Selective Immunohistochemical Staining in the Paraneoplasdc Retinopathy Syndrome

Selective Immunohistochemical Staining in the Paraneoplastic Retinopathy Syndrome Joseph F. Rizzo III, MD/ John W. Gittinger, Jr., MD2 Background: The mechanism leading to visual loss in paraneoplastic retinopathy is not known. An autoimmune process has been imputed based on immunologic investigations of several patients and by analogy to certain other paraneoplastic syndromes. Methods: Two patients with documented small cell carcinoma of the lung who had clinical evidence of paraneoplastic retinopathy are described. Histopathologic examination of the retina from one patient and immunohistochemical staining of human retina with serum from control subjects and both patients were performed. Results: Electroretinograms demonstrated dysfunction of photoreceptors in both patients, with predominant loss of rod function in one patient. Post mortem examination showed patchy loss of photo receptors of the extramacular retina and relative sparing of cones, findings consistent with the clinical and electrophysiologic test results. Serum from both patients stained the retina in an identical manner, with restriction of the stain to the outer retina. Stain was present over the outer plexiform layer, the outer nuclear layer, and the inner and outer segments of most photoreceptors. A sharp demarcation was present between those areas that did and did not stain. All rod inner and outer segments appeared to stain, and many cone inner segments were not stained. Immunologic tests obtained elsewhere did not show serum antibody to the 23 kD protein. Conclusion: These findings support the concept of an autoimmune pathogenesis by showing selectivity of the immune response and correlation between the apparent target of the immune response and the clinical and pathologic findings. The mechanism by which cell loss occurs in the retina is not answered by this study. The absence of antibody to the 23 kD protein does not exclude the diagnosis of paraneoplastic retinopathy. Ophthalmology 1992;99: 1286-1295

Visual loss occurring as a consequence of paraneoplastic retinopathy was first described in 3 patients in 1976,1 and another 15 cases have been added to the literature subsequently.2-15 An autoimmune process has been imputed, based on immunologic investigations of several patients2-4,7-9,14 and by analogy to certain other paraneoOriginally received: September 25, 1991. Revision accepted: February 17, 1992. I Department of Ophthalmology, Harvard Medical School, and the Neuro-ophthalmology service, Massachusetts Eye and Ear Infirmary, Boston. 2 Division of Ophthalmology, University of Massachusetts Medical School, Worcester. Reprint requests to Joseph F. Rizzo III, MD, Massachusetts Eye and Ear Infirmary, 243 Charles St, Boston, MA 02114.

1286

plastic syndromes. 16- 18 The favorable response to corticosteroid therapy noted in some patients also is compatible with an autoimmune pathogenesis. 4.6,15 The mechanism of retinal dysfunction, however, has yet to be delineated. Clinical presentations have often been interpreted as reflecting selective damage to the retina, hence the designation paraneopiastic photoreceptor retinopathy. Indeed pathologic examinations of retinas from affected patients have shown a loss of photoreceptors in all but one case. I ,4,5,8 Immunologic data, however, have either not demonstrated selective labeling of the retina, or the site demonstrated by laboratory methods has not been consistent with (or was not described together with) the clinical and electrophysiologic manifestations. We report nearly identical clinical findings in two patients with the visual paraneoplastic syndrome. These pa-

Rizzo and Gittinger . Paraneoplastic Retinopathy Syndrome tients had corroborating clinical and laboratory findings that indicated dysfunction of the outer retina.

