Lymphaticovenous Diferentiation in Kaposi's Sarcoma - NCBI - NIH

7 downloads 40 Views 2MB Size Report
mens that represented four main histologic stages. The early phase of Stage 1 contained lymphatic-like clefts lined by endothelial cells with thin, discontinuous.
American Joumal of Pathology, Vol. 130, No. 2, February 1988 Copyright X Amencan Association of Pathologists

Lymphaticovenous Diferentiation in Kaposi's Sarcoma Cellular Phenotypes by Stage

From the Department ofPathology, and the Immunopathology Laboratory, University Hospital, Lund, Sweden

MICHAEL DICTOR

and CHRISTINA ANDERSSON

The histogenesis of Kaposi's sarcoma was investigated by immunohistochemical staining of 20 skin specimens that represented four main histologic stages. The early phase of Stage 1 contained lymphatic-like clefts lined by endothelial cells with thin, discontinuous basement membranes shown by anti-laminin, absent Factor VIII-related antigen reactivity (FVIIIRAg), and only rare staining with dilute Ulex europaeus agglutinin I (UEA-I). In the late phase of Stage 1, the clefts developed into anastomosing, blood-filled channels, and the basement membrane became complete. Endothelial marker reactivities were not definitive, but weak stain-

ing with UEA-I and variable staining for FVIIIRAg characterized the spindle cells of Stage 2. However, spindle cells in monomorphic nodules were individually enclosed by immunoreactive laminin. The sequence of events, particularly in light of previous angiographic findings of lymphaticovenous union, suggests a disturbance in lymphaticovenous differentiation in Kaposi's sarcoma. Sclerotic closure of channels unable to maintain competent blood flow may select against lymphendothelial traits in the developing spindle cell nodule. (AmJ Pathol 1988, 130:411-417)

TWO CONVERSE paradoxes arise in attempting to understand the endothelial origin of Kaposi's sarcoma. The first concerns the inability to explain clinical' and angiographic2 evidence of lymphaticovenous connections in the Kaposi nodule if one assumes that the lesion is strictly an abnormal growth oflymphatic endothelium. Vascular invasion suggests a risk of metastatic disease in most other areas of tumor pathology; yet in spite ofintimate involvement of blood vessels with lymphatics from the inception of Kaposi's sarcoma (KS) to the development of the spindle cell intermediate stage,3 its course is characterized by multicentricity, rather than metastasis.4 Many pathologists view the lesion as a hyperplasia of endothelium.5~7 The second paradox arises if one ascribes the origin of KS solely to blood vessel endothelium.8 This would not explain the so-called lymphangiomatous form of KS,9 which in reality is the first phase of the potentially progressive Kaposi lesion. Again, the expectation of metastatic disease is not fulfilled, although in other solid tumors, lymphatic invasion often presages metastasis. In addition, recent electron-microscopic,'0 histochemical, and immunohistochemical studies have confirmed a role for lymphendothelium in the genesis of Kaposi's sarcoma.1'

A previous morphologic study of biopsy and autopsy material representing various stages in the development of the nodular lesion offered a framework in which these paradoxes may be resolved.3 Emphasis was placed on the differentiation, rather than malignancy, of endothelium; and coupling of venous and lymphatic channels appeared to precede the spindle cell stage. Reports of conflicting results with the use of special techniques to demonstrate origin in endothelium of either lymphatics or blood vessels have failed to consider the above paradoxes. As detailed previously,3 developmental Stage 1 is a coalescence of hyperplasic lymphatic and blood vascular endothelium. Bloodless clefts about central blood vessels produce so-called glomeruloid structures in the early (lymphangiomatous) phase of Stage 1; whereas the late (angiomatous) phase shows dimin-

411

Supported by grants from John och Augusta Perssons Stiftelse for Vetenskaplig Medicinsk Forskning, the Medical Faculty of the University of Lund, the Royal Physiographical Society of Lund, and Anna Lisa och Sven-Eric Lundgrens Stiftelse for Medicinsk Forskning. Accepted for publication October 7, 1987. Address reprint requests to Dr. Michael Dictor, Department of Pathology, University Hospital, 221 85 Lund, Sweden.

