[CANCER RESEARCH 56. 3004-3009. July I. 1996)

Expression of Vascular Permeability Factor/Vascular Endothelial Growth Factor inHuman Hepatocellular Carcinoma1

Kunio Suzuki, Norio Hayashi,2 Yasuhide Miyamoto, Masato Yamamoto, Kazuyoshi Ohkawa, Yoshiki Ito,

Yutaka Sasaki, Yumiko Yamaguchi, Hideyuki Nakase, Katsuhisa Noda, Norihiro Enomoto, Keiichi Arai,Yukinori \ ainada, Harumasa Yoshihara, Takashi Tujimura, Kiyoshi Kawano, Kiyoshi Yoshikawa, andTakenobu KamadaDepartment of Gastroenterologe, Osaka Rosai Hospital, 1179-3 Nagasone-cho, Sakai. Osaka 591 ¡K.S., Y. Yamag., H. N., K. N., N. £.,K. A.. Y. Yamail.. H. Y.]; First Departmentof Medicine. Osaka University School of Medicine. 2-2 Yamadaoka. Suita. Osaka 565 [N. H.. Y. M.. M. Y.. K. O.. Y. I.. Y. S., T. K.]; and Departments of Pathology ¡T.T.. K. K.Iand Surgery ¡K.Y.I. Osaka Rosai Hospital. 1179-3 Nagasone-cho. Sakai. Osaka 591. Japan

ABSTRACT

Vascular permeability factor (VPF)/vascular endothelial growth factor(VEGF) is unique in its ability to promote vascular permeability andendothelial cell growth, and its role in tumor development has receivedconsiderable attention. In this report, we describe the elevation of VIM-/

VEGF transcript expression in human hepatocellular carcinoma. Surgicalsamples of 23 patients with hepatocellular carcinoma were studied usingreverse transcription-PCR analysis. The oligonucleotide primers were

designed to amplify all four known splicing variants that could be expressed in the samples studied. Sixteen cases showed VPF/VEGF transcript expression in the tumor (16/23, 69.6%), whereas only 9 of the 23patients showed it in the corresponding nontumoral part. There was nodifference between the pattern of expression of VPF/VEGF isoforms intumoral and nontumoral tissues. VPF/VEGF mRNA expression in theliver tumors was associated with fibrous capsule formation and septalformation (/' < 0.05 respectively, Fisher's exact /' test). In situ hybridiza

tion confirmed the presence of VPF/VEGF mRNA expression in tumorcells and less intensely in hepatocytes of nontumoral liver. We also foundthat VPF/VEGF expression in the tumor cell was increased in the areaadjacent to necrotic regions (presumably hypoxic regions). As a regulatorof vascular permeability and endothelial cell growth factor, VPF/VEGFmay play an important role in the development of hepatocellular carcinoma.

INTRODUCTION

VPF' is a MT34,000-50,000 dimeric, disulfide-linked glycoprotein

synthesized by several human and animal cell types, both normal andneoplastic (1-4). VPF was originally recognized for its ability toincrease the permeability of the microvasculature to circulating mac-

romolecules (i.e., plasma proteins; Ref. 3). More recently, VPF hasalso been shown to be a selective endothelial cell mitogen, andtherefore has been alternatively called VEGF (5-8). By alternating

splicing of mRNA, four different molecular species with 121, 165,189, and 206 amino acids are determined (9), although the physiological significance of the multiple isoforms has not been fully established. VPF/VEGF has been reported to bind to at least two specificreceptors found on endothelial cells, flt-1 and KDR (10, 11).

VPF/VEGF was originally discovered as a tumor-secreted protein,

and its role in tumor development has been investigated (5, 12). It isconsidered to play an important role in tumor biology in at least twoways: as a vascular permeability factor and/or endothelial growthfactor. As a potent permeability factor, VPF/VEGF promotes extrav-

Received 8/25/95: accepted 5/1/96.The costs of publication of this article were defrayed in part by the payment of page

charges. This article must therefore be hereby marked advertisement in accordance with18 U.S.C. Section 1734 solely to indicate this fact.

1This work was supported in part by grants from the Cancer Research Scholarship

Program of the Osaka Cancer Foundation.2 To whom requests for reprints should be addressed. Phone: 81-6-879-3636: Fax:

81-6-879-3639.* The abbreviations used are: VPF. vascular permeability factor; VEGF, vascular

endothelial growth factor: RT, reverse transcription; HCC. hepatocellular carcinoma;SSC. saline sodium citrate.

asation of plasma fibrinogen, leading to the formation of a fibrinnetwork which serves as a substratum for cell migration duringangiogenesis. In addition, as an endothelial growth factor, VPF/VEGFstimulates endothelial cell proliferation and is likely to induce theformation of new blood vessels (13).

VPF/VEGF is synthesized and secreted by a variety of tumor cellsin tissue culture and by several transplantable animal tumors /';i vivo

(8, 14). Elevated expression of VPF/VEGF transcripts in humantumors was first demonstrated in tumors of the central nervous system(12, 15). Elevated expressions of VPF/VEGF have also been reponedfor tumors of the gastrointestinal tract, kidney, and breast in humans(1, 16-19). However, there has been only one report (20) on the

expression of VPF/VEGF in HCC, which is well known for itshypervascularity.

