THE AMERICAN JOURNAL of CANCER a Continuation of the Journal of Cancer Research

THE AMERICAN JOURNAL OF CANCER A Continuation of The Journal of Cancer Research

~ VOLUMEXXXIV DECEMBER,1938 NUMBER4

SYNOVIAL SARCOMAS IN SEROUS BURSAE AND TENDON SHEATHS

PROF. LOUIS BERGER, M.D. (From the Pathological Department, HBpital de I'Enfant-Jdsus, and the Anti-Cancer Center of Lava1 University, Quebec)

Progress in the knowledge of malignant tumors arising from synovial tissue has been slow. In spite of some recent and valuable contributions, this chapter is far from complete. The reasons for this are threefold: first, the want of knowledge concerning the normal features and nature of synovial tissue, which was long studied in articulations only, although it is common, also, to serous bursae and tendon sheaths; second, the relative-perhaps only apparent-rarity of cases; finally, the lack of precision and even vagueness of the reports in the literature., Most of the older authors, and even some contemporary ones, interested primarily in the clinical or surgical aspects of the question, have been satisfied with a purely topographical diagnosis and have either neglected the histologic aspects of their tumors or described them only briefly and superficially. We have had the opportunity of studying five cases of synovial sarcoma, differing more or less from one another but all originating outside of articulations, that is in serous bursae or tendon sheaths, where these tumors are less known, but perhaps easier to study than in the more intricate tissues of the joints. THENORMAL SYNOVIAL TISSUE The prototype of synovial tissue is encountered in the synovial membranes of the joints, but all histologists admit that the lining tissue of the serous bursae and tendon sheaths is homologous with articular synovialis. It consists of a dense connective tissue through which are scattered polymorphous cells. It is possible to distinguish an outer, vascular and fatty portion and an inner or superficial, lamellated layer containing dendritic cells. Some of the latter attain the surface, but the lining most often appears discontinuous, formed here 501 502 LOUIS BERCER by the bodies, there by the extensions of branching cells, between which are apparently bare portions of collagenous substance. In some recesses and on the surface of synovial villi the lining is more regular and quite continuous, built up by approximately cuboidal cells. The histologic structure of synovial tissue appears, therefore, quite simple, but it has been variously interpreted. The synovial membrane was first believed to be some sort of epithelium, but soon afterwards was classified as endothelium, a conception which long prevailed and is still held by many authors. Others, with Lubosch, regard it as a modified cartilaginous tissue, or, with Tourneux and Hammar, as of cartilaginous derivation. This opinion may appear convincing for the synovial membranes of the joints, where histo- genetically the anlage of synovial and cartilaginous tissue seems to be the same, but it is not easily acceptable for tendon sheaths, and even less so for bursae, which are very often independent of or remote from any cartilage. Although some histologists (Schaeffer ) and many pathologists (Kaufmann, Aschoff) adhere to the endothelial interpretation, others, as for example, Key, consider the lining too irregular to be consistent with an endothelial nature. They believe that the lining cells are ordinary fixed connective-tissue cells, the form of which-and perhaps also the function-is slightly modified by their superficial situation. Key, in particular, conceded only an endothelial (‘aspect ” to these cells. More recently Franceschini has brought forward a new conception of synovial membranes. He distinguishes two types of tissue: the first, a very simple one, resembling in appearance ordinary fibrous tissue, is encountered where the mechanical stress intrinsic to articular function directly exercises its effects; the second, which is more differentiated, is limited to areas less exposed to pressure and friction. The cells of the second type are surrounded by numerous reticular fibrils, and Franceschini concludes that these cells form a true reticulo-histiocytic meshwork, the superficial layer of which would give, after desquamation and regressive metamorphosis, the polymorphous elements encountered in the synovial fluid. Key objected to this hypothesis, but Sabraz&s, De Grailly and Salabartan confirmed and extended Franceschini’s conclusions. They believe that the superficial cells undergo a peculiar adaptation, without, however, the attainment of a truly regular structure. Some of the lining cells become flat and endothelium-like, adequate for a border r81e and perhaps incapable of differentiating further : others are transformed into histiocytes and functional monocytic cells which may emigrate and enter the synovial fluid. The most recent studies are those of Vaubel, who showed that cultures of synovial tissues differ entirely from cultures of other mesenchymal cells, in particular from those of the cutaneous and interstitial variety: “ The pe- culiarity of a synovial growth rests in part on the polymorphism of the cells, the countour of which ranges from round and spindle to polygonal and epithe- lioid, and especially on their tendency to form membranes composed of syncytial elements.” Although synovial cultures resemble to some degree those of growing osteoblasts or chondroblasts, Vaubel nevertheless regards synovial cells as different from either of these, forming an autonomous cell type with peculiar morphological and junctional properties; he even proposes for them SYNOVIAL SARCOMAS IN SEROUS BURSAE AND TENDON SHEATHS 503 the specific designation “ synovioblasts.” According to Vaubel, the essential function of these cells is the formation of synovial fluid, and the synovial mucin is analogous to the intercellular substance of cartilage, from which it differs chiefly in its lack of solidification. Vaubel avoids applying to this formation of mucin the term “secretion,” which he believes should be restricted to the enzyme-forming function of epithelial glands. Opinions as to the origin of the synovial fluid and the mechanism of its formation are not less varied. In the eyes of the older authors (Bichat, Todd) the articular fluid was a simple transudate. v. Koelliker believed that this transudate was modified by the influence of articular “ epithelium.” Frerichs (1846) was the first to show the presence of mucin and regarded it as coming from the “ serous ” lining through a continuous desquamation of cells and their subsequent dissolution in the alkaline “ serum.” Hueter, however, did not encounter any evidence which would favor such a desquamation and Kling saw in this hypothesis only an erroneous analogy with the disintegration of the horny layers of the epidermis. Soubbotine, Mayeda (according to Kling and Vaubel), Aschoff, and Raufmann believed that the mucus is secreted by synovial cells. For W. Mueller, on the other hand, the fluid is not-at least not exclusively-a secretory product of the membrane, but results in large part from a liquefaction of articular lining cells, through some sort of mucinous degeneration, and from a constant flow of liquid from the articular surface. To Sabrazks and his fellow-workers the mucus seems to develop from a premucoid extracellular substance, for only very exceptionally are minute droplets observed in the protoplasm. These authors admit the possibility of a mucinous degeneration of protoplasm and nucleus as a whole; the mucinoid and other (fatty) degenerations observed in synovial membranes represent, in their mind, a debased function through which the synovial fluids acquire some of the chemical elements which are necessary to the lubrication of the articular surface, but which cannot be furnished by transu- dation of plasma alone. King studied the Golgi apparatus in synovial cells and concludes that they participate by active secretion in the fluid formation. He also observed desquamation of lining cells, but the fact that these appeared to be alive and capable of growth and proliferation (Vaubel) seemed to him to contradict the hypothesis of cellular degeneration. King seems, however, in some measure to contradict himself, for he mentions the indistinctness of the outlines of the desquamated cells and the presence of cellular dkbris. Sabrazb, contrary to King and others, holds the presence of more or less degenerated or senescent desquamated cells to be unquestionable, and illustrates it by convincing evidence. But according to Vaubel, this desquamation, if one accepts it, cannot be linked with mucin, which is not due to such a degenerative process, being encountered only in young and never in degenerating cultures. This necessarily brief review shows that the conceptions of various authors concerning both the synovial tissue and the fluid are far from unanimous. The differences of opinion may be reduced to the following questions: (1) Is the synovial membrane a simple endothelium, a modified cartilage, a reticulo- histiocytic or a truly autonomous and specific tissue? (2) Is the synovial fluid the result of a direct or indirect secretion? Do desquamated and dissolved 504 LOUIS BERCER

cellular elements contribute to its formation, and what is the mechanism of mucin formation? It is far from our intention to draw from the study of our tumors definite conclusions as to the nature of their normal substratum. Indeed, this paper is, first of all, a discussion of the pathological aspects of malignant synovial tumors. During our study, however, we gained the impression that observations made in the pathological realm may perhaps be of aid in discriminating between the divergent opinions which are held by those interested in the normal histology of synovial membranes. We shall, therefore, later on, consider our results in the light of the questions just presented.

Flc:. 1. CASE I: ISDIFFERENT STRANDS; AS;’ECTIDENT~CAL WIT11 THATOF SOME KETICIJI.OSARCOMMAS OF BONEMARROW OR LYMPHNODES Left: routine staining. Right: reticulum impregnation.

CASEHISTORIES CASEI: No. 10.234. I.P.Q.: A white man, aged thirty, had a tumor of the posterior aspect of the left thigh, occupying the inferior third, but wholly independent of the capsule of the knee joint, from which its lowest point was at least 4 cm. distant. The tumor app- peared about six months before admission. When first seen, it was the size of a mandarin orange. bulging below the skin and grossly nodular; it was relatively superficial, but adherent in its deeper parts, without, however, involving the bone, over which it remained movable. The borders of the tumor were sharply defined, simulating a capsule. Surgical removal was followed by healing by first intention. Histologic examination showed the tumor to be of a peculiar but certainly malignant type, and five months later the patient returned with a recurrence. Amputation was proposed but refused. Macroscopically the excised tumor has an average diameter of about 6 cm. Opened, it appears to consist of a central mass surrounded by some smaller nodules partly fused together and partly adherent to the principal mass. Involving two thirds of the periphery, principally towards the skin, is a virtual cleft between the neoplastic and normal tissue, as if the tumor had originated in a cavity, the wall of which it had thrust back opposite its point of origin. Some shorter clefts penetrate here and there into the tumor itself. On section a little mucinous substance flows from these spaces. The deep border of the tumor is indistinct. Histologic Description: The tumor lies below the subcutaneous fat layer, from which it is separated by a band of dense connective tissue of fascia1 type. Between this aponeuro- SYNOVIAL SARCOMAS IN SEROUS BURSAE AND TENDON SHEATHS 505 sis and the neoplastic tissue is a small irregular cleft, the lower wall of which is formed by the peripheral elements of the tumor, whereas the upper lining rests on the aponeurosis. The upper wall differs from one point to the other: it is composed in some areas of branching cells embedded in a collagenous cement; the most superficial of these constitute with their bodies or their extensions a discontinuous protoplasmic fringe closely resembling normal synovial membrane ; elsewhere the border cells are pseudo-epithelial, cubical, or columnar, as in chronic hyperplastic synovitis or bursitis. The upper wall is therefore an irritated but not a neoplastic synovial membrane. In some points, however, the tumor tissue crosses the cleft, adheres to the upper wall and even invades it up to the aponeurotic tissue. On the sides and towards the depth the neoplasm invades the neighboring connective, fat, and muscle tissue. From a purely topographic point of view the tumor is therefore malignant, either

FIG.2. CASE 1: ASPECT RESEMBLISG COMhION FIBROSARCOMA Reticulum impregnation (right) revcals fibrillar strands between non-impregnated strands of clearer cells with slight silver precipitate. invading or originating in a subfascial serous bursa. It has destroyed the inferior wall and some parts of the lateral walls of the bursa, has proliferated in the bursa1 cavity and distended it, and partly repulsed and partly invaded the opposite upper wall which rested on the fascia. The neoplastic tissue offers essentially two aspects, one of which is limited to certain areas, whereas the other is encountered throughout the tumor. (1) The first type of tissue is formed by solid cellular strands which are so dense that even on 5 micra cuts the nuclei are overlapping. The latter are generally round and characterized by a very fine membrane, a granular chromatin network, and one or two distinct nucleoli; they appear therefore very pale. The protoplasm is scanty. amphophile, and without a membrane. Neither the van Gieson stain nor Masson’s trichrome shows the least connective tissue. These strands are truly indifferent, but Laidlaw’s silver method reveals an abundant reticulum of minute fibrils surrounding nearly every cell and forming a close meshwork (Fig. 1). 5 06 LOUIS BERGER