412

DICITOR AND ANDERSSON

ished glomeruloid structures, variable sclerotic regression, and frequently blood-filled anastomosing channels. Coupling of both vascular systems tends to proceed along progressively larger veins (Figure 1). The first discrete fascicles of spindle cells with vascular slits appear in the late phase. Note that spindle cells with slits by definition form a unit distinct from clefts and channels lined by elongated flattened cells. Stage 2 is the expansive spindle cell nodule. Stage 3 shows increasing atypia in spindle cells with loss of slits, and Stage 4 is represented by frank angiosarcoma. The present study uses immunohistochemistry to define the presence or absence of the following antigens in the cells or basement membranes ofKS at all stages of development: Factor VIII-related antigen (FVIIIRAg), 12 fucosyl moieties as reflected by binding with Ulex europaeus I lectin (UEA-I),13 and laminin.'4 The significance of the immunohistochemical results is interpreted on the basis of previous light-microscopic and ultrastructural observations, Palmer's angiographic findings, and other recent studies.

Materials and Methods Material was obtained by excisional or punch biopsy from patients with AIDS-related and classic KS. Formalin-fixed paraffin sections were subjected to a panel of commercially available polyclonal antisera against stable well-defined antigens as summa-

AJP * February 1988

Table 1 -Antisera and Technical Procedures*

Polyclonal antibody/lectin UEA-It

FVIII-RAgt laminin§

Dilution

Time

Temperature

1:6000 1:1000 1:200

2 hours 16 hours 16 hours

Room temperature 4C 4C

*Predigestion with 0.4% pepsin for 2 hours at 37 C prior to anti-laminin; prior to other primary reagents, 0. 1% trypsin/CaCI2 for 30 minutes at 37 C, pH 7.8. Endogenous peroxidase blocked in all sections by preincubation in 1% H202/methanol for 30 minutes. All washings between steps in PBS. tBoehnnger-Mannheim, Mannheim, FRG. tDakopatts, Copenhagen, Denmark. §Bethesda Research Laboratories, Rockville, Maryland.

rized in Table 1. Blocks were chosen on the basis of good tissue preservation, and in Stage 1 lesions, on a relative paucity of inflammatory cells. Secondary antibodies were swine anti-rabbit Ig diluted 1: 50 and followed by peroxidase-antiperoxidase (PAP) complex diluted 1: 50. In the case of UEA-I lectin, the secondary antibody was preceded by rabbit antiUEA-I at a dilution of 1: 2000. Depending on the need to easily distinguish hemosiderin from reaction product, either 3-amino-9-ethylcarbazol or 3,3'-diaminobenzidine was used as the chromagen and a weak solution of Mayer's hematoxylin as a counterstain. Positive controls within the sections included normal vessels for FVIII-RAg, epithelium and vessels for UEA-I and laminin. Negative controls were provided by substitution of an irrelevant primary antibody. Positive staining was interpreted on the basis of a granular precipitate regardless of intensity. Stages 1 and 2 of the developing lesion were represented in both groups, but Stage 4 was limited to the group with classic disease, as shown in Table 2. In addition, sections of 1 case each of formalin-fixed, paraffin-embedded normal skin and lymph node, dermatofibroma, cutaneous and splenic lymphangioma, cutaneous hemangioma, and fibrous granulation tissue were stained to provide a further frame of reference for interpretation ofresults in the KS tissue. Results in these comparative lesions conformed to those generally reported,2"15 but are only presented when they illustrate the variability of phenotypic markers in a specific context. No attempt was made to Table 2-Number of Cases Stained by Histologic Stage and Clinical Group (Total 20)

vein Figure 1-Scheme of lymphaticovenous union from level of capillaries to progressively larger veins. Venous coupling levels vary in Stage 1. Participation of arterial capillaries is uncertain, as indicated by the question mark.

Stage 1 Early phase Late phase Stage2 Stage3 Stage 4

Classic KS

AIDS-related KS

1 3 4 1 2

4 2 1 2

DIFFERENTIATION IN KAPOSI'S SARCOMA

Vol. 130 * No. 2

distinguish classic and AIDS-related KS on the basis of immunohistochemical staining.

