Absence of lymphangiogenesis and intratumoural lymph vessels in human metastatic breast cancer

Absence of lymphangiogenesis and intratumoural lymph vessels in human metastatic breast cancer
of 12
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Related Documents
   Journal of Pathology  J Pathol   2003;  200 : 195–206.Published online 14 March 2003 in Wiley InterScience (  DOI:  10.1002/path.1343 Original Paper  Absence of lymphangiogenesis and intratumoural lymphvessels in human metastatic breast cancer  Cory SM Williams, 1 Russell D Leek, 2 Alistair M Robson, 1 Suneale Banerji, 3 Remko Prevo, 3 Adrian L Harris 2 and David G Jackson 3 * 1 Nuffield Department of Clinical Laboratory Sciences, John Radcliffe Hospital, Headington, Oxford OX3 9DS, UK  2 ICRF Molecular Oncology Laboratory, Weatherall Institute of Molecular Medicine, John Radcliffe Hospital, Headington, Oxford OX3 9DS, UK  3  MRC Human Immunology Unit, Weatherall Institute of Molecular Medicine, John Radcliffe Hospital, Headington, Oxford OX3 9DS, UK *Correspondence to:Dr David G Jackson, MRC Human Immunology Unit,Weatherall Institute of Molecular  Medicine, John Radcliffe Hospital,Headington, Oxford OX3 9DS,  Received: 2 July 2002Revised: 11 September 2002Accepted: 12 December 2002 Abstract Early metastasis to lymph nodes is a frequent complication in human breast cancer.However, the extent to which this depends on lymphangiogenesis or on invasion of existinglymph vessels remains controversial. Although proliferating intratumoural lymphaticsthat promote nodal metastasis have been demonstrated in experimental breast tumoursoverexpressing VEGF-C, it has yet to be determined whether the same phenomena occurin spontaneous human breast cancers. To address this important issue, the present studyinvestigated the lymphatics in primary human breast carcinoma (75 cases of invasive ductaland lobular breast cancer) by quantitative immunohistochemical staining for the lymphaticendothelial hyaluronan receptor LYVE-1, the blood vascular marker CD34, and the nuclearproliferation marker pKi67. None of the breast carcinomas was found to contain dividinglymph vessels, even in areas of active haemangiogenesis. Furthermore, the majority of non-dividing lymph vessels were confined to the tumour periphery where their incidence waslow and unrelated to tumour size, grade or nodal status; rather, their density was inverselycorrelated with tumour aggressiveness as assessed by macrophage density (  p  = 0 . 009), andblood microvessel density (  p  = 0 . 05, Spearman Rank), as well as with distance from thetumour edge. Finally, a proportion of the peritumoural lymphatics contained tumour emboliassociated with hyaluronan, indicating a possible role for LYVE-1/hyaluronan interactionsin lymphatic invasion or metastasis. These results suggest that naturally occurring breastcarcinomas invade and destroy lymph vessels rather than promoting their proliferation; thatbreast tumour lymphangiogenesis may not always occur at physiological VEGF-C levels; andthat nodal metastasis can proceed via pre-existing lymphatics. Copyright © 2003 John Wiley& Sons, Ltd.Keywords: lymphangiogenesis; LYVE-1; metastasis; breast cancer; VEGF-C Introduction Breast carcinoma is one of the most frequently diag-nosed cancers in women. However, despite earlydetection, and apparently curative surgery, manypatients still die of metastatic disease that remainsundetected at presentation [1]. These hidden micro-metastases are thought to reside not only in distant,highly vascularized tissues such as the bone marrow,as the result of dissemination through the blood vascu-lature, but also in local sites such as the axillary lymphnodes as the result of dissemination through lymphaticvessels [2,3]. The lymphatic route is important andone of the most widely used prognostic