Spontaneous and iatrogenic spreading of liver-derived cells into peripheral blood of patients with primary liver cancer

Spontaneous and Iatrogenic Spreading of Liver-Derived Cells Into Peripheral Blood of Patients With Primary Liver Cancer MALEK LOUHA,1 KARINE POUSSIN,1 NATHALIE GANNE,2 HERVE ZYLBERBERG,3 BERTRAND NALPAS,1 JEROME NICOLET,4 FREDERIQUE CAPRON,6 OLIVIER SOUBRANE,5 CORINNE VONS,4 STANISLAS POL,3 MICHEL BEAUGRAND,2 PIERRE BERTHELOT,3 DOMINIQUE FRANCO,4 JEAN CLAUDE TRINCHET,2 CHRISTIAN BRE´CHOT,1,3 AND PATRIZIA PATERLINI1

The prognosis for patients with primary liver cancer (PLC) often depends on tumor recurrence and the development of extrahepatic metastases, particularly after liver transplantation. We have developed a sensitive test to detect both spontaneous circulation of tumor cells and the spread of liver cells due to chemoembolization and alcoholization. Reverse-transcription polymerase chain reaction was used to search for cells expressing a-fetoprotein (AFP) messenger RNA in the peripheral blood of 84 patients with PLC and 102 controls (55 patients with chronic hepatitis and/or cirrhosis, 10 patients with benign liver tumors or liver metastases from intestinal cancers, and 37 healthy individuals). By spiking the blood of healthy volunteers with HepG2 cells, we assessed the sensitivity limit: one HepG2 cell mixed with 107 leukocytes. All 102 controls tested negative. In contrast, 28 patients (33.3%) with PLC tested positive. Positivity for the test was significantly associated with portal thrombosis, tumor size, intravascular tumor emboli, serum AFP level, and extrahepatic metastases. Patients were followed up for a mean period of 39 { 51 weeks: the probability of developing extrahepatic metastases was significantly higher in positive than in negative patients. Eighteen negative patients with PLC were tested before, 1 hour after, and 24 hours after locoregional therapy: 9 tested positive either 1 or 24 hours after alcoholization or chemoembolization. In conclusion, we have developed a highly specific and sensitive test to detect circulating tumor cells in patients with PLC. This test is likely to be clinically useful in evaluating the risk of developing extrahepatic metastases and the possibility of iatrogenic spreading of liver-derived, possibly tumorous, cells. (HEPATOLOGY 1997;26:9981005.) Abbreviations: PLC, primary liver cancer; TACE, transarterial chemoembolization; RT, reverse transcription; mRNA, messenger RNA; PEI, percutaneous ethanol injection; HCC, hepatocellular carcinoma; AFP, a-fetoprotein; EDTA, ethylenediaminetetraacetic acid; PBS, phosphate-buffered saline; DEPC, diethylpyrocarbonate; RNasin, ribonuclease inhibitor; cDNA, complementary DNA; PCR, polymerase chain reaction; IGF2, insulin-like growth factor 2. From 1INSERM Unite´ 370, Institut Necker, Paris; 2Service de Gastroente´rologie, Hoˆpital Jean Verdier, Bondy; 3Service d’He´patologie, Hoˆpital Necker, Paris; 4Service de Chirurgie, Hoˆpital Antoine Be´cle`re, Clamart; 5Service de Chirurgie, Hoˆpital Cochin, Paris; and 6Service d’Anatomo-Pathologie, Hoˆpital Antoine Be´cle`re, Clamart, France. Received January 22, 1997; accepted May 20, 1997. Supported by grants from the Association pour la Recherche sur le Cancer, Communaute´ des Etats Europe´ens, Assistance Publique, Ligue Nationale contre le Cancer, and Institut National de la Sante´ et de la Recherche Me´dicale. Address reprint requests to: Patrizia Paterlini, M.D., Ph.D., INSERM Unit 370, Faculte´ Necker, 156 rue de Vaugirard, 75730 Paris Cedex 15, France. Fax: 33-140615581. Copyright q 1997 by the American Association for the Study of Liver Diseases. 0270-9139/97/2604-0031$3.00/0

Despite advances in the appraisal of the impact and role of etiologic factors, improvement in diagnostic tools and surgical and nonsurgical treatments, primary liver cancer (PLC), the eighth commonest tumor worldwide, still represents a therapeutic challenge.1-6 In Western countries, PLC generally develops on a background of cirrhosis; this hampers surgical and nonsurgical efforts to eradicate the disease because cirrhosis is a permanent source of tumorous nodules, and it complicates treatment with severe liver failure.7-11 Partial hepatic resection is limited by tumor characteristics such as size, multifocality, topography, and severity of associated cirrhosis. Only 10% of patients with PLC are considered acceptable for resection and, of these, the 3-year recurrence rate ranges from 60% to 80%, and the survival rate ranges from 40% to 70%.4,11-16 An alternative treatment for small tumors is alcoholization,17 but the advantages of this approach are currently being debated.18 The treatment of advanced liver cancers by transarterial chemoembolization (TACE), chemotherapy, and hormonal treatment with tamoxifen has provided discouraging results in previous published trials.2, 19 Despite this overall poor prognosis, a main feature of PLC is the rarity of extrahepatic metastases, which generally only become evident at a late tumor stage.20 Liver transplantation has, therefore, been proposed to cure both liver cancer and cirrhosis.1,21-24 Surgery-related mortality is higher with this approach, and intrahepatic tumor recurrence still occurs, especially for tumors larger than 3 cm.1,3,5,25-27 These observations have raised the hypothesis that tumorous liver cells can spread outside the liver either spontaneously or iatrogenically. In this context, recent reports on the feasibility of detecting circulating liver-derived cells through reverse-transcription (RT) and amplification of liver-specific messenger RNA (mRNA)28-31 have raised new hopes concerning prediction and early detection of tumor recurrence and extrahepatic metastases. This methodology might also allow a more rational choice between transplantation and nonsurgical treatments. However, some concerns still persist with the specificity of these tests, which showed positive results in patients with acute and chronic hepatitis and liver cirrhosis.28,29 In this study, we have developed and validated a sensitive and specific test to detect circulating liver cells derived from PLC. To assess its clinical impact, we have applied this assay to patients undergoing surgical and nonsurgical therapies and correlated results with clinical, pathological, and followup data. We have also looked for iatrogenic spread of liverderived cells in patients undergoing locoregional therapies,

998

AID

Hepa 0015

/

5p27$$$281

09-17-97 19:21:46

hepa

WBS: Hepatology

HEPATOLOGY Vol. 26, No. 4, 1997

LOUHA ET AL.

specifically transarterial chemoembolization and percutaneous ethanol injection (PEI).