In this study, we examined the VPF/VEGF expression in HCCusing the RT-PCR method and compared it with the pathological

features of HCC. We also examined for the first time the localizationof VPF/VEGF transcripts in HCC by means of the in situ hybridization technique. The aims of our study were to determine whetherVPF/VEGF is overexpressed in HCC compared with the nontumoralpart of the liver and its role in the pathogenesis of HCC.

MATERIALS AND METHODS

Subjects and Tissue Samples. The subjects in this study were 23 patientswith HCC who underwent surgical resections between June 1990 and September 1994 at Osaka University Hospital and Osaka Rosai Hospital. The studygroup comprised 18 men and 5 women ranging from 42 to 71 (mean age,60.9 ±7.6) years of age. None had apparent distant metastasis. None of thepatients had been previously treated for HCC before surgical resection, such aswith percutaneous ethanol injection or transcatheter arterial embolization. Atumor sample and a nontumoral part of the liver were obtained during thesurgical resection. The samples were each divided into two pieces. One wassubjected to fixation in formalin for histological examination, and the otherwas snap frozen in liquid nitrogen and stored at -80°C until use for VEGF/

VPF expression analysis. Pathological features of the patients were obtainedfrom the pathological reports and clinical records. Histological analysis of thetumors was done by experienced pathologists without knowledge of theanalytical results. Tumors were graded as well differentiated, moderatelydifferentiated, or poorly differentiated according to the grading system of theLiver Cancer Study Group of Japan (21). The differentiation grading was basedon the predominant findings.

The tumors were also examined for size, number of additional nodules,fibrous capsule formation, existence of capsule invasion, septal formation, andportal vein invasion.

RT-PCR Analysis of the VEGF Transcript. Total cellular RNA wasextracted from frozen samples using the single-step method (22), and 3 /¿gtotal RNA were dissolved in 8 jal RNase-free water. One fj.1of 0.01 absorbance

unit of random hexanucleotide primer. 1 nl of 200 units//*! Moloney murineleukemia virus-reverse transcriptase (Life Technologies, Inc., BRL. Gaithers-burg, MD), 4 /j.1 Moloney murine virus-reverse transcriptase buffer, 0.1 ß\

RNase inhibitor, 1 /il 0.1 M DTT, and 5 /xl of 5 mM dNTP mixture were addedto each tube. RT was performed at 37°Cfor 60 min.

3004

VPF/VEGF EXPRESSION IN HEPATOCELLULAR CARCINOMA

Five fil each of RT products were amplified by means of PCR. VEGF/VPFhas been reported to have four isoforms arising from alternative splicing of itsRNA transcript (9). In this study, oligonucleotide primers flanking the insertion/deletion site of VEGF/VPF (codon 115) were used to amplify cDNA. Thesense primer was 5'-CGAAGTGGTGAAGTTCATGGATG-3', and the anti-sense primer was 5'-TTCTGTATCAGTCTTTCCTGGT-3' (6, 9). cDNA was

amplified by 0.5 units Taq DNA polymerase (Takara, Kyoto, Japan) in a 20-^ilreaction mixture containing 10 mM Tris-HCl (pH 8.3). 50 m.M KC1, 1.5 rtiM

MgCl2, 0.2 mM deoxynucleotides, and 40 pM each primer. The PCR profileconsisted of a 7-min initial denaturation at 94°Cfollowed by 40 cycles of1-min denaturation at 94°C.1-min annealing at 60°C.3-min extension at 72°C,and finally a 10-min extension at 72°C.Five ¡ûeach of the PCR reaction

products were run on a 1% agarose gel and visualized using ethidium bromidestaining.

Preparation of Single-Stranded RNA Probes for in Situ Hybridization.

To amplify a VPF/VEGF cDNA fragment from mRNA. we used human lungpoly(A)+ RNA (Toyobo, Osaka, Japan), which is abundant in VPF/VEGF ( 1).RT was performed, and the cDNA produced was amplified with PCR. Wesynthesized the following oligonucleotide primers with BamHI (sense primer)and EcoRl (antisense primer) restriction sites. The sense primer was 5'-CGCGGATCCAGGAGTACCCTGATGAG-3', and the antisense primer was5'-CCGGAATTCACATTTGTTGTGCTGT-3' (1). PCR was performed withthe following thermocycle parameters: a 7-min initial denaturation at 94°Cfollowed by 40 cycles of 1-min denaturation at 94°C,2-min annealing at 55°C,2-min extension at 72°C, and finally a 10-min extension at 72°C. This

amplified 204-bp VPF/VEGF cDNA fragment, which is common to all known

VPF/VEGF splicing variants (9), was digested with ßamHIand EcuRl andsubcloned into the pBluescript II SK+ (Stratagene Inc.. La Jolla, CA). Theidentity of the cloned insert was confirmed by DNA sequencing.

