 (60) PH 6 (90) OLT >6 (50) PH 1 (100) OLT 2 (35) PH 1 (50) OLT 3 (30) OLT 1 (10) OLT 2 (15)

III II III III II III III II II II II II II II II III II I II

± ± ± ± + ± + + + ± ± + + + + ± + + ±

D3S1317a, D5S409, D13S153 D4S395, BAT 26 IFNA, p53 D5S409, D13S153, D16S402 IFNAa, D16S402 D2S123 D16S402, BAT 26 D4S395, D16S402 BAT 26

B 20 21 22 23 24 25 26 27 28 29

M/55 M/48 M/18 M/57 M/53 M/60 M/51 M/50 F/62 M/54

+ + ± ± + + + ± + +

HBV+HCV HBV+HCV ± alcohol HBV+HCV alcohol alcohol ± HCV HCV

± ± ± ± ± ± ± ± ± ±

OLT PH PH PH OLT OLT OLT PH OLT OLT

1 (20) 5 (22) 1 (200) 1 (150) 1 (30) >6 (45) 3 (38) 1 (150) 3 (35) 5 (30)

II II II II II II II II II II

± + + + ± ± ± + + ±

D4S395 IFNA IFNA p53 BAT 26

C 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46

M/52 M/61 M/65 M/40 F/46 M/45 M/57 M/53 M/59 M/72 M/53 M/50 M/75 M/68 F/44 M/64 M/57

+ ± + ± ± + + CH + CH + + CH CH + + +

HCV ± alcohol ± ± HBV alcohol ± HCV HCV HCV HCV HBV ± ± ± HCV

± ± ± ± ± + ± ± ± ± ± ± + ± ± ± ±

OLT PH PH OLT OLT OLT OLT PH OLT PH OLT OLT PH PH OLT OLT OLT

2 (22) 1 (50) 1 (75) 1 (140) 1 (200) 1 (60) 1 (20) 1 (90) 2 (30) 2 (55) 1 (40) 1 (40) 1 (60) 1 (80) 1 (14) 2 (40) 5 (18)

II III III II II I I II II II II II III II II III II

± + ± + + ± ± + ± + + ± + ± + + ±

Etiology

Sites positive for microsatellite instability Nontumour liver DNA Tumour liver DNA

D16S402, BAT 26 D16S402 D16S402 D4S395, D16S402 IFNA, D16S402 IFNA, D16402 IFNA, D16S402 D5S409 D2S123, IFNA

D16S402 D16S402a D16S402

D3S1317a, D5S409b, D13S153, p53 D3S1317, D4S395, BAT 26 IFNA, p53, D16S402, D5S409 D5S409, D13S153, D16S402b D3S1317, D13S153a, D16S402 D2S123, D5S409a, D16S402a BAT 26b, D16S402b, D5S409 D4S395, D16S402 D16S402, BAT 26 D2S123a, D16S402a D16S402b, BAT 26b D4S395, D16S402b D5S409, D16S402b D5S409 IFNA, D5S409 IFNAa,b, D16S402 D2S123, D5S409b D2S123, IFNA D4S395 IFNA IFNA p53 BAT 26 IFNA D16S402b D16S402a,b D16S402b D16S402

a

Loss of one allele; bDierent rearrangements in tumour and nontumour liver DNAs for a same microsatellite marker; CH: chronic hepatitis; PH: partial hepatectomy; OLT: orthotopic liver transplantation; *Tumor grade was according to Edmondson's classi®cation

status of the tumour, whereas the number of nodules was signi®cantly higher when D16S402 was altered (3.8+0.8 vs 1.9+0.4, P50.02). Most of the tumours were of grade II (31/46) and were equally distributed among the microsatellite instability positive and negative groups. Portal and/or hepatic invasion was present in 24/46 (52%) patients examined; no correlation was found with microsatellite instability of the tumour. Correlation between microsatellite instability and patient performance The follow up available for the 46 patients analysed ranged from 1 to 51 months after surgery. Twenty

