(CANCER RESEARCH57. 1554-1560. April 15. 1997]

ABSTRACT

Several lines of experimental evidence in in vitro and animal modelsystems suggest that the Integrin av@3 plays a role in the tumorigenicity

of human melanoma cells and that the blocking of aVIJ3 ligand bindingcan inhibit tumor progression. However, there Is only scanty informationabout the role of aVP3 in malignant melanoma in a clinical setting.

Therefore, in the present study, we have analyzed the distribution in

lesions of melanocyte origin and in normal tissues of the@ integrin

subunit and of the avfi3 complex and their association with histopathological and clinical parameters of malignant melanoma. We have used as

probes the monoclonal antibodies (mAbs) TP36.1 and VF27.263.15, whichwe have shown with a combination of serological and immunochemicalassays to be specific for the a@,subunit and for the a@@33complex,respectively. In immunohistochemical assays, mAb TP36.1 stained bothbenign and malignant lesionsof melanocyteorigin. In contrast, the mctivity of mAb VF27.263.15 was restricted to malignant lesions. Both mAbsdisplayed differential reactivity with primary melanoma lesions of differ

ent histotypes because they stained about 50% of acral lentiginous melanoma and superficial spreading melanoma lesions, at least 80% of nodularmelanoma lesions, and none of the uveal melanoma lesions tested. BothmAbs TP36.1 and VF27.263.15 stained about 60% of lymph node metestases and 80% of cutaneous metastases.

Expression of the a@P3 complex in melanocytic lesions resembles thatofintercellular adhesion molecule-i (ICAM-1) in several respects: (a) bothare expressed in a significantly (P < O.004)larger proportion of malignantthan of benign lesions; (b) expression of both molecules In primary melanoma lesions Is significantly (P < 0.05) associated with lesion thickness;and (c) expression ofboth molecules in primary lesions from patients withstage I melanoma Is significantly (P < 0.05) associated with an increased

probability of disease recurrence following surgical excision. aVfi3 and

ICAM-1 in primary melanoma lesions complement each other in predicting the outcome of the disease, because the association with prognosis wasenhanced when primary lesions were stained by both anti-a,,4J3 mAb

VF27.263.15 and anti-ICAM-1 mAb CL203.4 or by neither mAb.

Because aVIJ3 has been suggested as a potential target of immunother

apy, its distribution in normal tissues was investigated. aV@I3expression isrestricted because it was only detected in ductal epithelium of parotidglands, thyrocytes, basal glands of the stomach, colonic and rectal epithe.hum glomerull, Bowman's capsules and proximal and distal tubules ofkidneys, and endometrial epithelium. These findings suggest that renal

function will be a critical clinical parameter to monitor in therapies ofmalignant diseases relying on systemic administration of anti-a@fi3 mAb.

Received I 1/25/96; accepted 2/22/97.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.

I This work was supported by the Consiglio Nazionale delle Ricerche Progeuo Final

izzato (Applicazioni Cliniche della Ricerca Oncologica), by Associazione Italiana per IaRicerca sul Cancro, by the Italian Ministry of Public Health, and by USPHS GrantsCA37959, CA5 I8 14, CA67 108, and CA67988 awarded by the National Cancer Institute,Department of Health and Human Services.

2 To whom requests for reprints should be addressed, at Department of Microbiology

and Immunology. Basic Science Building, Room 308, New York Medical College,Valhalla, NY 10595. Phone: (914) 993-4175; Fax: (914) 993-4176.

INTRODUCTION

Human melanoma cells have been found to express a number ofadhesion molecules (1). Among them, the avf33 integrin and theICAM-13 have been suggested to play a role in the tumorigenicity ofmelanoma cells by several experimental and/or clinical findings (1, 2).First, tumorigenicity of human melanoma cell lines in athymic nudemice strongly correlates with avf33 integrin expression by malignantcells (3) and with the extent of ICAM-l release (4). Furthermore, theexpression of both markers by cells of the melanocyte lineage dramatically increases with tumor progression in patients (5—12).Lastly,there is convincing evidence that ICAM-l expression in primarylesions is correlated positively with their thickness and negativelywith disease-free survival in patients with malignant melanoma (6—9).

There is only scanty information about the relationship between theaV1@3complex expression in primary melanoma lesions and theirhistopathological characteristics. Immunohistochemical staining of alimited number of primary lesions has surprisingly showed no association between their thickness and a@(33 complex expression (10).Furthermore, to the best of our knowledge, it is not known whetheraVP3 complex expression in primary melanoma lesions is associated

with the clinical course of the disease and whether aV(33 complex andICAM-l expression in primary melanoma lesions complement eachother in predicting the prognosis of the disease. Therefore, in thepresent study, we have addressed these questions with mAbs we haveshown to recognize the aVf33 complex and a@ subunit.

MATERIALS AND METHODS

Cell Lines, Purified Receptors, and Tissue Samples. Cultured humanmelanoma cells Cob 38 and cultured human B-lymphoid cells LG-2 weregrown in RPMI 1640 (Life Technologies, Inc., Grand Island, NY) supplemented with 10% FCS (Life Technologies, Inc.) and 2 mML-glutamine(LifeTechnologies, Inc.). Cultured human melanoma cells M2l and M2lLIIb (13)and cultured human lung adenocarcinoma cells UCLA-P3 were grown inDMEM (Life Technologies, Inc.) containing 10% FCS, 20 inst HEPES, and 1mM pyruvate. The av@33receptor (14) and the a211b133receptor (15) proteinswere purified from human placenta and platelets, respectively, as described.

Normal tissues and benign and malignant lesions of melanocyte origin wereobtained from the surgical pathology section of the Regina Elena Cancer

Institute. Tissue samples were snap frozen in liquid nitrogen. Four-sm-thickcryostat sections of each sample were fixed in absolute acetone for 10 ruin.Fixed sections were either immediately used in immunohistochemical assaysor kept frozen at —70°Cfor at least 6 months with no loss of serologicalreactivity. Fixed sections were stained with 1% toluidine blue to evaluate thehistological features of the lesions. Histological diagnoses were made accord

ing to Clark et a!. (16), and tumor thickness was evaluated according to

Breslow (17).mAb and Conventional Antisera. mAbs VF27.263.15 and TP36.l, both

IgG2b, were obtained from BALB/c mice immunized with SDS-treated melanoma cell membranes and with IFN--y-treatedmelanoma cells, respectively,

3Theabbreviationsusedare:ICAM-l,intercellularadhesionmolecule-I;mAb,monoclonal antibody; MAA, melanoma associated antigen; ALM, acral lentiginous melanoma;SSM, superficial spreading melanoma; HF, indirect immunofluorescence; lIP, indirectimmunoperoxidase; LMM, lentigo maligna melanoma; NM. nodular melanoma.