The pBluescript II SK+ construct with the VPF/VEGF insert was linearizedwith BamHl and transcribed in vitro from the T7 polymerase promoter to yieldthe antisense RNA probe. The same construct was linearized with EcoKl andtranscribed from the T3 promoter to yield the sense (control) probe. Thetranscription reaction was performed in the presence of 0.35 mM digoxigenin-

UTP (Boehringer Mannheim GmbH, Biochemica, Mannheim, Germany) toyield the digoxigenin-labeled RNA probe.

In Situ Hybridization. In situ hybridization was performed in three casesof HCC using a digoxigenin nucleic acid detection kit (Boehringer MannheimGmbH, Biochemica) with some modifications (23). Frozen sections were cutat 8-10 ^irn, fixed for 20 min in 4% paraformaldehyde/0.l M phosphate buffer,

and rinsed in PBS (0.02 Mphosphate buffer and 0.15 M NaCl). Specimens werethen treated with 0.001% proteinase K, fixed in 4% paraformaldehyde/0. l Mphosphate buffer again, denatured in 0.2 N HC1, acetylated in 0. l M trietha-

nolamine and 0.25% acetic anhydride, and dehydrated through a gradedethanol series. Sections were prehybridized for 30 min at 50°Cwith 100 fil of

prehybridization fluid containing 50% formamide, 0.3 M NaCl, 20 mM Tris-HCl (pH 8.0), 2 mM EDTA, 10% dextran sulfate, 0.2% sarcosyl, single-strength Denhardt's solution (0.2 g/liter BSA, 0.2 g/liter Ficoll, and 0.2 g/liter

polyvinyl pyrrolidone), and 0.02% salmon sperm DNA in a humidified chamber. After the prehybridization, hybridization was performed in buffer containing prehybridization fluid and 1.0 ng/ml of either antisense or sense probeat 50°Cfor 14 h. Next, sections were washed at 60°Cfor 20 min in 5X SSC

(150 mM NaCl and 15 mM sodium citrate) and 30 min in 2x SSC/50%formamide. digested with RNase (10 /¿g/ml)at 37°Cfor 30 min, and washedagain at 60°Cfor 30 min in 2X SSC/50% formamide. Then, blocking was

performed in blocking buffer containing 1.5% BSA for 30 min. After theblocking buffer was removed, the solution containing 1.5 units/ml sheepanti-digoxigenin Fab fragment conjugated to alkaline phosphatase and 1.5%

BSA was placed on each section. Sections were placed in a humidifiedchamber and incubated at room temperature for 1 h. After the sections wererinsed, a coloring reaction was performed using 230 ¿ig/mlnitroblue tetrazo-lium salt and 180 /ng/ml 5-bromo-4-chloro-3-indolyl-phosphate, and then

sections were counterstained with methyl green.Statistical Analysis. For comparison of the histopathological features be

tween VEGF/VPF expression and nonexpression groups, qualitative variableswere analyzed with the Fisher's exact test; quantitative variables were evaluated with Student's t test.

RESULTS

RT-PCR Analysis of VPF/VEGF Transcript in HCC Tissuesand Normal Liver Tissues. RT-PCR analysis revealed that

VPF/VEGF mRNA was expressed in the liver tumors of 16 of 23patients examined (16/23, 69.6%). In 9 of 16 patients, the transcriptswere also found in nontumoral liver, although the bands corresponding to VPF/VEGF in nontumoral liver were faint in 3 of these cases.In the remaining 7 patients, VPF/VEGF transcripts were detected onlyin tumor tissues. No VPF/VEGF transcript was found in either nontumoral or tumoral tissues of the livers in seven patients. Pathologicalfeatures of HCC and the summary of VPF/VEGF analysis is shown inTable 1.

Mature VPF/VEGF transcripts are known to arise from alternativesplicing of the primary gene transcript (9). The oligonucleotide primers were designed to amplify all four known splicing variants that

Table I A summary of pathological features and VPF/VEGF expression nf patients with HCC

Patient1234567891011121314151617181920212223"NI,''NDBackgroundCirrhosisChronic

hepatitisChronichepatitisCirrhosisCirrhosisCirrhosisCirrhosisNormalCirrhosisChronic

hepatitisChronichepatitisNormalChronic

hepatitisCirrhosisCirrhosisCirrhosisCirrhosisChronic

hepatitisChronichepatitisCirrhosisCirrhosisChronic

hepatitisChronichepatitisnot

informative..notdone.Tumor

size(cm)1.72045.941363.53.33.652.85544.8831.93.221.5Additionalnodules10001100000000011010000Portalinvasion0NI01000000NI00NI000000000CapsuleformationNI"NI+++++NI+NINI++NI++NINI+—+--CapsuleinvasionNINI+++++NI—NINI++NI—+NINI———-—SeptalformationNINI++NI—+NI+NINI—+NI—+NINI———-—DegreeofdifferentiationWellPoorlyModeratelyModeratelyModeratelyWellModeratelyModeratelyModeratelyModeratelyModeratelyModeratelyPoorlyModeratelyModeratelyModeratelyModeratelyPoorlyModeratelyWellModeratelyModeratelyModeratelyVascularityin VEGFmRNAangiographynon-tumorHypervascular