patients (43%) had detectable recurrence assessed by abdominal ultrasound, thoracoabdominal CT scan and bone scintiscan in case of boneache; 16 of them had microsatellite instability in their tumour. No correlation was found between recurrence of the disease and overall genomic instability of the tumour. However, when the risk of recurrence was analysed for each microsatellite tested, a high association between the alteration of D16S402 and the risk of recurrence was found (P50.005); no signi®cance was observed with other microsatellite loci used in this study. Indeed, of the 19 patients who exhibited a D16S402 rearrangement in their primary tumour, 14 had HCC recurrence. Figure 2 shows the Kaplan ± Meier curves of survival rate without recurrence for the patients analysed. The

183


184

530 mm a correlation with recurrence was found (P50.05). The association of the two factors (D16S402 and nodule size) showed no in¯uence on recurrence of the disease. Discussion

Figure 1 Alteration of microsatellite markers in hepatocellular carcinoma. Examples of alterations include loss of one allele in the tumour DNA (patient 6), gain of new shorter repeats in both the tumour and nontumour DNAs (patient 15) and loss of one allele in the tumour DNA and loss of one allele plus shift of the other in the corresponding nontumour DNA (patient 27). (A) normal DNA; (B) nontumour liver DNA; (C) tumour DNA

Figure 2 Cumulative disease free survival after surgery for hepatocellular carcinoma: D16S402 rearranged or not

median delay of recurrence was 11.6 months in the group of patients with an altered D16S402 whereas it was 39.0 months in the other group (P50.005, log rank analysis). The risk was similar for both the resected (n=18) and transplanted (n=28) patients. Using the Cox model analysis, the risk ratio of recurrence was 2.4 when D16S402 was rearranged as compared to patients exempt of D16S402 alteration in their tumour. We also analysed the survival rate without recurrence according to the histopathological characteristics of the tumour. When the size of the nodules was

Microsatellite instability was analysed in 46 cases of HCC and the surrounding nontumour tissue using eight highly polymorphic microsatellite markers known to be altered in digestive diseases (Tamura et al., 1995) and the poly(A)tract BAT26 (Hoang et al., 1996). TGFbRII mutations were assessed in parallel in the tumours. Microsatellite instability was observed in 29/ 46 cases (63%); no mutation of the TGFbRII gene was detected. The most striking feature was the frequent alteration of the D16S402 locus (19/46, 41%) and its association with a higher risk of recurrence (P50.005). We have previously reported an incidence of 60% of microsatellite instability in cirrhosis. This incidence was of 86% in HBV positive livers. In the current study, we extend this observation to HCC by showing that the group of patients with HBV positive serology present a higher rate of microsatellite alterations in their tumour than the others. Among dierent studies, there is considerable variation in the frequencies of microsatellite instability reported for dierent tumour types and within a same tumour type from one series to another. Such variability may partly depend on the number of cases and microsatellites examined and on the sensitivity of the loci analysed. Using nine microsatellite markers, we report here a high incidence of genomic instability in HCC (63%). It is possible that the incidence we determined underestimates the actual frequency, since we analysed only nine of the many thousands of microsatellites sequences in the genome. The incidence of microsatellite instability in a series of 17 HCCs was examined in a previous study. The authors reported the absence of genomic instability by means of four microsatellite markers (Hann et al., 1993). On another hand, our results are in agreement with recent data analysing alterations of another type of DNA repeats. This study indicates frequent rearrangements at minisatellite loci D1S7 (1p33-35), D7S22 (7q36-ter) and D12S11 (12q24.3-ter) in a series of 26 hepatitis B virus-positive hepatocellular carcinomas from Thai patients (Kaplanski et al., 1997). In our series, of the 29 patients exhibiting microsatellite instability, ten had one alteration and 19 at least two. Of the eight chromosomes investigated, in both the tumour and nontumour tissue, loci on chromosomes 16, 9 (IFNA) and 5 were preferentially altered (41%, 20% and 17% respectively). In our previous study on cirrhosis, IFNA and D5S409 were the most frequently rearranged loci, although D16S402 had not been investigated. Frequent allelic loss on chromosomes 4q and 16q reported in HCC from Taiwan (Yeh et al., 1996) led us to include microsatellite markers on these loci in the present study. In our series of European HCCs, only ®ve patients had alterations on chromosome 4 but 19 had rearrangements on chromosome 16. However, overall allelic losses were found in only 3/46 cases (7%) for this latter marker which is far less than the 70%