Patients. We have studied 149 patients, including 84 with PLC

(group A), 55 with chronic hepatitis and/or cirrhosis of various etiologies (group B), and 10 with benign or secondary liver tumors (2 with adenoma, 5 with nodular focal hyperplasia, and 3 with liver metastases of intestinal cancers) (group C) (Table 1). Histopathological analyses of tumors were obtained in 65 patients after resection (n Å 39) or biopsy (n Å 26). Hepatocellular carcinoma (HCC) was found in 60 patients, cholangiocarcinoma in 3 patients, and fibrolamellar-type liver cancer in 2 patients. In 19 patients, the diagnosis was based on the presence of one or more nodules with arterial hypervascularity detected by CT scanning and/or magnetic resonance imaging in a cirrhotic liver, and/or on a serum a-fetoprotein (AFP) level of ú500 ng/mL. Thirty-eight patients with PLC underwent surgical resection of their tumor, and 1 underwent liver transplantation. Sixteen patients underwent PEI, 10 were treated by TACE, and 9 were treated with hormones. Ten patients remained untreated. Tumorous and nontumorous liver samples were collected from all surgically treated patients, quickly frozen in liquid nitrogen, and stored at 0807C. An extensive search for intravascular tumor emboli was made in 32 of 39 surgically treated patients. Extrahepatic metastases were checked by ultrasonography, computed tomography, and chest radiography. The 84 patients with PLC were followed up for a mean period of 39 { 51 weeks. The x2 test or Fisher’s Exact Test was used for statistical analysis of between-group frequencies. The cumulative overall and metastasis-free survival rates were calculated by the Kaplan-Meier method.32 Survival curves were compared by means of the log-rank test.33 Univariate analyses were made using BMDP statistical software (BMDP Statistical Software Inc., Los Angeles, CA). A P value of õ.05 was considered significant. Informed consent was obtained from each patient participating in the study. The study protocol conformed to the ethical guidelines of the 1975 Declaration of Helsinki. Isolation of Circulating Cells. Peripheral blood samples were obtained before treatment and before any transfusion. In 14 patients undergoing PEI and 4 undergoing TACE, peripheral blood samples were also obtained 1 and 24 hours after treatment. Peripheral blood samples (15 mL) were collected on buffered ethylenediaminetetraacetic acid (EDTA) in both PLC and control

TABLE 1. Clinical and Pathological Data in Patients With PLC (Group A) and Patients With Chronic Liver Disease Without Cancer (Group B) Group A

Group B

84 71/13 63 15 6

55 40/15 24 31 0

37* 26 5 7 9

4 38 11 0 2

67 17

55† 0

* Ten patients are also anti-HCV positive. † Five patients had high serum AFP levels (48 ng/mL; 81 ng/mL; 115 ng/mL; 161 ng/mL; and 137 ng/mL).

AID

Hepa 0015

/

5p27$$$281

TABLE 2. Oligonucleotide Sequences Used As Primers and Probes Code

Sequence

Position

37

PATIENTS AND METHODS

n M/F Cirrhosis Chronic Hepatitis Minimal lesions Etiology Alcohol Anti-HCV/ HBsAg/ Hemochromatosis Unknown Serum AFP õ500 ng/mL ú500 ng/mL

999

09-17-97 19:21:46

AFP gene AFP 1 (OF) AFP 2 (OR) AFP 3* (IF) AFP 4 (IR) b-actin gene35 b ACTIN1 (F) b ACTIN2 (R) Albumin gene36 ALB1* (F) ALB2 (R) IGF-2 gene38 IGF-2 1* (F) IGF-2 3 (R)

5* 5* 5* 5*

CTC TTC CAG AAA CTA GGA GAA 3* CTC TTC AGC AAA GCA GAC TT 3* GCT GAC ATT ATT ATC GGA CAC-3* AGC CTC AAG TTG TTC CTC TGT-3*

5* ACA ATG AGC TGC GTG TGG CT 3* 5* TCT CCT TAA TGT CAC GCA CGA 3*

Exon Exon Exon Exon

11 13 12 13

Exon 2 Exon 3

5* CTT GAA TGT GCT GAT GAC AGG-3* Exon 7 5* GCA AGT CAG CAG GCA TCT CAT C-3* Exon 8 5* CAGAGGAGTGTCCGGAGGA-3* 5* CAGCACTCCTCAACGATGCCA-3*

Exon 5 Exon 8

NOTE. IGF-2 primers specifically amplify the 6-kilobase IGF-2 mRNA, driven by the IGF-2 promoter P3, and expressed in human fetal liver. Abbreviations: OF, outer forward; OR, outer reverse; IF, inner forward; IR, inner reverse; F, forward; R, reverse. * Oligonucleotide sequences used as probes.