This aspect does not recall any normal tissue, but is identical with that of some reticulosarcomas of bone marrow or lymph nodes (second variety of Oberling and of Craciun; immature variety of Koulet), this comparison having been made on personal cases of reticulosarcomas which were fixed and stained by exactly the same methods. (2) Most of the neoplastic tissue is less uniform, and its aspect varies from point to point and with various stains. The tissue forms large strands which are either solid or contain spaces. Aiter routine staining with hematein-eosin the solid areas appear to be composed of small, partly elongated, partly irregular cells with indistinct outlines. The cells form small threads and more or less regular whorls or dense and altogether irregular masses; there is no collagenous substance between them. The general aspect resembles a somewhat irregular fihrosarcoma or rather a histiocytoma. Masson’s hematoxylin-ponceau red-anilin blue

FIG. 3. CASE I: “ EPITIIELIO-SARCOMATOUS” ASPECT WITH DEXDRITIC FEATURES

trichrome stain reveals two intimately intermingled cell types: the first is represented by elongated, approximately fusiform cells with scanty, moderately deep staining protoplasm ; the second is a clearer staining, branching cell with less scanty protoplasm and often syncytial arrangement. Cells of this latter type form irregular clear spots or strands in the more deeply staining tissue. Laidlaw’s silver method for reticulum further accentuates these types: it reveals the presence of a very fine and abundant fibrillar meshwork around the elongated cells, where routine methods gave an ambiguous fibroblastic impression, but is negative in the clearer staining cells. This method showed in our Bouin-fixed material still another difference; the protoplasm of the branching elements. which are devoid of fibrils, showed a very fine perinuclear silver precipitate, whereas that of the elongated cells did not (Fig. 2). In areas containing small cavities the elongated cells are outnumbered by the clearer SYNOVIAL SARCOMAS IN SEROUS BURSAE AND TENDON SHEATHS 507 staining elements; the latter are larger than in the dense areas and polyhedral. When located near the spaces, they swell, becoming cuboidal and even columnar-like and pseudo- epithelial. They rest upon the nearby elongated cells and give these points an “epitheliosarcomatous ” aspect; elsewhere they protrude into the cavities and acquire dendritic features (Fig. 3). Between the cells which surround the spaces on the one hand, and the branching cells of histiocytic type and the elongated cells on the other hand, exist numerous intermediate forms, which demonstrate their unique origin and the endothelial quality of the pseudoglandular structures. Indeed, neither the cuboidal nor the cylindrical cells ever attain the perfect features of normal or even neoplastic glandular epithelium : their lateral outlines are less distinct and more irregular, there is no basal membrane, and from one space to another-and sometimes even around the same space-are encountered the most various shapes, from flat

FIG.4. CASEI: SMALLCAVITIES FILLED WITII CALCOSPHERITES endothelial-like cells, through swollen and protruding elements, to nearly cuboidal or cylindrical cells. None of these cells shows any inclusions. The spaces are active pro- liferating centers, for many of the lining cells show mitoses; in the dense areas cell divisions are generally direct and scarce. The origin of the spaces is indistinct on routinely stained sections, but is shown particularly well with Laidlaw’s silver stain followed by hematein-metanil yellow and mucicarmin. The spaces, often so small as to be hardly recognizable, appear only in the clearer staining spots, which are devoid of reticular elements. It may be concluded that in the dense areas the clearer staining cells represent a beginning endothelial evolution, which is not yet accompanied by appearance of spaces. It may be added that with Laidlaw’s method the lining cells show the same perinuclear silver precipitate as the clearer staining cells in the dense areas. Space-formation begins often with the appearance of very minute deep 508 LOUIS BERGER

red (mucicarmin) mucous droplets, which are never intracellular. Between these droplets and the larger spaces there is an uninterrupted series of cavities, progressively increasing in size. In the large, pseudoglandular cavities the much is thick and plentiful, but it is often mixed with an albuminous substance, or magma, which may contain nuclear dCbris. Small strips of mucus parallel the free borders of some cells, but seem to originate outside. Some spaces contain desquamated cells either intact or degenerating. The magma may be squamous, its elements being of the same shape and size as some of the lining cells. Des- quamation and much formation are, however, often independent of each other, and the mucus seems not to result from dissolution of cells, but rather to originate, in contact with the lining cells, from an unstainable substance which is present in the cavities. The magma

may become homogeneous and form small rhizoid corpuscles or become saturated with calcium salts and give rise to calcospherites (Fig. 4). Some neoplastic cells show phagocytic properties : they may become larger, vacuolar or foamy, manifestly lipophagic (Fig. S), or may incorporate hemosiderin granules when located near some small hemorrhagic area. Diagnosis: From a purely morphological point of view the tumor gives the impression of a reticulo-endothelio-histiocytosarcoma. Its reticular nature is attested not only by the extremely developed fibrillar network shown by special silver stains, but above all by the presence of very immature cells which are surrounded by delicate reticulated fibrils such as have hitherto been seen exclusively in reticulosarcomas of the hematopoietic tissues. The spaces strewn throughout the tumor are due to the endothelial modification of the common mesenchymal stock, which is neither vascular nor lymphatic, but is accompanied by a pseudoglandular arrangement of the border cells, by desquamation and disintegration of SYNOVIAL SARCOMAS IN SEROUS BURSAE AND TENDON SHEATHS 509 some of them, and by the appearance of mucin outside the cell bodies. The elongated or branching mesenchymal cells resemble more the cells characterizing histiocytomas than the fusiform elements of common fibrosarcoma, for they only seldom give rise to collagenous substance and then in scant amount. A minority of cells show lipophagic or siderophagic properties. By its topographical relations with a serous bursa on the one hand, and on the other by the structural and functional resemblances of its elements to synovial cells, as they have been described in normal or pathological articulations and bursae, this tumor gives the impression of a typical synovial sarcoma of a serous bursa. CASE11. No. 6.555. I.P.Q.: In a white man of thirty-eight a tumor appeared in the upper third of the left thigh six months prior to his admission to the hospital. When first

FIG, 6. CASE 11: REMNANTOF NORMAL,SLIGIITLY IRRITATED AND HYPERPLASTICSYNOVIAL MEM- BRANE OF SEROUSBURSA seen, the tumor seemed to be embedded in the antero-lateral muscle group but was movable over the bone and did not show any connection with the hip joint. The tumor bulged flatly below the skin and was approximately 10 cm. long and 6 cm. broad, though its borders were indistinct. A biopsy showed its sarcomatous nature, but the patient refused amputation, and the tumor was therefore widely extirpated. Healing was rapid and uneventful, but there was early recurrence and death occurred five months later with symptoms of inguinal, abdominal, and pulmonary metastases. Autopsy was refused. Macroscopically the general aspect of the tumor is like that in the case just described. It, too, is situated immediately below the fascia, but infiltrates the surrounding tissues more extensively. It also is composed of several large firm nodules, between some of which are small clefts. Here and there the tumor tissue is softened or hemorrhagic. Histologic Description: On the whole the tissue is less polymorphous than that of the first tumor. Topographically the tumor adheres firmly to the fascia. Instead of the long subfascial cleft seen in the first case, we found several smaller ones, the borders of which are either formed by tumor tissue or show the features of a normal or irritated synovial membrane (Fig. 6). Most of the tumor tissue consists of densely packed and almost undifferentiated mesenchymal elements, with great numbers of overlapping nuclei and little protoplasm. The nuclei are pale and minutely nucleolated and contain a very fine granular chromatin with a distinct but irregularly notched membrane; they are elongated or oval. The cells are branching and in part syncytial. After saffron staining the cell strands appear homogeneous and suggest very strongly the aspect of histiocytoma which has become sarcomatous, but in some areas Masson’s trichrome with anilin blue shows capillaries and a few collagenous fibers. Laidlaw’s silver method for reticulum reveals a rich network between the cells (Fig. 7). In some isolated and widely separated spots are pseudoglandular structures the con- 5 10 LOUIS BERGER tinuity of which with the surrounding sarcomatous tissue gives an " epitheliosarcomatous " aspect. The pseudoglandular spaces are either tubular, branching, or papillary (Fig. 8'). The lining cells are flat, globular, or cubical. hut most often approximately columnar. There is no basal membrane, the cells resting directly on the neighboring sarcomatous cells, which may form small protruding axes. The nuclei of the lining cells are identical with those of the other cells. The cavities are often empty, but some contain isolated desquamated border cells or a little magma which recalls that of the first case. There is generally not much mucin, but shreds staining unmistakably with mucicarmine are occasionally seen. Desquamation and

FIG.7. CASE11: SOLIOPART OF TUMOR With routine staining (left) small threads of elongated, moderately deep-staining cells are secn running between threads or spots of slightly clearer and more irregular cells. With rcticulum impregnation (right) the clearer spots do not show any reticular fibrils.

appearance of mucus seem independent of each other; the latter originates in the lumen of the spaces in contact with the lining cells, independently of any cell destruction. Mitoses are frequent in the tumor and are as numerous in the sarcomatous areas as in the cells surrounding the spaces. Diugtiosis: This tumor is of the same nature as the first: it consists equally of solid areas of cells resembling histiocytes, between which is an extremely rich argyrophile reticulum, and of pseudoglandular endothelial spaces. The persistence of some normal or irritated synovial fragments or clefts characterizes it as a synovial sarcoma of a subfascial bursa. It differs, however, from the preceding tumor in its lack of very indifferent reticulo- sarcomatous elements, preponderance of the solid mesenchymal areas over endothelial structures, Iiresence of some fibroblastic and collagen-forming cells and scarcity of mucin. From a purely descriptive point of view it appears, nevertheless, as a reticulo-endothelio-histiocytosarcoma. CASE111: No. 19.030. I.P.Q.: A white man of twenty-six entered the hospital with a tumor in the right axilla which had appeared three months before. A tentative diagnosis of lymphosarcoma was made, but on operation the tumor was seen to invade neither the lymph nodes nor the articular capsule of the shoulder. Excision was difficult and the nro- plastic tissue was removed in fragments. Healing was delayed by suppuration and was complete only after six months. The patient was subsequently lost sight of, but on inquiry it was found that he died a year after operation, two years from the onset of the tumor, with symptoms which were thought by his lay associates to be tuberculous-the man was a SYNOVIAL SARCOMAS IN SEROUS BURSAE AND TENDON SHEATHS 5 11 lumberjack-but which may as well be attributed to pulmonary metastases, inasmuch as no sign of tuberculosis was present during his prolonged sojourn in the hospital. Histologic Exatnitdon: Only a few small and in part poorly fixed pieces of the tumor were available for study. Because of this and because of the presence of necrosis, a topographical study as in the first two cases was impossible. One of the fragments, however, was bordered by a dense collagenous layer of fascia1 or tendon-sheath type. For the most part the tissue consists of strands and whorls of small elongated cells separated from each other by strands of slightly more voluminous syncytial elements. The elongated cells show only exceptionally some fine collagenous fibrils, but are always sur-