indicators inbreast cancer is the presence of tumour within a drain-ing axillary node. Yet in comparison with the tumourblood vasculature, which has been extensively studiedin breast cancer [4], relatively little is known abouthow lymph vessels are generated or invaded in neo-plastic disease. Thus, in breast, as in other cancers,it is not clear whether tumours promote lymphangio-genesis or whether pre-existing lymphatics provide themain conduit for nodal metastasis [5,6].Of the few lymphangiogenic factors identified sofar, the best characterized are members of the vascularendothelial growth factor (VEGF) family. In particu-lar, VEGF-C has been shown to promote lymphangio-genesis both  in vitro  and  in vivo  [7–9] through bindingto its primary tyrosine kinase-linked receptor VEGFR3expressed on lymphatic endothelium. The possibilitythat VEGF-C might also promote tumour lymphangio-genesis has received much attention. Indeed, a numberof clinical studies have reported VEGF-C expressionin tumours [10] and have demonstrated a significantassociation between tumour VEGF-C levels and eitherlymph node metastasis [11–15] or vessel invasion invarious human cancers [16,17]. Although these stud-ies appear to confirm that VEGF-C promotes tumourlymphangiogenesis  in vivo  [18–20], other reports havebeen contradictory [11,21,22] and the evidence for Copyright 󰂩 2003 John Wiley & Sons, Ltd.  196 CSM Williams  et al lymphangiogenesis in human cancer has remained cir-cumstantial (reviewed in ref 23). Just recently, theartificial overexpression of VEGF-C by orthotopicallytransplanted MDA-435 and MCF-7 breast carcinomashas been shown to induce both intratumoural lym-phangiogenesis and subsequent lymph node metastasisin nude mice [24–26]. These groundbreaking experi-ments provide proof of principle that lymphatic vesselscan proliferate within tumours and serve as a conduitfor lymphogenic dissemination. However, it has yet tobe satisfactorily resolved whether the same phenom-ena occur at physiological levels of VEGF-C in breastcancers that arise spontaneously.In this study we have used antibodies to the lym-phatic endothelial hyaluronan receptor LYVE-1 tocharacterize lymphatic vessels in a large panel of human invasive breast cancers. In contrast to the ani-mal studies, we found few, if any, intratumoural lymphvessels in our study; only a low density in the peritu-moural areas; and no indication of lymphangiogenesiswithin either site. Rather, our evidence suggests thatbreast cancers disseminate to lymph nodes by invadingand destroying existing lymph vessels, implying thatlymphangiogenesis may not necessarily be involved. Materials and methods Paraffin-embedded tissue samples Breast carcinoma tissues from cases with detailed clin-ical, research, and survival data were retrieved fromthe archives of the Nuffield Department of ClinicalLaboratory Sciences, University of Oxford (Table 1).For fixation, tissues were treated overnight with 10%formal saline prior to paraffin embedding in an auto-mated process (NHS Pathology Laboratory, John Rad-cliffe Hospital, Headington, Oxford) involving passagethrough graded alcohols, xylene, and molten paraf-fin wax. Sections (5  µ m) were cut using a micro-tome (Leica Microsystems UK) and mounted on glassmicroscope slides for subsequent immunohistochemi-cal staining as described below. Antibodies Rabbit polyclonal antisera specific for human LYVE-1 were generated against soluble LYVE-1 Fc fusionprotein [27] and antibodies (Abs) reactive with thehuman IgG1 Fc region removed by chromatographyon human IgG-Sepharose columns. LYVE-1 reactiveAbs were then affinity-purified by chromatography oncolumns of human LYVE-1 Fc Sepharose and storedat  − 20 ◦ C prior to use.The CD34 monoclonal antibody (MAb) Qbend 10,the CD31 MAb JC/70a, and the CD68 MAb KP-1 were kindly provided by Professor DY Mason,Nuffield Department of Clinical Laboratory Sci-ence, Oxford, UK. The CD44v3-specific mouse MAb3G5 (IgG2b) has been described previously [28]. Table 1.  