patients. Mononuclear and tumor cells were isolated from peripheral blood samples by density gradient (Ficoll Paque Plus; Pharmacia Biotech, Uppsala, Sweden) on preadapted tubes (Leucosep; Bernas Medical, Paris, France). Eleven milliliters of Ficoll was layered on the synthetic foam of Leucosep tubes and then rapidly spun down under the foam. Fifteen milliliters of blood diluted with 7.5 mL of NaCl solution (0.9%) was layered onto the foam. Tubes were spun at 800g for 10 minutes, and the supernatant (liquid phase over the foam) was collected and mixed with the 2 mL of NaCl solution (0.9%), which had been used to wash the tube walls. After centrifugation at 1,850g for 5 minutes, the supernatant was discarded, and cells were washed two times by resuspending them in 10 mL of 11 phosphate-buffered saline (PBS) solution and spinning at 1850g for 5 minutes. After the last centrifugation, cells were resuspended in 1.5 mL TRIzol B solution (Gibco BRL, Bethesda, MD) for RNA extraction and stored at 0807C. RNA Extraction, RT, and Polymerase Chain Reaction. RNA was extracted from tumorous and nontumorous liver samples as previously described.34 RNA was extracted from isolated circulating cells according to TRIzol-supplied instructions and resuspended in 3 mL of diethylpyrocarbonate (DEPC)-treated water for each 3 mL of peripheral blood processed. Six microliters of RNA solution was denatured with 10 U of ribonuclease inhibitor (RNasin) (RNAguard; Pharmacia Biotech) and 40 ng of hexamers random primer (Pharmacia Biotech) in 10 mL final volume at 657C for 5 minutes and then reverse-transcribed in 20 mL final volume, with an additional 10 U of RNasin, 11 buffer supplied with the enzyme, 40 mmol/L of four deoxynucleotides, and 20 U of Moloney murine leukemia virus reverse transcriptase (Life Technologies, Gibco BRL) at 377C for 60 minutes. To evaluate the amplifiability of the RNA, 3 mL of complementary DNA (cDNA) was amplified with b-actin specific primers.35 Ten microliters of cDNA, corresponding to RNA sequences extracted from 3 mL of blood, were then amplified in a final volume of 100 mL containing 10 mmol/L TRIS HCl, 50 mmol/ L KCl, 2 mmol/L MgCl2 , 0.01% gelatin, 250 mmol/L each of four deoxynucleotides (deoxyuridine triphosphate was used instead of deoxythymidine triphosphate, allowing uracyl N-glycosylase [Life Technologies, Gibco BRL] degradation of polymerase chain reaction [PCR] product in case of PCR product carryover), 10 pmol of primers AFP 1 and 2 (Table 2), 2.5 U of Taq polymerase (Perkin-Elmer Cetus, Emeryville, CA), and the PCR buffer for 40 cycles (947C, 1 min; 557C, 1 min; 727C, 1 min). Two microliters of the amplification product was reamplified with the inner primers AFP 3 and 4 with the same protocol for a further 40 cycles.

hepa

WBS: Hepatology

1000 LOUHA ET AL.

HEPATOLOGY October 1997

TABLE 3. AFP mRNA–Positive Circulating Cells in PLC and Control Patients

Group A B C D Total (B / C / D)

n

AFP / Cells

84 55 10 37 102

28 (33.3%) 0 0 0 0

NOTE. Group A, patients with PLC; group B, patients with chronic hepatitis and/or cirrhosis; group C, patients with benign or secondary liver tumors; and group D, healthy individuals.

Ten microliters of the final PCR product was analyzed by 2% agarose gel electrophoresis and Southern blot hybridization with an AFP-specific probe. For liver and tumor samples, cDNA was synthesized as described with hexamer random primers starting from 1 mg of tissue RNA. One fifth of the cDNA, corresponding to 200 ng of RNA, was serially diluted (1:10) five times and then amplified for 40 cycles. Choice of Primers. Three different sets of primers were tested, amplifying albumin,36 AFP,37 and fetal insulin-like growth factor 2 (IGF-2)38specific mRNAs (Table 2). Amplification was first carried out on RNA extracted from the reference cell line HepG2. To test the specificity of the primers and to evaluate the possibility of illegitimate transcription39 of the three genes in peripheral blood cells, RNA extracted from peripheral blood cells obtained from 10 healthy volunteers was analyzed. To test whether the AFP gene is expressed in a high proportion of tumorous tissues, we analyzed 55 tumorous tissues by AFP RT-PCR. Cell Culture Conditions. The HepG2 cell line, derived from a human HCC, was used to establish the sensitivity of the method. HepG2 cells were cultured in Dulbecco’s minimal essential medium (Life Technologies, Gibco BRL) supplemented with 10% fetal bovine serum (Life Technologies, Gibco BRL). Medium was changed every second day, and cells were subcultured after 7 days and plated at a dilution of 1:10. Cell counting was performed in a hemocytometer. To ensure the reliability of this counting and the precision of the sensitivity test, several cell dilutions were counted by different operators and compared. Sensitivity Test. To define the sensitivity of our test, HepG2 cells were precisely counted, and serial dilutions (from 1,000 to 5 cells) were mixed with peripheral blood containing 107 leukocytes/mL obtained from a healthy volunteer. Three different methods to isolate circulating cells were compared: Ficoll-Leucosep isolation as described, lysis of erythrocytes by DEPC-treated water,40 and NH4Cl.41 Two different methods of RNA extraction were compared: Chomiczynski-Sacchi acid/guanidium thiocyanate/phenol/chloroform extraction42 and TRIzol. Three different PCR protocols were compared: simple PCR (one

single round of 40 cycles), nested PCR performed in one tube,43 and nested PCR performed in separate tubes. Controls of Specificity and Sensitivity. To test the specificity of the assay, 37 healthy individuals (blood donors and volunteers) were tested as controls (group D, Table 3). To determine the absence of PCR or sample contamination, five buffers without samples and a blood donor sample were included at the extraction step and run to the end of each test. Finally, PCR products were transferred to a nylon membrane (Gene Screen Plus Nylon membrane; New England Nuclear Corp., Boston, MA) and hybridized with a specific 32 P-labeled probe. Sensitivity was tested in each assay by including two positive control samples corresponding to one HepG2 per 106 and one HepG2 cell per 107 leukocytes. Precautions to Avoid Carry-Over of PCR Products. Precautions to avoid PCR product carryover have been previously described.44 Buffer and reaction mix preparation, RNA extraction, and PCR product loading on agarose gel were performed using barrier tips in separate rooms located on different floors of the same building. In each of these different rooms, specific coats were used and specific rules of work were observed; in particular, entry to the mix room and to the extraction rooms was forbidden to persons who had entered the PCR products loading room on the same day. Extensive utilization of UV light and 0.1 mol/L HCl solution was applied after each test to clean the bench and instruments. RESULTS Choice of the Marker Gene: Analysis of Peripheral Blood From Normal Volunteers. We first evaluated three different candi-