reported by Yeh et al. (1996). Two recent studies have examined loss of heterozygosity (LOH) in HCC using a collection of 195 ± 275 microsatellites loci distributed along the autosomes (Boige et al., 1997; Nagai et al., 1997). Among the chromosome segments most frequently aected by deletion, LOH of speci®c loci on chromosome 16q represented 39% in aneuploid tumours (Boige et al., 1997) and 31% in tumours of various geographical origin (Nagai et al., 1997). The dierences between these studies and ours could be explained by the use of only one microsatellite marker on chromosome 16 in the present study, the grade of the tumours analysed and/or by dierences in the ethnic origin of the studied populations. Our results shown in Table 1 indicate that most of the microsatellite markers rearrangements found in the tumour DNA were identically present in the nontumour liver DNA, highlighting both the earliness of these alterations and the risk of using the nontumour liver tissue as a reference in these studies. Of the 29 cases listed as microsatellite instability positive, seven would have been scored negative if we had compared the tumour to the nontumour liver DNA (see Figure 1, patient 15, IFNA). In 11 patients, alterations found in the tumour DNA were dierent from that found in the nontumour DNA for a same microsatellite marker, irrespective of the cirrhotic or noncirrhotic character of the underlying liver disease. By investigating the clonality of microsatellite alterations at several locations within a same cirrhotic liver, we have previously shown that a polyclonal pattern of instability compatible with the heterogeneous cell populations in a precancerous tissue was observed in some cases (Salvucci et al., 1996). In others, a monoclonal pattern suggested that microsatellite alterations had occurred in a progenitor cell ancestral to the majority of the cells in the samples analysed and argued for a high rate of regeneration in these particular specimens. In the present study, either identical or dierent alterations of a same microsatellite marker are found when comparing tumour and nontumour DNAs from a same liver to the reference DNA, underlying the monoclonal character of the peritumourous tissue in the former cases. In four cases (patients 2, 3, 6 and 9), the similarity between the tumour and nontumour DNA in the alteration pro®le of one or two microsatellite markers and the presence of additional rearranged microsatellites in the tumour DNA are in agreement with the concept of accumulation of genetic alterations during the progression of tumorigenesis. Mutation has been reported to occur in the coding poly(A) tract of the TGFbRII gene in numerous gastrointestinal tumours with microsatellite instability (Parsons et al., 1995; Souza et al., 1997; Myero et al., 1995) resulting in production of truncated RII protein. Since TGFb inhibits cell proliferation, the loss of response to TGFb due to inactived RII may be an important tumour progression step. In our study, no mutations were found in both the poly(A) tract and (GT) repeat of the TGFbRII gene suggesting that this gene only plays a small role, if any, in liver carcinogenesis. Two other studies have already reported the absence of TGFbRII gene mutation in HCC (Lu et al., 1995; Vincent et al., 1996). However, we show for the ®rst time the absence of correlation