Case Reports Case 1. A 67-year-old man noticed a "balo" in the peripheral

vision of his right eye while watching television on July 15, 1987. His medical history was significant for a 20-year history of systemic hypertension and a IO-year history of diet-controlled diabetes mellitus. He smoked approximately 11/2 packs of cigarettes per day for 40 years and was taking hydrochlorothiazide, alphamethyldopa, and I aspirin tablet per day. He did not experience pain, flashes oflight, dizziness, or other neurologic symptoms in association with the change in vision. The left eye was not affected. Results of an ophthalmologic examination were normal. The symptom persisted, and 1 week later he returned to the same ophthalmologist and glasses were prescribed. He was referred to a neurologist who found no abnormalities. Acomplete blood count and routine blood chemistries were normal, except for a nonfasting serum glucose of 149. An erythrocyte sedimentation rate was 32. Tuberculin (i.e., PPD) skin test was not reactive. Noninvasive carotid studies were normal. Computed tomography of the brain, performed with and without contrast, showed calcified carotid arteries but was otherwise normal. Six weeks after the initial symptom, he noted a cbange in vision of the left eye on awakening. A second ophthalmologist found an afferent pupillary defect in the right eye and suggested the diagnosis of optic neuritis. A repeat computed tomography scan of the brain did not show a cerebral lesion. The patient began to feel that his vision was "darkening" in each eye, as if he had been "looking through sunglasses." Three days before presenting at the Massachusetts Eye and Ear Infirmary he noted a "worsening" of vision in both eyes. Neuro-ophthalmologic examination on September 4 showed Snellen acuities of20/80 in the right eye, 20/200 in the left, and an inability to correctly identify any Ishihara color plates with either eye, including the control plate. There was a subtle afferent pupillary defect in the right eye. Goldmann visual fields showed paracentral and midperipheral scotomas and constriction of the peripheral isopters in both eyes (Fig I). Dilated funduscopy was normal in both eyes. Prednisone therapy (80 mg/day) was initiated 5 days later and vision improved to 20/40 in the right eye, 20/80 in the left within 8 days of treatment.

Blood chemistries, including liver function tests and serum protein electrophoresis, and urinalysis were normal. Serum VDRL and FTA were negative. Vitamin B\2 level was 322 pg/ ml (normal, 171 to 953). Chest x-ray showed markedly enlarged hilar lymph nodes on the right side. A full-field electroretinogram (ERG) performed on September 15, 1987, showed reduced amplitudes (82 ,." V in the right eye, 141 ,." V in the left) to white light in both eyes, nondetectable recordings to blue light in both eyes, and 5.3 ,."V amplitudes with an implicit time of 40.6 msec in response to 30 Hz white light stimulus in the left eye. The cerebrospinal fluid was acellular, under normal pressure, with a total protein of 36 mg/dl, a glucose of94 mg/dl, and a negative VDRL. Cerebrospinal fluid cytology was negative. Bronchial washings showed numerous degenerating hyperchromatic cells similar to those observed in small cell lung carcinoma. Biopsy of the lung showed "oat-cell" carcinoma. Further work-up showed metastases to the liver and bone marrow. Chemotherapy (Cytoxan, Adriamycin, VP-16) was initiated on September 30, 1987. Nine days later, visual acuity was 20/ 25 in the right eye and 20/50- in the left, visual fields were similar to those of the initial examination, and prednisone was tapered to 40 mg/day. Adiagnosis of para neoplastic retinopathy was considered likely and blood was drawn for immunohistochemical testing. Blood was drawn on February 22, 1988 (the patient still was receiving chemotherapy, but had been offprednisone since December 7) and sent to the University of California, Davis for other immunologic analyses that included enzymelinked immunosorbent assay (ELISA) and Western blot tests. The ELISA test of reaction between the patient's serum and retina showed a higher than normal titer with a serum dilution of I: I00, and normal results with less concentrated sera (dilutions of I :200 and higher). Western blot analysis of the patient's serum and retina "may" have shown a "faint reaction" with a 23 kD protein. These latter two analyses were described as not revealing "anything worthy of comment." Visual acuity of 20/25 in the right eye and 20/25 in the left was recorded 31/2 months after prednisone was discontinued; visual fields at that time showed slightly more constriction than the prior examination. The patient remained on chemotherapy without change in vision until he died several months later. An autopsy was not performed. Case 2. A 59-year-old man noticed a "shimmering curtain" over his vision in mid September 1987. He had a history of emphysema and tachyarrhythmia, and was taking digoxin, triamcinolone inhaler, and alprazolam (Xanax).

'I ~F~~' ",'''',{::'J

B

c9=_ I

.,'

"

Figure 1. Case 1. Goldmann visual field test results. Paracentral and midperipheral scotomas, and constriction of the peripheral isopters were present in both eyes. (A, right eye; B, left eye.)