1554

Clinical Significance of aV1@3Integrin and Intercellular Adhesion Molecule-iExpression in Cutaneous Malignant Melanoma Lesions'

Pier Giorgio Natali, Carl V. Hamby, Brunhilde Felding-Habermann, Bitao Liang, Maria R. Nicotra,

Franco Di Fiippo, Diana Giannarelli, Massimo Temponi, and Soldano Ferrone2

Departments of Immunology (P. G. N.J and Surgery (F. D., D. G.J. Regina Elena Cancer Institute, 00158 Rome, italy: Department of Microbiology and Immunology, New YorkMedical College. Valhalla, New York 10595 (C. V. H., M. T., B. L, S. F.J; Department of Molecular and Experimental Medicine, The Scripps Research Institute, La Jolla,California 92037 (B. Fl: and Consiglio Nazionale della Ricerche Institute of Biomedical Technologies. 00161 Rome, Italy (M. R. N.J

on April 23, 2020. © 1997 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

0V@3 INThGRIN, ICAM- I . AND CLINICAL COURSE OF MELANOMA

Fig. 1. SDS-PAGE analysis of antigens immunoprecipitated from cultured human melanomacells Cob 38 by mAb VF27.263.l5, mAb TP36.l,anti-a@ chain mAb LM142, and anti-a@133 com

plex mAb LM609. NP4O extracts from 25I4abeled (left panel) and [35S]methionine-labeled(rightpanel) Cob 38 cells were incubated for 16 hat 4*C on a rotator with protein A-Sepharose insolubilized mAb LM142 (Lane A), LM609 (LaneB), VF27.263.l5(LaneC),andTP36.l (LaneD).After washing, antigens were eluted from immunoadsorbent beads by boiling with 50 @slof electrophoresis sample buffer and separated by SDSPAGE under reducing (left panel) or nonreducing(rightpanel) conditions in 3—15%polyacrylamidegradient gels using the Laemmli buffer system(23). After drying, the gel containing ‘251-labeledantigens was autoradiographed using a KOdakXAR-5 film (Eastman KOdakCompany, Rochester, NY), and the gel containing 35S-labeledantigens was processed for fluorography as describedby Bonner and Laskey (28).

ABCD

@ — — @.

Serological Assays. The ELISA with cells, the binding assay with 25Ilabeled mAb, the Scatchard plot analysis of the binding of radiolabeled mAbto cells, and the cross-blocking assay with radiolabeled mAb were performedin 96-well polyvinyl chloride microtiter plates (Dynatech Laboratories, Alex

andria, VA), as described elsewhere (21, 26). The ELISA with purifiedreceptor proteins was performed using 96-well microtiter plates coated withavps or aIIb@3 receptor proteins by an overnight incubation at 4°Cwith 100pi of a 1 mg/mI receptor protein solution in NaHCO3, pH 9.5. Plates were

blocked with 2% BSA in Tris-buffered saline (10 mMTris, 150 mMNaC1,pH7.4) supplemented with I mM CaCl2, 1 mM MgCl2, and 1 mM MnC12.

Increasing concentrations of mAbs were incubated in plates for 2 h at 4°C.

Following three washings, plates were incubated with horseradish peroxidaseconjugated anti-mouse IgG xenoantibodies, washed, and developed with ophenylene diamine as the substrate. Plates were read in an ELISA reader at 490nm, and the results were expressed as absorbance values.

hF staining and flow cytometry of cells was performed as describedelsewhere (13). hF staining of acetone-fixed cryostat tissue sections with mAb(concentrations ranging from 10 to 30 .tg/ml) was performed as described (7).lIP staining oftissues with purified mAb (concentrations ranging from 10to 50@tg/ml)was performed using the ABC Vectastain kit according to the manu

facturer's instructions. Slides were developed using 3-amino-9-ethylcarbazole

as the chromogenic substrate and were counterstained with Mayer's hematox

IMMUNODEPLETEDWITHmAb

376.96 VF27.263.15 LM609

as described in detail elsewhere (18). The anti-a@chain mAbs LM142 (2) andAV-8 (13), both IgGl, the anti-@33chain mAb AV-lO, an IgGI (13), theanti-a@j33 complex mAb LM609, an IgGl (2), the anti-aIIb@3 complex mAb

LJ-CP8, an IgGl (19), the anti-lOO,000molecular weight MAA mAb 376.96,an IgG2a (20), and the anti-ICAM-l mAb CL203.4, an IgGl (21), weredeveloped and characterized as described previously.

mAbs were purified from ascitic fluid either by sequential precipitation withcaprylic acid and ammonium sulfate (22) or by affinity chromatography onprotein A-Sepharose (Pharmacia Biotech, Inc., Uppsala, Sweden). The purityof mAb was analyzed by SDS-PAGE (23). Antibodies were labeled with 25!using the lodo-Gen method (24). Radiolabeled mAb preparations used inbinding assays had at least 80% immunoreactivities, as measured by themethod of Lindmo et a!. (25).

Rabbit anti-mouse immunoglobulin antibodies were purchased from Jackson Immunoresearch Laboratories (West Grove, PA). FITC-conjugated, antimouse IgG (H+L chain) xenoantibodies were purchased from Zymed (SanFrancisco, CA). FITC-conjugated F(ab')2 fragments of anti-mouse and antirabbit immunoglobulin xenoantibodies were purchased from Cappel (OrganonTeknika, Tumhout, Belgium). Prior to use, the latter preparations were extensively adsorbed with washed human AB, Rh@RBCs and with insolubilizedpooled normal human plasma. The ABC Vectastain kit for immunoperoxidasestaining was purchased from Vector Laboratories, Inc. (Burlingame, CA).