between microsatellite instability and TGFbRII gene mutation in this tumour type. Genomic instability is probably here dierent from that observed in colorectal cancer. HCC is not the ®rst tumour type to be found exempt of TGFbRII gene mutations in spite of demonstrated genomic instability. Indeed no mutations were detected in endometrial, pancreatic and lung cancers with microsatellite instability (Myero et al., 1995; Abe et al., 1996; Takenoshita et al., 1997). We have analysed the clinicopathological characteristics of our series of tumours and found no correlation between the presence of microsatellite instability and the grade, the size of nodules, the underlying liver disease (cirrhosis or not) and etiological factors. However, ten of the 12 (83%) HBV studied positive patients exhibited microsatellite instability, extending our previous observation in cirrhosis (Salvucci et al., 1996). We also analysed the survival rate without recurrence for the 46 patients of our study (follow up 1 to 51 months). Microsatellite instability was not associated with a higher risk of recurrent disease that occurred in 20/46 patients (43%). However, when each microsatellite locus was analysed separately, recurrence was signi®cantly more frequent in patients with an altered D16S402. Indeed, 14/20 (70%) patients who had recurrent disease exhibited D16S402 alteration in their tumour. Their median delay of recurrence was 11.6 months whereas it was 39.0 months for the patients without D16S402 instability (P50.005). In conclusion we have shown a frequent microsatellite instability not associated with TGFbRII mutation in European HCC. Of the nine microsatellite markers analysed, D16S402 was the most frequently rearranged and correlated with a higher risk of recurrence. In light of our ®ndings, it may be particularly informative to investigate whether D16S402 rearrangement is involved in the subset of HCC patients who have poor disease outcome. In addition to the potential prognostic value, identification of D16S402 positive tumours could lead to more eective therapies.

Materials and methods Patients and samples A retrospective study on liver tissues harvested during the course of surgery and kept embedded in paran from 1993 to 1997 was performed. Histology was reexamined by the pathologist for 186 patients. Necrotic tumours were excluded from the study and samples areas containing more than 70% tumourous cells were selected. Finally, 46 pairs of tumour and nontumour liver samples from 46 Caucasian patients (39 males, 7 females, aged 18 ± 75 years) were analysed. All patients were tested serologically for hepatitis B, D and C viruses as published elsewhere (Samuel et al., 1995; Feray et al., 1994). In six of them (13%), HCC was associated with hepatitis B, in one (2%) with hepatitis B+D, in ®ve (11%) with hepatitis B+C and in 13 (28%) with hepatitis C virus infection alone. In ten (22%) patients, HCC was consecutive to excessive alcoholic consumption. In 30 (65%) patients the nontumour liver tissue was cirrhotic and ®ve (11%) patients had a chronic hepatitis (Table 1). We analysed formalin®xed paran-embedded (FFPE) specimens from tumour and nontumour liver tissues obtained during surgery and

185


186

paired peripheral blood samples obtained at distance from surgery. This was performed to avoid blood contamination by hepatocytes which is always possible during the course of surgery (Lemoine et al., 1997). DNA was extracted from blood as already published (Salvucci et al., 1996). When blood samples were not available, FFPE specimens of nonhepatic tissue (hilum for three patients, hilar connective tissue for one patient, hilar lymph node for ®ve patients and gallbladder for 18 patients) were used. Liver or other tissues sections (approx. 2 ± 4 cm2, with a thickness of 8 mm) were cut on a microtome with careful cleaning of the blade between each handling. Tumour sections were placed in 1.5 ml microtubes and DNA was extracted in a ®nal volume of 200 ml as previously described (Bourdon et al., 1995). As a normal control, DNA was prepared from both biopsies of normal liver routinely performed to assess the degree of steatosis during harvesting of six dierent livers for further transplantation and the corresponding blood lymphocyte DNA from the donors. Presence of HBV DNA was checked in all liver tissue samples by polymerase chain reaction (PCR) using primers on the X region as already reported (Bourdon et al., 1995) with slight modi®cations. Microsatellite analysis Microsatellite alterations at poly(CA) microsatellite loci on chromosomes 2p (D2S123), 3p (D3S1317), 4q (D4S395), 5q (D5S409), 9p (IFNA), 13q (D13S153), 16q (D16S402) and 17p (p53) and the poly(A) tract BAT26 were examined in liver (tumour and nontumour) and blood or nonhepatic DNAs processed simultaneously. One ml of extracted DNA from liver tissue specimens and 50 ng of blood DNA or 1 ml of DNA extracted from non liver tissue specimens were ampli®ed in parallel with an appropriate pair of primers. In each case, one of the paired primers was labeled with a ¯uorescent dye (Perkin-Elmer, Courtaboeuf, France). PCR was performed in a 25 ml volume of a mixture containing 10 mM Tris buer (pH 8.3), 1 mM MgCl2, 50 mM KCl, 0.2 mM of each deoxynucleotide triphosphate, 0.5 units of Taq DNA polymerase (Appligene-Illkirch, France) and 5 pmoles of each primer using a gene Amp PCR system 2400 (Perkin-Elmer, Courtaboeuf, France). Ampli®cation consisted of one cycle of 2 mn at

