, . 185: 184–187 (1998)

QUANTITATIVE ASSESSMENT OF BASEMENTMEMBRANES IN SOFT TISSUE TUMOURS.

COMPUTERIZED IMAGE ANALYSIS OF LAMININ ANDTYPE IV COLLAGEN

. *

Department of Pathology, Mater Misericordiae Hospital and University College Dublin, Ireland

SUMMARY

Basement membranes (BMs) in 201 soft tissue tumours were quantified using computerized image analysis of tissues immunostainedfor laminin and type IV collagen. The purpose of the study was to compare and quantify the extent of BM deposition in a large andvaried group of benign and malignant tumours. Laminin and type IV collagen gave similar results. The difference between benign andmalignant was statistically highly significant (P=0·0001), with greater deposition in benign tumours. BM deposition was homogeneousin benign tumours and heterogeneous in sarcomas and appeared to correlate with the degree of differentiation. Some poorly differentiatedsarcomas showed cytoplasmic laminin staining but little or no extracellular BM. Immunohistochemical evaluation of BM has someadvantages over electron microscopy; specialized equipment is not needed and since large samples can be studied with little samplingerror, heterogeneity can be studied more readily. Subjective visual assessment gives a good overall indication of the extent of BMdeposition and in many situations is likely to be a suitable alternative to image analysis. Because of staining heterogeneity, BMimmunohistochemistry is unlikely to be of significant value in the diagnosis of specific types of sarcoma. ? 1998 John Wiley & Sons,Ltd.

J. Pathol. 185: 184–187, 1998.

KEY WORDS—immunohistochemistry; image analysis; basement membranes; laminin; type IV collagen; soft tissue tumours

INTRODUCTION

Basement membrane (BM) provides a platform orfoundation for epithelium and endothelium. Malignantepithelial tumours must transgress BM before theymetastasize and volumes have been written about theinterplay between BM and epithelial tumour invasion.However, the relationship between mesenchymal cells,their tumours, and BM has been less well documented,possibly because the paradigm of BM protecting thehost from invasive cancer does not exist with sarcomas.However, many normal mesenchymal cells and theirneoplastic counterparts produce BM, includingSchwann cells, smooth and striated muscle cells,endothelial cells, mesothelial cells, synovial cells, andadipocytes.

Most of our information on BM in connective tissueand its tumours comes from electron microscopystudies. With the recent availability of reliable anti-bodies, immunoreactive with constituent proteins ofBM, it has become possible to study BM with the lightmicroscope, initially in frozen sections and morerecently in paraffin sections. In normal tissue, type IVcollagen is present primarily in the lamina densaand laminin is the major component of the laminalucida.1–3

Hospital Research and Development Trust.

CCC 0123–3417/98/060212–07 $17.50? 1998 John Wiley & Sons, Ltd.

The purpose of this study was to documentand quantify BM deposition in a large and variedgroup of connective tissue tumours and to compare theextent of BM deposition in benign and malignanttumours. We were also intrigued by the possibilitythat immunostaining of BMs might have diagnosticapplications.

MATERIALS AND METHODS

Two hundred and one soft tissue tumours werestudied, 84 malignant and 117 benign. In each case, thediagnosis was established primarily on haematoxylinand eosin sections, with supplementary immunohisto-chemistry or electron microscopy where appropriate.Four-micrometre, formalin-fixed paraffin sections weremounted on poly--lysine-coated slides and incubated at37)C overnight. The sections were stained with haema-toxylin and eosin, Gomori’s silver methenamine, andimmunostained4 for collagen type IV (Dako-collagen,dilution 1/50) and laminin (Sigma L6145, dilution1/1000), after pepsin digestion (Sigma P6887, 0·4 percent). Normal blood vessel basement membrane acted asa built-in positive control. Negative controls consisted ofsections stained with normal goat serum instead ofprimary antibody.

Image analysis

Two or, where necessary, more representativephotomicrographs of each slide were taken at #200

*Correspondence to: Professor Peter Dervan, Department ofPathology, Mater Misericordiae Hospital, Eccles Street, Dublin 7,Ireland.

Contract grant sponsors: Health Research Board, Ireland; Cappagh

185SOFT TISSUE BASEMENT MEMBRANES

Fig. 1—Glomus tumour. Thick BM encasing individual glomus cells.Laminin

? 1998 John Wiley & Sons, Ltd.