date genes for which mRNA detection should allow identification of circulating liver-derived cells. Albumin and AFP are specifically expressed at different stages of liver cell differentiation.45 IGF-2 is a growth factor whose fetal mRNA pattern is reexpressed during PLC development.46 We looked first for albumin, AFP, and fetal IGF-2 specific mRNAs in peripheral blood from 10 healthy volunteers. Five of 10 samples tested positive for albumin, 9 of 10 were positive for fetal IGF-2, and none were positive for AFP mRNA (Fig. 1). We then performed the AFP test in 27 additional samples obtained from healthy volunteers. These 27 samples yielded negative results (group D, Table 3) and confirmed the absence of illegitimate AFP gene transcription in peripheral blood leukocytes. Detection of AFP mRNA in Tumorous and Nontumorous Liver Tissues. The second step in the choice of the marker gene

was to ensure that it is expressed and is detectable by RTPCR in 100% of PLCs. The AFP test was performed on a panel of 55 RNAs extracted from stock human PLC. AFP mRNA was clearly expressed in all tumorous tissues (data not shown). This result, which is in agreement with a recent report from Niwa et al.,47 contrasts with previously reported

FIG. 1. Choice of marker gene. Total RNA was extracted from peripheral blood of healthy subjects, reverse-transcribed, and amplified with (A) bactin primers to determine its amplifiability, (B) albumin, (C) fetal IGF-2, and (D) AFP primers. Lane 1: HepG2 cells (positive control); lane 2: negative control (only buffer without RNA); and lanes 3-6: peripheral blood from healthy subjects. Agarose gel electrophoresis and ethidium bromide staining of PCR products. The size of the amplification products is indicated.

AID

Hepa 0015

/

5p27$$$281

09-17-97 19:21:46

hepa

WBS: Hepatology

HEPATOLOGY Vol. 26, No. 4, 1997

LOUHA ET AL.

1001

FIG. 2. Sensitivity test. HepG2 cells were carefully counted, and serial dilutions (lane A, 1,000 cells; lane B, 100 cells; lane C, 10 cells; and lane D, 5 cells) were mixed with 10 mL of blood containing 108 leukocytes from a healthy volunteer. Tumor cell isolation, RNA extraction, RT, and AFP PCR were performed on the different mixes. Amplification products were then analyzed by gel electrophoresis, stained with ethidium bromide (upper section), and hybridized by Southern blot with a 32P-labeled AFP-specific probe (lower section). Two independent tests are shown for each dilution. A positive response was found for dilutions corresponding to (lane A) one HepG2 cell per 105 leukocytes, (lane B) one HepG2 cell per 106 leukocytes, and (lane C) one HepG2 cell per 107 leukocytes. (Lane D) The dilution corresponding to one HepG2 cell per 2 1 107 leukocytes produced a negative result.

data showing that protein expression of the AFP gene is only detectable by immunohistochemistry in about 35% of human HCC tissues.48 The high sensitivity of specific RNA detection by RT-PCR, compared with the sensitivity of protein detection by immunohistochemistry, accounts for this discrepancy in the evaluation of AFP gene expression. AFP-specific RNAs were also detected in the two tested cholangiocarcinoma tumorous tissues and in one tested fibrolamellar-type liver tumor. These three samples were obtained from patients analyzed in the peripheral blood for the presence of AFP RNA expressing circulating cells (1 patient with cholangiocarcinoma tested positive). The results show that malignant cells from cholangiocarcinonoma and fibrolamellar-type liver tumors can express AFP-specific RNAs, a result consistent with the liver developmental pattern. Sensitivity of the Test. As reported in Patients and Methods, we tested different techniques to isolate circulating tumor cells, to extract mRNA, and to perform RT-PCR in order to obtain the greatest sensitivity. Serial dilutions of HepG2 cells

(from 1,000 to 5 cells) were mixed with 10 mL of human blood containing 107 leukocytes/mL. By using our assay, we were able to consistently detect, in three independently performed tests, one HepG2 cell per 107 leukocytes (Fig. 2). In these experiments, the strong signal obtained when testing one HepG2 cell per 107 leukocytes was followed by an abrupt disappearence of RT-PCR products when the following dilution point (1 tumor cell per 2 1 107 leukocytes) was tested (Fig. 2). This pattern is due to the highly sensitive nested PCR protocol, using 40 cycles twice; in these conditions, when reaching the sensitivity limit, an all-or-nothing result can be shown. Detection of AFP mRNA in the Peripheral Blood of Patients With Chronic Hepatitis and/or Cirrhosis With Benign or Secondary Liver Tumors. Blind analyses of blood samples from 55 patients

with chronic hepatitis without PLC (group B), 7 patients with benign liver tumors, and 3 patients with liver metastases of gastrointestinal tumors (group C) were made. None of these samples tested positive. (Table 3).

FIG. 3. Detection of AFP mRNA expressing cells in the peripheral blood circulation. Lanes 1, 3, and 4, peripheral blood from patients with liver cancer. Lane 2, peripheral blood from a patient with liver cirrhosis without PLC. Lane 5, negative control (buffer without RNA). Lanes 6 and 7, positive controls of sensitivity: one HepG2 cell per 107 leukocytes (lane 6) and one HepG2 cell per 106 leukocytes (lane 7). M, molecular size marker (100-base pair ladder). (Upper panel) Agarose gel electrophoresis and ethidium bromide staining; (lower panel) Southern blot hybridization with a 32P-labeled AFPspecific probe of PCR products.