Clinico-pathological data for breast cancer cases Age, years—median/range 53/28–80Pre-menopausal ( < 50) 32Post-menopausal ( > 50) 43Lymph node status—negative/positive 56/19Tumour size, cm—median/range 2.5/1.0–5.5Small ( < 2) 5Large ( > 2) 70Histology—node-negative/node-positive (total)Ductal 44/15 (59)Lobular 5/0 (5)Mixed 5/2 (7)Other 2/2 (4)Grade–node-negative/node-positive (total)I 4/4 (8)II 21/7 (28)III 19/6 (25)ER—median/range 6.9/0–695ER-positive ( < 5 fmol/ µ l) 43ER-negative (  5 fmol/ µ l) 32CLVC—mean/range 2.33/0–6.5Frequency Low ( < 2.5) 35High ( ≥ 2.5) 40CBVC—mean/range 5.7/3–8.7Frequency Low ( < 4.5) 54High ( ≥ 4.5) 20Macrophage Index—median/range 8/0–57Frequency Low ( ≤ 12) 25High ( > 12) 48Intralymphatic tumour—No of negative/positive 65/10 CLVC = Chalkley lymph vessel count; CBVC = Chalkley blood vesselcount; ER = oestrogen receptor. The MAb MIB-1, specific for the pKi67 prolifer-ation marker; FITC-conjugated goat anti-mouse Ig;biotin-conjugated goat anti-rabbit Ig; and anti-mouseIg/alkaline phosphatase/anti-alkaline phosphatase(APAAP) complex were obtained from DAKO. Alexa488-conjugated goat anti-rabbit IgG, Alexa 568-conjugated goat anti-mouse IgG, and Alexa 568-conjugated streptavidin were from Molecular Probes.Texas red-conjugated goat anti-mouse Ig was fromSouthern Biotech. FITC-conjugated avidin was fromSigma.Biotinylated hyaluronan-binding protein (bHABP)was a kind gift from Dr Raija Tammi, University of Kuopio, Finland. Immunohistochemical staining Preparation of slides Paraffin-embedded breast cancer tissue sections (5  µ mthickness) were dewaxed in xylene, rehydrated throughsequential changes of alcohol and distilled water, andsubjected to antigen retrieval by pressure cooking in0.01  M  citrate buffer, pH 6.0, or microwave treatmentin DAKO antigen retrieval buffer, pH 6.0, followed byincubation in H 2 O 2  to quench endogenous peroxidaseactivity. Sections were then treated with 5% (v/v)normal swine serum (30 min). Copyright 󰂩 2003 John Wiley & Sons, Ltd.  J Pathol   2003;  200 : 195–206.  Absence of lymphangiogenesis in breast cancer 197 Single immunoperoxidase staining for LYVE-1 andCD34 For single staining, slides were incubated (30 min)with antibody to LYVE-1 (1  µ g/ml affinity-purifiedLYVE-1 Ig) or CD34 (1/5 diluted QBEND-10 tis-sue culture supernatant), followed by horseradishperoxidase-conjugated anti-rabbit or anti-mouse Igpolymer (DAKO peroxidase Envision kit). Thesewere then developed with diaminobenzidine (DAB),counterstained with haematoxylin, and mounted formicroscopy. Double immunoperoxidase/APAAP staining for Lyve-1/CD34, LYVE-1/MIB-1, and CD34/MIB-1 To distinguish between lymph and blood vessels, tissuesections were subjected to immunoperoxidase stain-ing with LYVE-1 affinity-purified Ig as describedabove, followed by incubation (30 min) with mouseCD34 (MAb QBEND-10) and detection with anti-mouse Ig/alkaline phosphatase anti-alkaline phos-phatase (APAAP) complex and new fuschin substrate.To detect proliferating cells, sections were firststained (30 min) with MIB-1 MAb and peroxidase-conjugated anti-mouse Ig polymer (Envision kit,Dako) prior to development with diaminobenzidine(DAB) using nickel chloride enhancement. Sectionswere then stained (30 min) with either LYVE-1(10  µ g/ml purified Ig or 1/100 diluted serum) orCD34 (QBEND-10, 10  µ g/ml) and developed with theappropriate anti-rabbit or anti-mouse Ig/alkaline phos-phatase anti-alkaline phosphatase (APAAP) complexand new fuchsin substrate. Double-immunofluorescence staining for LYVE-1/CD34 and LYVE-1/pKi67 Sections were subjected to double-immunofluore-scence staining with either LYVE-1/CD34 or LYVE-1/MIB-1, to