FIG.8. CASE11: PAPILLARY" EPITIIEI.IO-SARCOMATOZ~S " ASPECT rounded by a rich network of argyrophil reticular fibrils. The syncytial strands contain no reticulated elements, but are frequently interspersed with cavities which are progressively enlarging, while their flat lining cells are swelling, approaching a cubical or cylindrical shape (Fig. 9). The spaces are often branching and some groups of elongated sarcomatous cells protrude and form irregular, epithelial, dendritic villosities. The cavities are most often empty, but some contain desquamated cells or a little magma similar to that of the two preceding cases. There is no mucin. Diagnosis: This tumor was at first believed to be a reticulo-endotheliosarcoma of a somewhat exceptional type, arising in a lymph node. A more thorough study, a considera- 512 LOUIS BERCER tion of supplementary details obtained from the surgeon, and experience gained through the first two cases caused us to change our diagnosis. The exclusively mesenchymal and endothelial, even pseudo-epithelial, aspect of the growth, the absence of lymphatic cells in the tumor and of lymph node tissue around it, the presence of the peculiar magma and of desquamated cells in some of the pseudoglandular cavities, were incompatible with a lymph node origin. On the other hand, the abundance of reticulum, the similarities to the preceding cases, and its contact with a fascia1 or tendon sheath layer characterize the tumor- in the absence of any articular connections-as a synovial sarcoma arising in an axillary serous bursa or tendon sheath and having the features of a reticulo-endothelio-sarcoma. CASEIV. No. 1428. H.E.J.: A white female, seventy-five years old, had a small tumor on the dorsal side of the right wrist. When she first noticed the growth, she applied

FIG.9. CASE111: ‘I EPITHELIO-SARCOMATOUS” ASPECT WITH HIGHCOLUMNAR-LIKE CELLS

the popular procedure of crushing it, but without success. Growth continued slowly for seven years, but later became more rapid, inducing the patient to consult a surgeon. At that time the tumor was the size of a prune, flatly bulging, nodular, and quite hard. It was situated over the 3d and 4th metacarpals and was adherent to the tendon sheath. At operation there was found a central mass occupying the site of the common extensor tendon sheath and surrounded by some smaller nodules. The entire tumor was enucleated and the surrounding tissues were curetted to the bone. The tendons were not invaded. The wound healed uneventfully, but three weeks later a recurrence appeared, which grew rapidly and in two months attained the size and the shape of the primary tumor at the time of operation. In spite of progressive growth, the patient did not return until ten months later. At that time the tumor covered the two proximal thirds of the metacarpal area and continued over the styloid process of the radius. At this latter point it was ulcerated. Move- FIG. 10. CASE Iv: SOLID STRANDS OF LARGEPALE AXD POLYMORPliOUS CELLS, WITH INDEFINITE OUTLINES,BECOMING SYNCYTIAL AND PLASMODULELEMENTS

FIG. 11. CASE IV: ASPECT LIKE THATIN FIG. 10 AFTER RETICULUMIMPREGNATION

513 5 14 LOUIS BERCER ments of the wrist were only slightly impeded; the fingers were in semi-flexion. The center of the growth lay over the common extensor. Histologic examination of the material removed at the earlier operation and the subsequent evolution showing without doubt that the growth was malignant, the patient agreed to amputation of the forearm at its upper third, which was done without further delay. The tumor forms a swelling 2.5 cm. thick. In the metacarpal region the transverse diameter is 6.5 cm., the longitudinal 4.5 cm. The expansion towards the radius measures 2.5 cm. In the depth the tumor occupies the place of the common extensor tendon sheath; it does not involve the tendons but widely invades the deep layers of the skin. When cut, the tumor is seen to be composed of numerous more or less oval nodules, of an average diameter of 1 cm., bathed by a viscous and mutinous fluid. The peripheral nodules are rather firm; those toward the center are softened. There are no clefts.

FI~.12. CASEIV: ISTERCEI.I.UI.ARDROPLETS AND THREADSOF MGCIN(DARK GRAY)

Histologic Description: The features of the primary tumor and of the recurrence are identical and may therefore be described together. The neoplastic tissue is grossly and irregularly lobulated; the nodules are in part fused together and elsewhere are separated by small connective-tissue strands. Towards the skin the limits are generally quite shalgly defined but in a few places peripheral cells are clearly infiltrating the adjacent tissue, having penetrated small lymph vessels or formed isolated groups around sweat glands. Here and there lymph vessels of the corium are distended and filled by a fluid which takes a moderate mucicarmine stain, but are devoid of cells. Towards the depth the tumor respects the tendons, but invades the neighboring connective and fat tissue. The tumor is interspersed with necrotic and hemorrhagic foci. The neoplastic tissue has two principal aspects which are often found together but may appear separately. One type is formed by pale polymorphous elements, of the average size of malpighian cells, but without definite boundaries; the grouping is indeed often syncytial and the cells may form SYNOVIAL SARCOMAS IN SEROUS BURSAE AND TENDON SHEATHS 515 large multinucleated plasmodia1 masses (Fig. 10) ; the nuclei may be monstrous and budding. Mitoses are numerous and often abortive or multipolar. The cell masses are irregularly subdivided by small connective-tissue and vascular septa. Laidlaw’s silver method reveals an abundant reticular meshwork which often surrounds each individual cell (Fig. 11). The second type of neoplastic tissue is characterized essentially by the occurrence of small cell groups or isolated cells through a homogeneous or stratified substance which stains heavily with mucicarmine. These areas have, at first view, the aspect of a mucous or colloid epithelioma, but there is no mucus in the protoplasm. The quantity of much varies from one area to another: sometimes it forms large threads which are poor in cellular elements; sometimes it is reduced to small intercellular filaments or droplets (Fig. 12). The cells resemble somewhat those of the first type: they are equally pale, polymorphous, and indefinitely outlined, but are of a larger average size and more distinctly syncytial. The reticulum is less developed and may be broken up and reduced to a fibrillar debris. What appear to be intracellular mucous droplets are sometimes seen. Actually mucin is always extracellular. though some may be engaged in small peripheral depressions of cells and appear intracellular in tangential sections. Between the two types of tissue described are numerous transitional aspects which prove a common origin. The mucous type seems to develop from the reticular, through progressive disappearance of the reticulated fibrils. Diagnosis: The malignancy of this tumor is evident; it is shown clinically by the recurrence and rapid growth, histologically by absence of a capsule, peripheral infiltration, and invasion of lymph vessels. The mesenchymatous nature of the tumor is certain, but its aspect does not recall any classical sarcoma. It is composed in part of polymorphous and voluminous cells with indistinct boundaries and a tendency towards syncytial arrangement, which are surrounded by a very fine reticular meshwork. This aspect recalls in some points that of certain histiocytomas, but is progressively giving way to a mucous or colloid tumor. The mucous parts, which present an intense mucicarmine reaction, resemble much more a mucous epithelioma than a myxosarcoma, but actually correspond to neither. They are truly specific, but their mesenchymal affiliation is still evident. The tumor is thus characterized by a kinship with histiocytomas on the one hand and by muciparous properties on the other hand. The much does not, however, seem to be properly secreted by the tumor cells, but appears between them, at their points of contact. The surgical and microscopic observations show the common extensor tendon sheath to be the point of origin of the tumor. We do not hesitate to identify it as a synovial sarcoma of a peculiar type, which is nevertheless akin to our three other cases in its reticulo- histiocytic aspect and the presence of extracellular mucin. From a purely morphological point of view it is a reticular or histiocytic muciparous sarcoma. It differs from the preceding cases in the entire lack of endothelial or ‘I pseudoglandular ” structures. Rtsumd of the Four Tumors: These four tumors form, by virtue of their peculiar anatomo-clinical and above all by their histologic features, a distinct oncologic entity. The resemblance between the first three is particularly impressive; the fourth appears on first sight to be different, but it is nevertheless akin to the others, the fundamental characters of which it reproduces in a different form. The first three tumors, histologically so similar to one another, appeared in males of twenty-six, thirty, and thirty-eight years, and were removed by surgery. All three recurred a few months later and two led to death one and two years after onset, very probably through pulmonary metastases. The rapidity of recurrence in the patient still living suggests a poor prognosis. These three tumors are regarded as originating in serous bursae. In Cases I and 11, where fragments of the normal synovial covering were found, this origin is unquestionable. It is only probable in Case 111, where a tendon- sheath origin cannot be entirely eliminated. The bursae of the first two cases were subfascial and found in places where bursae are inconstant or not 516 LOUIS BERCER

regularly reported: the first in the upper and interior third, the second in the lower and posterior third of the thigh; the third was in the axilla. The histologic aspect of these neoplasms suggests at first view (‘epitheliosarcoma,” for they seem to be formed by an association of “ fibro-epithelial ” elements. A closer analysis shows them to be of a single tissue origin, mesenchymal in nature, but presenting various morphological modifications. The tissue consists essentially of dense cellular masses giving rise to pseudoglandular, simple or ramifying cavities. The cell masses are formed by branching, curved, sometimes polyhedral, more rarely spindle-shaped cells. Their scanty protoplasm stains well; its expansions appear to anastomose with those of the neighboring cells, producing a syncytial aspect. The nuclei are most often elongated, but irregular and relatively large. The cells may form irregular threads or whorls or may be piled up without any order. They give rise only rarely to collagenous fibers, which always remain slender. The aspect of these dense masses recalls some histiocytomas rather than the classical ljbrosarcoma, as may be seen in Fig. 13. (The distinction between the histiocytic and fibroblastic tumor entities is of a quite recent date; it is above all due to G. Levy, Woringer, SPzary and to Pautrier. In the American literature a very good resum6 of their work with new original observations may be found in the recent paper of Senear and Caro. For our part, we are the more inclined to accept this distinction because we could verify it in several personal observations.) In the dense masses of our tumors the aspect differs slightly from that of histiocytomas, but the difference is essentially due to the fact that, whereas histiocytomas are benign, our tumors are malignant. We do not hesitate to consider them as histiocytic sarcomas. This diagnosis is supported by the presence of an argyrophil reticulum, demonstrable by Laidlaw’s method, of great richness, forming very small meshes, the fibrils of which in some places surround each individual cell, and by the functional differentiation of some of the neoplastic elements. The histiocytic concept has, indeed, primarily a physiological basis and rests upon manifestations of phagocytic properties. Some of the tumor cells contain lipoid or hemosiderin inclusions and recall the foam cells of xanthoma. These features sustain their histiocytic nature. This nature reveals itself further in Case I, where some peculiarly undifferentiated cell masses have exactly the same features as some reticulosarcomas of the bone marrow or of lymph nodes. The cavities are lined by cells which may be flat, cuboidal, or approximately columnar, resting directly on elements of sarcomatous aspect. There is an uninterrupted series of intermediate aspects between the highest pseudo- epithelial and the flattest frankly endothelial cells. On the other hand, the latter derive directly from sarcomatous elements, as is shown by very small clefts or spaces which are lined in part by sarcomatous, in part by swollen endothelial or cuboidal cells. The pseudoglandular structures are therefore endothelial and result from a differentiation of histiocytic sarcomatous elements. This unique neoplastic matrix appears with peculiar distinctness after impregnation of the reticulum; in some places where routine methods show only sarcomatous clusters, the Laidlaw method demonstrates clear spots or threads of cells which are scarcely larger and correspond to a beginning endothelial evolution still devoid of spaces. SYNOVIAL SARCOMAS IN SEROUS BURSAE AND TENDON SHEATHS 517 Some spaces present peculiar and characteristic features: there may be a desquamation of cells, which degenerate and form a partly squamous magma. The frequency with which this albuminous magma appears and its systemati- sation suggest that it may be the equivalent of a normal process. Secondarily it may become transformed into calcospherites. Some spaces contain mucin, staining deeply with mucicarmine. The mucus seems to be of extracellular origin, arising from a colorless premucoid substance. Mucin droplets which were unquestionably intracellular were never found, though in some rare instances an intraprotoplasmic aspect is simulated. This, however, is due to tangential cutting of cells and to the fact that the mucin is often precipitated and moulded upon the cellular surface, adhering to its depressions. Rarely mucin and the desquamative magma are mixed together, but each may occur independently of the other.