1287

Ophthalmology

Volume 99, Number 8, August 1992

He was evaluated by an ophthalmologist within days of onset of his visual symptoms and Snellen acuities were 20/15 in the right eye and 20/25 in the left. Computed tomography of the brain was normal, as was a fluorescein angiogram. Over the next few days, his vision began to deteriorate, especially in the left eye. A presumptive diagnosis of optic neuritis was made and 80 mg/day of prednisone was prescribed. The patient was examined at the University of Massachusetts Medical Center on October 9, 2 days after prednisone was initiated. Snellen acuities were 20/25 in the right eye and 20/200 in the left. He described a "doughnut of blackness" in the left eye. Goldmann visual fields demonstrated an arcuate defect in the right eye and a ring scotoma in the left (Fig 2). He missed 5 AO pseudoisochromatic plates in the right eye and was unable to see the control plate in the left. A rare cell was noted in the left anterior chamber, and intraocular pressures were 16 mmHg in both eyes. Dilated ophthalmologic examination was normal. A full-field, flash ERG showed no response. A chest x-ray showed hilar enlargement on the right side, and fiberoptic bronchoscopy showed complete occlusion of the right lower bronchus. Biopsy specimens obtained during bronchoscopy showed small cell undifferentiated carcinoma of the lung. Visual acuity on October 30, 23 days after prednisone therapy was initiated, was 20/20 in both eyes. Radiation therapy was given to the lungs and brain, and three cycles of chemotherapy (Adriamycin, Cytoxan, Cisplatin, Etoposide, VP-16) were initiated. A diagnosis of paraneoplastic retinopathy was considered likely and blood was drawn for immunologic studies while the patient was receiving chemotherapy and prednisone, 10 mg/ day. Tests performed at the University of California, Davis included an ELISA test between the patient's serum and retina that showed titers within normal range with serum dilutions of I :400 or higher, and a Western blot analysis that showed reaction with a 48 kD protein. The serum did "not bind" the 23 kD component. Results of an ELISA test against optic nerve were normal at all titers (lowest dilution, I: 100). The tumor of the lung initially responded to therapy, and in June 1988, a repeat bronchoscopy showed no abnormalities. In July, the patient presented with chest pain and shortness of breath from a malignant pleural effusion. He reported darkening and shimmering of vision, although visual acuity remained normal. He did not respond to further chemotherapy, and on August I lost central vision in his left eye. He was not receiving prednisone when his vision failed again.

A

The next day he died. Autopsy showed small cell undifferentiated carcinoma with extensive interstitial and lymphangitic spread.

Methods Both eyes from patient 2 were enucleated I hour after death. The eye that we studied was kept on ice for 6 hours and then put into 4% paraformaldehyde for 16 hours. Pieces of peripheral, midperipheral, and central retina were placed in JB4 embedding medium (Polysciences), cut into 3-p,m sections, and stained with 0.25% cresyl violet.

Immunohistochemistry Control retinal tissue was obtained through the New England Eye Bank from a 50-year-old man who had no history of eye disease. The donor eye was placed in 4% paraformaldehyde for 4 hours. The retina was isolated and placed in 25% sucrose solution overnight. The tissue was then flattened on a glass slide and covered with OCT embedding solution (Miles). The slide was placed on dry ice until the embedding medium solidified. The retina was separated from the slide, placed into a block containing OCT, and frozen. The retina was cut into 5-p,m sections. In preparation for staining, the slides were placed in 0.25 M TRIS buffer for 10 minutes. Subsequent reactions were performed in a closed, moist box, and care was taken not to allow the slides to dry. Endogenous peroxidase activity was reduced by pretreatment with a 5: I mixture of 100% methanol and 3% hydrogen peroxide for 30 minutes. A blocking solution of 3% goat serum with 0.5% Triton in 0.25 M TRIS buffer was placed on the slides for I hour. Serum obtained from both patients while still on prednisone and chemotherapy was tested by overnight incubation on the retinal sections in the following concentrations: 1:100, 1:250, 1:500, 1:1000, and 1:2000. After incubation, the slides were washed in a large volume of

B

Figure 2. Case 2. Goldmann visual field test results. An arcuate scotoma was present in the right eye (B), and a ring scotoma was present in the left eye (A). Constriction of the peripheral isopters was present in both eyes.