TP36.1 LM142

I4

Fig. 2. Structural relationship among the antigensrecognized by mAb VF27.263.l5. mAb TP36.l. anti-a,,, chain mAb LM142, and anti-av@s complexmAbLM609.NP4Oextractsfrom‘25l-labeledCob38 cells were depleted of antigens by five rounds ofincubation with protein A-insolubilized mAbsVF27.263.l5, TP36.l, LM142, or LM609. Extractswere then divided into aliquots, and each aliquotwas immunoprecipitatedseparately with mAbsVF27.263.l5, TP36.l, LM142, or LM609. Afterwashing, antigens were eluted by boiling with 50 @lof electrophoresisbuffer and subjectedto SDSPAGE under reducing conditions in 10% polyacrylamide gels. The anti-lOO,000molecularweightMAA mAb 376.96 was used to monitor the specificity of the depletion procedure. After drying, gelswereautoradiographedusinga KOdakXAR-5film(EastmanKOdakCompany).

,@v.— -

av.@ .@@I@3I35―-@ ... ..

I555

137 Kd-112Kd@

32 Kd,@30KdS

A B CD

@@ —@ -l6OKd

@@@ . 100 Kd

‘4 p.

-. I

@@ ;@@!@

<r@rc,i <r-ir <r@r

@I@

j@@P4 @jCO..aP@)aI j,.:@Pii)°@

@ @ah)@@ @P@àg@ g@

:.Ih @I @@Jb@ ..lh

a ui UI us

IMMUNOPRECIPITATEDWITHmAb

on April 23, 2020. © 1997 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

Table 1 Mapping of the determinants recognized by mAbs VF27.263. 15, TP36. 1. l..M142,and LMÔO9on the cell surface of cultured human melanoma cells Cob38°Unlabeled

mAbSpecificity‘25I-labeled

mAbVF27.263.l5TP36.lLM142LM609VF27.263.l5

TP36.lLMI42LM609aj33

a@

av

aj3397

1431960

958407

85861599

02

85

tions of antigens immunoprecipitated by mAbs VF27.263.l5 and0 TP36. 1 from radiolabeled cultured melanoma cells Cob 38 resemble

those of antigens immunoprecipitated by anti-a@(33 complex mAbLM609 and by anti-a@ chain mAb LM142 (Fig. I). Immunodepletionexperiments defined the structural relationship among the antigenicpopulations recognized by the four mAbs. As shown in Fig. 2, mAbsVF27.263.l5 and LM609 did not immunoprecipitate any componentfrom Cob 38 melanoma cell extracts that had been immunodepletedwith either mAb TP36.l or with mAb LM142. The latter two mAbs,which immunodepleted each other, immunoprecipitated a subpopulation of molecules from the Cob 38 cell extracts that had beenimmunodepleted with mAbs VF27.263.l5 or LM609. The specificityof the immunodepletion was monitored with the anti-100,000 molecular weight MAA mAb 376.96, which immunoprecipitated a 100,000molecular weight antigen from Cob 38 melanoma cell extracts thathad been immunodepleted with any of the four mAbs. mAb 376.96failed to immunodeplete the antigens immunoprecipitated by theabove four mAbs. These results suggest that mAb TP36.l reacts withthe a@ subunit and that mAb VF27.263.l5 recognizes a determinantexpressed by the 133chain or by the avP3 complex, because Cob 38cells are likely to express more than one a@,integrin. aIIb@33is notinvolved in the reactivity of mAb VF27.263.15 with Cob 38 cellsextracts, because these cells were not stained in 1W by the mAbLJ-CP8. The latter mAb stained aIIb@3-transfected M21-LIIb humanmelanoma cells (13) and human platelets (results not shown). Theconclusions about the specificity of mAbs TP36.l and V27.263.15were corroborated by the following findings. mAb TP36.l reactedwith melanoma cells M21, which express both GV(33 and aV135complexes (2), with lung adenocarcinoma cells UCLA-P3, whichexpress a@1@5but no 133 integrins (29), and with purified avf3scomplex. In contrast, mAb TP36.l did not react with melanoma cellsM2l-LIIb, which express a11bf33 complex, but do not express a@integrins (13), and with purified crIIbf33complex (Fig. 3 and 4). mAb

av@ INTEGRIN. ICAM-I. AND CLINICAL COURSE OF MELANOMA

1.0

0.5

B

100 101

E

Iio-@ 10.1 101 io-@ 1Ô.1

M21ay+1@3+

UCLA-P3U@+

1@3+ I@3-

Fluorescence IntensIty

Fig. 4. Differential 1W staining by mAbs TP36.l and VF27.263.l5 of cultured humancell lines with different integrin expression. Cultured human melanoma cells M21(av@Ps@) and M21-Lllb (a@f33') and cultured human lung adenocarcinoma cellsUCLA-P3(av@@s) were stained in LWwith mAb TP36.l (—)and with mAbVF27.263.15 (—). Cell-associated fluorescence was analyzed using a Becton DickinsonFACScan.Anti-a@mAbAV-8(—) andanti-f33mAbAV-lO(—)wereusedascontrols.

a Cob 38 cells (2 X b0@cells/well) were incubated with unlabeled mAb (10 @sg/well)in a total volume of 100 pi of PBS-BSA. After a 1-h incubation at 4°Con a rotator, ‘25I-labeledmAb (2 x lO@cpm/well) was added to each well, and the incubation was prolonged for an additional 2 h at 4°C.Plates were then centrifuged, and cell-bound radioactivity was measuredin a gamma counter. Results are expressed as percentage of inhibition compared to binding performed in the presence of mouse IgG.

1556

mAb (pglml)

Fig. 3. Differential reactivity ofmAb TP36.l and VF 27.26315 with purified av@s andaIIb@3receptorproteins.mAb TP36.l (—) and VF27.263.l5(---) were added to96-well microtiter plates coated with av@3i (left panel) or aIIbf33 (right panel) receptorprotein. Binding of mAb was quantified following incubation with horseradish peroxidase-conjugated anti-mouse IgO xenoantibodies and o-phenylenediamine substrate reaction using an ELISA reader at 490 nm. Anti-@3 mAb AV-lO (I @.tg/well;0) was used asa positive control.

ylin. Sections incubated with mouse immunoglobulin instead of the primary

mAb served as controls in both IIF and lIP. No difference in sensitivitybetween IIF and lIP assays was observed; therefore, the assays were usedinterchangeably. Slides were read by independent readers, and the stainingpatterns were classified as described in the figure legends.