948C, 35 cycles of 1 mn at 928C, 30 s at the annealing temperature, 45 s at 728C plus a ®nal extension step of 7 mn at 728C. The annealing temperature was 508C for BAT26, 528C for D4S395 and D16S402 and 558C for the other microsatellites. The PCR products were separated by capillary electrophoresis using the ABI PRISM 310 (Perkin-Elmer, Applied Biosystems, Courtaboeuf, France). Direct sequencing of TGFbRII gene To detect mutations in the poly(A) sequence (nucleotides 709 ± 718) and GT repeat sequence (nucleotides 1931 ± 1936) in RII, we used the protocol described by Togo et al. (1996), with slight modi®cations: the PCR conditions were those described above (annealing temperature: 568C) and direct sequencing was performed by a terminator cycle sequencing ready reaction kit (Perkin-Elmer, Applied Biosystems, Courtaboeuf, France) and a ABI Prism 310 DNA sequencing system (Perkin-Elmer, Applied Biosystems) according to the manufacturer's recommendations. Statistical analysis Statistical analysis was performed using the SAS software system (SAS Institut Inc, Cary, NC, USA) and the Student t-test was used to determine the statistical signi®cance of dierences of clinical data, according to the presence or absence of microsatellite instability. The survival probabilities without recurrent disease were determined using the Kaplan ± Meier technique according to the presence of genomic instability and the signi®cance of the dierence was analysed by the log rank test. A Cox regression analysis was performed to identify which independent factors would have a signi®cant in¯uence on recurrence and survival. A P value less than 0.05 was considered to have statistical signi®cance.

Acknowledgements We thank Dr E May for helpful advice and discussion and V Caillez for statistical analysis. This work was supported by grants from the FaculteÂ de MeÂdecine Paris-Sud.

References Abe T, Ouyan H, Migita T, Kato Y, Kimura M, Shiiba K, Sunamura M, Matsuno S and Horii A. (1996). Eur. J. Surg. Oncol., 22, 474 ± 477. Akashi Y, Koreeda C, Emomoto S, Uchiyama S, Mizumo T, Shiozaki Y, Sameshima Y and Imoue K. (1991). Hepatology, 14, 262 ± 268. Arzimanoglou IA, Lallas T, Osborne M, Barber H and Gilbert F. (1996). Carcinogenesis, 47, 1799 ± 1804. Bismuth H, Chiche L, Adam P, Castaing D, Diamond T and Dennison A. (1993). Ann. Surg., 218, 145 ± 151. Boige V, Laurent-Puig P, Fouchet P, FleÂjou JF, Monges G, Bedossa P, Bioulac-Sage P, Capron F, Schmitz A, Olschwang S and Thomas G. (1997). Cancer Res., 57, 1986 ± 1990. Bourdon JC, D'Errico A, Paterlini P, Grigioni W, May E and Debuire B. (1995). Gastroenterol., 108, 1176 ± 1182. Brentnall TA, Chen R, Lee JG, Kimmey MB, Bronner MP, Haggitt RC, Kowdley KV, Hecker LM and Byrd DR. (1995). Cancer Res., 55, 4264 ± 4267. Brentnall TA, Crispin DA, Bronner P, Cherian SP, Hueed M, Rabinovitch PS, Rubin CE, Haggitt RC and Boland R. (1996). Cancer Res., 56, 1237 ± 1240. Chung YJ, Song JM, Lee JY, Jung YT, Seo JE, Choi SW and Rhyu MG. (1996). Cancer Res., 56, 4662 ± 4665.