Results Endothelial Markers The earliest phase of Stage 1 showed granular cytoplasmic staining for FVIIIRAg and strong membrane staining with UEA-I in vascular endothelial cells that were often found in the center of glomeruloid structures. Cells lining the surrounding clefts were rarely stained by UEA-I (Figure 2A,) and lining cells showed spotty positivity for FVIIIRAg only in effacing, but not fully preserved, glomeruloid structures (Figure 2B). The late phase of Stage 1 was characterized by a similar intense central endothelial staining in "nodes" created by a small cuff offrequently negative elongated endothelial-like cells. Anastomosing angiomatoid channels failed to show definitive staining for either marker. In Stage 2 lesions, FVIIIRAg staining of hypertrophic endothelial cells was often prominent within and on the perimeter of smaller nodules, but less so in larger ones, despite strong positivity in overlying granulation tissue in a case with ulceration. However, in most lesions, focal granular bipolar staining of spindle cells was present, although the intensity was variable (Figure 3). This contrasted with continued minimal UEA-I staining, which was present only in occasional spindle cells, in addition to intranodular blood vessels. Scattered spindle cells positive for endothelial markers were difficult to distinguish from tangentially cut vascular endothelium, a problem noted even in the separate sections of fibrous granulation tissue. One case in transition to Stage 3 showed diffuse moderately intense staining for FVIIIRAg, whereas another Stage 3 case was largely negative for this antigen. FVIIIRAg was not identified in malignant cells in Stage 4, although strong UEA-I reactivity developed at this stage in both cases studied (Figure 4). Among the relevant comparative structures, nor-

mal blood vessels and some cells lining the cutaneous lymphangioma were positive for FVIIIRAg and reacted even more strongly with UEA-I. Occasional skin lymphatic capillaries showed weak focal staining with UEA-I. Basement Membrane In both normal and pathologic tissues, crisp staining for laminin was noted in the basement membrane

of epithelium, blood vessel endothelium, smooth muscle, and the Schwann cells of small dermal nerves.

413

Interrupted weak or absent staining of a thinned basement membrane was seen about lymphatic clefts in early Stage 1 of KS. Glomeruloid structures, particularly, showed intense staining of central vessels and extremely attenuated, discontinuous staining of surrounding clefts in most cases (Figure 5A). Rarely, in these locations the staining became continuous, although barely thicker. By comparison, lymphatics in normal skin and cutaneous and splenic lymphangioma mostly lacked the basement membrane protein, although short stretches of attenuated staining with the above fenestrated character were visible. Lymphendothelium lining the marginal sinus of lymph nodes also showed positive basement membrane staining, as previously reported. 14 Vessels in the cutaneous hemangioma were enclosed by solid la-

minin rings. In the late phase of Stage 1, the situation became more complex. With glomeruloid structures diminished in number, and clefts narrowed and appearing more like blood capillaries, the basement membrane tended to become complete and thicker (Figure 5B). In contrast, lymphatic clefts that had collapsed were marked by hemosiderin deposition and sclerosis of surrounding collagen bundles; they were largely de-

void of laminin staining. All Stage 2 monomorphic nodules showed prominent laminin staining, with a marked tendency to surround individual spindle cells. This produced a honeycomb appearance in cross-section (Figure 6).

Discussion Previous angiographic and histologic findings indicate that both venous and lymphatic endothelium proliferate and merge to form anatomic pathways between the vascular systems.3 Careful stepwise analysis of our data in conjunction with these findings explains contradictory interpretations of the lym-

phatic9-"' versus blood vascular8"16 origin of KS. En-

dothelial markers FVIIIRAg and UEA-I were inconsistently detected, often at low intensity, in benign lymphatic capillaries in the comparative lesions studied. For UEA-I, this is due to high dilutions of lectin and antibodies, which maintained strong staining of blood vascular endothelium, but weak, focal staining of normal skin lymphatics. Differentiation toward lymphatic or blood vessel endothelium in KS is thus indicated by a spectrum of staining intensities for such markers, in addition to changes in basement membrane integrity. The earliest recognizable phase in the Stage 1 lesion shows jagged, bloodless dermal clefts lined by cells with extremely thin cytoplasm, which may some-

414

DICITOR AND ANDERSSON

AJP * February 1988

2A

2B

3

4

;@

)8xi