+Hypervascular+Hypervascular+Hypervascular+Hypervascular+Hypervascular+ND''±Hypervascular

±Hypervascular±HypervascularHypervascularHypervascularIsovascularHypervascularHypervascularHypervascularHypervascularNDHypervascularHypervascularHypervascularHypervascularHypervascularVEGF

mRNAtumor+++++++++++++±-t--t-—-———-—

3005

VPF/VEGF EXPRESSION IN HEPATOCELLULAR CARCINOMA

607-535-

403-

Case NO. 10Fig. 1. RT-PCR analysis of VPF/VEGF transcript in HCC and nontumoral liver.

Matched nonlumoral liver and lumor RNAs from the patients with HCC were processedas described in "Materials and Methods." Left, size of the cDNA fragments. M, DNA size

marker (Hindi digest of <¡>174).The bands of 607. 535. and 403 correspond toVPF/VEGF,,,,,. VPF/VEGF„,5.and VPF/VEGF,,,. respectively. Samples of cases 10, 8,and 6 are shown. N, nontumoral part of the liver; T. tumoral part of the liver.

could be expressed in the samples studied. Fig. 1 shows the results ofthe electrophoretic analysis of PCR products in representative cases.In patient 10 (Fig. 1. Lane 2. nontumoral part; Lane 3, tumoral part),PCR products of VPF/VEGF transcripts were detected only in thetumoral part. In patient 8 (Fig. 1, Lane 4, nontumoral part; Lane 5,tumoral part) and patient 6 (Fig. 1, Lane 6, nontumoral part; Lane 7,tumoral part), VPF/VEGF transcripts were detected in both the tu-

moral and nontumoral parts. In the majority of the cases, threeamplified bands of 403, 535, and 607 bp corresponding to 121, 165,and 189 amino acid isoforms of VPF/VEGF (VPF/VEGFI21,VPF/VEGF16V and VPF/VEGF1K9), respectively, were observed (Fig.1, Lanes 3 and 5-7). In the samples with faint bands, only two bands

(VPF/VEGF121 and VPF/VEGF165) were detected (Fig. 1, Lane 4).VPF/VEGF levels in these cases with faint bands were designatedas ± in Table 1. There seemed to be no difference between thebanding pattern in nontumoral and tumoral tissues. Notably, no sample was positive for the band corresponding to VPF/VEGF-,()6.

Comparison of Pathological Features of Liver Tumors betweenVPF/VEGF Expression and Nonexpression Groups. VPF/VEGFmRNA expression in HCC was compared with the clinical and his-

tological features of the tumor. The comparison between VPF/VEGFexpression groups and nonexpression groups is shown in Table 2. Forcomparison between VPF/VEGF expression and capsule invasion,cases without capsule formation were eliminated from the analysis,since capsule invasion cannot take place in the absence of capsule. Itis well known that vascularity is not homogeneous in the tumors ofHCC. HCC often shows in homogeneous enhancement, which is oftendescribed as "mosaic-like," on enhanced CT (24). Therefore, we

evaluated the vascularity by a rather gross method, angiography.Tumors that demonstrated tumor staining were designated as hyper-

vascular. Twenty patients were found to have hypervascular tumors,and VPF/VEGF transcript expression was observed in liver tumors of14 of 20 patients. Only one patient (patient 13) was found to beisovascular, whereas the liver tumor in this patient was found to theexpress VPF/VEGF transcript. VPF/VEGF gene expression in livertumors was not related to increased vascularity of the tumors accord

ing to Fisher's exact P test. However, because patient 13 had a rather

small tumor (2.8 cm in diameter), it might have been difficult to detecttumor staining using angiography. In fact, in computed tomography,this tumor exhibited enhancement by bolus injection of contrastmedium. Thus, this case should be regarded as hypervascular, although the difference in vascularity between the tumor and the normalpart was not sufficient to be detected by angiography. Taken together,all 21 cases subjected to angiography had hypervascular tumors, and15 of them expressed VPF/VEGF transcript, whereas 6 cases did not.

Interestingly, VPF/VEGF mRNA expression in the liver tumorswas related to fibrous capsule formation and septal formation of thetumor (P < 0.05, respectively. Fisher's exact P test). VPF/VEGF

mRNA expression in the liver tumors did not show any relationshipwith other histopathological features of the tumors, namely, tumorsize, number of additional nodules, differentiation degree, capsuleinvasion, and portal invasion.

VPF/VEGF mRNA expression in the nontumoral part of the liver wasalso compared with hemoglobin levels, serum alanine aminotransferaselevels, and pathological diagnosis of the background liver disease. However, no significant relationships were found (data not shown).