De Marchis L, Contegiacomo A, D'Amico C, Palmirotta R, Pizzi C, Ottini L, Mastranzo P, Figliolini M, Petrella G, Amanti C, Battista P, Bianco AR, Frati L, Cama A and Costantini RM. (1997). Clin. Cancer Res., 3, 241 ± 248. Duggan BD, Felix JC, Uderspach LI, Tourgeman D, Zheng J and Shibata D. (1994). J. Natl. Cancer Inst., 86, 1216 ± 1221. Feray C, Gigou M, Samuel D, Paradis V, Wilber J, David MF, Urdea M, Reynes M, Brechot C and Bismuth H. (1994). Hepatology, 20, 1137 ± 1143. Gonzalez-Zulueta M, Ruppert JM, Tokino K, Tsai YC, Spruck III CH, Miyao N, Nichols PW, Hermann GG, Horn T, Steven K, Summerhayes IC, Sidransky D and Jones PA. (1993). Cancer, 53, 5620 ± 5623. Hann HJ, Yanagisawa A, Kato Y, Park JG and Nakamura Y. (1993). Cancer Res., 53, 5087 ± 5089. Hoang JM, Cottu PH, Thuille B, Salmon RJ, Thomas G and Hamelin R. (1996). Cancer Res., 53, 300 ± 303. Honchel R, Halling KC, Schaid DJ, Pittelkow M and Thibodeau SN. (1994). Cancer Res., 54, 1159 ± 1163. Ionov Y, Peinado MA, Malkhosyan S, Shibata D and Perucho M. (1993). Nature, 363, 558 ± 561. Kaneko H, Horiike S, Taniwaki M and Misawa S. (1996). Leukemia, 10, 1696 ± 1699.


Kaplanski C, Srivatanakul P and Wild CP. (1997). Int. J. Cancer, 72, 248 ± 254. Lemoine A, Le Bricon T, Salvucci M, Azoulay S, Pham P, Raccuia JS, Bismuth H and Debuire B. (1997). Ann. Surg., 226, 43 ± 50. Lin HY, Wang XF, Ng-Eaton E, Weinberg RA and Lodish HF. (1992). Cell, 68, 775 ± 785. Lothe RA, Peltomaki P, Meling GI, Aaltonen L, NystrooÈmLahti M, PylkkaÈnen L, Heimdal K, Andersen TI, Moller P, Rognum TO, Fossa SD, Haldorsen T, Langmark F, Brogger A, de la Chapelle A and Borresen AL. (1993). Cancer Res., 53, 5849 ± 5852. Lu S, Akiyama Y, Nagasaki H, Saijo K and Yuasa Y. (1995). Biochem. Biophys. Res. Commun., 216, 452 ± 457. Meltzer SJ, Yin J, Manin B, Rhuy MG, Cottrell J, Hudson E and Redd BJ. (1994). Cancer Res., 54, 3379 ± 3382. Merlo A, Mabry M, Gabrielson E, Vollmer R, Baylin SB and Sidransky D. (1994). Cancer Res., 54, 2098 ± 2101. Myero LL, Parsons R, Kim SJ, Hedrick L, Cho KR, Orth K, Mathis M, Kinzler KW, Lutterbaugh J, Park K, Bang YJ, Lee HY, Park JG, Lynch HT, Roberts AB, Vogelstein B and Markowitz SD. (1995). Cancer Res., 55, 5545 ± 5547. Nagai H, Pineau P, Tiollais P, Buendia MA and Dejean A. (1997). Oncogene, 14, 2927 ± 2933. Parsons R, Myerho LL, Liu B, Wilson JKV, Markowitz SD, Kinzler KW and Vogelstein B. (1995). Cancer Res., 55, 5548 ± 5550. Paulson TG, Wright FA, Parker BA, Russack V and Wahl G. (1996). Cancer Res., 56, 4021 ± 4026. Polyak K. (1996). Biochim. Biophys. Acta., 1424, 185 ± 199. Rhyu MG, Park WS and Meltzer SJ. (1994). Oncogene, 9, 29 ± 32. Salvucci M, Lemoine A, Azoulay D, Sebagh M, Bismuth H, ReyneÁs M, May E and Debuire B. (1996). Oncogene, 13, 2681 ± 2685. Samuel D, Zigneco D, Reynes M, Feray C, Arulnaden JL, David MF, Gigou M, Bismuth A, Mathieu D, Gentilini P, Benhamou JP, Brechot C and Bismuth H. (1995). Hepatology, 21, 333 ± 339.