distinguish lymph and blood vesselsand to detect dividing lymph vessel endothelialcells, respectively. For LYVE-1/CD34 staining, slideswere incubated with both primary antibodies together(LYVE-1, 10  µ g/ml purified Ig; CD34 MAb 1/5diluted hybridoma supernatant) for 20 min, followedby FITC-conjugated goat anti-mouse Ig, biotinylatedgoat anti-rabbit Ig (1/300), and Alexa 568-conjugatedstreptavidin (each at 1/100 dilution) for the sameperiod. For LYVE-1/MIB-1 staining, slides were incu-bated with both antibodies (10  µ g/ml each) for 20 min,followed by Alexa 488-conjugated goat anti-rabbit Ig,and Alexa 568-conjugated goat anti-mouse Ig (bothdiluted 1/250). Slides were in each case mounted inVectashield and viewed under a Zeiss Axioskop orAxiovert S100 fluorescence microscope equipped witha Hamamatsu C4742-95 digital camera and Improvi-sion OpenLab software. Double- and triple-fluorescence staining withLYVE-1/CD44v3/biotinylated HA-binding protein Sections were double-stained with antibodies toLYVE-1 and a tumour-associated splice variant (v3)of the CD44 HA-receptor to visualize intralymphatictumour emboli. Alternatively, sections were triple-stained with antibodies to LYVE-1, CD44v3, and withbiotinylated hyaluronan-binding protein (bHABP) toassess the association of intralymphatic tumourcells with HA. For double-fluorescence, slides wereincubated with a mixture of LYVE-1 antiserum(1/100 diluted) and CD44v3 MAb 3G5 (10  µ g/ml) for20 min, followed by a mixture of FITC-conjugatedgoat anti-rabbit Ig and Texas red-conjugated goatanti-mouse Ig (both at 1/100 dilution). For triple-fluorescence, slides were first incubated with bHABP(3  µ g/ml) overnight, followed by a mixture of LYVE-1 antiserum (1/100 diluted) and CD44v3 MAb 3G5(10  µ g/ml) for 20 min. This was followed by FITC-conjugated avidin, AMCA-conjugated goat anti-mouseIg, and Texas red-conjugated goat anti-rabbit Ig for20 min (1/100 diluted). In each case, the slides weremounted and viewed as described above. Immunostaining controls Paraffin sections of normal human small intestine wereused as a positive control for LYVE-1, CD34, andCD44 antibody staining. In breast carcinoma tissue,the level of background staining was assessed foreach marker by replacement of the relevant primaryantibody with equivalent concentrations of isotype-matched control MAbs or non-immune rabbit Ig asappropriate. Estimation of lymph vessel, blood microvessel, andmacrophage counts Slides were scored for lymph vessel count (CLVC)and blood vessel count (CBVC) using the guidelinesdescribed previously [29]. Briefly, LYVE-1- or CD34-stained tumour sections were scanned at low powerand three lymphatic or blood vessel hotspots were cho-sen. Each hotspot was examined in turn at 250 × mag-nification using an eyepiece-mounted Chalkley PointArray graticule (PYSER-SGI Ltd) and the numberof points on the eyepiece that coincided with posi-tively stained vessel endothelium for each hotspot wascounted. The average score was then determined foreach section and measurements were compared withclinical and prognostic data already available for thecases studied.Tumour-infiltrating macrophage density (macro-phage index, MØI) was determined in the three areasof densest macrophage infiltration, as assessed byCD68 immunostaining using a method analogous tothat for vascular enumeration [30]. For each area, theChalkley array graticule was used to assess the MØIusing a  × 25 objective lens with a  × 10 eyepiece. Thegraticule was rotated so that the maximum number of points was coincident with stained macrophages anda count was taken. The mean of the three Chalkleycounts was used for the subsequent statistical analysis. Copyright 󰂩 2003 John Wiley & Sons, Ltd.  J Pathol   2003;  200 : 195–206.  