Immunochemical Assays. Radiolabeling of cells with 125!using the lactoperoxidase method, metabolic labeling with [35S)methionine,solubilizationwith NP-40, and indirect immunoprecipitation with mAb using protein A-

Sepharose CL-4B coated with rabbit anti-mouse immunoglobulin antibodieswere performed as described previously (27). One dimensional SDS-PAGEwas performed under reducing and nonreducing conditions in 3—15%polyacrylamide gradient gels using the Laemmli buffer system (23). Gels containing 25I-labeled samples were processed for autoradiography using KOdakXAR-5 films (Eastman Kodak Company, Rochester, NY). Gels containing[35Sjmethionine-labeledsamples were fluorographed as described by Bonnerand Laskey (28).

Statistical Analysis. Fisher's Exact test was used to analyze differences in

expression of a@chain and rtv@33complex in benign melanocyte lesions andthe distribution of these two molecules in primary and metastatic melanomalesions. The relationship of a@ chain, rtv@33complex, and ICAM-l expressionin primary lesions with their thickness was analyzed using the Mann-Whitney

test. The significance of the association of a@j33and !CAM-l expression inprimary lesions with patients' disease-free survival was analyzed by applyingthe Log-Rank test to Kaplan-Meier estimates of survival curves.

RESULTS

Specificity of mAbs VF27.263.15 and TP36.1. Both mAbsVF27.263. 15 and TP36.l reacted in ELISA with cultured humanmelanoma cells Cob 38 and did not react with cultured humanB-lymphoid cells LG-2. The association constants of mAbsVF27.263.15 and TP36.l for Cob 38 cells are 1.8 X l0@ M1 and3.3X l0@M1,respectively.

The SDS-PAGE profiles under reducing and nonreducing condi

M21-Lllb

av

on April 23, 2020. © 1997 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

Table2 Expressionof integrina,, chain and a@j3@complexin benign andmalignantlesionsof melanocyteorigin―Lesionsa,@

chainaj33 complex

13/160/150/40/42/42/41/10/15/54/511/2111/210/80/8

ay@3 INTEGRIN. ICAM-l. AND CLINICAL COURSE OF MELANOMA

Fig. 5. Immunoperoxidase staining with anti-a,., chain mAb TP36.l andwith anti-a@f33 complex mAb VF27.263.l5 of frozen sections of surgicallyremovedbenignandmalignantlesionsofmelanocyteorigin.Cryostatsections of surgically removed intradermal nevi (A and B) and primary (Cand D) and metastatic(E and F) melanomalesionswere stainedwithanti-a@@3 ch5ln mAb TP36.l (A, C, and E) or anti-avP3 complex mAb

VF27.263.l5 (B, D, and F) in an avidin-biotin complex immunoperoxidasereaction.mAbTP36.l (A)stainsbasalkeratinocytes(thinarrows)andnevic cells (large arrows), whereas mAb VF27.263.lS (B) stains neithercell type. The two mAbs stain primary melanoma lesions with variableintensity (C and D) but display a strong and homogeneous staining ofmetastatic lesions (E and F). Slides were developed with aminoethylcarbazole as the chromogenic substrate and were counterstained with Mayer'shematoxylin. A and B, x 120; C and D. X200; E and F, X300.

VF27.263.15 reacted only with M2l melanoma cells and with purifiedaVP3 complex (Figs. 3 and 4). The results of cross-blocking expen

ments suggest that mAb TP36.l recognizes a determinant that isexpressed on the cell surface in close proximity, if it is not identicalto that recognized by mAb LM142, because the two mAbs competedfor binding to Cob 38 cells (Table 1). The determinant(s) recognizedby mAbs TP36.l and LM142 is distinct and spatially distant from thatdefined by mAbs VF27.263.15 and LM609, because these mAbsfailed to interfere with the binding of mAbs TP36.l and LM142 toCob 38 cells. mAbs VF27.263. 15 and LM609 recognize spatiallyclose, if not identical, determinants because they inhibit each other'sbinding to target cells (Table 1).

Expression of av Chain and aVI@3Complex in Benign andMalignant Lesions of Melanocyte Origin. Fig. 5 shows representative staining patterns by mAbs VF27.263.15 and TP36.l of benignand malignant lesions of melanocyte origin. The results are summarized in Table 2 based on immunohistochemical evaluation of arepresentative panel of surgical biopsies. It is readily apparent that the

BenignIntradermal neviBluenevi

MalignantPrimary

ALMLMMNMSSM

MetastaticLymph node 14/24 14/24

Cutaneous @.!1@@_________________ _________aSurgicallyremovedbenignandmalignantlesionswerestainedinindirectimmu

noperoxidasereactionswith anti-a@chain mAbTP36.l and anti-a@P3complexmAbVF27.263.l5. Results are indicated as number of positive lesions/number of tested lesions.

b UM, uveal melanoma.

1557

.. ..@‘: -@@@ -:; , ..@ I ,@@@ @.‘

.a@ !J@ . @s5@@@V..@@

@*@Ec.@ :@“!:‘@ ..‘

reactivity pattern of mAb VF27.263.15 was associated with malignanttransformation of melanocytes, whereas that of mAb TP36. I was not.Over a wide range of mAb concentrations (5—50p.g/ml), mAbVF27.263.l5 stained none of the intradermal nevi tested, whereasmAb TP36. 1 stained 87% of them, a difference that was highlysignificant (P < 0.004). In contrast, the reactivity of the two mAbswith a panel of primary and metastatic melanoma lesions was concordant in 86% of the cases. Both mAbs stained approximately 45%of primary and 65% of metastatic melanoma lesions. The difference inthe proportion of primary and metastatic lesions stained by the twomAbs was not significant (P > 0.09). The two mAbs displayeddifferential reactivity with primary melanoma lesions of differenthistotype. Both mAbs stained about 50% of ALM and SSM lesionsand at least 80% of NM lesions. Neither mAb stained any of the uvealmelanoma lesions tested. Within the group of metastatic lesions, bothmAbs stained a higher percentage of cutaneous than of lymph node

metastases. However, the difference did not achieve statistical significance(P > 0.18).Thereactivitypatternsof thetwomAbswithautologous metastases are concordant. A summary of the results isshown in Table 3, which includes for comparison purposes the staining by anti-ICAM-l mAb CL203.4. A high degree of concordancewas found in the expression of a@ chain, aV@33complex, and ICAM- Iin the set of metastases analyzed.