1288

Rizzo and Gittinger . Paraneoplastic Retinopathy Syndrome

Figure 3. Case 2. Three micron sections of the post mortem retina stained with cresyl violet. A, taken from the midperipheral retina and shows a transition from normal (right side) to grossly abnormal (left side). Within the abnormal area there is a near absence of outer nuclear layer cells and the finger-like projections of the inner and outer segments. The outer plexiform layer is reduced in thickness. The inner retina appears similar across this section. In particular, the density of cells within the inner nuclear and ganglion cell layers appears normal for this region of retina. Horizontal bar: 60 J.Lm (original magnification, X 160). B, this section was taken from the peripheral retina and shows degenerating profiles of outer nuclear layer cells (arrow), and an almost complete loss of inner and outer segments. Only a sparse outer nuclear cell layer that is a single cell in thickness remains. Compare with C, in which the outer nuclear layer is approximately five to six cells in thickness. Horizontal bar, 30 J.Lm (original magnification, X250). C, section was taken just beyond the peripheral edge of the macula and appears nearly normal. Horizontal bar: 30 J.Lm (original magnification, X250). ONL = outer nuclear layer; INL = inner nuclear layer; GCL = ganglion cell layer.

0.25% TRIS buffer for 30 minutes. One of the three secondary antibodies (peroxidase-conjugated, goat anti-human IgG [light and heavy chain specific], IgM [light and heavy chain specific], or IgG-IgM-IgA [Cappel]) were used at 1:200 dilution and applied for 1 hour. The slides were then reacted with diaminobenzidine mixture (5 mg/IO ml 0.25 M TRIS, pH 7.4) combined with 3.3.\ of 30% hydrogen peroxide for approximately 1 minute. The slides were washed, placed through ascending alcohols, and finally treated with two 15-minute immersions in xylene. Slides were coverslipped in Krystalon (Diagnostic Systems). Several immunohistochemical controls were performed. In the first method, reactions were performed exactly as described above except that 0.25% TRIS buffer was used in lieu of the patient's serum. In the second

method, serum from three controls (dilution 1:50) was substituted for the patient's serum. Lastly, serum from three patients with autosomal recessive retinitis pigmentosa and from four patients with biopsy-proven oat-cell carcinoma of the lung without visual problems were tested, using serum dilutions of 1: 100 and 1:500. Sera from both patients also were tested against other tissues with ciliated epithelia, including bronchus, testes, and the kidney.

Results Histology Sections from the midperipheral retina stained with cresyl violet showed an abrupt transition between normal and

1289

Ophthalmology

Volume 99, Number 8, August 1992

abnormal appearing retina (Fig 3A). The most conspicuous abnormalities included a loss of inner and outer segments of the photoreceptors and a near total loss of cells of the outer nuclear layer. The outer plexiform layer was reduced in thickness. The inner retina appeared normal. Sections from the peripheral retina consistently showed complete absence of inner and outer segments, nearly complete loss of cells of the outer nuclear layer with degenerating profiles of some remaining cells, and a shrunken outer plexiform layer (Fig 3B). There was no evidence of inflammation. Migration of retinal pigment epithelial cells into the outer retina was observed occasionally. Sections from the peripheral edge of the macula showed nearly normal appearing retina in most sections (Fig 3C); there was a grossly normal ratio of cellular densities among the three nuclear layers.

Immunohistochemistry All control experiments showed an absence of staining, except for small areas within the inner retina that probably represented residual peroxidase activity within blood vessels (Fig 4B). This staining is expected because the endogenous enzyme activity normally found within blood vessels is not completely eliminated by the method described above. The presence of peroxidase reaction product within blood vessels was beneficial because it demonstrated that the reaction was performed sufficiently to result in some staining. It also is possible, but less likely,

that the residual peroxidase activity was caused by a reaction of the secondary antibody with serum within the blood vessels of the donor retina. Serum from both patients stained the retina in an identical manner, with restriction of the stain to the outer retina (Fig 4A). Stain was present over the outer plexiform layer, the outer nuclear layer, and the inner and outer segments of most photoreceptors. A sharp demarcation was present between those areas that did and did not stain. Occasionally, cellular appearing structures within the inner retina appeared to stain, although no particular cellular morphology or pattern of staining could be discerned. Usually small structures within the inner retina that appeared to stain could be recognized by through-focusing to be tangential sections of blood vessels. All rod inner and outer segments appeared to stain. Conversely, many cone inner segments were not stained (Fig 5). The degree to which the cone outer segments were stained could not always be determined confidently because of overlapping cellular profiles. Staining of the outer retina was visible, although faintly, to 1:2000 dilution. No staining was observed when antihuman IgM secondary antibody was used, but identical patterns were observed when either of the other two secondary antibodies were used. This was interpreted as indicating that an antibody of the IgG class was most likely responsible for the staining pattern that had been observed. Immunohistochemical tests using the patient's serum against cilia-containing tissue from other organs were negative.

Figure 4. Sections of the retina 5 Ilm in thickness were obtained from a healthy donor. Nomarski optics were used to visualize edges of cell bodies and other structures such as the inner and outer segments. These structures can be imaged with this technique without the need for standard histochemical stains. The stain is the insoluble reaction product of the immunohistochemical procedure (see Methods section). A, is the immunohistochemical test using the serum of case 2. Serum was diluted 1:100. Dark staining of the outer retina is visible, and involves the outer plexiform layer, outer nuclear layer, and inner and outer segments. There is a sharp demarcation at the junction of the inner nuclear layer and outer plexiform layer. Small areas of staining within the inner retina almost always occurred within blood vessels. Rarely a cellular-appearing structure within the inner retina appeared to stain. Similar staining could be observed at dilutions up to 1:2000. An identical pattern of staining occurred with serum from case 1. Horizontal bars: 30 Ilm (original magnification, X250). B, is a control for the immunohistochemistry in which normal serum (dilution 1:50) was substituted for the patient's serum. A concentration of the control serum greater than that used for the patients (A) is desirable because this will increase the possibility of discovering antiretinal antibodies in the normal controls. No staining of the retina is visible except in blood vessels, which is expected because of the endogenous peroxidase that is normally present within blood vessels.

1290

Rizzo and Gittinger . Paraneoplastic Retinopathy Syndrome

Figure 5. Case 2. Magnified views of retinal sections similar to those shown in Figure 4A. The relatively thin rod inner segments are uniformly stained in the three sections. Many of the wider cone inner segments do not appear to stain. In the picture to the right, two cone inner segments are present; the one to the left is stained. A similar pattern was observed with the serum from case 1. Horizontal bar: 10 /Lm (original magnification, X600).

Discussion In 1976, Sawyer and co-workers I described three cases in which photoreceptor degeneration appeared to be the remote effect of cancer. The first of these cases serves as a paradigm for diagnosis and evaluation of this syndrome: a 65-year-old patient experienced progressive visual loss with ring scotomas and had a markedly abnormal ERG. Undifferentiated small cell carcinoma of the lung was eventually discovered. Post mortem examination of the eyes showed disintegration of the inner and outer segments of the rods and cones and widespread degeneration of the outer nuclear layer. The inner retinal layers were preserved, and there was no other abnormality of the visual pathways. Two additional cases reported in this seminal paper demonstrated similar findings at autopsy but neither patient had an ERG performed. Hence, a paraneoplastic entity causing selective destruction of the outer retina was documented but the pathogenesis was not evident. The next description of this syndrome was by Kornguth and co-workers2 in 1982, working at the University of Wisconsin. The primary subject in their report had ring scotomas and an abnormal ERG; an autopsy was not obtained. The researchers studied the serum of this and another patient and found antiganglion cell antibodies in their primary patient but nonspecific staining of retinal nuclei in the other. Additional immunologic studies of the sera from these two patients, reported in 1985, 7 showed serum immunoglobulins that reacted with proteins of the neurofilament triplet and with a 20 to 24 kD protein. Kornguth et al cautiously suggested that the antiretinal immunoglobulins were responsible for the loss of vision. This hypothesis created an apparent inconsistency. How could visual loss associated with photoreceptor degeneration, as documented by an ERG, be the result of immunoglobulins directed against ganglion cells? In 1987 Grunwald and co-workers8 described the second patient mentioned in the article of Kornguth et al 2 in more detail. This patient had undergone retinal detachment surgery and developed decreased central acuity, arcuate visual field defects, and optic atrophy. He was diagnosed and treated for oat-cell carcinoma of the lung. An ERG was not performed. Immunohistochemical

staining of the retina was reported to be somewhat specific and directed against large ganglion cells and large cells in the inner nuclear layer; serum from the other patient showed a similar pattern. Histopathologic examination of the retina showed diffuse loss of ganglion cells; photoreceptors were preserved. These presentations resemble those reported by Sawyer et aI, but they are dissimilar when compared at a cellular level. Between the first two articles published from the University of Wisconsin, Keltner et at,4 working at the U niversity of California, Davis, and Buchanan et al 5 had described histopathologic findings in two patients with outer retinal degeneration and normal optic nerves. Both patients had had extinguished ERGs. The case reported by Keltner et al showed "diffuse hyperfluorescence more intense in outer retinal layers" when the serum was incubated with normal human retina. They concluded that the loss of photoreceptors, with the sparing of other cells, resulted from the serum antibodies that had produced the "diffuse" staining pattern. They were the first to propose that the visual paraneoplastic syndrome might have an autoimmune basis. An autoimmune basis also was supported by the steroid responsiveness that they,4 and later others/,15 reported in patients with this syndrome. The University of California, Davis group then reported four patients whose sera were studied by ELISA and Western blot testing. 9 Immunoglobulins from their sera bound a retinal antigen with a molecular weight of 23 kD, which they named cancer-associated retinopathy (CAR) antigen. Antibodies against CAR were found only in patients with cancer who experienced visual loss. The investigators suspected that the antibodies directed against the CAR antigen contributed to loss of vision, and this notion strengthened the autoimmune hypothesis that they previously had suggested. In the most recent case reported by the University of California, Davis group, there were extensive immunologic studies, but neither ERG nor autopsy, leaving the clinical diagnosis uncertain. 14 Their patient had antibodies directed against small cell carcinoma and ocular antigens, including the CAR antigen. Extensive demyelination in the optic nerve of a guinea pig was induced by passive transfer of the patient's serum into the nerve, further sup-

1291

Ophthalmology

Volume 99, Number 8, August 1992

porting the notion of an autoimmune pathogenesis. Immunohistochemical testing of this patient's serum showed "heavy labeling of the three retinal nuclear layers, the nuclei of the retinal pigment epithelium, and the large choroidal blood vessels." This lack of specificity differs from the histopathologic results of most of the autopsy cases, which had shown selective degeneration of the outer retina (see belOW).1,4,5 Two other clinical reports have included immunologic studies. The patient of Crofts et al l3 had antibodies to an approximately 50 kD antigen. Jacobson et al 15 reported 2 patients whose immunoblot tests were interpreted as showing reactions to retinal antigens of 48 and 23 kD in the first patient and to a 23 kD antigen in the second patient. Although there have been other clinical cases,6,lO,11 these have neither come to autopsy nor had immunologic studies. One striking aspect of the literature reviewed herein is the lack of correlation between level of retinal dysfunction suspected on clinical grounds and the distribution of immUllologic and pathologic abnormalities. In fact, at times the results have been contradictory. For example, a patient with neurofilament antibodies directed against ganglion cells, especially larger ganglion cells, had an abnormal ERG suggestive of photoreceptor dysfunction and loss of central acuity, neither of which would be expected if damage had been limited to larger ganglion cells. 2.3,7,8 We describe two patients with documented small cell carcinoma of the lung who had the visual paraneoplastic syndrome, evinced clinically by dense bilateral scotomas and abnormal ERGs. Both patients were studied immunologically by the University of California, Davis group, The first patient had no detectable abnormal reactions on either ELISA or Western blot tests, which were performed when he was not receiving prednisone but still receiving chemotherapy. The second patient had normal ELISA test results against optic nerve and retina, at least with dilutions of I :400 or higher in the latter case (i.e., only the most concentrated sera produced results outside of the normal range). Western blot testing in the second patient when he was still receiving prednisone and chemotherapy demonstrated a reaction with a 48 kD antigen (possibly the S antigen 19- 2l ) and no reaction with a 23 kD antigen (CAR antigen), Our immunohistochemistry demonstrated selective labeling of the outer retina by the serum from both patients. The staining respected the anatomic boundary formed by the junction of the outer plexiform and inner nuclear layers (Fig 4). Not all structures within the outer retina were labeled, however. Although there was relatively uniform staining of rods, there was frequent absence of staining from cone inner segments (Fig 5). This latter finding correlated well with the predominant dysfunction of the rod system revealed by electrophysiologic testing in the first case. Staining was present even at high dilutions (up to I :2000), was organ-specific to the extent that the sera did not stain ciliated tissues that were examined, and was probably related to the presence of an IgG antibody with affinity for the outer retina. It is noteworthy that these positive immunohistochemical results were obtained