AID

Hepa 0015

/

5p27$$$281

09-17-97 19:21:46

hepa

WBS: Hepatology

1002 LOUHA ET AL.

HEPATOLOGY October 1997

TABLE 4. Circulating Tumor Cells According to Tumor Characteristics and Serum AFP Level

Edmonson’s tumor grade (n Å 60) I-II III-IV Intravascular tumor emboli (n Å 32) Absent Present Tumor nodules (n Å 84) 1 2 or more Tumor size (n Å 84) õ3 cm ú3 cm Portal thrombosis (n Å 70) Absent Present Serum AFP (n Å 84) õ500 ng/mL ú500 ng/mL

n

AFP/ Cells

33 27

11 14

NS

13 19

0 9

õ.02

52 32

15 13

NS

23 61

3 25

õ.05

55 15

10 8

õ.02

66 18

16 12

P

õ.01

NOTE. Number of available samples for statistical analysis is indicated in parentheses. NS, not statistically significant.

Detection of AFP mRNA Positive Cells in the Peripheral Blood of Patients With PLC. AFP-positive circulating cells were de-

tected in 28 of the 84 patients (33.3%) with PLC (group A). None of the controls, including 37 healthy individuals (group D), tested positive (Table 3; P õ .0001). This result indicates that our test is highly specific for detection of circulating tumorous liver cells (Fig. 3). As shown in Table 4, the detection of circulating tumorous liver cells was associated with the size of tumorous nodules (P õ .05), the presence of intravascular tumor emboli (P õ .02), and portal thrombosis (P õ .02). In contrast, no association was shown with the number of tumorous nodules or Edmondson’s grade of differentiation. The presence of circulating tumor cells correlated with the serum concentration of AFP (P õ .01); however, a positive result was also found in patients with low serum levels of AFP (Fig. 4).

FIG. 4. Distribution of positive and negative results for the detection of circulating liver-derived tumor cells in 84 patients with PLC according to serum AFP level. (●) mRNA AFP /; and (s) mRNA AFP 0.

AID

Hepa 0015

/

5p27$$$281

TABLE 5. AFP mRNA–Positive Circulating Cells and Extrahepatic Metastases

09-17-97 19:21:46

Extrahepatic metastases Absent Present*

n

AFP / Cells

P

68 16†

18 10‡

õ.02

* Present at the time of the first test or developed during follow-up. † Five patients with extrahepatic metastases at the time of the first test. ‡ Four patients with extrahepatic metastases at the time of the first test.

Four of five patients with extrahepatic metastases at the time of sample collection tested positive. The subject who showed a negative result had a very high level of serum AFP (57,400 ng/mL); nevertheless, this patient consistently tested negative when samples collected at two different times were tested twice in independent assays. Eleven patients developed extrahepatic metastases during follow-up, 6 of them having tested positive. Overall, the detection of circulating tumorous liver cells was associated with the presence or development of extrahepatic metastases (P õ .02) (Table 5). This association was also confirmed by the rate of development of metastases according to time; namely, this rate was significantly higher in patients with positive test results when compared with those with negative results (Fig. 5) (P õ .0067). This result indicates that the probability of the absence of metastases decreases more rapidly in patients who tested positive than in those who had negative results for the test. Iatrogenic Spreading of Liver-Derived Cells in Patients Undergoing Locoregional Therapy. The hypothesis that locoregional

therapy may, through its direct impact on the liver, spread liver cells into the peripheral blood circulation was tested. Eighteen negative patients with PLC were tested before, 1 hour after, and 24 hours after locoregional therapy (14 after PEI and 4 after TACE). Nine patients tested positive (6 after PEI and 3 after TACE) either 1 hour or 24 hours after locoregional therapy (Table 6). In an attempt to determine whether our test can specifically recognize the circulation of tumor cells after iatrogenic intervention in the liver and the possible spread of nontumor as well as tumor cells, we estimated the amount of AFP transcripts in tumorous and nontumorous liver tissues. For

FIG. 5. Actuarial rate of development of metastases in the 84 patients with PLC. The probability of the absence of extrahepatic metastases decreases significantly more rapidly in the group of 28 patients who tested positive than in the 56 patients who tested negative (P õ .0067).

hepa

WBS: Hepatology

HEPATOLOGY Vol. 26, No. 4, 1997

LOUHA ET AL.

TABLE 6. AFP mRNA–Positive Circulating Cells in Patients With PLC Tested Before and Immediately After Locoregional Therapy AFP-Positive Cells n

Before

1 h After

24 h After

9* 6† 3‡

0 0 0

0 / 0

0 0 /

* Eight patients treated by PEI and 1 by TACE. † Five patients treated by PEI and 1 by TACE. ‡ One patient treated by PEI and 2 by TACE.

this purpose, we performed the AFP test on serial dilutions of total RNA (from 200 ng to 0.02 ng) extracted from five tumorous and five nontumorous liver tissues. We found a variable amount of AFP mRNA in PLC tissue with a positive result up to 0.2 ng in three tumors and up to 2 ng in two tumorous tissues. However, nontumorous tissues also showed high amounts of AFP transcripts, with a positive result up to 2 ng in four cirrhotic and one noncirrhotic (chronic hepatitis) tissue samples (Fig 6). This result indicates that the amount of AFP mRNA in tumor and nontumor cells is clearly a variable and unpredictable parameter in vivo. In this context, it is difficult to define the ability of our test to distinguish between tumor and nontumor cells. Namely, after locoregional therapy, the positivity of the test will depend both on the number of liver-derived cells spread into the blood circulation and on the number of AFP transcripts expressed per circulating cell.