198 CSM Williams  et al Statistical analysis Statistical analyses were performed using StatView4.5. Correlations were assessed using the Spearmanrank correlation test for two continuous variables andthe Mann–Whitney  U  -test when one variable wascontinuous and the other categorical. Differences wereconsidered statistically significant for  p  values lessthan or equal to 0.05. Results LYVE-1-positive lymph vessels are mostly confinedto the periphery in invasive breast carcinoma The specificity of LYVE-1 for lymph vessels in themammary gland was first assessed by immunostainingnormal breast tissue. This revealed positive staining of vessels with the characteristic irregular morphology,empty lumina (devoid of red blood cells), and thinendothelium of lymph vessels around lobular glands,and within adipose tissue, smooth muscle, and dermis(Figure 1A). No vessels containing erythrocytes werereactive with LYVE-1 (not shown). These resultsconfirm the overall specificity of LYVE-1 for lymphvessels that we observed previously in a range of normal human tissues [27,31].Staining of a panel of 76 invasive breast carcinomasfor LYVE-1 revealed a generally sparse arrangementof LYVE-1-positive vessels in the vicinity of bothductal and lobular tumours (Figures 1 and 2). Wherepresent, these were morphologically similar to lymphvessels in normal breast tissue (Figure 1A), in somecases proximal to the tumour rim (Figure 1C), or morefrequently distal to the tumour mass in interveningregions of stromal tissue (Figures 1B, 2A, and 4B).The identity of the vessels was further investigatedby differential staining for LYVE-1 and the vascularendothelial mucoprotein CD34 (Qbend10 MAb), pre-viously shown to be a reliable marker for quantitatingtumour blood vessels [32]. Two-colour immunohisto-chemical staining (Figure 3) and immunofluorescencestaining (Figure 4) with these markers confirmed thatthe LYVE-1-positive vessels were distinct from CD34-positive blood vessels, but also revealed differences inthe organization of the two vessel types. For exam-ple, CD34-positive blood capillaries often formed anecklace around the circumference of the tumour massand made intimate contacts with the tumour mass(Figures 4B and 4C). In contrast, most of the LYVE-1-positive lymphatic vessels were found at a distancefrom the tumour rim.Virtually no LYVE-1-positive intratumoural lymphvessels were present within either ductal or lobularbreast carcinomas, as assessed by CD34 immunohis-tochemical staining (Figure 2A) and LYVE-1/CD34double-immunofluorescence (Figures 4B and 4C anddata not shown). This was in marked contrast totumour blood vessels, which permeated the inte-rior of the tumour mass (Figure 2B) and surround-ing peritumoural tissues (Figures 4B and 4C) in bothtypes of carcinoma. Significantly, the peritumourallymph vessels in some cases contained tumour emboliwithin the lumen (Figures 1D, 5A, 5B, and Table 1)that stained positively with antibody to a tumour-associated isoform (v3) of the CD44 hyaluronan recep-tor (Figures 5A and 5B). A proportion of the intralym-phatic tumour cells were also decorated with hyaluro-nan, as assessed by staining with biotinylated HA-binding protein (bHABP), a complex of the cartilageproteoglycans Aggrecan and Link protein that has beenused as a probe for tissue hyaluronan. These latter find-ings raise the interesting possibility that tumour cellswithin draining lymphatics may bind HA and thusinteract with the vessel wall via CD44/HA/LYVE-1interactions (Figure 5B). Lymph vessel density and its relationship totumour proximity, blood vessel density,macrophage index, and nodal metastasis To quantitate lymphatic vessel density in the differentbreast carcinoma cases, we used the Chalkley countingmethod, which scores for vessel ‘hotspots’ in the samemicroscopic field ( × 250 magnification) as the tumour.The analyses revealed low lymph vessel counts (meanCLVC 2.33) compared with blood vessel counts (meanCBVC 5.7), as shown in Table 1. These results confirmthe relative absence of lymph vessels in the immediatetumour margin apparent from the immunohistochemi-cal analyses. Intriguingly, when the mean CLVC wasplotted against proximity to the tumour edge (mini-mum vessel to tumour distance), the two variables dis-played an inverse relationship (Figure 6A). This rela-tionship was highly significant regardless of whetherCLVC was treated as a continuous (  p  < 0 . 