Fig. 2—Low-grade leiomyosarcoma. This pattern of immunostainingis indistinguishable from that seen in leiomyomas. Type IV collagen

Fig. 3—Liposarcoma. In this tumour, some well-differentiated lipob-lasts in the lower portion of the photomicrograph are surrounded bycontinuous BM. Elsewhere BM is ragged and ill-defined. Laminin

magnification and these were selected randomly. Witheach film, a 1 mm grid was included to provide astandard reference for enlargement of the photographs.Each photomicrograph was enlarged to 3+#7+, usingthe grid as a reference. Black and white photographicprints were processed to provide maximum contrastbetween the BM and adjacent stroma. BMs were quan-tified from the photographic prints using a Quantimet570 image analyser (Leica, Cambridge, U.K.). For eachphotomicrograph, the grey level threshold was adjustedto give the best contrast between the BM and adjacenttissue. To ensure similar thresholding for each case,previously measured cases were included as standards ateach image analysis session. Normal blood vessel BMsand occasional obvious artefacts were selected ‘manu-ally’ and excluded from the counts. BM areas weremeasured in mm2 and converted to a percentage of thetissue section. Non-tumour tissue and areas of necrosiswere avoided.

RESULTS

BM was deposited as thick or thin immuno- orsilver-positive lines. In some tumours, these appeared assingle fine lines, while in others, there were arrays ofextensive BM, consistent with BM reduplication.Schwannomas, neurofibromas, and leiomyomascontained the greatest amount of BM.

Apart from dermatofibromas, varying quantities ofBM deposition were present in all benign tumours.These showed single cells or small groups of cellssurrounded by continuous BM (Fig. 1). Within eachindividual benign tumour, there was little BM hetero-geneity. Areas of fat necrosis within a few lipomas werenegative.

In contrast, the staining pattern in malignant tumourswas highly variable (Figs 2 and 3). Because the numberof cases in each group was small, a formal statisticalanalysis of correlation was not meaningful, but in gen-eral, the extent of BM formation paralleled tumourgrade and degree of differentiation. High-grade sarco-

mas had little or no BM surrounding tumour cells(Fig. 4).

Ewing’s sarcoma, malignant fibrous histiocytoma,epithelioid sarcoma, chondrosarcoma, fibrosarcoma,and dermatofibrosarcoma protuberans were devoid ofBM. Some malignant tumour cells showed cytoplasmicimmunopositivity, sometimes even in the absence ofextracellular BM deposition. This appeared as diffusecytoplasmic positivity with varying intensity from cell tocell. In all tumours, normal blood vessels stained promi-nently and were excluded from measurements; theyacted as excellent internal controls.

Quantitative assessment of each tumour group isshown in Table I. When only tiny strands of BMdeposition were visible, these were arbitrarily recordedas less than 1 per cent. There was little differencebetween the laminin and type IV collagen results. Bothantibodies showed a greater amount of BM in benigntumours than in malignant tumours. The results(recorded as a percentage of the tumour area) for benigntumours were collagen IV—mean 29, STD 15·4, 95 percent confidence interval (CI) 23·5–30 andlaminin—mean 29, STD 16·8, 95 per cent CI 24–32.

, . 185: 184–187 (1998)

186 E. NIGAR AND P. A. DERVAN

Malignant tumours gave the following results: collagenIV—mean 13, STD 14·4, 95 per cent CI 8–17·5;laminin—mean 14, STD 15, 95 per cent CI 9–19. Thedifferences between benign and malignant for collagenIV and laminin were statistically highly significant(P=0·0001).

Fig. 4—Malignant fibrous histiocytoma. Many tumour cells showcytoplasmic laminin immunopositivity despite the absence of a sur-rounding BM. Laminin

Table I—Percentage of tumour tissue occupied by basement membranes, using collagen type IV and laminin immunostaining

Benign tumoursCollagen

%Lam

% Malignant tumoursCollagen

%Lam

%

Lipoma N=20 8 8 Liposarcoma N=6 15 17Leiomyoma N=24 39 42 Leiomyosarcoma N=9 28 32Haemangioma N=21 18 19 Rhabdomyosarcoma N=12 1 1Glomus tumour N=2 23 23 Angiosarcoma N=3 22

142420

Neurofibroma N=15 40 45 Malignant Schwannoma N=1 33 34Schwannoma N=7 43 37 Synovial sarcoma N=5 39 37Dermatofibroma N=28 0 0 Mesothelioma* N=3 24 26

Alveolar soft part sarcoma N=1 8 8MFH N=22 0 0DFP* N=4 <1 <1Ewing’s N=3 0 0Epithelioid N=3 0 0Chondrosarcoma N=5 0 0Fibrosarcoma N=7 0 0

*DFP, dermatofibrosarcoma protuberans, and mesothelioma (spindle cell variant) are arbitrarily included with the sarcoma group.LAM=laminin.