FIG. 13. LEFT:CONMON FIBROSARCOMA-LIKEFEATURES IN SYNOVIALOSARCOMA (CASE 11). RICIIT: COMMONFEATURE i~ A FIBROUSAREA OF AN ORDINARYXANTHOMA (HISTIOCYTOMA) OF SKIN (No. 1358. I.P.Q.)

The quantities of the various neoplastic elements vary greatly, not only from one tumor to another but even in the same tumor. In Cases I and I11 the reticulo-histiocytic and endothelial elements are about equally abundant and the cavities are numerous, but in Case I1 the histiocytic aspect predomi- nates and the endothelial evolution is apparent only here and there. This last observation is of importance in diagnosis: there must be numerous sections or, better, some very large sections made at various levels of the tumor, if an inadequate and therefore incorrect diagnosis is to be avoided. Mucin, so much more characteristic as it appears in the absence of epithelial cells, is also variable: it was abundant in the first, scarce in the second, and entirely lacking in the third case. Desquamation was intense in the first, moderate in the third, and discrete in the second. In comparison with these tumors the fourth seems at first to be of a distinct variety, for in it all evidence of endothelial evolution is lacking. It appeared, furthermore, in a female very much older than the three males; its evolution 518 LOUIS BERGER was much slower, and its origin was in a tendon sheath. It is nevertheless akin to the preceding tumors, though constituting a peculiar variety from the point of view of evolution. Like them, it is built up by a fundamentally reticulo-histiocytic tissue, but its cells are less elongated, much richer in protoplasm and more polymorphous. Interstitial collagenous matter is lacking, but silver-impregnation reveals a very fine and abundant reticular meshwork. It contains large quantities of intercellular and deeply staining mucus. It is therefore a muciparous reticulo-histiocytosarcoma without endothelial evolution. This tumor shows, among other points, that mucin production may be independent of any endothelial disposition or differentiation. All four tumors reveal, therefore, through their structure and function, certain characters which are attributable to synovial membranes, of which they are to be considered the malignant neoplasms. Our cases show two varieties: a first and more frequent is formed by the endothelial synovial sarcomas, with or without mucin production; the second by the mucin-forming synovial sarcomas. All four contain features recalling the fundamental characters of the synovial matrix, but only the first is highly typical. Its organoid structure is, in spite of its malignancy, so pronounced that it seems capable of making a new contribution to the problem of the nature of normal synovial tissue.

SYNOVIALSARCOMAS AND NORMALSYNOVIAL MEMBRANES It may seem questionable to utilize pathological findings, particularly in the realm of neoplasms, in the determination of the nature of normal tissue, although such a proceeding has yielded substantial results in the physiological as well as in the histologic field. In the present case we do not seek to recon- struct from our neoplasms the nature of normal synovial tissue, but only to determine with which of the different conceptions of normal tissue these neoplasms are most compatible, or at any rate least incompatible. The histologic structure of our four tumors sustains strongly the ideas of Franceschini and of Sabrazes and his fellow-workers, who conceive of the synovial membrane as a reticulo-endothelial or reticulo-histiocytic tissue. Indeed, from a strictly morphological point of view these tumors are fundamentally reticulo-histio-endotheliosarcomas. The coordination of the synovial tissue and its tumors with the normal and neoplastic reticulo-endothelial system may seem surprising, but it must be recalled that the first conception of this system was much more physiological and functional than anatomical and that since the time of Aschoff, who restricted it to the lymph nodes, spleen, bone marrow and Kupffer cells, the system has been enlarged through the addition of histiocytic elements in the loose connective, peculiarly perivascular, tissue (Kiyono). According to the various localizations the cells of this system undergo particular differentiations : sometimes predominantly myelo- cytic or lymphocytic, sometimes essentially phagocytic. These properties are reflected in its tumors : Oberling’s medullary reticulosarcomas with myeloid evolution and Craciun’s reticulosarcomas of lymph nodes with lymphoid evolution ; the xanthomas and xanthelasmatized histiocytomas of G. Levy. With Franceschini and Sabrazks one may consider the synovial tissue as another part of this system, endowed with peculiar properties. SYNOVIAL SARCOMAS IN SEROUS BURSAE AND TENDON SHEATHS 519

In our tumors not only do we find a very rich reticular network which surrounds cells with histiocytic features, but in one of them (Case I) there are masses of indifferent syncytial cells which are indistinguishable from the second variety of Oberling’s medullary reticulosarcoma and Craciun’s lympho- reticulosarcoma or with the immature form of Roulet’s retothelsarcoma. The phagocytic property, pertaining to the whole reticulo-endothelial system, mani- fests itself here and there, where some neoplastic elements contain lipoid or hemosiderin inclusions. We do not, however, agree entirely with Sabra& when he says that the superficial cells of the normal synovial membrane, though they present a peculiar adaptation, never realize a truly endothelial structure, and that the formative potentialities of this reticulo-histiocytic tissue are directed primarily toward the production of monocytes. While we shall not discuss this monocytic function, which appears nowhere in our tumors, our observations seem to prove the exquisite endothelial potcntiality of synovial cells, which appears in the numerous pseudo-epithelial structures. It may be that under normal conditions its development is generally impeded, although it may be apparent in some folds or on some villi, but the endothelial modification which appears very often deep in sarcomatous cell masses, independent of any morphogenetic action depending upon mechanical factors, is highly in favor of an intrinsic endothelial potentiality of synovial cells. The endothelial conception of synovial membranes seems to be with some authors a question of definition. While some (Schaffer, Aschoff, Kaufmann) do not hesitate to speak of a synovial endothelium, others refuse that term on account of the discontinuity of the lining and perhaps also because they prefer to restrict its use to the vascular system. Today, however, we distinguish (Hueck) even in the circulatory system between ‘‘ closed ” or continuous and “ open ’’ or discontinuous endothelium: the ordinary vascular endothelium is “ closed,” that of the sinuses of lymph nodes and bone marrow is “ open.” The designation endothelium may therefore be equally applied to synovial linings, the more so, since it is- at least in some places and under certain circumstances-continuous, unless one prefers the term synoviothelium proposed by Hueck, which is analogous to the retothel of Roulet. The synovial endothelium has, however, peculiar features which distinguish it from other endothelium, namely its desquamative properties and the appearance of mucus in the spaces which it lines. Our tumors seem also to permit of some conclusions regarding the formation of synovial fluid. In normal tissues (articulations, bursae and tendon sheaths) this liquid forms slowly and in spaces too large to be studied easily. In the small spaces of our tumors, which are equivalent to the normal cavities, the activity of the lining cells is intensified and the magma and mucin seem to form a sort of concentrated synovial fluid, due to the lack of liquid diffusing from the blood stream. Our slides show the magma resulting from desquamated lining cells and the mucin appearing from an unstained premucoid substance which seems to be present in the spaces. The production of mucin seems independent of desquamation, for it may appear without the latter. Our findings, therefore, favor the opinion of those authors who believe that desquamation of cells occurs in normal synovial cavities and that the albuminoid substance resulting from their disintegration plays a r61e in the 5 20 LOUIS BERGER formation of synovial fluid. With regard to mucin our observations favor the hypothesis of an extracellular origin and seem to be inconsistent with the view which attributes it to cell lysis, for on the one hand we never found it in the protoplasm of cells, and on the other hand it was often entirely free of albuniinoid particles or formed intercellular droplets or threads too small for derivation from disintegrated cells. The genesis of synovial mucin appears very different, therefore, from that in epithelial cells. It is probable that the synovial cells play a r6le in its formation, but this r61e is less one of secretion proper than that of a cellular interaction which may be compared to that of collagenous matter in the sense of Nageotte or to that of prechondroid or preosseous substance. It is worthy of note that, notwithstanding the production of mucin, our tumors never presented a myxoid or myxomatous aspect. In rCsum6, it may be stated that our tumors constitute, by virtue of their endothelial as well as their mucous features, a group of clearly defined neoplasms without parallel in the oncological realm, though they are in some manner akin to the reticulo-endothelio-histiocytictumors as a whole, showing peculiar evolutionary features. They favor the conception that the synovial tissue is a tissue sui geneiis, being a part of the reticulo-histiocytic system in a broad sense,’ but unique in its pronounced endothelial potentiality, its functional cellular desquamation and peculiar extracellular production of mucin.

CRITICALSURVEY OF THE LITERATUREON SYNOVIALTUMORS The literature of synovial tumors presenting the gross features of our own cases is surprisingly limited, and many tumors which are classified as synovial are inadequately described, especially from the histologic point of view. Some of the reported tumors are common sarcomas with only a presumptive synovial origin, and many cases-perhaps the majority-are of the “ xanthomatous giant-cell tumor ” variety, which we shall later ‘discuss. In some cases of true synovial neoplasms, moreover, it is not always made clear whether the point of origin was in a joint, in a serous bursa, or in a tendon sheath. The endothelial neoplastic entity which we found in serous bursae was first described in articulations. It seems highly probable that the ‘‘ adenosarcoma ” of the elbow reported by Stuer in 1893 is the first instance of the kind, but it was not until 1910 that the first proved case of “endothelio- sarcoinatous ” tumor was recorded by Lejars and Rubens-Duval, who called it a synovial endothelioma. The cases of Ruediger-Rydygier (1916) and of Hannemuller (1909) seem doubtful to us. A small number of cases of the same type have since been reported and are summarized in the accompanying table. A more detailed review may be found in the papers of Sabrazits and of Knox, who reported three cases, one of which, however, in the ankle joint The classing of synovial tissue with the reticulo-endothelial system may seem radical. It may be that some day we shall recognize that the fundamental feature of that system is not so much its phagocytic and hematopoietic potentiality as its plasticity. Some authors express this thought by attributing embryonal qualities to the system. For many reasons we dislike this formula, for the most suggestive of its features, the storing capacity of its cells, appears most often in pathological conditions, quite apart from the rigid determination of embryonal developmmt. Aschoff-Kiyono’s system impresses us primarily as an indifferent “ waiting ’’ or “ slumbering ” tissue with several potentialities, the manifestation of which depends upon various, often accidental influences. With regard to our actual question, this tissue may, for example, give rise to irregular, occupational serous bursae. SYNOVIAL SARCOMAS IN SEROUS BURSAE AND TENDON SHEATHS 521 or a nearby tendon sheath, did not, like her other cases, present convincing synovial characters. We must disagree with Knox, also, as to the cases of Faccini, Tavernier, Hodgson and Bishop presenting indisputable synovial characters, for the histologic descriptions in these cases are either too brief or too indefinite. The most comprehensive and detailed study of synovial tumors of the joints is that of Sabrazb and his co-workers, which may justly be considered as classical. Their two cases were characterized by the coexistence of a sarcomatous tissue which was composed of elementary, elongated or branching connective-tissue cells with scanty intercellular substance, and of flat, cuboidal or columnar endothelial cells which originated from the former and surrounded pseudoglandular cavities; the two features were often intermingled and suggested “ epitheliosarcoma.” The cavities contained either an irregularly staining magma or mucin. Sabrazb also reports cellular desquamation and insists upon the reticular aspect of the sarcomatous strands; without having impregnated the reticulum, he considers them as reticulosarco- matous cell cords. The similarity of our first three cases in bursae to the articular tumors of Sabrazks is impressive. The case published by Razemon and Bizard in their excellent review of articular tumors is not specifically synovial. Fievez’s recent paper is difficult of evaluation, for it contains a rather fundamental contradiction which is worth quoting. The author states, first, that ‘‘ in the places where the structure is well recognizable it [the tumor] is an endothelioma constituted by elongated cells for the most part similar, but here and there forming cavities around which they take the cuboidal form of a lining endothelium ”; in his conclusions, however, he says: “ We are confronted with a more atypical neoplasm [than that of Lejars and Rubens-Duval] , in which there is nothing [ ! 1 recalling an endothelial lining, i.e. with an atypical endothelioma.” The lack of figures makes impossible a choice between these contradictory statements, and Fievez’s case therefore remains doubtful. Records of synovial tumors in serous bursae are still more rare. The often mentioned but seldom analyzed work of Adrian is considered classical, but is primarily clinical and surgical. Adrian reports on a preceding paper by Ranke, eliminates the cases he considers doubtful and, after completing a review of the literature, accepts 17 cases as true tumors of bursae. None of these, however, can actually be admitted as synovial sarcomas inasmuch as most of them were innocent or even inflammatory growths. Two personal cases added by Adrian, examined by Lubarsch and M. B. Schmidt, were briefly described as round-cell sarcomas, but their origin may equally as well have been in connective as in synovial tissue. Even the malignancy of these sarcomas is doubtful, for in Adrian’s series only his own first case proved fatal through metastases. After the appearance of Adrian’s paper Martina reported on a myxofibro- sarcoma of the posterior Achilles bursa which appeared six years after a fracture of the malleolus in an officer of twenty-nine. The histologic description is in accord with the diagnosis, which leaves the truly synovial nature of the tumor problematic. Becker’s case of a tumor of the prepatellar bursa in a woman of thirty-six, with recurrence and swelling of the inguinal lymph nodes, is simply diagnosed as a spindle-cell sarcoma and must be eliminated. Sex anc Author age Locat ion Diagnosis Course 1 Remarks (years) I