Relationship of Staining by anti-a@I33 Complex mAbVF27.263i5 and by Anti-ICAM-1 mAb CL203.4 of Primary Lesions with Their Thickness and with the Clinical Course of the

Disease. The Mann-Whitney test was applied to determine whetherthe staining of primary lesions from 32 patients by mAbsVF27.263.l5 and CL203.4 was associated with lesion thickness measured according to Breslow ( 17). The reactivity of mAbs VF27.263. 15and CL203.4 with primary tumors was significantly (P = 0.047 andP —0.009, respectively) associated with lesion thickness. This histopathological parameter was more strongly associated (P < 0.005)with the simultaneous expression of both markers (Table 4). Theassociation between disease-free survival and expression of aV133and

.@@

@-::

@:T―b::@

on April 23, 2020. © 1997 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

PatientMetastasesa@ chaina)f3@complexICAM-lAI

2var varvar varvarvarBI

234var

varvarNT'@var

varvarvar++

++++

++C1

23++

++++

++NT

var—DI

2var —var —+ NT

av133INTEGRIN, ICAM-I. AND CLINICAL COURSE OF MELANOMA

Table 3 Expression of integrin a@chain, aj3@ complex and ICAM-l in autologousmelanoma metastases―

were stained by both mAbs or by neither was compared. The probability of recurrent disease was significantly (P —0.001) greater forpatients whose primary lesions were stained by both mAbs than forthose whose lesions were stained by neither mAb (Fig. 6C). Notably,all patients whose primary lesions lacked both avf33 and ICAM-lexpression were disease-free 7 years following surgery.

Expression of a@ Chain and aVP3 Complex in Normal Adult

Tissues of Nonlymphoid Origin. Because the aV/33 complex hasbeen proposed as a target for mAb-mediated immunotherapy of malignant diseases (30), we analyzed the distribution of the aV133complex in normal adult tissues and compared it to that of the av chain.The results of this analysis are summarized in Table 5. Both moleculeshave a restricted tissue distribution. Notably, the glomeruli and Bowman's capsule of the kidney exhibited staining of both a@ chain and

aV1@3complex. In all other organ systems, staining by anti-cry chainmAb TP36.l and by anti-a@P3 complex mAb VF27.263.lS was eithernegative or restricted to the basal or basolateral membranes of thetissues. Two exceptions to this general pattern were the staining ofductal epithelium of the pancreas and biliary ducts of the liver byanti-a@ chain mAb TP36.l. The distribution of the aV(33 complex wasmore restricted than that of a@ chain. Epidermis, lung bronchi, stomach surface epithelium, urinary bladder, prostate, and tubal epitheliumwere not stained by anti-a@433 complex mAb VF27.263.l5 but werestained by anti-a@ chain mAb TP36.l.

DISCUSSION

Immunohistochemical staining of surgically removed benign andmalignant lesions of melanocyte origin with mAbs TP36.l and

a Nonconsecutive sections of metastalic melanoma lesions were stained in indirectimmunoperoxidase reactions with anti-a@ chain mAb TP36. 1, anti-a@(33 complex mAbVF27.263.15, and anti-ICAM-l mAb CL203.4. The staining patterns were classified as:—, when no staining was detected; var, when the intensity of staining was variable; +,

when the staining was homogenous; and + + , when the staining was strong and homogeneous.

b NT, not tested.

ICAM-l in primary tumors was examined in 20 patients with stage Imelanoma who underwent excisional surgery and had at least 7 yearsof follow-up. Kaplan-Meier estimates of the survival curve for eachpatient group were tested for significant differences using the LogRank test. Patients bearing a@433-positive tumors had a significantly(P = 0.024) greater probability of disease recurrence within 7 yearsthan those with primary lesions without detectable aV(33 expression(Fig. 6A). ICAM-l expression was also associated with a significantly(P = 0.040) greater probability of disease recurrence (Fig. 6B). Theassociation with clinical outcome became even more pronounced,when the disease-free survival of the patients whose primary lesions

Table 4 ICAM-! and a@@3sexpression. priinar@melanoma lesion thickness, and disease-free interval ofpatients with stage II melanoma―

RecurrentThickness DFIb disease

Patient no. Histotype (mm) (mos) and site complexc ICAM@lc

1 SSM 0.2 108 — — —2 SSM 0.4 53 — + —3 SSM 0.5 60 — + —4 SSM 0.5 84 — — —5 SSM 0.8 NA― NA + +6 SSM 0.8 7 +vise + —7 SSM 1.0 74 +brain — +8 SSM 1.0 53 — — +9 SSM 1.0 62 — — —

10 ALM 1.0 98 — — —11 LMM 1.2 97 — — —12 SSM 1.5 69 — + —13 SSM 1.5 72 — + +14 SSM 1.5 150 — — +15 SSM 1.8 NA NA + +16 SSM 1.9 103 — + +17 SSM 2.0 84 — — —18 SSM 2.0 NA NA — +19 SSM 2.0 144 — + —20 SSM 3.2 NA NA — +21 ALM 3.5 9 +skin + +22 NM 3.5 19 +brain + +23 SSM 3.6 120 — + +24 NM 4.0 134 — + —25 NM 4.0 7 +brain + +26 SSM 4.5 NA NA + +27 SSM 4.5 12 +vis + +28 ALM 4.5 26 — + +29 NM 5.0 11 +vis + +30 ALM 5.0 42 +skin — +31 NM 6.0 16 +lung + —32 NM 6.0 10 +skin + +

a Surgically removed primary melanoma lesions were stained in indirect immunoperoxidase reactions with anti-aj33 complex mAb VF27.263.l5 and with anti-ICAM-l mAbCL203.4. Lesions containing at least 20% of tumor cells stained by each mAb were scored positive. The percentage of stained cells ranged between 20 and 100.

b DR. disease-free interval.

C The association between marker expression and lesion thickness was analyzed by the Mann-Whitney test. The Ps were 0.047, 0.009, and 0.005 for@ for ICAM-l, and for

@ and ICAM-l, respectively.for a one-sidedtestof significance.NA, no information is available about the disease-free interval and the recurrence of disease.

C vis, visceral.