DISCUSSION

BM and BM markers have been studied extensively inmalignant epithelial tumours, i.e., carcinomas.1,5–11

Most studies have focused on host BM as a barrier toinvasion. However, it is well recognized that malignantepithelial cells may also produce BM.12 In general, theability to produce BM is related to the degree ofdifferentiation of the tumour and possibly also to slowgrowth. In soft tissue tumours, BM studies have notgenerated the same interest, perhaps because there is no

? 1998 John Wiley & Sons, Ltd.

normal BM barrier between the tumour and stroma.However, electron microscopic studies have shown BMin many soft tissue tumours.

Previous immunohistochemical studies have shownstrong BM positivity in Schwannomas and neurofibro-mas, while fibrosarcomas and neurogenic sarcomas werenegative.1,13 Similarly, leiomyomas and moderately dif-ferentiated leiomyosarcomas were found to producelaminin. Our study supports many of these observations.In addition, we have attempted to quantify the extent ofBM production by soft tissue tumours.

Benign tumours such as lipomas, leiomyomas, hae-mangiomas, Schwannomas, neurofibromas, and glomustumours were all found to have abundant BMs. In somebenign tumours such as lipomas, haemangiomas, andglomus tumours, the continuous pericellular distributionwas easily recognized. In other tumours which werehighly cellular, with overlapping nuclei, such as Schwan-nomas and neurofibromas, the localization of BMssurrounding individual cells could not readily be ident-ified. The dimorphic differentiation of synovial sarcomaswas typified by the presence of BM surrounding gland-like spaces and by its absence in the spindle cell areas.Loss of BM appears to correlate with the degree ofdifferentiation in many sarcomas.

While Gomori’s silver staining technique often pro-duced intense BM staining, in general it was lesssatisfactory than the laminin and type IV collagenimmunohistochemistry. In many tumours, the silverproduced an unacceptable level of background staining.For practical purposes, subjective visual semiquantita-tion of the degree and intensity of laminin and type IVcollagen staining reflected accurately the amount of BMdetected by image analysis.

We have demonstrated BM deposition in many differ-ent types of connective tissue tumours. In circumstanceswhere electron microscopy (EM) is not available,immunohistochemical detection of BM is an acceptablealternative. Light microscopic immunohistochemistry

, . 185: 184–187 (1998)

187SOFT TISSUE BASEMENT MEMBRANES

overcomes one of the inherent disadvantages of electronmicroscopy, namely, sampling error. Using antibodiesdirected against BM proteins, it is possible to study largetissue sections. Immunohistochemistry also allows BMtumour heterogeneity to be studied in greater detail. Itallows quantification of the extent of BM production invarious tumours and correlation of BM production withdifferentiation, tumour grade, proliferative indices, orwhatever other variable may be of interest.

Some tumours show varying degrees of cytoplasmiclaminin or type IV collagen staining. In general, cyto-plasmic laminin staining was present more frequentlythan type IV collagen. We did not pursue the question ofwhether or not cytoplasmic laminin immunopositivityrepresents true protein production or is a false-positiveimmunoreaction. Other studies have shown cytoplasmiclaminin synthesis in sarcoma cells, demonstrated bymRNA in situ hybridization.14 Absent BM productionin the presence of cytoplasmic production must be dueto failure to assemble or secrete BM, or to increased BMdegradation. It is notable that a number of poorlydifferentiated tumours showed cytoplasmic staining,whereas their better differentiated counterparts had onlyextracellular BM deposition. Overall, there appeared tobe an inverse relationship between BM deposition andtumour grade.

We encountered some difficulties common to allimmunohistochemical image analysis studies—problemsof sampling and thresholding. However, we believe thatour results reflect reasonably accurately the extent ofBM production in a wide variety of soft tissue tumours.The inclusion of previously measured cases with eachnew batch helped to maintain similar thresholdinglevels.

Alternative quantitative strategies could be used, suchas biochemical estimation of extracted BM proteins.However, this could suffer from serious sampling errors,since, for example, normal blood vessel BM wouldconfound the results. Digestion and removal of tumourcells, leaving behind the scaffolding of stroma for scan-ning EM study, might provide good three-dimensionalviews and some quantitative information.15 This tech-nique produces no discrimination between BM andother stromal matrix components. It is likely that con-focal microscopy would give elegant three-dimensionalarchitectural information.

Initially we had hoped, somewhat naively, that thepattern or quantity of BM deposition might be diagnos-tic for different tumour categories. A number of authorshave suggested that this may be true.13,16–18 However, itis obvious that there is considerable overlap in BMdeposition between the different tumour groups, somuch so that BM deposition is not likely to have usefuldiagnostic discriminatory power. Most pathologists willfind the wide variety of immunohistochemical differen-tiation markers (now available commercially) far moreuseful. In addition, new molecular biological techniquesare available for detecting chromosomal translocationsspecific for certain sarcomas, such as synovial sarcoma,

? 1998 John Wiley & Sons, Ltd.