1. Stuer 1893 3 21 .\rt icular (elbow " Adenosarcoma " Iiemoval: slow recovery; no recurrence after four so follow-up months but swelling of cubital I>.mph nodes (metastasis?) 2. Lejars and d 22 .4rticular (left Synovial endothelionia Removal: recurrence after five months; amputation; Rubens-Duval knee) of "epitheliosarcoma- patient well Seven months later but subsequent in- 1910 tous" aspect quiry by Kazemon and Bizard revealed death from metastases 3. Chenot and 0 38 .+rticular (astrag S~movialendothelionia So clinical follow-up Tzanck 1912 alo-scaphoid) v, 4. Smith (I) 1927 sot Bursa (of Hun- Malignant synovionia Removal; death six months later, probably from pul- Description brief, but N N ;pecifiec ter's canal) monar!' metastases figures convincing 5. Smith (11) 1927 0 24 Bursa (upper in- hlalignant synovionia Removal; death two years after appearance of symp- Author does not en- ner aspect ol toms, from pulmonary metastases tirely eliminate ova- thigh) rian origin 6. Smith (Ill) 1927 8 35 Articular (knee) Alalignant sjmovioma Removal; recurrence after wnie months; x-ray and radium therapy without result; amputation 15 cm. above knee; pulmonary metastases; death two and a half years after onset 7. \\Pgelin 1928 8 28 Articular (right S!movial sarco-endotheli- So recurrence after five months So late follow-up knee) oma (organoid tumor) 8. Prym (1930) 0 66 Articular (right Fibro-endothelial tumor liemoval; recurrence after seven months; amputa- So autopsy knee) with mucin production tion; inguinal metastasesafter twoyears;death nine years after onset without evident metastases 9. \\'agner (1) 1930 3 35 Articular (right Synovial fibro-endot heli- Removal and x-ray therapy; recurrence two years Diagnosis confirmed knee) oma later; mid-thigh amputation; lived eight years with by Ewing, who had no recurrence seen one of Smith's ' cases and two others 10. IYagner (11) 1930 0 15 Articular (tibio- Synovial fibro-endotheli- Removal and x-ray therapy; untraced (unpublished) tarsal) oma Discussion SYNOVIAL SARCOMAS IN SEROUS BURSAE AND TENDON SHEATHS 523

In 1927 Smith reported three instances of synovioma which closely re- sembled our first three cases. The first, in a patient whose age and sex are not given, was located in the fascia of Hunter’s canal and was characterized by clefts which were lined by epithelial-like cuboidal cells and were separated from each other by solid cords of spindle cells, devoid of fibrils and intercellular substance. Wolbach, who studied the slides, confirmed the diagnosis of synovioma, which seems sustained by convincing figures, in spite of Prym’s contrary opinion. Smith’s second case is that of a tumor of the superior and inner third of the thigh, extending above Poupart’s ligament, in a woman of twenty-four. Although Smith himself does not entirely rule out the possibility of an ovarian origin, he considers this tumor as a probable synovioma, and here, too, the diagnosis seems to be sustained by his figures and description and the likeness to the first case. The third tumor was an indisputable synovial sarcoma of the knee-joint in a man of thirty-five. Schwamm (1930) reported on a tumor in the subacromial bursa in a girl of nineteen, which was studied by Maresch. It was encapsulated, grew slowly for three years, and was interspersed with cavities which presented papillary projections. The flat lining cells sometimes became cubical or cylindrical and the spaces contained a granular or filamentous substance staining indiffer- ently with hematein and eosin. Maresch considered this tumor as “ a papillary neoplasm toward which the synovial and connective tissue and the synovial lining all contributed.” Schwamm himself regarded the tumor as a papil- lomatous bursitis, but Maresch’s interpretation, being that of an experienced pathologist, must prevail. In his opinion this case is an instance of benign synovial tumor. The five cases in bursae reported by Geschickter and Lewis (1935) were chondromas (2), enchondromas (2), and a spindle-cell sarcoma, none of which presented specific synovial features. In the same year Zwahlen described a synovioma of the tendon sheath and one of a serous bursa, which were accepted as such by Knox in her bibliographic review. We must confess to being less convinced. In his case reports proper Zwahlen makes no men- tion of epithelial-like cells, but in his discussion states: “ Although our cases lack true glandular structures, clefts and spaces are encountered which are lined in part by epithelial-like cells and definite transitions between epithelioma and sarcomatous elements are seen.” He mentions King’s synovial spaces with which he identifies some of the cavities in his own tumors, but none of his figures shows any specific feature. It is therefore difficult to decide on the true nature of Zwahlen’s tumors. Knox (1936) reported three cases of synovial tumors, the precise point of origin of which could not be ascertained: the first being probably in a tendon-sheath, the third (mentioned above and eliminated) in the ankle-joint or a nearby tendon sheath, and the second in a tendon sheath or bursa of the popliteal space. This last was formed by fusocellular masses which contained dilated sinuses bordered by flat, cuboidal, polyhedral or cylindrical cells and containing mucoid substance. Brunner (1936) restudied a tumor of a man of forty-seven, which was first erroneously regarded as a heterotopic adamantinoma ( ! ). It arose in a serous bursa of the bicipital space and was formed by cell-cords of epithelial-like TABLEI-Continued

~~ - Sex and .4uthor aw Location Diagnosis I Course Remarks

(years) ~

~~ 11. Schwanini 1930 0 19 Bursa (subacro- Synovialtumor (encapsu- Growing for three years; no recurrence after three Histologic diagnosis mial) lated) months by Xlaresch; considered benign 12. Diez 1931 6 33 Articular (right Synovial endothelioma Amputation refused; untraced knee) 13. Sabrazcs, Loubat, 0 18 .4rticular (left Malignant synovialonia Removal and postoperative s-ray therapy; pain three de Grailly and knee) years later and x-ray therapy; recurrence one year hlagendie (1932) later; removal and resection of upper part of fibula; further recurrence after sis months; amputation at hip: death eight months later with pulmonary nie- 01 tastases 11. SabrazPs Bon- 3 50 .4rticular (left llalignant synovialonia Growing for five years, with recent acceleration ; re- nard, Baillis. and elbow) moval; death shortly afterward probably from pul- de Grailly 1932 monary metastases

15. Knox (I) 1936 0 22 Tendon sheath Synovial sarcoma Removal; recurrence fifteen months later; amputa- (forearm) tion of arm at upper third; seven years later metastasis in shoulder, of same histologic aspect: death ten years after onset, probably from pleuropul- monary metastases 16. Knox (11) 1936 d 33 Bursa (popliteal Synovial sarcoma Biopsy and amputation of hip: pulmonary metas- space) tases a year and a half later; death two and a half years from onset 17. Brunner 1936 8 17 Bursa (bicipital Malignant synovioma Growing Seven years; removal; metastasis after two Tumor first diagnosed space) months: removal and s-ray therapy; asillary recur- as heterotopic adam- rence six months later; death from cerebral metas- antinoma , tases ten years after onset SYNOVIAL SARCOMAS IN SEROUS BURSAE AND TENDON SHEATHS 525

and reticular aspect, which were folded and interspersed with cavities. The neoplastic spindle cells formed a single or double row of palisades around the cavities and became columnar, less often cuboidal; there were some dendritic structures. The cellular tissue had a reticular and syncytial aspect and contained a very fine fibrillar network. The case of Bonne and Collet ( 1936) seems to have been a true synovial tumor of a bursa of the external side of the knee, but their original paper could not be consulted. The literature of tumors in tendon sheaths, exclusive of xanthomatous giant-cell tumors, which will be discussed later, and of tumors which may equally well be derived from the nearby supporting tissue, is even less precise and more complicated than that dealing with tumors of the joints and bursae. In 1931 King attempted a synthetic study. He believes the importance of giant and xanthomatous cells has been over-estimated and describes and illustrates tumors containing spaces which he considers as equivalent to normal synovial cavities. These spaces are lined by spindle-shaped or spherical cells and sometimes contain papillary projections; endothelial cells are said to be lacking. Unfortunately King’s well documented paper is too general and too condensed-there is no individual case history-for a clear understanding of its arguments. Although the author mentions instances of true malignancy in his tumor cases, he seems to be dealing chiefly with xanthomatous giant-cell tumors. We do not consider that his classification favors the understanding of synovial tumors as a whole, but his discovery of peculiar spaces in the apparently xanthomatous variety of growth is of interest. Tumors in tendon sheaths, resembling more or less our own, are particularly scarce. Most of the reported cases are mentioned above, under the heading of articulations and serous bursae, because the tumors had simul- taneously invaded these structures or were so large that their exact point of departure could not be determined, as in the cases of Knox. We found in the literature to which we had access only one case, that of Black, occurring in the web of the thumb. The cellular and mucoid features in this instance corresponded very closely to those in our fourth case, but the tumor contained also spaces lined by epithelial-like cells. The point of departure is probably a tendon sheath or a bursa. Black believed the tumor to be innocent. Our own cases included, there are actually 23 cases in all presenting specific synovial features, of which 11 are articular, 9 in bursae, and 3 in tendon sheaths. These respective numbers are approximate, for in some cases the authors did not or could not discriminate between a tendon sheath or bursa origin. Sixteen cases were without doubt clinically or histologically malignant. Twenty-one were of the type we call endothelial, and 2 were of the mucous variety; of the first type, 15 were certainly and 5 probably malignant and 1 benign; of the second type 1 was malignant and the other innocent.