0001, Spear-man rank correlation) or a categorical variable (  p  < 0 . 0001, Mann–Whitney  U  -test, Table 2). When thetumour panel was divided into two groups accord-ing to lymph vessel density (high, CLVC  ≥ 2.5; low,CLVC  < 2.5), it was clearly apparent that in tumourswith a low CLVC, the lymph vessels were much far-ther from the tumour rim than those with a high CLVC(Figure 6B). Indeed, in tumours with the lowest vesseldensity, the lymphatics were separated from tumourby distances as large as 600  µ m. Further compar-isons revealed that CLVC was also inversely relatedto blood vessel density (CBVC,  p  = 0 . 05, Spearmanrank correlation, Table 2 and data not shown) and tothe number of infiltrating inflammatory macrophagesmeasured as the macrophage index or MØI (Figure 7),regardless of whether CLVC was treated as a contin-uous (  p  = 0 . 009, Spearman rank correlation) or a cat-egorical variable (  p  = 0 . 03, Mann–Whitney  U  -test,Table 2). As these latter variables are associated withincreased tumour aggressiveness and poor prognosis inbreast cancer [30], it is possible that low lymph vesseldensity and apparent loss of tumour-proximal lymphat-ics are consequences of tumour invasion. Some focalloss of LYVE-1-positive endothelium indeed appearedto coincide with tumour invasion in lymph vessels Copyright 󰂩 2003 John Wiley & Sons, Ltd.  J Pathol   2003;  200 : 195–206.  Absence of lymphangiogenesis in breast cancer 199 Figure 1.  Immunostaining of lymph vessels in normal breast and invasive breast carcinoma. Paraffin-embedded sections of humanbreast tissue were subjected to immunoperoxidase staining with an antibody to human LYVE-1 to detect lymph vessels asdescribed in the Materials and methods section. In panel A, normal breast tissue shows LYVE-1-positive lymph vessels (arrows)arranged around breast lobules, and within adipose tissue and smooth muscle (srcinal magnification  × 50). In panels B andC, LYVE-1-positive lymph vessels (arrows) are confined to the tumour (T) periphery (srcinal magnification  × 100) in ductalcarcinoma. In panel D, an invasive ductal tumour is seen invading an LYVE-1-positive peritumoural vessel (note extensive loss of vessel endothelium; srcinal magnification × 200). The inset shows carcinoma cells within the lumen of a peritumoural lymph vessel Figure 2.  Intratumoural blood vessels, but not lymph vessels, are present in ductal breast carcinoma. Paraffin sections of breastcarcinoma were subjected to immunoperoxidase staining to detect either lymph vessels (human LYVE-1 Ab) or blood vessels(CD34 MAb), followed by peroxidase staining as described in the Materials and methods section. The figure depicts ductalcarcinomas  in situ  of papillary or cystic type containing numerous clefts. In panel A, LYVE-1-positive lymph vessels (brown; arrows)are completely absent from the tumour mass and immediate peritumoural areas in each case, but are present some distance fromthe tumour margins (srcinal magnification × 50). In panel B, extensive CD34-positive blood capillaries are present intratumourally,at the tumour edge and in the peritumoural regions (srcinal magnification × 50) Copyright 󰂩 2003 John Wiley & Sons, Ltd.  J Pathol   2003;  200 : 195–206.
Similar documents
View more...
Related Search
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks