von Hippel Angiomatosis

A Light, Electron Microscopic, and Immunoperoxidase Characterization

LINDA MOTTOW-LIPPA, MD, MARK O. M. TSO, MD, GHOLAM A. PEYMAN, MD, GREGORIO CHEJFEC, MD

Abstract: A 25-year-old Iranian man had undergone eye wall resection of a large von Hippel angioma to alleviate an exaggerated macular response, af­fording study by light and electron microscopy and immunohistochemistry before the obfuscatory effects of long-standing exudative retinal detachment, gliosis, or iatrogenic ablation supervened. We used this vantage point to as­sess the interrelation between the component endothelial cells, pericytes, and stromal foam cells. On the basis of staining with glial fibrillary acidic protein, factor VIII, the C3 fraction of complement, fibrinogen, and lysozyme, it is unlikely that stromal foam cells derive from glial precursors, but may represent de­generating cells, perhaps arising from a common vasoformative stem cell under hypoxic stress. [Key words: angiomatosis retinae, factor VIII, foam cells, glial fibrillary acidic protein, immunoperoxidase, von Hippel angiomatosis.] Ophthalmology 90:848-855, 1983

von Hippel angiomas are such infrequently seen le­sions that few specimens become available for histologic study before massive exudative retinal detachment and ablative therapy supervene. 1

-3 The eye wall resection of

a large peripheral angioma in the treatment of macular exudation afforded a rare opportunity for electron mi­croscopic study and immunohistochemical character­ization.

CASE REPORT

A detailed history is concurrently being reported in a com­panion article on the surgical excision of large von Hippel tumors.4 Briefly, the patient was a 25-year-old Iranian man

From the Departments of Ophthalmology and Georgiana Theobald Lab­oratory of Ophthalmic Pathology, University of Illinois Eye and Ear Infir­mary, Chicago, Illinois; Departments of Ophthalmology and Pathology, Hines Veterans Administration Hospital, Hines, Illinois; and Loyola Uni­versity Stritch School of Medicine, Maywood, Illinois.

Supported in part by a Heed Foundation Fellowship (Dr. Mottow-Lippa) and training grant PHS EY 7038-05S1, core grant 1 P3EY01792, and PHS EY1903 from the National Institutes of Health, Bethesda, Maryland.

Reprint requests to Mark O. M. Tso, MD, University of Illinois Eye and Ear Infirmary, 1855 W Taylor St, Chicago, IL 60612.

who complained of seeing floaters, and was diagnosed a year previously as having a vascular tumor in his left eye. The paternal family history was notable for retinal disease, a brain tumor, and seizures. His vision was 6/6 (20/20) right eye and finger counting at 4 ft left eye; intraocular tensions were normal.

Ocular examination revealed a superotemporal retinal mass, 3 disc diameters in size in the left eye. The mass, served by dilated, tortuous afferent and efferent vessels, was locally sur­rounded by an exudative retinal detachment (Fig 1), and was accompanied by massive accumulations of subretinallipoidal material in the macula and in scattered extramacular sites. The right eye was uninvolved. Fluorescein angiograms re­vealed massive leakage from retinal vessels, both locally and diffusely, including the macula. An extensive medical work­up revealed no extraocular involvement. An eye wall resection was performed following perilesional photocoagulation, and the tissue was submitted in glutaraldehyde-formaldehyde fix­ative for microscopic studies.

MATERIALS AND METHODS

The vascular tumor was bisected longitudinally. The half containing the draining vein was submitted for par­affin embedding, sectioned for routine histopathologic study, and stained with hematoxylin-eosin, periodic

848 0161-6420/83/0700/0848/$1.20 © American Academy of Ophthalmology

MonOW-UPPA, et al • VON HIPPEL ANGIOMATOSIS

Fig L A tortuous dilated feeder artery (below) and distended draining vein approach the retinal angioma_ Patchy exudates, especially at the arterovenous crossing (above) are scattered throughout the retina_ Fig 2. In a low power microscopic view of the eye wall resection specimen, the draining vein (V) is seen amidst the anomalous capillary mass, which arises from normal retina (R)_ The overlying vitreous contains neovascular tufts (arrowhead) (hematoxylin-eosin, original magnifi­cation, X50)_ Fig 3_ Stromal foam cells cluster around vascular chan­nels within a fibrous interstitium punctuated by occasional plasma cells (arrowheads) (hematoxylin-eosin, original magnification X 10)_ Fig 4. Within the hamartomatous mass, capillary spaces of variable size are lined by flattened endothelial cells (arrows) and are adjoined by characteristic stromal "foam cells" (F) (I-micron plastic section, Mallory, original magnification X750)_ Fig 5_ Glial cells in the tumor

heavily stain with immunoperoxidase-labeled glial fibrillary acidic protein (GFAP)_ Adjacent foam cells show no such affinity (GFAP, original magnification X 100)_

Schiff reagent, and reticulin stains_ The 5-micron tissue sections selected for immunoperoxidase evaluation were incubated in a vacuum oven at 60 C for 48 hours to enhance tissue adherence to the slide, then deparaffin­ized_ An immunoperoxidase kit for glial fibrillary acidic protein (GFA) was used (DAKO Corporation, Kit# K507), and the procedure recommended by DAKO was

modified by reducing Tris buffer rinses to five minutes_ The substrate and peroxide were applied as directed, and the tissue was counterstained with Gill Baker hematox­ylin, blued in Scott's tap water, and mounted with DAKO glycergei mounting medium. Corroborative GF A staining was also employed using the triple layer technique.5 The procedure was repeated with factor

849

OPHTHALMOLOGY • JUL Y 1983 • VOLUME 90 • NUMBER 7

VIII, lysozyme, fibrinogen, C3 complement, and alpha­i-antitrypsin.

The tissue fragment containing the afferent feeder artery was processed for electron microscopy, fixed in Dalton's chrome osmium, embedded in Fluka epoxy resin, and stained with uranyl acetate and lead citrate.

OBSERV ATIONS

By light microscopy, the retinal architecture was markedly distorted by a mass of hamartomatous vas­cular channels surrounding a dilated draining vein, as-

Fig 6. The variable thickness of the vessel endothelial lining is featured in this low power electron micrograph. Note plump endothelial cells of cap­illary (one with indented nucleus, right), as opposed to the flattened ones (above and left). Organelles are fairly sparse. Some pinocytotic vesicles are seen, and villous processes jut into the lumen. In contrast, a macrophage (M), with its characteristic phagolysosomes is seen within the perivascular space, adjacent to a stromal foam cell whose nucleus is indented by abundant cytoplasmic vacuoles (X8500).

sociated with gliosis and eosinophilic exudate. The in­ternallimiting membrane was disrupted, and the over­lying condensed vitreous contained neovascular tufts (Fig 2). The capillaries composing the mass were ofvari­able caliber and were interspersed with stromal foamy cells,2-4 and rarely by scattered plasma or erythrocytes (Figs 3, 4). The subjacent choroid was congested, and there was sheet-like proliferation of the retinal pig­mented epithelium. Abundant reticulin fibers were re­vealed by Wright's reticulin stain.

After incubation with immunoperoxidase stains, GF A was detectable in glial cells of both the tumor and control brain tissue. The stromal foamy cells remained distinctly

)

Fig 7. Overview of a characteristic hamartomatous capillary. The endothelial lining is punctuated by numerous fenestrations highlighted in inset I. Underlying the endothelium is a focally redundant basement membrane seen at higher magnification (inset 2). Subjacent to these layers is a degenerating vacuolated pericyte (P), with indistinct organelles (below at right). Contrast the appearance of this cell with the stromal foam cells at left (Fig 7, X8250; inset 2 X29,500; inset 1 X36,700).

850

OPHTHALMOLOGY • JULY 1983 • VOLUME 90 • NUMBER 7

unstained by the GFA-immunoperoxidase complex (Fig 5). Markers for factor VIII, C3 , fibrinogen, lysozyme, and alpha-I-antitrypsin had no affinity for the stromal foam cells, in the face of markedly positive controls.

By transmission electron microscopy, the vascular channels composing the mass were lined by endothelial cells, which varied from flat to plump and cuboidal. Some contained bilobed nuclei and were laden with mitochondria and rough endoplasmic reticulum (Figs 6, 7). Some pinocytotic vesicles could also be identified. Villous projections extended toward the vessel lumen (Figs 6-8). Unlike normal retinal vessels, which are un­fenestrated, definite attenuations were present in the endothelial cells of these aberrant vessels (Fig 7). Focal areas of hydropic degeneration of the endothelium were also identified (Fig 8).

The endothelial basement membrane showed focal reduplications. Pericytes, with their oval nuclei and lightly stippled heterochromatin pattern, were sparse; when present, they showed degenerative necrotic changes, including watery vacuolated cytoplasm and sparse or­ganelles (Fig 8). Their surrounding basement mem­brane, like that of the endothelium, was focally multi­laminar.

The nuclei of the stromal foam cells were indented by abundant, large electron-lucent nonosmophilic vac­uoles undelineated by a unit membrane (Fig 9). Inter­posed between the vacuoles were scattered cytoplasmic filaments and few organelles (Fig 9). No basement mem­brane was intimately associated with these stromal cells, although they lay apposed to the basement membrane of adjacent endothelial cells and pericytes. No evidence for interconversion among endothelial cells, pericytes, and stromal cells was noted. Scattered ' macrophages, nestled within the perivascular space distant from ~he lumens, contained characteristic lysosomes and cellylar debris (Figs 3, 6). .