CLASSIFICATIONOF SYNOVIALTUMORS Synovial tumors may be classified from two points of view, one essentially topographical, the other exclusively histological. Up to now, only the former has gained expression, and the classifications of King and of Morton comprise TABLEI-Continued

Sex and Author age Location Diagnosis Course Remarks (years)

18. Bonnet and Collet 0 25 Bursa (external Synovioma Tumor attained diameter of 20 cm. in ten months. 1936 aspect of knee) so follo\v-up

19. Black 1936 d 36 Tendon sheath Synovioma (mucous) Removal; no recurrence Considered benign (or bursa?) (web of thumb) 20. Berger (I) 1938 3 30 Bursa (supra- Endothelial synovialo- Preoperative duration six months: removal: recur- Still under observation rence after four months; amputation refused I4 popliteal) sarcoma 21. Berger (11) 1938 8 38 Bursa (inner as- Endothelial synovialo- Preoperative duration six months; removal and x-ray Autopsy refused pect, upper sarcoma therapy; death four months later, probably from third of thigh) pulmonary metastases 22. Berger (111) 1938 d 26 Bursa (or tendon Endothelial synovialo- Preoperative duration three months; incomplete re- So autopsy sheath) (axilla) sarcoma moval; death one year later probably from pulmonary metastases 23. Berger (IV) 1938 0 82 Tendon sheath Mucous synovialo- Duration seven years, with recent acceleration of Still under observation (common ex- sarcoma growth; removal: recurrence two months later; am- tensor. right putation of forearm hand) 24. Berger (V) 1938 d 50 Bursa (middle Histiocytic synovialosar- Preoperative duration one year; removal; recurrence Still under obsenation anterior aspect coma two years later; second recurrence after four months of thigh) disarticulation of hip SYNOVIAL SARCOMAS IN SEROUS BURSAE AND TENDON SHEATHS 527 fibromas, lipomas, chondromas, myxomas, angiomas, xanthomatous giant-cell tumors, synoviomas and their corresponding malignant forms. They contain, indeed, all tumors surgically attributed to synovial tissues and are rather classifications of tumors encountered in articulations, bursae, and tendon sheaths than of truly synovial neoplasms. The intricate relationship existing between synovial membranes and their supporting or nourishing, connective, or even fatty, and vascular tissue does not allow any macroscopic discrimination concerning the point of origin of their respective growths. Histologic grouping is therefore the only rational procedure. Only those tumors should be considered truly synovial which present features characteristic of normal synovial tissue. Contrary to Morton, we think the number of actual cases sufficient to permit an attempt at histologic classification. In the first place, such tumors as lipomas (Straws; White), chondromas (Lewis and Geschickter), hemangiomas and lymphangiomas (Bastos; Faldini; Huguenin and Oberling), myxomas (Martina; Becker), and their corresponding malignant types may be disregarded, as these lack specific features and may be found in other tissues of mesenchymal nature or origin. With regard to the angiomas and lipomas in particular, it is more reasonable to attribute their origin to the nourishing blood or lymph vessels and to interspersed fat tissue respectively than to synovial cells proper. The conception of chondromas and chondrosarcomas as synovial tumors is apparently due to the widespread opinion that synovial cells are more or less modified cartilaginous elements. This conception, as stated earlier in this paper, is attributable to the fact that the first studies of synovial tumors were made in joints, where the two cell families are contiguous or even intermingled, apparently arising from a common anlage. Such a hypothesis, however, is difficult to sustain in the case of dermic or subfascial bursae, which are independent of and sometimes remote from articulations and from any cartilage, but where chondromas nevertheless have been reported. These chondromas we consider as arising in the parasynovial connective tissue, being equivalent to chondromas appearing elsewhere, as for example in the breast, where they would never be considered a variety of mammary neoplasm. According to the investigations of Vaubel, who showed that chondroblasts and synovialoblasts were unrelated to each other, a more logical derivation, in particular with regard to articular chondromas, seems to be from articular cartilage, or, if the tumor appears independent from the articular surface, from aberrant cartilaginous cell rem- nants, which are frequently encountered in the vicinity of joints during routine examinations. Theoretically chondronias may arise from synovial elements through metaplasia, but such a mechanism we believe to be very difficult of proof in particular instances. As to myxomas and myxosarcomas, these tumors may also arise elsewhere in tissues of mesenchymal origin and are not peculiarly frequent in synovial tissues. They are to be sharply distinguished from the mucous variety of synovialoma and synovialosarcoma (case of Black; our Case IV). With regard to fibromas and fibrosarcomas (round-cell and spindle-cell sarcomas), the question is a different one. These tumors represent the most elementary form of connective-tissue tumors and may be found in any mesenchymatous derivative, following some process of dedifferentiation. Fibromas 528 LOUIS BERCER and fibrosarcomas may therefore be derived from synovial cells proper and may be considered an entirely indifferent form of synovialoma and synovial sarcoma. Thus, Sabra& described the recurrence of a typical endothelial (or pseudo-epithelial) synovial sarcoma as looking like a common fibrosarcoma with no resemblance whatever to the synovial matrix. It may be that some of the round-cell or spindle-cell sarcomas of the literature which were considered of synovial origin are really synovial, but owing to the narrowness of the synovial membrane, we believe it impossible to state that a primarily indifferent fibrosarcoma arising in synovial tissue originates from dedifferentiated synovioblasts rather than from a common supporting connective tissue. On the other hand, some cases reported as spindle-cell sarcoma may be histiocytic in nature. Indeed, owing to the kinship of synovial tissue with the reticulo- histiocytic system histiocytosarcomas originating in synovial membranes are to be expected. The diagnosis of these tumors, however, is still a delicate matter. Specific silver methods may sooner or later allow better differentiation between monomorphous histiocytic and common fibroblastic sarcomas ; but, as the question stands today, the former can be definitely diagnosed only if some of the elements are differentiating and grouping themselves charac- teristically or present specific functional characters. Xanthomatotis Giant-Cell Tumors: Of all tumors which have been described in synovial tissues, those known as xanthomatous tumors with giant cells occupy the first place, particularly in the tendon sheaths, where they form the overwhelming majority of published cases. Their frequency in synovial tissues is such that they may be considered peculiar-at least to some degree-to that tissue. They merit, therefore, a more detailed discussion. For this we shall utilize, besides the cases recorded in the literature, 11 personal cases which we have had the opportunity of studying, without however describing them in detail, since their general features are well known (Broders; Ely; King; Leche and Moulonguet; Tourneux; Razemon and Bizard; Morton, and others). These tumors are composed of more or less xanthelasmatized or lipoid- carrying histiocytic elements and giant cells, the proportion of which may vary considerably from one case to the other and even from one area to another in the same tumor, sometimes lagging behind and sometimes dominating the foam cells. The latter are identical with the cells composing xanthomas of the skin. The giant cells were long considered identical with the myeloplaxic or osteoclastic cells of certain bone tumors. The resemblance may be strik- ing, but since in many instances neither the structure and disposition of nuclei nor the staining properties of the cytoplasm are quite the same, we believe that giant cells in bone and tendon sheath tumors represent two different cell families. The resemblance is due to the fact that in both cases the plasmodia1 evolution expresses a phagocytic or resorptive function or at least potentiality. Indeed, we consider giant cells of bone tumors not to be endotheliovascular elements (Malassez; Lubarsch) , but osteoclasts (Robin; Virchow), i.e. cells with osteophagic potentialities or properties, while those in tendon sheath tumors are derived directly from the other histiocytic cells, the fundamental characteristic of which is also their phagocytic function or property (Aschoff; Kiyono). The giant cells in synovial tumors appear to us, therefore, as SYNOVIAL SARCOMAS IN SEROUS BURSAE AND TENDON SHEATHS 529 plasmodial, but not xanthelasmatized histiocytes of the same nature and origin as the large niultinucleated xanthoniatous cells which are often encountered in these tumors. We believe, from a more general point of view, that the plasmodial transformation of cells expresses-if not exclusively, at least most often-an actual or potential resorptive or phagocytic character, examples being the placental syncytium, the multinucleated osteoclasts in normal or pathological bone resorption, tuberculous, foreign body and lipophagic giant cells. The more or less pronounced resemblance of plasmodial cells, even of different nature and origin, is easily explained by that common property. On the other hand, one must beware of over-estimating the significance of the

FIG.14. BEGINNINSPACE FORMATIOX IN A COM~IOSBESIGN XASTIIOMATOUS CWNT-CELL TUMOR (No. 966C. H.S.S.) non-xanthelasmatized giant cells in synovial and peculiarly tendon sheath tumors. Indeed, in our own 11 cases they were present in substantial number in only 7 cases, were scarce in 2, and entirely lacking in 1-all other features being strictly identical. The nature of xanthomatous giant-cell tumors is much debated. In this respect these tumors share the fate of xanthomas of skin, corresponding entirely neither to hyperplastic nor to neoplastic growth. Some authors believe them to be granulomas of inflammatory origin, others dyscrasic tumors resulting from a locally or generally disturbed lipoid metabolism, others benign but true neoplasms. With G. LCvy we believe the fundamental cell of the cutaneous xanthoma to be of histiocytic nature and regard xanthomas, therefore, as xanthelasma- 5 30 LOUIS BERGER tized histiocytomas. Since we have shown that the synovial tissue is in some measure akin to the reticulo-endothelio-histiocytic system, the appearance of xanthomas here is not surprising. These tumors are due to the histiocytic lipopexic potentialities of synovial tissue, which may dominate and even suppress the endothelial and mucous aspects, a phenomenon very common in the realm of oncology. In that sense one may conceive of the xanthomatous giant-cell tumor as a variety of synovial tumor-with the restriction, however, that it does not belong exclusively to synovial tissues but is shared by other components of the histiocytic system, although the presence and eventual frequency of giant cells resembling those in osteoclastomas may produce a distinctive aspect. The synovial spaces described by King, when present, may render this variety still more characteristic and enhance its specificity. King’s evidence for the synovial nature of these spaces is convincing and some

FIG. 15. CASEV: VERY POLYMORPEIOUSCELLS WITH ISDEFISITE OUTLISES

of King’s figures appear conclusive, but we encountered such spaces only once in our series of 11 cases. King believes that the spaces are preceded by a condensation or accumulation of cells along a definite line, followed by cleavage and cleft formation. Our own observation points to another mechanism: at several widely separated points branching histiocytic cells become loosened as if their protoplasm were retracted, and later the cells themselves undergo a slow evanescence and disappear (Fig. 14). The “ degenerating ” area is, however, sharply limited, and the most peripheral cells persist, become cuboidal or even columnar, ready for lining spaces, which are rather round or oval cavities than clefts. As mentioned above, these cavities are often absent, but when present their significance is obvious. A reasonably extended survey of the literature concerning xanthomatous giant-cell tumors shows them to be invariably innocent, whatever the interpretation of their nature. It is a surprising fact that there is no indisputably malignant case on record. We had the opportunity of observing a tumor SYNOVIAL SARCOMAS IN SEROUS BURSAE AND TENDON SHEATHS 531 which, we believe, is of the latter variety and which we shall report, inasmuch as it logically completes the classification proposed below. (Two similar cases were found in our slide collection, but in the absence of complete details we shall refrain from utilizing them in this paper.) CASEV. Nos. 663 and 2035. H.E.J.: A white man of fifty years had a tumor of the anterior aspect of the left thigh, which he had first noticed about a year before. The tumor grew slowly and was never painful. When seen by a surgeon, it was bulging, surrounded by distended vessels, and occupied the entire middle third of the thigh. The overlying skin was thin and glistening; the tumor seemed only slightly adherent to the muscle and to have developed at the level of the fascia. It appeared to be relatively sharply delimited and was therefore removed locally. On excision, however, the underlying quadriceps was found to be invaded. The tumor had apparently grown between that muscle and the fascia, and had attained the size of a fattened orange. It was formed by a few grossly nodular and hard parts. In the upper level a few small cavities filled with serous liquid were present.