1558

on April 23, 2020. © 1997 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

av@sINTEGRIN,ICAM.I.ANDCLINICALCOURSEOFMELANOMA

100

Fig. 6. Relationship of av@s complex andICAM-l expression in primary lesions with disease-free interval in patients with stage I melanoma. The expression of av(3s complex andICAM-l in primary lesions as variables associatedwith the disease-free interval of patients with stageI melanoma was analyzed by the Log-Rank test ofsignificance between Kaplan-Meier estimates ofsurvival curves. The probability of remaining cbsease free was significantly lower for patientswhoselesionswerestainedby anti-ICAM-lmAbCL203.4 (A; P = 0.04) by anti-a,433 complexmAb VF27.263.15 (B; P = 0.02) and by antiICAM-l mAb CL203.4 and anti-a@@3 complexmAb VF27.263.lS (C; P = 0.01; continuous linein each panel) than for those whose lesions werenot stained by the corresponding mAb (broken linein each panel).

I

5

Years

VF27.263.l5, which we have shown to be specific for the a@ chainand aVj33 complex, respectively, has corroborated some of the datapublished previously and has provided novel information. The a@chain and the aVP3 complex are highly concordant in their expressionin primary and metastatic melanoma lesions but display a differentialexpression in benign and malignant lesions. In agreement with the

information in the literature (5, 10—12),we have found that the a@j33complex is a marker of malignant transformation of melanocytes,because it is not detectable in benign lesions but is expressed in atleast 50% of primary cutaneous melanoma lesions and in at least 60%of metastatic lesions. In contrast, the a@ chain is expressed in at least50% of both benign and malignant lesions, suggesting that a@,subunits associate with @3subunits other than@ in benign lesions of

Table 5 Distribution of integrin a,, chain and aj33 complex in normal adult humantissues of nonlymphoid origin

Tissue a@ chain a,j33 complex

Epidermal keratinocytes Basal —Brain — —Parotid

Ductal epithelium Basolateral BasolateralAcinar epithelium Basal -

Thyroid thyrocytes Basolateral BasolateralMammary epithelium var —Lung

Alveoli — —Bronchi + —Bronchial glands Basal -

Esophagus — —Stomach

Basal glands Basal/cytoplasmic BasalSurface epithelium Basolateral —

Colon-rectum var BasalLiver biliary ducts + NT@'Pancreaticductalepithelium + —Kidney

Proximal tubules Basal BasalDistal tubules Basal/cytoplasmic Basal/cytoplasmicGlomeruli + +Bowman's capsule + +

Urinary bladder Basal —Prostate Basal -Testis - -Ovary - -Endometrialepithelium Basolateral BasolateralTubal epitheium Basolateral —

a Surgically removed normal tissues were stained in indirect immunoperoxidase reactions with anti-a,@chain mAb TP36.l and with anti-aj3@ complex mAb VF27.263.15. Thestaining patterns were classified as: —,when no staining was detected; Basal, when thestaining was restricted to basal membranes; basolateral, when both basal and lateral cellmembraneswere stained;cytoplasmic,when cytoplasmwas stained;var, when thestaining intensity was variable; and +, when membrane and cytoplasm were stained.

b NT, not tested.

melanocyte origin. Both the a@ chain and the avf3i complex aredifferentially expressed in primary lesions of different histotype. Inagreement with Ten Berge et a!. (3 1), we have found that neithermolecule is expressed in uveal melanoma lesions. The biologicalsignificance of the latter finding remains to be determined.

The distribution of the aVf33 complex in lesions of melanocyteorigin resembles that of ICAM-l in some characteristics but alsodisplays distinct features. Like ICAM-l (6—9),avf33 has a significantly higher expression in malignant than in benign lesions. Furthermore, the expression of both markers in primary melanoma lesions isassociated with their thickness. However, the expression of ICAM- 1by melanoma cells is modulated by cytokines such as IFN-'y, interleukin 1, and tumor necrosis factor-a (32, 33), whereas that of theaVP3 complex is not (18). The latter difference may have a bearing onthe mechanism(s) underlying the association of the expression of theaVP3 complex and ICAM- I with the thickness of primary melanoma

lesions. The association for ICAM- I may reflect the secretion ofcytokines by T cells infiltrating the melanoma lesions, whereas theassociation for the a@j33 complex may reflect an intrinsic characteristic of malignantly transformed melanocytes. Furthermore, ICAM-l

has a significantly higher expression in metastases than in primarylesions (6—9),whereas the a@f33complex has a similar distribution inboth types of lesions. The latter finding parallels that of Danen et a!.(12), who reported a lack of association between a@43@complexexpression by melanoma cell lines and their metastatic potential innude mice.

Experiments in animal model systems have demonstrated the role

of czVf33in tumorigenicity and dissemination of melanoma cells tolymph nodes (5, 10—12,34). Our data in a clinical setting do not arguein favor of lymph node dissemination but strongly suggest that aV@33expression plays a role in the aggressive phenotype of melanomacells. We have demonstrated for the first time a significant associationbetween a@P3 expression in primary melanoma lesions and diseaserecurrence in long-term follow-up of patients with stage I disease whohad undergone surgical excision of their primary lesions. ICAM-lexpression was also associated with poor prognosis, consistent withour findings from a previous study (7). Several mechanisms derivedfrom animal model systems can account for these findings. Nip et a!.(34) have shown that the aj33 complex mediates the attachment andmigration of tumor cells over extracellular matrix proteins, and morerecently, Brooks et a!. (35) have reported a direct interaction of thematrix-degrading metalloproteinase MMP-2 with the aj33 complex

on the surface of melanoma cells. The latter association led to enhanced levels of proteolytically active MMP-2 and facilitated tumor

1559

on April 23, 2020. © 1997 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

@V@3INTEGRIN. ICAM- I . AND CLINICAL COURSE OF MELANOMA

14. Smith, J. W., and Cheresh, D. A. The Arg-Gly-Asp binding domain of the vitronectinreceptor photoaffinity cross-linking implicates amino acid residues 61—203of the 13subunit. J. Biol. Chem., 263: 18726—18731, 1988.

15. Fitzgerald, L. A., Leung, B., and Phillips, D. R. A method for purifying the plateletmembrane glycoprotein lIb-Illa complex. Anal. Biochem., 151: 169—177,1985.

16. Clark, W. H., Jr., Ainsworth, A. M., Bernardino, E. A., Yang, C-H., Mihm, M. C., Jr.,and Reed, R. S. The developmental biology of primary cutaneous malignant melanoma. Semin. Oncol., 2: 83—103,1975.