DISCUSSION

Routine paraffin sections corroborated the clinical diagnosis of angiomatosis retinae, documenting aberrant capillary channels interspersed with stromal foam cells. By electron microscopy, the capillary lining featured sometimes flattened but also cubodial endothelium, in­dicative of a relatively young metabolically active cell.

While retinal capillaries do not normally demonstrate fenestrations,6 the endothelial cells in this tumor were frequently punctuated by attenuations (Figs 6, 7). This feature corroborates the findings of Jakobiec2 but is at variance with those described by Nicholson et al3 in a similarly young angioma unscathed by gliosis or de­tachment. The electron microscopic documentation of fenestrations in our case is consistent with the clinical

and angiographic features of intraretinalleakage accom­panied by exudate and fluid dissection into the macula. Further, the endothelial hydropic degeneration seen also attests to some degree of vascular distress.

The abnormal fenestrations suggestive of leakage might also lend a clue to the etiology of the focally thick­ened basement membrane seen in this case. Chronic leakage might stimulate basement membrane produc­tion by irritating endothelial cells and pericytes, and passive imbibition of plasma lipids by stromal cells has been proposed by Jakobiec et al to be a response to the massive exudation imposed by incontinent blood ves­sels.2 Alternatively, thickened, reduplicated basement membrane has also been postulated to reflect repeated cycles of cell death and regeneration,7 and has been ob­served in such diverse entities as diabetes, hypertension, collagen vascular disease, and granular cell tumors. On the basis ofa number of human and animal studies, the pathologic changes in diabetes and hypertension at least have been associated in the exudation and ischemia8,9

and secondary neovascularization, which is mirrored in our lesion. Whatever the stimulus, thickening of the basement membrane would affect the flow dynamics of these vascular hamartomas. Capillary basement mem­brane thickness directly contributes to the rigidity of the vascular channel. 10 It is tempting to speculate that the reduplication process might also either directly compro­mise the lumen or limit diffusion of nutrients to adjacent cells, with attendant exudation and cellular degenera­tion.

The paucity of pericytes and the degenerative changes seen in those few surviving might perhaps also parallel the pericytic dropout in the diabetic fundus, in . which both exudation and ischemia maybe operant.

The genealogy of stromal foam cells has for many years been a controversial issue,2,11 Microglia, neuro­epithelium, macrophages, pericytes, endothelial cells, and pial cells have all been implicated as cells of orjgin. II It has been suggested that in the immature human cor­pus callosum 4nder hypoxic stress, glia may u~dergo fatty metamorphosis.12 Some investigators have used this argument,in conjunction with the presence of in­tracytoplasmic filaments, to support a glial origin for the foam cells.2 The conqensed chromatin, prop1inent fil­aments, and relative dearth of organelles have prompted the suggestion that they may belong to the spectrum of fibrous astrocytes, II however, no characteristic intracel­lular filamentous whorls could be found to corroborate that theory in our case.

The lack of foam cell affinity for glial fibrillary acidic protein in the face of considerable staining of adjacent glia casts some doubt upon the theory of a glial origin. These findings are in concordance with those recently

- reported by Jurco and co-workers. 13 Although few fibrils were found between the prominent vacuoles in the foam

Fig 8. One hamartomatous capillary wall is hallmarked by hydropic degeneration of its endothelial cells (EC) adjacent to a degenerating pericyte (P). Note that this pericyte lies adjacent to both the plump viable endothelial cell (right) and multilaminar aberrant basement membrane (arrow) (X 15,6(0). Inset shows another degenerating pericyte with its striped electron-lucent nucleus (X 12,6(0). .

852

OPHTHALMOLOGY • JULY 1983 • VOLUME 90 • NUMBER 7

cells on routine electron microscopy, GFA has been claimed to localize astrocytes even in the absence of detectable filaments by electron microscopy.14 Electron microscopic immunoperoxidase technique might be helpful in further confirmation of this finding.

Like Kawamura's, our foam cells lacked a basement membrane, weakening the argument for a pericytic or­igin. These foam cells, like those described by Spence, failed to resemble fibroblasts or macrophages by electron microscopyl5; neither lysosomes nor degenerative inclu­sions could be demonstrated. The failure to localize ly­sozyme by immunohistochemistry is also adjunctive evidence against amacrophagic origin, although lyso­zyme stores could be exhausted by prolonged stimula­tion. One could argue that the contrasting appearance of foam cells and macrophages (Fig 6) might depend on cell stage, degree of stimulation, and commensurate state of lysosomal depletion. However, the lack of lo­calization of alpha-I-antitrypsin or fibrinogen would argue against intrinsic phagocytic properties ofthe foam cells. Rather, prolonged exudation might stimulate the migration of wandering phagocytic cells into the ha­martoma.

Recently, Jurco and co-workers described a striking but inconstant affinity of stromal foam cells for factor

854

Fig 9. Multilaminar base­ment membrane separates stromal foam cells (s) from a degenerating pericyte (p). The foam cell nucleus is in­dented by electron-lucent material devoid of delimit­ing membrane structures. Fibrillary component (F) of foam cell cytoplasm is high­lighted in inset A ( X 18,500, inset X 18,750).

VIII.13 While our studies indicated heavy localization in endothelial cells, no factor VIII was traceable to the interstitial foam cells, making a metamorphosis from mature endothelial cells unlikely in our case. However, the absence of endothelial cell markers need not rule out the concept of incompletely differentiated stem cells degenerating and then accumulating cholesterol and serum products from adjacent differentiated but incon­tinent capillary endothelial cells. The lack of evidence for interconvertability between vasoformative cells in our specimen might support a common mesenchymal origin for both hamartomatous capillaries and stromal cells, as suggested by SpenceY The stromal cell is pos­tulated to somehow restrict its own ability to differen­tiate fully toward mature endothelial status. 15 Thus, a pre-existing angioblastic anlage could give rise to a vas­cular hamartoma in much the same way as pre-existing capillaries bud off in granulation tissue. II

While the origin of the foam cells and the von Hippel lesion itself remains speculative, the immunohistochem­ical evidence from the case we present seems to favor the differentiation of a pi uri potential vasoformative stem cell toward both an "immature" stromal cell and its more mature but functionally limited endothelial cell counterpart. Electron microscopic immunoperoxidase

MonOW-LlPPA, et al • VON HIPPEL ANGIOMATOSIS

techniques may ultimately elucidate the origin of these stromal cells.

ACKNOWLEDGMENT

Special thanks are extended to Mr. Robert Milewski and Mrs. Irena Suvaizdis and Mariann Bernacki for their technical assistance and Mrs. Diann W. Robinson for typing this manu­script.

REFERENCES

1. Wing GL, Weiter JJ, Kelly PJ. et al. von Hippel-Lindau disease; an­giomatosis of the retina and central nervous system. Ophthalmology 1981; 88:1311-4.

2. Jakobiec FA, Font RL, Johnson FB. Angiomatosis retinae; an ultra­structural study and lipid analysis. Cancer 1976; 38:2042-46.

3. Nicholson DH, Green WR, Kenyon KR. Light and electron micro­scopic study of early lesions in angiomatosis retinae. Am J Ophthal­mol 1976; 82:193-204.

4. Peyman GA, Rednam KRV, Mottow-Lippa L, Flood T. Treatment of a large von Hippel tumor by eye wall resection. Ophthalmology 1983: 90:840-7.

5. Heyderman E. Immunoperoxidase technique in histopathology; ap· plications, methods, and controls. J Clin Pathol 1979; 32:971-8.

6. Hogan MJ, Alvarado JA, Wedell JE. Histology of the Human Eye: An Atlas and Textbook. Philadelphia: WB Saunders, 1971; 514.

7. Vracko R. Basal lamina layering in diabetes mellitus; evidence for accelerated rate of cell death and cell regeneration. Diabetes 1974; 23:94-104.

8. Tso MOM, Jampol LM. Pathophysiology of hypertensive retinopathy. Ophthalmology 1982; 89:1132-45.

9. Kishi S, Tso MOM, Hayreh SS. Malignant hypertensive choroidopathy, retinopathy and optic neuropathy in monkeys. ARVO Abstracts. In· vest Ophthalmol Vis Sci 1982; 22(Suppl):195.

10. Murphy ME, Johnson PC. Possible contribution of basement memo brane to the structural rigidity of blood capillaries. Microvasc Res 1975; 9:242-5.

11. Kawamura J, Garcia JH, Kamijyo Y. Cerebellar hemangioblastoma; histogenesis of stroma cells. Cancer 1973; 31 :1528-40.

12. Leech RW, Alvord EC Jr. Glial fatty metamorphosis; an abnormal response of premyelin glia frequently accompanying periventricular leukomalacia. Am J Pathol 1974; 74:603-12.

13. Jurco S III, Nadji M, Harvey DG, et al. Hemangioblastomas; histo­geneSis of the stromal cell studied by immunocytochemistry. Hum Pathol 1982; 13:13-8.

14. Braak E, Drenckhahn D, Un sicker K, et al. Distribution of myosin and the glial fibrillary acidic protein (GFA protein) in rat spinal cord and in the human frontal cortex as revealed by immunofluorescence microscopy. Cell Tissue Res 1978; 191 :493-9.

15. Spence AM, Rubinstein LJ. Cerebellar capillary hemangioblastoma; its histogenesis studied by organ culture and electron microscopy. Cancer 1975; 35:326-41 .

855