FIG. 16. CASEV: IRREGULARFOAM CELLS, SOME OF WHICII BECOMEMONSTROUS

Histologic Description: The tumor had invaded the underlying muscle tissue, whereas the skin and laterally situated tissues were intact. The aspect is polymorphous and varies much from one place to another. Several different but often intermingling types of tissue are distinguishable. About one-third of the tumor has the features of a common fibrosarcoma: the cell. are small, spindle-shaped, with scanty protoplasm ; they have generally two long narrow expansions surrounded by extremely thin and delicate collagenous sheaths. They are separated from each other by small, clear clefts or by albuminoid fluid, so that the tissue appears loose. In a few places the cells are larger, sometimes branching, and have rather indistinct limits. They are generally interspersed with coarse collagenous fibers. The remainder of the tumor is more polymorphous. There are four principal cell types. The first is represented by elongated cells of very irregular size: some are still smaller than those of the fibrosarcomatous part, but most are large and some even reach giant dimensions (Fig. IS). These cells have no long expansions; cell limits are less sharp, and collagenous fibrils are scarce. Between some cells are very narrow empty spaces, but 532 LOUIS BERGER

most often the cells are crowded together. The giant cells may have one or several irregular budding nuclei. The protoplasm is moderately basophilic. There are many, often irregular, indirect cell divisions. A second type has a smaller average size and the principal feature is the indistinctness of the cell outlines, the aspect becoming definitely syncytial ; this type, too, is rich in giant elements. The third type consists of small or large foam cells and reproduces in a more irregular or atypical manner the classical aspect of xanthoma (Fig. 16). Here, too, are giant cells, but they are-at least in part-foamy; some of them either contain numerous fine basophilic and peripheral granulations or are filled with peculiar, coarse and very acidophilic granules. A fourth type of neoplastic tissue has a very loose pseudomyxomatous aspect, but contains a substantial number of giant xanthoma cells. The fundamental features of the tumor are without doubt those of fibrosarcoma, but the exceptionally high number of giant cells, the frequently syncytial arrangement of the

FIG. 17. CASEV: CWSSICAL“ GIANT-CELL”FEATURES tumor cells, and the numerous foam cells point to a less simple matrix than the common connective-tissue cells. Polymorphous myosarcorna (myoblastoma) was ruled out, as there were no tonofibrils and the xanthomatous aspect was difficult to reconcile with a muscle origin. Considering the syncytial tendency of the tumor cells and the xanthomatous evolution of some of them, we arrived at the tentative diagnosis of a partially dedifferentiated histiocytic sarcoma. Laidlaw’s silver method neither sustained nor contradicted this diagnosis. Indeed, in some areas peculiarly of the xanthomatous type it revealed a loose reticular network; elsewhere it showed coarser fibrils, which ran approximately all in the same direction and corresponded more to young collagenous fibers than to truly reticular fibrils. The problematic diagnosis of histiocytosarcoma was, however, later confirmed. Two years after the operation the patient returned to the hospital with a recurrence of about the same size and shape as the original tumor. He refused amputation and the second tumor was therefore removed. Its gross appearance was exactly like that of the SYNOVIAL SARCOMAS IN SEROUS BURSAE AND TENDON SHEATHS 533 first tumor. Below and laterally it invaded the muscle and connective tissues; towards the skin it reached the dermis in some places but elsewhere fragments of coarse connective tissue of fascia1 type extended between that layer and the tumor tissue. The latter is more polymorphous than in the primary tumor: it shows not only almost all of the above described types, but two more features, which are highly important with regard to its nature. In the first place, there are several areas where the tissue recalls very closely the aspect of the classical giant-cell tumors of joints and tendon sheaths (Fig. 17). The giant cells are strictly like those in such tumors, hut the small cells are more irregular, the last character being very probably due to the malignant nature of the neoplasm. In the second place, the tumor contains--especially towards the skin-large irregular spaces, generally blood- filled, the borders of which suggest a hyperplastic synovial membrane. (Some contain granular blood pigment.) In fact, their neoplastic nature is evident: their cells are more

FIG. 18. CASEV: SYNOVIAL-LIKEBORDER, LINING A BLOOD-FLLLEDSPACE The deeper lying cells contain hemosiderin granular inclusions. irregular than in chronic hyperplastic or villous synovitis, but the general arrangement and outlines of the cells are the same, as may he seen in Fig. 18. Although the spaces contain blood, they do not seem to result from a necrotic or hemorrhagic process, for the lining has too organoid an appearance; on the other hand, they differ entirely from truly necrotic cavities which may be found elsewhere in the tumor. The blood seems rather to have invaded them secondarily from small nearby vessels, which have been destroyed or ruptured by infiltrating tumor cells. The spaces do not, however, contain mucin. In the presence of the synovial-like lining one is reminded of King’s synovial spaces and those encountered in one of our own benign xanthomatous giant-cell tumors. We believe that the presence of typical giant cells, of xanthomatous cells, and of cavities highly favors a synovial-and therefore histiocytic-nature for this tumor. The giant cells are identical with those in synovial tumors, particularly of tendon sheaths. The neoplasm is far from any articular or tendinous tissue, but lies directly beneath the aponeurosis, where serous bursae seem to he fairly common, as is shown by our first two cases as well as by others. With regard to the cavities, their appearance in the recurrence only and the lack of mucin may seem to render their synovial nature doubtful, but they are surrounded by too organoid a tissue to be considered accidental. The absence of mucin may be due either to lack of differentiation or to its disappearance through the inflowing of blood. From a general point of view the absence of the more differentiated or more specific structures in the primary tumor and their appearance in the recurrence only, is a curious fact, but not without analogy in other neoplasms. 534 LOUIS BERGER Summary of Case V: The general histiocytic features, the presence of xanthomatous foam cells, of peculiar giant cells and of synovial-like cavities, the partly connective-tissue aspect of the tumor, its polymorphous, irregular and often dividing cells and, finally, its points of resemblance to the innocent xanthomatous giant-cell tumors of synovial tendon sheath origin, characterize our fifth case as the malignant, sarcomatous counterpart of these synovial tumors. Our case does not seem, however, to have originated in a tendon sheath, but in a subfascial serous bursa. We believe, therefore, that this case proves the existence of a truly malignant xanthomatous giant-cell tumor, hitherto undescribed.

Critically reviewing the literature and taking into consideration our own observations, we are led to the conclusion that only those tumors are specifically synovial in which there appears one of the morphological or functional characters pertaining to normal synovial membranes. These characters are ( 1 ) essentially reticulo-histiocytic, (2) endothelial, (3) or predominantly mucous. The first of the three corresponding tumor types is the so-called xanthomatous giant-cell tumor which we consider as a more or less xanthelasmatized histiocytoma and which includes a xanthomatous (lipopexic) and an essentially plasmodia1 form (with giant cells); both types may be and are most often associated in the same tumor. In some cases the presence of peculiar spaces further confirms the synovial origin of this tumor type. The second type of neoplasm is that containing more or less numerous spaces which are lined with columnar, pseudo-epithelial cells; it may contain mucin, and shows solid, rather indifferent looking cell strands through which runs an abundant meshwork of reticulum fibrils. These cells resemble the histiocytic more closely than the common connective-tissue cell type and may occasionally show phagocytic, lipopexic or sideropexic, properties, but also may dedifferentiate and become fibroblastic in aspect. The third type of tumor is that with a heavily mucicarmin-staining mucin between polymorphous and often voluminous histiocytes which are interwoven with a rich reticulum. Each of these three tumor types may be either benign or malignant, One may, finally, conceive of the existence of a dedifferentiated type, looking like a comnion fibrosarcoma, but the truly synovial origin of which may be ascertained only exceptionally.' Nomenclature: What is the most appropriate designation of malignant synovial tumors? We have shown that from an essentially histologic point of view these tumors are reticulo-endothelio-histiocytic sarcomas of variqus types. This term sounds cumbersome, however. Smith calls his tumors synoviomas, Sabra& malignant synoviomas or synovialosarcomas. The term

2 We had the opportunity of seeing a case most probably of that form in P. Masson's labora- tory. The entire tumor tissue presented the aspect described in our first case outside of the pseudoglandular structures, where routine methods showed compact features, but where the reticulum method reveals small threads of cells which form clear spots and correspond to a beginning endothelial evolution, which is not yet accompanied by appearance of spaces. A still more indifferent type, entirely undistinguishable from common fibrosarcoma, was encountered by Sabrazbs in the recurrence of a primary endothelial synovial sarcoma. SYNOVIAL SARCOMAS IN SEROUS BURSAE AND TENDON SHEATHS 535 synovioma is etymologically wrong, deriving as it does from synovia instead of from synovialis; it is further inapplicable to malignant cases, for it corresponds to fibroma, etc. Prym dislikes it for philological reasons and, in particular, because a tumor should be called according to its mother-tissue only if its differentiation is high enough to reproduce the general aspect of the normal tissue; he calls his own case a synovial sarcoendothelioma. Now, even in the malignant tumors the organoid features are often pronounced enough to justify a terminology in accord with the tissue of origin, a terminology which has, besides, many precedents. The term endothelioma, on the other hand, is generically wrong, because many tumors lack endothelial features. For the same reason a name derived from Hueck’s proposed “ synoviothelium ” (for example, synoviothelioma or synoviotheliosarcoma) seems difficult of acceptance. The most convenient designations seem to be synovialoma for the innocent, and synovialosarcoma for the malignant variety, the more so because since Vaubel’s conception of the synovioblast (or, better, synovialoblast ) these terms no longer possess an organoid significance, but complete the general oncological terminology by analogy and are in line with fibroma and fibrosarcoma. For the xanthomatous giant-cell tumors (sometimes without giant cells! ) we favor the generic name synovial histiocytoma or histiocytosarcoma, for this type is not peculiar to synovial tissue and is encountered in it only in so far as this tissue is akin to or a variety of the histiocytic cell system in general. According to these considerations of classification and nomenclature the following table of synovial tumors may be established: I. Innocent tumors A. Synovial histiocytomas (with predominating histiocytic features and of undetermined hyperplastic or neoplastic nature) 1. Xanthelasmatized: pure xanthomas 2. Plasmodial: with giant cells 3. Mixed : xanthomatous giant-cell tumor Synovial spaces may be present (King; own case) B. Synoviulomas (neoplastic) 1. Endothelial (synoviothelial) (case of Schwamm) 2. Mucous (case of Black) 11. Malignant neoplasms Synovialosarcomas 1. Histiocytic; polymorphous; plasmodial, xanthomatous, with giant cells (Case V) 2. Endothelial (synoviothelial) : “ epitheliosarcomas ” with cavities (a) containing niucin (Case I and 11) (6) without much (Case IIIj 3. Mucous (Case IV) 4. Indifferent: (a) Compact structure, but with clear spots after reticulum impregnation, corresponding to a beginning endothelial evolution (6) Compact structure, entirely indistinguishable from common fibrosarcoma 536 LOUIS BERGER

HISTOLOGICDIAGNOSIS AND ITS PRACTICALIMPORTANCE Practically all cases of synovialosarcomas which could be observed long enough proved to be fatal although the removal of the first tumor seemed most often complete. Recurrences are very frequent and may appear early. Death is generally due to pulmonary metastases, from one to five years after the first symptoms. The evolution of synovialosarcomas is the same in the joints, bursae, or tendons, but is perhaps, on an average, somewhat more rapid in the bursae. The prognosis depends therefore in a large measure on the correctness of the histologic diagnosis. The latter may be easy when the tumor is articular and typical. It may be much more difficult when the tumor is extra-articular or only in part characteristic. Our Cases I, I1 and V appeared at sites where synovial tumors are not commonly expected, originating in serous bursae. In some cases, as in Case 11, such specific features as pseudoglandular or “ epitheliosarconiatous ” aspects may be scarce and the greater part of the tumor tissue may suggest a rather common sarcoma. The diagnosis should therefore be based on very numerous sections or, better still, on several very large sections taken at different levels, so as to eliminate the possibility of omitting specific parts. Reticulum and mucin stains may be very helpful for the diagnosis. An established diagnosis of synovialosarcoma calls for a prompt and high amputation of the affected extremity. Sabrazks wrote: “ It is amputation which should impose itself from the first examination. The family, the physician, the surgeon, the pathologist hesitate proposing such a sacrifice in the presence of a tumor which seems to be so slightly malignant in the first stage of its development. Later, one is forced to this intervention; one undertakes it, alas! too late.” With regard to our own cases we can sub- scribe without restriction to this statement of Sabrazks.