17. Breslow, A. Tumor thickness, level of invasion and node dissection in stage Icutaneous melanoma. Ann. Surg., 182: 572—575,1975.

18. Hamby, C. V., Kageshita, T., Wang, Z., Liang, B., Temponi, M. and Ferrone, S.Analysis of aJ133 integrin expression by cultured human melanoma cells. In: Y. Hon.V. J. Hearing, and J. Nakayama (eds.), Melanogenesis and Malignant Melanoma:Biochemistry, Cell Biology, Pathophysiology, Diagnosis and Treatment, pp. 185—194. New York: Elsevier, 1996.

19. Niiya. K., Hodson, E., Bader, R., Byers-Ward, V., Koziol, J. A., Plow, E. F., andRuggen, Z. M. Increased surface expression of the membrane glycoprotein lIb/lIlacomplex induced by platelet activation relationship to the binding of fibnnogen andplatelet aggregation. Blood, 70: 475—483,1987.

20. Imai, K., Wilson, B. S., Bigoni, A., Natali, P. G., and Ferrone, S. A 94,000-Daltonglycoprotein expressed by human melanoma and carcinoma cells. J. Natl. CancerInst., 68: 761—769,1982.

21 . Matsui, M., Temponi, M., and Ferrone, S. Characterization of a monoclonal antibodydefined human melanoma-associated antigen susceptible to induction by immuneinterferon. J. Immunol., 139: 2088—2095, 1987.

22. Temponi, M., Kageshita, T., Perosa, F., Ono, R., Okada, H., and Ferrone, S. Purification of murine IgG monoclonal antibodies by precipitation with caprylic acid:comparison with other methods of purification. Hybndoma, 8: 85—95,1989.

23. Laemmli. U. K. Cleavage of structural proteins during the assembly of the head ofbacteriophage 14. Nature (Land.), 227: 680—685, 1970.

24. Fraker, P. J., and Speck, i. C., Jr. Protein and cell membrane iodinations with asparingly soluble chloroamide l,3,4,6-tetrachloro-3a,6a-diphenyl-glycolutil. Biochem. Biophys. Res. Commun., 80: 849—857, 1978.

25. Lindmo, T., Boven, E., Cunina, F., Fedorko, J., and Bunn, P. A., Jr. Determination ofthe immunoreactive fraction of radiolabeled monoclonal antibodies by linear extrapolalion to binding at infinite antigen excess. J. Immunol. Methods, 72: 77—89,1984.

26. Scatchard, G. The attractions of proteins for small molecules and ions. Ann. NYAced. Sci., 51: 660—672,1949.

27. Zweig, S. E., and Shevach, E. H. Production and properties of monoclonal antibodiesto guinea pig Ia antigens. Methods Enzymol., 92: 66—85,1983.

28. Bonner, W. M., and Laskey, R. A. A film detection method for tritium-labelledproteins and nucleic acids in polyacrylamide gels. Eur. J. Biochem., 46: 83—88,1974.

29. Wayner, E. A., Orlando, R. A., and Cheresh, D. A. Integrins av133 and o-43@contribute to cell attachment to vitronectin but differently distribute on the cellsurface. J. Cell Biol., 113: 919—929, 1991.

30. Brooks, P. C., Montgomery, A. M. P., Rosenfeld, M., Reisfeld, R. A., Hu, T., Klier,G., and Cheresh, D. A. Integrin avPs antagonists promote tumor regression byinducing apoptosis of angiogenic blood vessels. Cell, 79: 1157—1164, 1994.

31. Ten Berge, P. J. M., Danen, E. H. J., van Muijen, G. N. P., Jager, M. J., and Ruiter,D. J. Integrin expression in uveal melanoma differs from cutaneous melanoma. Invest.Ophthalmol. Visual Sci., 34: 3635—3640.1993.

32. Maio, M., Gulwani, B., Langer, J. A., Kerbel, R. S., Duigou, G. J., Fisher, P. B., andFerrone, S. Modulation by interferons of HLA antigen, high-molecular-weight melanoma-associated antigen, and intercellular adhesion molecule I expression by cultured melanoma cells with different metastatic potential. Cancer Res., 49: 2980—2987, 1989.

33. Maio, M., Gulwani, B., Tombesi, S., and Ferrone, S. Modulation by cytokines ofHLA antigens, intercellular adhesion molecule I and high molecular weight melanoma associated antigen expression and of immune lysis of clones derived from themelanoma cell line MeM 50—10.Cancer Immunol. Immunother., 30: 34—42,1989.

34. Nip, J., Shibata, H., Loskutoff, D. J., Cheresh, D. A., and Brodt, P. Human melanomacells derived from lymphatic metastases use integrin a@f33to adhere to lymph nodevitronectin. J. Clin. Invest., 90: 1406—1413,1992.

35. Brooks, P. C., Stromblad, S., Sanders, L. C., von Schlascha, T. L., Aimes, R. 1.,Stetler-Stevenson, W. G., Quigley, J. P., and Cheresh, D. A. Localization of matrixmetalloproteinase MMP-2 to the surface of invasive cells by interaction with integrinav133. Cell, 85: 683—693,1996.

36. Vuori, K., and Ruoslabti, E. Association of insulin receptor substrate-I with integnns.Science (Washington DC), 266: 1576—1578, 1994.

37. Montgomery, A. M. P., Reisfeld, R. A., and Cheresh, D. A. Integrin av133 rescuesmelanoma cells from apoptosis in three-dimensional dermal collagen. Proc. NatI.Acad. Sci. USA, 91: 8856—8860, 1994.

38. Aced, P. A., Nakajima, M., and Welch, D. R. The role of polymorphonuclearleukocytes (PMN) on the growth and metastatic potential of 13762NF mammaryadenocarcinoma cells. Int. J. Cancer, 42: 748—759,1988.

39. Becker, J. C., Dummer, R., Hartmann, A. A., Burg, G., and Schmidt, R. E. Sheddingof ICAM-l from human melanoma cell lines induced by !FN-y and tumor necrosisfactor-a. J. Insmunol., 147: 4398—4401,1991.

40. Altomonte, M., Gloghini, A., Bertola, G., Gasparollo, A., Carbone, A., Ferrone, S.,and Maio, M. Differential expression of cell adhesion molecules CD54/CDI Ia andCD58/CD2 by human melanoma cells and functional role in their interaction withcytotoxic cells. Cancer Res.. 53: 3343—3348, 1993.