In Situ Hybridization. By using the VPF/VEGF antisense probe,we observed specific labeling in the tissue sections from HCC. Ingeneral, VPF/VEGF mRNA was distributed homogeneously and intensely throughout the tumor. Microscopic examination revealed thatVPF/VEGF is expressed by tumor cells (Fig. la). No specific elevation of hybridization was observed in the area adjacent to fibroticregions or in the vascular stroma in HCC. However, it is noteworthythat, in the periphery of necrotic foci of the HCC, labeling with theVPF/VEGF antisense probe was distinctly intensified in the tumorcells (Fig. 2c). Hybridization of the antisense VPF/VEGF probe wasalso evident, although less intensely, in the normal hepatocytes of thenontumoral part of the liver. Distribution of VPF/VEGF mRNA innontumoral liver was also homogeneous (Fig. 2e). No significanthybridization was noted in the vascular endothelial cells, either in thetumoral tissue or in the nontumoral liver tissue. No specific cellular

Table 2 Comparison of the clinical and pathological features of HCC between VPF/VEGF expression groups and nonexpression groups"

CharacteristicsVascularityHypervascularIsovascularHypovascularTumor

size(cm)cNo.

of additionalnodules01Degree

ofdifferentiationWellModeratelyPoorlyCapsule

formation+-Capsule

invasion+—Septal

formation+—Portal

veininvasion+-With

VPF/VEGFexpression(n =16)14104.8

±4.312421221008263112Without

VPF/VEGFexpression

(n =7)6003.5

±2.35215123020507PNS*NSNSNS<0.05NS<0.05NS

" VPF/VEGF mRNA expression in HCC was associated with fibrous capsule formation and septal formation of the cancer (P < 0.05, respectively. Fisher's exact P test).

NS, not significant.' Data expressed as mean ±SD.

3006

VPF/VEGF EXPRESSION IN HEPATOCELLULAR CARCINOMA

«mmi_

a

fFig. 2. Analysis of VPF/VEGF mRNA expression with in .vii«hybridization. Digoxigenin-labeled RNA probes complimentary (antisense) or identical (sense) to VPF/VEGF were

hybridized to tissue sections from HCC (a and b). HCC with necrotic region (r and d), and nontumoral liver (e and/). Hybridization with antisense probe (a. c. and e) or sense probe(b. d. and f) are shown. Specific labeling of tumor cells in HCC is observed (a). Strong labeling for VPF/VEGF mRNA is observed in tumor cells adjacent to tumor necrosis (r).Hybridization of VPF/VEGF probe is also shown, although less intensely than tumor cells, in the normal hepatocytes of nontumoral liver (e). Tissue sections probed with the senseprobe show that no specific labeling is present and that background is low (b, d. and/). Bar. 200 pm. X100.

labeling was seen with the sense control probes, and the nonspecific of those patients undergoing liver resection for colorectal cancerbackground was low (Fig. 2, b, d, and/).

DISCUSSION

In the literature, VPF/VEGF expression in normal liver has beenreported, although the level of expression is not clarified. Berse et al.(1) found VPF/VEGF mRNA expression in a single case of normalhuman liver by means of Northern blot analysis and reported thatVPF/VEGF transcripts were "readily detected" in human liver. On the

other hand. Warren et al. (25) examined grossly normal human liver

metastasis. They observed a detectable signal for VPF/VEGF in threeof five livers using Northern blot analysis. Furthermore, the functions

of VPF/VEGF in normal liver are poorly understood. In the presentstudy, under the RT-PCR conditions used, VPF/VEGF mRNA was

detected in nontumoral liver in only 9 of 23 patients. Therefore, the

detection of the VPF/VEGF transcript in 16 of 23 HCCs suggests thatVPF/VEGF mRNA is overexpressed in HCC compared with thenontumoral part of the liver. The in situ hybridization techniqueconfirmed the presence of VPF/VEGF mRNA in the tumor cells and

3007

VPK/VEGF EXPRESSION IN HEPATOCELLULAR CARCINOMA

less intensely in nontumoral hepatocytes. To our knowledge, this is growth factor, transforming growth factor ß,and tumor necrosisthe first report to reveal the localization of VPF/VEGF mRNA intumor cells of HCC. Recently, Mise et al. (20) reported overexpres-

sion of VPF/VEGF in HCC by means of Northern blot analysis.However, they examined the localization of VPF/VEGF with immu-

nohistochemistry. They found VPF/VEGF protein in HCC cells andthe vascular endothelial cells, but they could not determine whichcells synthesize VPF/VEGF. Taken their study into account, it seemsthat VPF/VEGF is synthesized by HCC cells and accumulates in thevascular endothelial cells, its target of action.

Three forms of VPF/VEGF mRNA have been shown to be expressed in HCC, with an abundance of VPF/VEGF,21 and VPF/VEGF,6V Nontumoral liver shows a similar pattern of VPF/VEGFisoform expression, although the number of VPF/VEGF-expressing

cases is small. In literature, central nervous system neoplasms andnormal brain have been reported to express VPF/VEGF121,VPF/VEGF,65, and VPF/VEGF189 (26). Our results are consistentwith the studies on other human tumors, with no unique isoform totumor tissue being observed. The VPF/VEGF2()6 isoform was notdetected either in HCC or nontumoral liver in this study. To date, theVPF/VEGF-,()ft transcript has been observed only in the human fetal

liver cDNA library (9). Although HCC often produces embryonicprotein such as a-fetoprotein, it appears that HCC does not express thefetal liver-type VPF/VEGF.

Solid tumors are composed of two distinct compartments, themalignant cells themselves and the vascular and connective tissuestroma that they induce and in which they are dispersed. The stromaprovides the vascular supply that tumors require for obtaining nutrients, gas exchange, and waste disposal. Fibrin is considered to serveas a provisional stroma that is gradually replaced by granulation tissueand then by mature stroma (27). Tumor cells are not known tosynthesize fibrinogen. Therefore, for fibrin to deposit at the extravas-

cular site, the permeability of microvasculature must increase significantly. VPF/VEGF is thought to be responsible for the vascularhyperpermeability and consequent fibrin deposition in tumor (13). Inthis study. VPF/VEGF transcript expression in HCC was related tofibrous capsule formation and septal formation. The origin of capsuleformation in HCC is unclear. It has been suggested that capsuleformation is a result of compression and collagenization of the adjacent stroma (28, 29). However, this mechanism has been doubted,since the tumor size of HCC did not correlate with either incidence ofcapsule formation or thickness of the capsule (30, 31 ). These findingssuggest that capsules might be formed by active fibrosis rather than bytumor compression on the adjacent stroma (30, 31). The origin offibrous septa in a tumor also remains to be clarified. Nakashima el al.(29) suggested a possibility of fibrogenesis in the interface of twotumor nodules of different properties. If these fibrotic changes,namely, capsule formation and septal formation, are the result ofactive fibrosis, they require fibrin deposition in their initial stage (27).In breast cancer. Brown et al. (19) have reported the strong expressionof VPF/VEGF mRNA in infiltrating ductal carcinoma, which characteristically induces a desmoplastic stromal reaction. Their findingimplies a correlation between increased expression of VPF/VEGF andthe induction of a stromal response (19). Our finding offers additionalsupport for the VPF/VEGF involvement in stroma formation.

To investigate the role for VPF/VEGF in angiogenesis, we compared VIM-/VEGF expression with vascularity of HCC. All 21 cases

subjected to angiography were revealed to have hypervascular tumorsby angiography or computed tomography. Although VPF/VEGF expression was observed in 15 cases, it was not detected in 6 cases. Thissuggests that angiogenesis is not simply controlled by the presence ofVPF/VEGF but is mediated by several angiogenic inducers. To date,many angiogenesis factors have been found, including basic fibroblast

factor a (32). In fact, there has been a report on acidic and basicfibroblast growth factor expression in HCC (33). Therefore, it is likelythat vascularization in some tumors is related to other angiogenesisfactors.

The in situ hybridization technique revealed that VPF/VEGFmRNA was expressed in tumor cells and that the labeling was ratherhomogeneous in the tumors. Elevated VPF/VEGF expression was notevident in the area adjacent to the capsule or blood vessels, whereasVPF/VEGF is thought to be associated with stroma formation andincreased vascularity. Our finding is consistent with other humantumor studies, including those on renal cell carcinoma (17), breastcancer (19), and adenocarcinoma of the gastrointestinal tract (16),which have reported that most tumor cells express VPF/VEGFmRNA. Specific overexpression of VPF/VEGF transcripts by tumorcells was not noted around the vessels in any of the studies. Alternatively, some studies observed elevated VPF/VEGF receptor (flt-1 and

KDR) expression in the vascular endothelial cells (12,16,17,19).Therefore, the distribution of fibrotic change or vascularization maybe determined by VPF/VEGF receptor(s) expression, not by VPF/VEGF expression. Further study into the cellular distribution of VPF/VEGF receptors may be needed to clarify the role of VPF/VEGF inthese pathological changes.

It is noteworthy that VPF/VEGF mRNA expression in the tumorcell was locally elevated in the area adjacent to tumor necrosis. Thiselevation around the necrotic area of the tumor has been reported inseveral other human cancers (12, 15-17, 19). Rapid cell proliferation

in the center of a tumor can lead to increased interstitial pressure,which may lead to compression closure of capillaries and consecutivetumor necrosis (34). Tumor hypoxia has been reported to increaseexpression of VPF/VEGF mRNA in a variety of cultured tumor cells(15). Therefore, it is likely that, in the development of necrosis,hypoxia may increase the VPF/VEGF mRNA expression.

In this study, we observed VPF/VEGF mRNA expression in nontumoral liver in 9 of 23 patients. We sought to determine under whatcircumstances VPF/VEGF was expressed in nontumoral liver. However, VPF/VEGF expression in nontumoral liver was not related tohemoglobin levels, serum alanine aminotransferase levels, or pathological diagnosis of the background liver disease. Further study shouldclarify the mechanisms of VPF/VEGF transcripts expression in nontumoral liver.

We determined that VPF/VEGF is overexpressed in HCC, and thatits expression in HCC was associated with capsule formation andseptal formation and hypoxia of the tumor. As a regulator of vascularpermeability and endothelial cell growth factor, VPF/VEGF may playan important role in the development of HCC.

REFERENCES

1. Berse. B., Brown. L. F., Van De Water. L.. Dvorak. H. F.. and Senger. D. R. Vascularpermeability factor (vascular endothelial growth factor) gene is expressed differentially in normal tissues, macrophages, and tumors. Mol. Biol. Cell. 3: 211-220. 1992.

2. Senger. D. R.. Galli. S. J.. Dvorak. A. M.. Peruzzi. C. A.. Harvey. V. S., and Dvorak,H. F. Tumor cells secrete a vascular permeability factor that promotes accumulationof ascites Huid. Science (Washington DC). 2/9: 983-985. 1983.

3. Dvorak. H. F., Orenstein, N. S.. Carvalho. A. C. Churchill. W. H.. Dvorak. A. M..Galli. S. J.. Feder. J.. Bitzer. A. M.. Rypysc. J.. and Giovinco. P. Induction of afibrin-gel investment: an early event in line 10 hepatocarcinoma growth mediated bytumor-secreted products. J. Immunol.. /22: 156-174. 1979.

4. hum.i. K.. Yoshikawa, N.. Connolly. D. T.. and Nakamura. H. Human mesangial cellsand peripheral blood mononuclear cells produce vascular permeability factor. KidneyInt., 44: 959-966, 1993.

5. Ferrara, N.. and Henzel, W. J. Pituitary follicular cells secrete a novel heparin-bindinggrowth factor specific for vascular endothelial cells. Biochem. Biophys. Res. Commun.. 161: 851-858. 1989.

6. Leung, D. W., Cachianes, G.. Kuang, W. J.. Goeddel, D. V.. and Ferrara. N. Vascularendothelial growth factor is a secreted angiogenic mitogen. Science (WashingtonDC). 246: 1306-1309. 1989.

3008

VPF/VEGF EXPRESSION IN HEPATOCELLULAR CARCINOMA

7. Keck, P. J.. Hauser, S. D.. Krivi, G.. Sanzo. K.. Warren. T., Feder, J.. and Connoly,D. T. Vascular permeability factor, an endothelial cell mitogen related to PDGF.Science (Washington DC). 246: 1309-1312, 1989.

8. Connoly, D. T., Heuvelman, D. M.. Nelson, R., Olander. J. V., Eppley, B. L.. Delfino,J. J., Siege], N. R., Leimgruber, R. M., and Feder. J. Tumor vascular permeabilityfactor stimulates endothelial cell growth and angiogenesis. J. Clin. Invest.. 84:1470-1478. 1989.

9. Houck. K. A.. Ferrara. N.. Winer. J., Cachianes. G.. Li. B., and Leung, D. W. Thevascular endothelial growth factor family: identification of a fourth molecular speciesand characterization of alternative splicing of RNA. Mol. Endocrinol., 5: 1806-1814,

1991.10. de Vries, C, Escobedo. J. A., Ueno, H., Houck. K.. Ferrara, N., and Williams. L. T.

The fms-like tyrosine kinase, a receptor for vascular endothelial growth factor.Science (Washington DC). 259; 989-991, 1992.

11. Terman. B. I.. Dougher-Vermazen, M.. Carrion, M. E., Dimitrov. D.. Armelino, D. C.,

Gospodarowicz, D.. and Boehlen. P. Identification of the KDR tyrosine kinase as areceptor for vascular endothelial growth factor. Biochem. Biophys. Res. Commun..¡87: 1579-1586. 1992.

12. Plate. K. H.. Breier, G.. Weich, H. A., and Risau, W. Vascular endothelial growthfactor is a potential tumour angiogenesis factor in human gliomas in vivo. Nature(Lond.), 359: 845-848, 1992.

13. Senger, D. R.. Van De Water, L.. Brown, L. F.. Nagy. J. A., Yeo, K. T., Yeo, T. K.,Berse, B., Jackman. R. W, Dvorak, A. M. and Dvorak. H. F. Vascular permeabilityfactor in tumor biology. Cancer Metastasis Rev., 12: 303-324. 1993.

14. Dvorak. H. F., Sioussat. T. M., Brown, L. F., Berse. B.. Nagy. J. A., Sotrel, A.,Manseau, E. J., Van De Water, L., and Senger. D. R. Distribution of vascularpermeability factor (vascular endothelial growth factor) in tumors: concentration intumor blood vessels. J. Exp. Med., ¡74: 1275-1278, 1991.

15. Shweiki. D., hin. A., Soffer, D., and Keshet, E. Vascular endothelial growth factorinduced by hypoxia may mediate hypoxia-initiated angiogenesis. Nature (Lond.).559: 843-845. 1992.

16. Brown, L. F.. Berse, B.. Jackman, R. W.. Tognazzi, K.. Manseau. E. J., Senger, D. R.,and Dvorak, H. F. Expression of vascular permeability factor (vascular endothelialgrowth factor) and its receptors in adenocarcinomas of the gastrointestinal tract.Cancer Res., S3: 4727-4735. 1993.

17. Brown, L. F., Berse, B., Jackman, R. W., Tognazzi, K.. Manseau, E. J.. Dvorak. H. F.,and Senger. D. R. Increased expression of vascular permeability factor (vascularendothelial growth factor) and its receptors in kidney and bladder carcinomas. Am.J. Pathol.. 143: 1255-1262, 1993.

18. Takahashi, A.. Sasaki. H.. Kim. S. J., Tobisu, K.. Kakizoe. T., Tsukamoto, T.,Kumamoto. Y.. Sugimura. T., and Terada. M. Markedly increased amounts ofmessenger RNAs for vascular endothelial growth factor and placenta growth factor inrenal cell carcinoma associated with angiogenesis. Cancer Res.. 54: 4233-4237,

1994.19. Brown, L. F., Berse, B., Jackman. R. W., Tognazzi, K.. Guidi, A. J.. Dvorak, H. F.,

Senger, D. R., Connoly, J. L., and Schnitt, S. J. Expression of vascular permeability

factor (vascular endothelial growth factor) and its receptors in breast cancer. Hum.Pathol., 26: 86-91, 1995.

20. Mise, M., Arii. S., Higashituji, H., Furutani, M., Niwano, M., Harada. T., Ishigami.S., Toda, Y.. Nakayama. H.. Fukumoto, M.. Fujita. J.. and Imamura. M. Clinicalsignificance of vascular endothelial growth factor and basic fibroblast growth factorgene expression in liver tumor. Hepatology, 23: 455-464. 1996.

21. Liver Cancer Study Group of Japan. The General Rules for the Clinical and Pathological Study of Primary Liver Cancer, 3rd ed, pp. 34-35. Kyoto: Kanehara. 1992.

22. Chomczynski. P.. and Sacchi, N. Single-step method of RNA isolation by acidguanidinium thiocyanate-phenol-chloroform extraction. Anal. Biochem.. 762: 156-

159. 1987.23. Horimoto, M.. Hayashi. N.. Sasaki, Y., Ito, T., Ito, Y., Wada, S., Tanaka, Y., Kaneko,

A.. Fusamoto. H.. Tohyama, M., and Kamada, T. Expression and phosphorylation ofrat c-met/hepatocyte growth factor receptor during rat liver regeneration. J. Hepatol.,23: 174-183. 1995.

24. Yamashita, Y., Takahashi, M.. Baba. Y., Kanazawa. S.. Charnsangavej, C.. Yang. D.,and Wallace, S. Hepatocellular carcinoma with or without cirrhosis: a comparison ofCT and angiographie presentations in the United States and Japan. Abdom. Imaging.18: 168-175. 1993.

25. Warren, R. S., Yuan, H., Math, M. R.. Gillett, N. A., and Ferrara, N. Regulation ofvascular endothelial growth factor of human colon cancer tumorigenesis in a mousemodel of experimental liver metastasis. J. Clin. Invest.. 95: 1789-1797, 1995.

26. Berkman, R. A., Merril. M. J., Reinhold. W. C., Monacci. W. T., Saxena, A., Clark,W. C., Robertson, J. T.. Ali. I. U.. and Oldfield, E. H. Expression of Ine vascularpermeability factor/vascular endothelial growth factor gene in central nervous systemneoplasms. J. Clin. Invest., 91: 153-159, 1993.

27. Nagy, J. A., Brown, L. F., Senger, D. R., Lanir, N.. Van De Water. L.. Dvorak, A. M.,and Dvorak, H. F. Pathogenesis of tumor stroma generation: a critical role for leakyblood vessels and fibrin deposition. Biochim. Biophys. Acta, 948: 305-326, 1988.

28. Okuda. K.. Nakashima. T.. Obata, H.. and Kubo. Y. Clinicopathological studies ofminute hepatocellular carcinoma. Gastroenterology, 73: 109-115. 1977.

29. Nakashima, T.. Okuda, K., Kojiro, M., Jimi, A., Yamaguchi, R.. Sakamoto, K., andIkari. T. Pathology of hepatocellular carcinoma in Japan. 232 consecutive casesautopsied in ten years. Cancer (Phila.). 5!: 863-877, 1983.

30. Okuda, K., Peters, R. L., and Simson, I. W. Gross anatomic features of hepatocellularcarcinoma from three disparate geographic areas. Proposal of new classification.Cancer (Phila.), 54: 2165-2173, 1984.

31. Irene. O. L., Path, M. R. C., Lai, E. C. S., Matthew, M. T. N., and Sheung, T. F.Tumor encapsulation in hepatocellular carcinoma. Cancer (Phila.), 70: 45-49, 1992.

32. Folkman. J., and Shing, Y. Angiogenesis. J. Biol. Chem., 267: 10931-10934, 1992.

33. Motoo. Y., Sawabu. N., and Nakanuma, Y. Expression of epidermal growth factorand fibroblast growth factor in human hepatocellular carcinoma: an immunohisto-chemical study. Liver, //: 272-277, 1991.

34. Jain, R. K. Determinants of tumor blood flow: a review. Cancer Res., 48: 2641-2658,

1988.

3009