Ewing’s sarcoma, myoxid liposarcoma, and alveolarrhabdomyosarcoma.

Immunohistochemical detection of BM in softtissue tumours is specific, simple, and easy to evaluate.Subjective visual impression gives a good overallassessment of BM production.

ACKNOWLEDGEMENTS

We gratefully acknowledge the support of the HealthResearch Board, Ireland and the Cappagh HospitalResearch and Development Trust. We also gratefullyacknowledge the photographic assistance of FlorenceGrehan, Photography Department, Mater Hospital andthe secretarial assistance of Geraldine O’Driscoll.

REFERENCES1. Martinez Hernandez A, Amenta PS. The basement membrane in pathology.

Eur J Cardiothorac Surg 1983; 10: 105–108.2. Foidart JM, Bere EW Jr, Yaar M, et al. Distribution and immunoelectron

microscopic localization of laminin, a noncollagenous basement membraneglycoprotein. Lab Invest 1980; 42: 336–342.

3. Laurie GW, Leblond CP, Martin GR. Localisation of type IV collagen,laminin, heparin sulphate proteoglycan and fibronectin to the basal laminaof basement membranes. Cell Biol 1982; 95: 340–344.

4. Colvin RB, Bhan AK, McCluskey RT. Diagnostic Immunopathology. NewYork: Raven Press, 1995.

5. Bosman FT, Havenith M, Cleutjens JPM. Basement membranes in cancer.Ultrastruct Pathol 1985; 8: 291–304.

6. Barsky SH, Siegal GP, Janotta F, Liotta LA. Loss of basement membranecomponents by invasive tumours but not by their benign counterparts. LabInvest 1983; 49: 140–147.

7. Birembaut P, Caron Y, Adnet JJ. Usefulness of basement membranemarkers in tumoral pathology. J Pathol 1985; 145: 283–296.

8. Abrahamson DR. Recent studies on the structure and pathology ofbasement membranes. J Pathol 1986; 149: 257–278.

9. Richards CJ, Furness PN. Basement membrane continuity in benign,premalignant and malignant epithelial conditions of the uterine cervix.Histopathology 1990; 16: 47–52.

10. Willebrand D, Bosman FT, De Goiej AFPM. Patterns of basementmembrane deposition in benign and malignant breast tumours. Histopathol-ogy 1986; 10: 1231–1241.

11. Leu FJ, Engvall E, Damjanov I. Heterogeneity of basement membranes ofthe human genitourinary tract revealed by sequential immunofluorescencestaining with monoclonal antibodies to laminin. J Histochem Cytochem1986; 34: 483–489.

12. Bonkhoff H, Wernert N, Dhom G, Remberger K. Distribution of basementmembranes in primary and metastatic carcinomas of the prostate. HumPathol 1992; 23: 934–939.

13. Miettinen M, Foidart JM, Ekblom P. Immunohistochemical demonstrationof laminin, the major glycoprotein of basement membranes, as an aid in thediagnosis of soft tissue tumours. Am J Clin Pathol 1983; 79: 306–311.

14. Soini Y, Autio-Harmainen H. Tumor cells of malignant fibrous histiocyto-mas express mRNA for laminin. Am J Pathol 1991; 139: 1061–1068.

15. Kelly D, O’Donnell MD, Dervan P, McGeeney KF, Fitzpatrick JM. Theskeletal framework of human kidney and renal cell carcinoma. A scanningelectron microscopic study. Urol Res 1990; 18: 241–244.

16. Leong AS, Vinyuvat S, Suthipintawong C, Leong FJ. Patterns of basallamina immunostaining in soft-tissue and bony tumors. Appl Immunohisto-chem 1997; 6: 1–7.

17. Ogawa K, Oguchi M, Yamabe H, et al. Distribution of collagen type IV insoft tissue tumors. An immunohistochemical study. Cancer 1986; 58:269–277.

18. Nerlich AG, Haraida S, Wiest I. Basement membrane components asdifferential markers for mesenchymal tumors of various origin.. AnticancerRes 1994; 14: 683–692.

19. Sreekantaiah C, Ladanyi M, Rodriguez E, Chaganti RS. Chromosomalaberrations in soft tissue tumors. Relevance to diagnosis, classification, andmolecular mechanisms. Am J Pathol 1994; 144: 1121–1134.

20. Bown NP, Reid MM, Malcolm AJ, Davison EV, Craft AW, Pearson AD.Cytogenetic abnormalities of small round cell tumours. Med Pediatr Oncol1994; 23: 124–129.

, . 185: 184–187 (1998)