1560

cell degradation of collagen matrix. Tumor cell proliferation may also beinfluenced by the (Xj33complex because it has been shown to bind withinsulin receptor substrate-l and may synergize with growth factors instimulating cell proliferation (36). Finally, the a@I33complex has beenshown to protect melanoma cells against apoptotic cell death in vitro (37).These mechanisms are distinct from those suggested for the associationof ICAM-l with a poor prognosis. ICAM-1 can possibly mediate tumorcell invasion by promoting aggregate formation of neutrophils and tumorcells in the circulation with subsequent lodging and invasion into tissue(38), and/or it may have an immunoprotective effect due to sheddingfrom tumor cells, thereby preventing effective contact and destruction bycytotoxic T cells (39, 40). The different roles of a@@33and ICAM-l in the

biology of melanoma cells may account for their complementary effect inthe clinical course of disease.

In view of the suggested application of anti-a@P3 mAb for immunotherapy of malignant diseases (30), the distribution of this moleculein normal tissues deserves some comment: (a) the aVf33 complex hasa more restricted distribution in normal tissues than the a,,, chain; and(b) theexpressionof thea@P3complexon glomeruliandon distaltubules in kidney suggest that renal function will be a critical clinicalparameter to monitor in therapies of malignant diseases relying onsystemic administration of anti-a@j33 mAb.

ACKNOWLEDGMENTS

We acknowledge the excellent secretarial assistance of Harriett V. Harrison,Donna Jones, Edwina L. Jones, Maria V. Sarcone, and Kathleen Williams.

REFERENCES

I . Johnson, J. P. Cell adhesion molecules of the immunoglobulin supergene family andtheir role in malignant transformation and progression to metastatic disease. CancerMetastasis Rev., 10: 11—22,1991.

2. Cheresh, D. A., and Spiro, R. C. Biosynthetic and functional properties of anArg-Gly-Asp-directed receptor involved in human melanoma cell attachment tovitronectin, fibrinogen, and von Willebrand factor. J. Biol. Chem., 262: 17703—17711,1987.

3. Felding-Habermann, B., Romerdahl, C. A., and Cheresh, D. A. Involvement ofintegrin a@gene expression in human melanoma tumorigenicity. J. Clin. Invest., 89:2018—2022, 1992.

4. Giavazzi, R., Chirivi, R. G. S., Garofalo, A., Rambaldi, A., Hemingway, I., Pigott, R.,andGearing,A.J. H.SolubleintercellularadhesionmoleculeI is releasedbyhumanmelanoma cells and is associated with tumor growth in nude mice. Cancer Res., 52:2628—2630,1992.

5. Albelda, S. M., Mette, S. A., Elder, D. E., Stewart, R., Damjanovich, L., Herlyn, M.,and Buck, C. A. Integrin distribution in malignant melanoma: association of the@subunit with tumor progression. Cancer Res., 50: 6757—6764, 1990.

6. Johnson,J. P., Stade, B. G., Holzmann, B., Schwäble,W., and Riethmtiller, G. Dcnova expression of intercellular-adhesion molecule I in melanoma correlates withincreased risk of metastasis. Proc. NatI. Acad. Sci. USA, 86: 641—644,1989.

7. Natali, P. G., Nicotra, M. R., Cavaliere, R., Bigotti, A., Romano, G., Temponi, M.,and Ferrone,S. Differentialexpressionof intercellularadhesionmolecule-Iin primary and metastatic melanoma lesions. Cancer Res., 50: 1271—1278,1990.

8. Carrel, S., Dorè,i-F., Ruiter, D. I., Prade, M., Lejeune, F. J., Dleeberg, U. R., RUmke,P., and Brocker, E. B. The EORTC melanoma group exchange program: evaluationof a multicenter monoclonal antibody study. Int. I. Cancer, 48: 836—847,1991.

9. Kageshita, T., Yoshii, A., Kimura, T., Kuriya. N., Ono, 1., Tsujisaki, M., Imai, K.,and Ferrone, S. Clinical relevance of ICAM-l expression in primary lesions andserum of patients with malignant melanoma. Cancer Res., 53: 4927—4932, 1993.

10. Si, Z., and Hersey, P. Immunohistobogical examination of the relationship betweenmetastatic potential and expression of adhesion molecules and “selectins―on melanoma cells. Pathology, 26: 6—15,1994.

I I. Danen, E. H. J., Ten Berge, P. J. M., Van Muijen, G. N. P., Van't Hof-Grootenboer,A. E., Brocker, E. B., and Ruiter, D. J. Emergence of a5@l fibronectin- and av@33vitronectin-receptor expression in melanocytic tumour progression. Histopathology,24: 249—256,1994.

12. Danen, E. H. J., Jansen, K. F. I., Van Kraats, A. A., Cornelissen, I. M. H. A., Ruiter,D. J., and Van Muijen, G. N. P. a@-integrins in human melanoma: gain of a@@33andloss of a@35 are related to tumor progression in situ but not to metastatic capacity ofcell lines in nude mice. Int. J. Cancer, 61: 491—496, 1995.

13. Felding-Habermann, B., Habermann, R., SaldIvar, E., and Ruggeri, Z. M. Role of f3@integrins in melanoma cell adhesion to activated platelets under flow. I. Biol. Chem.,271: 5892—5900,1996.

on April 23, 2020. © 1997 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

1997;57:1554-1560. Cancer Res   Pier Giorgio Natali, Carl V. Hamby, Brunhilde Felding-Habermann, et al.   LesionsMolecule-1 Expression in Cutaneous Malignant Melanoma

Integrin and Intercellular Adhesion3βvαClinical Significance of

  Updated version

  http://cancerres.aacrjournals.org/content/57/8/1554

Access the most recent version of this article at:

   

   

   

  E-mail alerts related to this article or journal.Sign up to receive free email-alerts

  Subscriptions

Reprints and

  .pubs@aacr.orgDepartment at

To order reprints of this article or to subscribe to the journal, contact the AACR Publications

  Permissions

  Rightslink site. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC)

.http://cancerres.aacrjournals.org/content/57/8/1554To request permission to re-use all or part of this article, use this link

on April 23, 2020. © 1997 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from