1003

tients who tested positive than in the 56 patients who tested negative. These results indicate that the circulation of tumorous liver cells correlates with the following: 1) histological and biological features of tumor invasion; and 2) the presence and risk of developing extrahepatic metastases. From a clinical point of view, this correlation is relevant and can provide useful information for the choice of treatment. However, from a biological point of view, our results raise questions about the mechanisms regulating the circulation of tumor cells, their molecular phenotype, and the development of more performant tests to detect them. One might argue that the strong correlation of our test results with serum levels of AFP might restrain the usefulness of the assay to patients with high serum AFP levels. Previously published results of AFP RNA expression in tumorous liver tissue showed a correlation between amount of AFP transcripts and AFP serum level. Using a competitive RTPCR, Niwa et al.47 have shown that the number of AFP mRNA copies per milligram of tissue ranges from 106 to 1010 in tumorous tissues, 105 to 107 in nontumorous tissues, and 105 in normal liver tissue, and is correlated to the serum AFP level. Our test also gave positive results in patients with liver cancer and normal serum AFP levels, whereas it remained negative in patients with liver cirrhosis (without liver cancer) and high serum AFP levels. This raises the hypothesis that, in some cases, only the tumor cells that become able to circulate express enough AFP transcripts to be detected. One patient consistently tested negative despite the pres-

DISCUSSION

We have developed a PCR-based, highly specific, and sensitive test to identify circulating liver tumor cells through the expression of liver-specific AFP mRNA. Albumin and fetal IGF-2 mRNA were excluded as markers of circulating liver-derived cells on evidence of illegitimate transcription39 in leukocytes from healthy subjects. The high limit of sensitivity of the assay (one AFP mRNA expressing cell per 107 leukocytes) was consistently obtained in three independent tests. This assay was used to test 84 patients with liver cancer, 39 of whom had undergone tumor resection, 35 who had undergone nonsurgical therapy (PEI or TACE or hormonal therapy), and 10 who had undergone no treatment. Circulating tumor cells were found in 33% of patients with PLC. All 102 controls tested (including patients with chronic hepatitis without cancer, benign liver tumors, liver metastases of gastrointestinal, nonhepatic tumors, and healthy individuals) tested negative. These data clearly indicate that a positive result specifically identifies circulating tumor-derived liver cells in patients tested before any locoregional and surgical treatment. Circulation of liver-derived tumor cells significantly correlates with the size of tumorous nodules, the presence of intravascular tumor emboli, portal thrombosis, and serum AFP level. Moreover, a statistically significant correlation was found between the presence of extrahepatic metastases at the time of the test, or development of extrahepatic metastases during follow-up, and positivity of the test. In the 4-year actuarial analysis of development of extrahepatic metastases, the probability of the appearance of extrahepatic metastases increased significantly more rapidly in the 28 pa-

AID

Hepa 0015

/

5p27$$$281

09-17-97 19:21:46

FIG. 6. Estimation of the amount of AFP transcripts in tumorous and nontumorous liver tissues. The AFP test was performed on serial dilutions of total RNA (lanes 1-5) extracted from (A and B) two tumorous samples, (C) one cirrhotic sample, and (chronic hepatitis; D) one noncirrhotic sample. Lane 1, 200 ng; lane 2, 20 ng; lane 3, 2 ng; lane 4, 0.2 ng; and lane 5, 0.02 ng. Agarose gel electrophoresis and ethidium bromide staining of PCR products.

hepa

WBS: Hepatology

1004 LOUHA ET AL.

HEPATOLOGY October 1997

ence of extrahepatic metastases and high serum AFP levels. Several explanations are plausible: 1) the circulation of highly undifferentiated tumor cells that would not express AFP mRNA; 2) the circulation of tumor cells expressing mutated AFP mRNAs; and 3) the absence or intermittent circulation of tumor cells. In any case, this point illustrates the limits of a test based on a single marker. Following the first report of Carr et al.49 on the possibility of detecting circulating liver-derived cells through RT and amplification of a liver-specific mRNA, contrasting results have been reported in the literature.28-31 It must be taken into account that apparently similar techniques may be very different in terms of sensitivity and specificity and, if both these parameters are not tested and proved to be the same, results are not expected to be comparable. It may be that the issue of illegitimate transcription includes several distinct points: 1) a real illegitimate transcription detectable by RTPCR; 2) different sensitivity of the whole technique applied to detect circulating cell transcripts; and 3) variable specificity of this technique, including specificity of selected primers and control of false-positive results due to carryover of PCR products. Like other investigators,50,51 and in contrast with some reports,28,30 we found illegitimate transcription of the albumin gene in peripheral blood leukocytes. This discrepancy may be related to the primer sets used30 as well as to the absence of a circulating cell isolation step before RNA extraction,28 which can result in a lower sensitivity of the whole test.51-53 Other investigators29 have found positive results using AFP primers in 52% of patients with PLC, 15% of patients with cirrhosis, and 12% of patients with chronic hepatitis. The sensitivity and specificity controls of that study do not seem to be the same as in this research and make comparisons difficult. The specificity of our PCR test and the ability to ascertain false-positives by cleaning up the cDNA with uracil N-glycosylase treatment allow us to obtain consistent results. Finally, we have applied our test to investigate whether locoregional therapies can spread tumor cells into peripheral blood. In 9 of 18 patients who were negative before therapy, we found circulating AFP mRNA-positive cells either 1 hour or 24 hours after locoregional therapy. However, because nontumorous liver cells also may express AFP mRNA detectable by RT-PCR, our results only indicate that locoregional therapies spread liver-derived cells into the peripheral blood circulation without ascertaining their tumor origin. Clearly, more work needs to be done on the development of tests specifically detecting mRNAs selectively expressed in tumor cells and absent or different in nontumor cells. In conclusion, we have developed a specific, reproducible, and sensitive test to show the circulation of tumor cells in patients with liver cancer. The positivity of the test is significantly associated with the presence and the risk of developing extrahepatic metastases. However, the sensitivity of the test in vivo seems to vary with the number of circulating tumor cells and the amount of AFP transcripts per cell. Finally, having shown that liver cells may be spread into peripheral blood upon locoregional therapies, the issue of their potential clinical consequences is raised. Taken together, our results indicate the clinical relevance and the limits of a well-defined test to detect circulating tumorous liver cells and offer the promise of the development and application of these tests in clinical oncology.

AID

Hepa 0015

/

5p27$$$281

09-17-97 19:21:46

Acknowledgment: The authors thank Dr. de Pagter-Holthuizen for help in defining primers specifically amplifying fetal insulin-like growth factor 2 messenger RNAs. REFERENCES 1. Mazzaferro V, Regalia E, Doci R, Andreola S, Pulvirenti A, Bozzetti F, Montalto F, et al. Liver transplantation for the treatment of small hepatocellular carcinomas in patients with cirrhosis. N Engl J Med 1996; 334:693-699. 2. Trinchet J-C, Rached A, Beaugrand M, Mathieu D, Chevret S, Chastang C. A comparison of lipiodol chemoembolization and conservative treatment for unresectable hepatocellular carcinoma. N Engl J Med 1995; 332:1256-1261. 3. Ringe B, Pichlmayr R, Wittekind C, Tusch G. Surgical treatment of hepatocellular carcinoma: experience with liver resection and transplantation in 198 patients. World J Surg 1991;15:270-285. 4. Zhou X-D, Tang Z-Y, Yu Y-Q, Yang B-H, Lin Z-Y, Lu J-Z, Ma Z-C. Long-term results of surgery for small primary liver cancer in 514 adults. J Cancer Res Clin Oncol 1996;122:59-62. 5. Bismuth H, Chiche L, Adam R, Castaing D, Diamond T, Dennison A. Liver resection versus transplantation for hepatocellular carcinoma in cirrhotic patients. Ann Surg 1993;218:145-151. 6. Marcos-Alvarez A, Jenkins R, Washburn W, Lewis D, Stuart K, Gordon F, Kane R, et al. Multimodality treatment of hepatocelluar carcinoma in Hepatobiliary Speciality Center. Arch Surg 1996;131:292-298. 7. Colombo M. Hepatocellular carcinoma. J Hepatol 1992;15:225-236. 8. Kew MC. Role of cirrhosis in hepatocarcinogenesis. In: Bannasch P, Keppler D, Weber G, eds. Liver Cell Carcinoma. Dordrecht, The Netherlands: Kluwer Academic, 1989:37-45. 9. Zaman S, Johnson P, Williams R. Silent cirrhosis in patients with hepatocellular carcinoma. Cancer 1990;65:1607-1610. 10. Colombo M, De Franchis K, Del Ninno E, Sangiovanni A, De Fazio C, Tommasini M, Donato MF, et al. Hepatocellular carcinoma in Italian patients with cirrhosis. N Engl J Med 1991;325:675-680. 11. Adachi E, Maeda T, Matsumata T, Shirabe K, Kinukawa N, Sugimachi K, Tsuneyoshi M. Risk factors for intrahepatic recurrence in human small hepatocellular carcinoma. Gastroenterology 1995;108:768-775. 12. Dusheiko G, Hobbs K, Dick R, Burrough S. Treatment of small hepatocellular carcinomas. Lancet 1992;340:285-288. 13. Gozzetti G, Belli L, Capussotti L Di Carlo, V, Gennari L, Maffei Faccioli A, Mazziotti A, Spina P. Liver resection for hepatocellular carcinoma in cirrhotic patients. Ital J Gastroenterol 1992;24:105-110. 14. Tsuzuki T, Sugioka A, Ueda M, Iida S, Kanai T, Yoshii H, Nakayasu K. Hepatic resection for hepatocellular carcinoma. Surgery 1990;107:511520. 15. Franco D, Capussotti L, Smadja C, Bouzari H, Meakins J, Kemeny F, Grange D, et al. Resection of hepatocellular carcinoma. Results in 72 European patients with cirrhosis. Gastroenterology 1990;98:733-738. 16. Belghiti J, Panis O, Benhamou P, Fekete F. Intrahepatic recurrence after resection of hepatocellular carcinoma. Ann Surg 1991;214:114-117. 17. Vilana R, Bruix J, Bru C, Ayuso C, Sole´ M, Rode`s J. Tumor size determines the efficacy of percutaneous ethanol injection for the treatment of small hepatocellular carcinoma. HEPATOLOGY 1992;16:353-357. 18. Livraghi T, Buscarini L, Cottone M, Mazziotti A, Morabito A, Torzilli G. No treatment, resection and ethanol injection in hepatocellular carcinoma: a retrospective analysis of survival in 391 patients with cirrhosis. J Hepatol 1995;22:522-526. 19. Okuda K, Ohtsuki T, Obata H, Tomimatsu M, Okazaki N, Hasegawa H, Nakajima Y, et al. Natural history of hepatocellular carcinoma and prognosis in relation to treatment. Study of 850 patients. Cancer 1985; 56:918-928. 20. Yuki K, Hirohashi S, Sakamoto M, Kanai T, Shimosato Y. Growth and spread of hepatocellular cacinoma. Cancer 1990;66:2174-2179. 21. Mazzaferro V, Rondinara G, Rossi G, Regalia E, De Carlis L, Caccamo L, Doci R, et al. Milan Multicenter experience in liver transplantation for hepatocellular carcinoma. Transplant Proc 1994;26:3557-3560. 22. Selby R, Kadry Z, Carr B, Tzakis A, Madariaga J, Iwatsuki S. Liver transplantation for hepatocellular carcinoma. World J Surg 1995;19: 53-58. 23. Penn I. Hepatic transplantation for primary and metastatic cancers of the liver. Surgery 1991;110:726-735. 24. Calne S, Yamanoi A, Oura S, Kawamura M. Liver transplantation for hepatocarcinoma. Surg Today 1993;23:1-3. 25. Iwatsuki S, Starzl T, Sheahan D, Yokoyama I, Demetris A, Todo S, Tzakis A, et al. Hepatic resection versus transplantation for hepatocellular carcinoma. Ann Surg 1991;214:221-229.

hepa

WBS: Hepatology

HEPATOLOGY Vol. 26, No. 4, 1997

LOUHA ET AL.