Shridhar V, Siegfried J, Hunt J, Del Mar Alonso M and Smith DL. (1994). Cancer Res., 54, 2084 ± 2087. Souza RF, Lei J, Yin J, Appel R, Zou TT, Zhou X, Wang S, Rhyu MG, Cymes K, Chan O, Park WS, Krasna MJ, Greenwald BD, Cottrell J, Abraham JM, Simms L, Leggett B, Young J, Harpaz N and Meltzer SJ. (1997). Gastroenterol., 112, 40 ± 45. Tamura G, Sakata K, Maesawa C, Suzuki Y, Terashima M, Satoh K, Sekiyama S, Suzuki A, Eda Y and Stodate R. (1995). Cancer Res., 55, 1933 ± 1936. Takenoshita S, Hagiware K, Gemma A, Nagashima M, Ryberg D, Lindstedt BA, Bennett WP, Haugen A and Harris CC. (1997). Carcinogenesis, 18, 1427 ± 1429. Thibodeau SN, Bren G and Schaid D. (1993). Science, 260, 816 ± 819. Togo G, Toda N, Kanai F, Kato N, Shiratori Y, Kishi K, Imazeki F, Makuuchi M and Omata M. (1996). Cancer Res., 56, 5620 ± 5623. Uchida T, Wada C, Wang C, Egawa S, Ohtani H and Koshiba K. (1994). Cancer Res., 54, 3682 ± 3685. Vincent F, Hagiwara K, Ke Y, Stoner GD, Demetrick DJ and Bennett WP. (1996). Biochem. Biophys. Res. Commun., 223, 561 ± 564. Wada C, Shionoya S, Fujino Y, Tokuhiro H, Akahoshi T, Uchida T and Ohtani H. (1994). Blood, 83, 3449 ± 3456. Wooster R, Cleton-Janser, Collins N, Mangion J, Cormelis RS, Cooper CS, Gusterson BA, Ponderr BAJ, Von Deimling A, Wiestler OD, Cornelisse CJ, Devilee P and Stratton MR. (1994). Nat. Genet., 6, 152 ± 156. Yeh SH, Chen PJ, Lai MY and Chen DS. (1996). Gastroenterol., 110, 184 ± 192. Zborowska E, Kinzler KW, Vogelstein B, Brattain M and Wilson JKV. (1995). Science, 268, 1336 ± 1338. Zhu J, Guo SZ, Beggs AH, Maruyama T, Santarius T, Dashner K, Olsen N, Wu JK and Black P. (1996). Oncogene, 12, 1417 ± 1423.

187