SUMMARY A study of five personal cases and a review of the literature lead us to formulate our knowledge of synovialosarcomas as follows : Histologically only those tumors should be considered of synovial nature which recall one or more specific features of the normal synovial tissue. Syno- vialosarconias are fundamentally reticulo-histiocytosarcomas containing a rich network of reticular fibrils. The neoplastic tissue may be so indifferent as to reproduce exactly the aspect of immature lymphoid or medullary reticulosarcoma (some forms of Ewing’s sarcoma), but is generally formed by more or less anastomosing or syncytial branching or polymorphous cells. A first variety corresponds to the benign xanthomatous giant-cell tumors. As in these, the essential features are the syncytial aspect of the cells, the protoplasm of which is quite abundant; the tendency to form plasmodia1 elements, sometimes very irregular and monstrous, but sometimes identical with the so-called myeloplaxic giant cells; the presence of xanthomatous cells some of which may be multinucleated; the occasional phagocytic ( lipophagic and siderophagic) properties, and the appearance of peculiar cavities, which seem to correspond to King’s synovial spaces of the innocent variety, which are surrounded by cells strongly suggesting irregular synovial tissue. SYNOVIAL SARCOMAS IN SEROUS BURSAE AND TENDON SHEATHS 537

The second variety first gives the impression of “ epitheliosarcoma,” due to the presence of small pseudoglandular cavities of endothelial or synoviothelial nature. The lining cells are either cuboid or cylindrical and often desquamate, forming finally a peculiar magma. In some tumors these cavities contain a mucicarmin-staining substance, which may be mixed with the magma, but which is never intracellular. The tumors may show secondary changes or deviations from the fundamental cell type: one may find lipophagic or siderophagic cell nests, calcospherites, or even hemangiosarcoma- tous features. The third tumor type is formed by polymorphous, often large cells surrounded by reticular fibrils, devoid of any characteristic arrangement, but between which a deep-staining extracellular mucin provides nevertheless a ‘somewhat specific aspect; the mucin forms droplets, small threads, or large strands. The different structures occur in various proportions not only in different tumors but even in different areas of the same tumor. The polymorphism is therefore great and several sections should be examined if a complete picture is to be obtained. Regarding the normal histology of synovial tissue, our Case I is so typical and the intensity of cell differentiation is so regular and so pronounced, that it seems to be a simple enhancement of a normal process and therefore to allow, in some measure, a discrimination between the still divergent opinions regarding this tissue. Our observations seem, indeed, to favor strongly the reticulo-endothelio-histiocyticnature of synovial membranes, which was formu- lated by Franceschini and supported by Sabra& and his fellow-workers. Because these tumors always show, at least in part, a rich reticular frame- work, because they may, on the other hand, reproduce most typically some lymphoid or medullary reticulosarcomas, and, on the other hand, more or less xanthelasmatized histiocytomas, it seems logical to conclude that their mother tissue is akin to that which gives origin to these neoplastic forms elsewhere. But in spite of this kinship, the synovial tissue has some peculiar characters setting it apart as a variety sui gcneris of the reticulo-histiocytic system as a whole. These features are, above all, the frequency and the intensity of the endothelial aspect and the appearance of true mucin. We must concede the endothelial nature of synovial membranes in the presence of the sometimes extraordinary expansion of pseudoglandular or pseudoepithelial structures in many of its tumors, structures which are as typical as those of pleural or peri- toneal mesotheliomas and develop without any mechanical morphogenetic influence. The majority of writers, in view of the apparently discontinuous internal lining, reject an endothelial nature, claiming that silver methods for endothelium are negative or at least give other results than in true endothelium. The fact that in some folds and on some villi the lining cells are continuous and cubical they generally explain as the result of a “ negative pressure.” We are tempted to reverse this line of reasoning, claiming for the synovial intima a fundamental endothelial potentiality, which under normal conditions is most often hindered in its expression by the mechanical influence of pressure, which thus acts as a negative morphogenetic factor. This conception 538 LOUIS BERGER would satisfactorily explain the endothelial structures in synovial tumors, where mechanical influences are lacking. To those who prefer to reserve the generic term endothelial for vascular and sinus linings, we propose the designation synoviothelial which is employed by Hueck and is in entire accordance with Vaubel’s conception of synovialoblasts. The intensity of desquamation in some tumors with magma formation seems to favor the hypothesis of Sabrazb and older authors that degeneration products of cells participate in forming synovial fluid. In tumors the magma may be considered in some measure as a concentrated synovial fluid. As it may be devoid of mucin, desquamation seems to be independent of mucino- genesis. The much formation accentuates the specificity of the mother tissue. It may accompany endothelial differentiation, but may appear without it between reticulo-histiocytic cells which lack any definite arrangement. We never found mucin inside of protoplasm; some seemingly intracellular droplets appeared to be artefacts or could be explained by tangential cutting of cells. The mucin tends to condense around the cell bodies. Synovial niucinogenesis is therefore entirely different from mucous secretion in glandular cells and seems to be comparable rather to intercellular substances appearing in mesenchymatous tissues. Synovial mucin would then be analogous to cartilaginous and osseous intercellular substances, but with the difference that in synovial tissues the substance remains liquid. The study of our synovial tumors thus favors Vaubel’s conception. From a topographical point of view our cases show that synovialosarcomas may arise outside of and far from articulations, in the midst of soft tissues, where they originate in serous bursae, normal or irregular from a purely anatomical point of view. With regard to evolution, synovialosarcomas are highly malignant tumors, although they seem at first to grow only slowly. Synovialosarcomas of serous bursae apparently run a more rapid course than those of articulations. Recur- rences develop in almost all cases operated upon. Death is generally due to pulmonary, exceptionally to cerebral metastases. Definite cure is exceptional and may be obtained only by early amputation. Therapeutic measures should therefore be prompt and radical. The study of our personal cases and a critical bibliographic review lead us to essay a new classification of synovial tumors as a whole.

BIBLIOGRAPHY ADRIAN,C.: Beitr. z. klin. Chir. 38: 459, 1903. BASTOS,E. S.: Bol. d. SOC.med. e cir. Sao Paolo 15: 475, 1032. UECKER: Deutsche Ztschr. f. Chir. 191: 300, 1925. BLACK,W. C.: Am. J. Cancer 28: 481, 1036. BONSE,C., ASD COLLET,A. M.: Geneesk. tijdschr. f. Neder1.-Inditr 75: 1354, 1935. Ab- stract in Am. J. Cancer 28: 455, 1936. BRODERS:Ann. Surg. 70: 574, 1919. BRUNNER.R.: These de Zurich, 1936. CHEKOTAND TZANCK:Bull. et mkm. SOC. anat. de Paris 8;: 293. 1912. CRACIUN,E. C., AND URSU,A.: Bull. Assoc. franc. p. 1’Ctude du cancer 22: 71 1, 1033. DIEZ.J.: Prensa med. argent. 18: 487, 1931. ELY,I,. W.: Ann. Surg. 68: 426, 1918. SYNOVIAL SARCOMAS IN SEROUS BURSAE AND TENDON SHEATHS 539

FACCINI,U.: Arch. ital. di chir. 7: 481, 1923. FIEVEZ:Bull. et mkm. SOC.nat. de chir. 61: 1034, 1935. FRANCESCHINI,I).: Arch. ital. anat. e di embriol. 27: 76, 1929. GESCHICKTER,C. F., AND LEWIS,D.: Am. J. Cancer 22: 96, 1934. HANNEMULLER:Beitr. z. klin. Chir. 63: 307, 1909. HODCSON,F. G., AND BISHOP,E. L.: J. Bone & Joint Surg. 17: 184, 1935. HUCUENIN.R., AND OBERLING,C.: Bull. Assoc. franc. p. I’ktude du cancer 20: 144, 1931. HUECK,W.: Morphologische l’athologie, G. Thieme, Leipzig, 1937, p. 27. KEY,J. A.: in Cowdry’s Special Cytology, Hoeber, New York, Vol. 11, p. 1062, 1932. KING,E. S. J.: J. Path. & Bact. 41: 117, 1935; Brit. J. Surg. 18: 594, 1930-31. KLING,D. H.: Arch. Surg. 23: 543, 1931. KNOX,L. C.: Am. J. Cancer 28: 461, 1936. LECENE,P., AND MOULOSCUET,P.: Ann. d’anat. path. 1: 393, 1924. LEJARSAND RUBENS-DUVAL:Rev. de chir. 41: 751, 1910. L~vY,G.: Ann. d’anat. path. 2: 247, 1925. MARTINA,A.: Deutsche Ztschr. f. Chir. 83: 317, 1906. MORTON,J. J.: Surg., Gyner. & Ohst. 59: 441, 1934. OBERLING,C.: Bull. Assoc. franc. 1). l’ktude du cancer 17: 259, 1928. PRYM,P.: Virchows Arch. f. path. Anat. 2i9: il,1930. RAZEMON.P., ASD BIZARD,G.: Rev. de chir. 50: 229, 1931. ROULET,F.: Virchows Arch. f. path. Anat. 286: 702, 1932. v. RUEDICER-RYDYGIER:Deutsche Ztschr. f. Chir. 82: 211, 1906. SABRAZES,J., AND DE GRAILLY,R.: Compt. rend. SOC. de biol. 106: 1155, 1931. SABRAZES,DE GRAILLY,AND SALABARTAN:Gaz. hebd. sc. mCd. Bordeaux 53: 385 and 401, 1932. SABRAZESET AL.:Gaz. hebd. sc. mkd. Bordeaux 53: 449, 481, 769, 1932. 55: 754, 772, 786, 802, 1934. 56: 36. 53. 83, 100, 1935. SCHWAMM,M.: Zentralbl. f. Chir. 57: 2478, 1930. SENEAR,F. E., AND CARO,M. R.: Arch. Dermat. & Syph. 33: 209, 1936. SEZARY,A., AND LEVY-COBLENTZ,G.: Bull. SOC.franc. dermat. et syph. 40: 798 and 1269, 1933. SMITH,L. K.:Am. J. Path. 3: 355, 1927. STRAWS, A.: Suig., Gynec. & Obst. 35: 161, 1922. TAVERNIER:Lyon chir. 27: 522, 19.30. TOURNEUX,J. P.: Rev. de chir. 47: 817, 1913. Progrhs mkd. 35: 215. 1920. VAUBEL,E.: J. Exper. Med. 58: 63 and 85, 1933. Circhows Arch. f. path. Anat. 289: 670, 1933. WAGNER.L. C. : Ann. Surg. 92 : 421, 1930. WEGELIN,C.: Schweiz. med. Wchnschr. 9: 722. 1928. WHITE,J. R.: Surg. Cynec. & Obst. 38: 489, 1924. WORINCER:Bull. SOC.franc. dermat. et syph. 38: 1401, 1931. WORINCER,F., ASD KWIATKOWSKI,S. L.: Ann. de dermat. et syph. 3: 998, 1932. ZWAHLEN,P.: Bull. Assoc. franc. 11. I’ktude du cancer 24: 682, 1935. These de Zurich, 1935. NOTE:All papers listed were consulted in the original, except that of Bonne and Collet. Many other papers concerning xanthomatous giant-cell tumors were read, but not listed.