26. Tan K, Rela M, Ryder S, Rizzi P, Karani J, Portmann B, Heaton N, et al. Experience of orthotopic liver transplantation and hepatic resection for hepatocellular carcinoma of less than 8 cm in patients with cirrhosis. Br J Surg 1995:253-256. 27. Schwartz M. Primary hepatocellular carcinoma: transplant versus resection. Semin Liv Disease 1994;14:135-139. 28. Hillaire S, Barbu V, Boucher E, Moukhtar M, Poupon R. Albumin messenger RNA as a marker of circulating hepatocytes in hepatocellular carcinoma. Gastroenterology 1994;106:239-242. 29. Matsumura M, Niwa Y, Kato N, Komatsu Y, Shiina S, Kawabe T, Kawase T, et al. Detection of alpha-fetoprotein mRNA, an indicator of hematogenous spreading hepatocellular carcinoma, in the circulation: a possible predictor of metastatic hepatocellular carcinoma. HEPATOLOGY 1994; 20:1418-1425. 30. Kar S and Carr B. Detection of liver cells in peripheral blood of patients with advanced-stage hepatocellular carcinoma. HEPATOLOGY 1995;21: 403-407. 31. Komeda T, Fukuda Y, Sando T, Kita R, Furukawa M, Nishida N, Amenomori M, et al. Sensitive detection of circulating hepatocellular carcinoma cells in peripheral venous blood. Cancer 1995;75:2214-2219. 32. Kaplan E and Meier P. Nonparametric estimation from incomplete observations. J Am Stat Assoc 1958:53: 33. Peto R and Peto J. Asymptotically efficient rank invariant test procedures. J R Stat Soc 1972;135:185-206. 34. Paterlini P, Flejou J-F, De Mitri M, Pisi E, Franco D, Be´rchot C. Structure and expression of the cyclin A gene in human primary liver cancer. Correlation with flow cytometric parameters. J Hepatol 1995;23:47-52. 35. Kakajima-Iijima S, Hamada H, Reddy P, Kakunaga T. Molecular structure of the human cytoplasmic b-actin gene: interspecies homology of sequences in the introns. Proc Natl Acad Sci U S A 1985;82:6133-6137. 36. Minghetti P, Ruffner D, Kuang W-J, Dennison O, Hawkins J Beattie, WG, Dugaiczyk A. Molecular structure of the human albumin gene is revealed by nucleotide sequence within q11-22 of chromosome 4. J Biol Chem 1986;261:6747-6757. 37. Morinaga T, Sakai M, Wegmann T, Tamaoki T. Primary structures of human a-fetoprotein and its mRNA. Proc Natl Acad Sci U S A 1983; 80:4604-4608. 38. de Pagter-Holthuizen P, Jansen M, van der Kammen R, van Schaik F, Sussenbach J. Differential expression of the human insulin-like growth factor II gene. Characterization of the IGF-II mRNAs and an mRNA encoding a putative IGF-II associated protein. Biochem Biophys Acta 1988;950:282-295. 39. Chelly J, Concordet J, Kaplan J-C, Kahn A. Illegitimate transcription: transcription of any gene in any cell type. Proc Natl Acad Sci U S A 1989;86:2617-2621. 40. Jaakkola S, Vornanen T, Leinonen J, Rannikko S, Stenman U. Detection

AID

Hepa 0015

/

5p27$$$281

09-17-97 19:21:46

41. 42. 43. 44.

45. 46.

47.

48.

49. 50. 51. 52.

53.

1005

of prostatic cells in peripheral blood: correlation with serum concentrations of prostate-specific antigen. Clin Chem 1995;412:182-186. Datta Y, Adams P, Drobyski W, Ethier S, Terry V, Roth M. Sensitive detection of occult breast cancer by the reverse-transcriptase polymerase chain reaction. J Clin Oncol 1994;12:475-482. Chomczynski P and Sacchi N. Single step method of RNA isolation by acid guanidium thiocyanate-phenol-chloroform extraction. Anal Biochem 1987;162:156-159. Fe´ray C, Samuel D, Thiers V, Gigou M, Pichon F, Bismuth A, Reynes AM, et al. Reinfection of liver graft by hepatitis C virus (HCV) after liver transplantation. J Clin Invest 1992;89:1361-1365. Paterlini P, Poussin K, Kew M, Franco D and Be´rchot C. Selective accumulation of the X transcript of hepatitis B virus in patients negative for hepatitis B surface antigen with hepatocellular carcinoma. HEPATOLOGY 1995;21:313-321. Camper S, Godbout R, Tilghman S. The developmental regulation of albumin and a-fetoprotein gene expression. Prog Nucleic Acid Res Mol Biol 1989;36:131-143. Cariani E, Lasserre C, Kemeny F, Franco D, Be´rchot C. Expression of insulin-like growth factor II, alpha-fetoprotein and hepatitis B virus transcripts in human primary liver cancer. HEPATOLOGY 1991;13:644649. Niwa Y, Matsumura M, Shiratori Y, Imamura M, Kato N, Shiina S, Okudaira T, et al. Quantitation of a-fetoprotein and albumin messenger RNAs in human hepatocellular carcinoma. HEPATOLOGY 1996;23:13841392. Heyward W, Bender T, Lanier A, Francis D, McMahon B, Maynard J. Serological markers of hepatitis B virus and alphafetoprotein levels preceding primary hepatocellular carcinoma in Alaskan Eskimos. Lancet 1982;2:889-891. Carr B and Kar S. An assay for hepatoma micrometastasis: albumin gene expression in peripheral blood from patients with advanced hepatocellular carcinoma [Abstract]. HEPATOLOGY 1991;14:242A. Chou H-C, Sheu J-C, Huang G-T, Wang J-T, Chen D-S. Albumin messager RNA is not specific for circulating hepatoma cells. Gastroenterology 1994;107:630-631. Mu¨ller C, Petermann D, Pfeffel F, Oesterreicher C, Fu¨gger R. Lack of specificity of albumin-mRNApositive cells as a marker of circulating hepatoma cells. HEPATOLOGY 1997;25:896-899. Paterlini P, Lallemand-Le Coeur S, Lallemand M, M’Pele´ P, Dazza M, Terre S, Moncany M, et al. Polymerase chain reaction for studies of mother to child transmission of HIV1 in Africa. J Med Virol 1990;30: 53-57. De Franchi R, Cross N, Foulkes N and Cox T. A potent inhibitor of Taq polymerase copurifies with human genomic DNA. Nucleic Acids Res 1988;16:10355.

hepa

WBS: Hepatology