1292

even though both patients were receiving prednisone and chemotherapy. Results of post mortem examination in case 2 showed a patchy loss of photo receptors of the extramacular retina and relative sparing of cones (Fig 3), a finding that was consistent with the clinical and electrophysiologic test results. Only six other autopsy cases have been reported in this syndrome (Table I), and five have had similar findings lA .5; "depopulation of ganglion cells" has been described in one case. 8 Similar to our findings, one autopsy study showed relatively less damage to cones. 5 Given the immunohistochemical findings, what was the mechanism for visual loss in our cases? The presence of serum antibodies bound to the outer retina supports the concept of an autoimmune process. It could be that the immune system ran amuck, or that there was a chance similarity between retinal antigens and those of the tumor to which the immune system responded. The mere presence of neuronal antibodies or other autoantibodies in the serum does not, however, necessarily result in disease. Indeed, healthy individuals may have such antibodies without apparent ill effect.22 ,23 Presumably, antibodies must have access to neurons to which they have an affinity in order for dysfunction to occur. The blood-retinal barrier is the most obvious impediment separating the offending antibody from the neuronal elements. This barrier must have been transgressed, or bypassed, for our patients to have lost vision, and our methods cannot address the question of how that might have occurred. Curiously, uveitis was present in case 2 and has been described in three other patients with this syndrome. 4,6. l3 Assuming the intraocular cells did not represent metastases, it is possible that concurrent uveitis and the increased vascular permeability that might have resulted permitted access of serum proteins to the retina through the blood-retinal barrier. Despite these considerations, the presence of the antiretinal antibodies might have been epiphenomenalthe antibodies might have developed in response to retinal damage initiated by some other mechanism. In support of this possibility is the recent study of Thirkill et al,24 who demonstrated antiretinal antibodies in patients with retinitis pigmentosa. It is noteworthy that our control patients with retinitis pigmentosa did not show the immunohistochemical findings of our patients with paraneoplastic visual loss. The mechanism of cellular destruction cannot be explained by our own or previous investigators' results. Immunologic attack almost always involves the participation of inflammatory cells, whether they are called to action by the humoral or cellular arms of the immune response. Inflammatory cells were not observed in our material and have not been prominent features of any autopsy to date, although foci of inflammatory cells have been reported in four of the six autopsy cases. 1.4.5.8 Hence, withering of the retina probably occurs by some noninflammatory means, assuming the disorder has an autoimmune pathogenesis, Other causes that have been considered in paraneoplastic visual loss include viruses, which can cause selective destruction of neurons,4,25.26 and elaboration of

Rizzo and Gittinger . Paraneoplastic Retinopathy Syndrome Table 1. Electrophysiologic and Laboratory Findings of All Patients Reported with Paraneoplastic Retinopathy Author Sawyer

Patient No.

1

Electrophysiology Flat/reduced ERG

2 3

Keltner

4*·t·t 5 6§

Flat ERG

Buchanan Klingele

7 8

Flat ERG; normal YEP Flat ERG

Grunwald II

9t·~

Abnormal YEP

Komguth

Thirkill

10 11

van der Pol Nunez

12

Berson Crofts Thirkill

14 15 16

Jacobson

17 18 19

Rizzo

Flat/reduced ERG

Flat ERG Flat ERG Flat ERG Flat ERG; low/reduced YEP Loss of rod signal on ERG Flat ERG

13

Almost flat ERG Flat ERG Predominantly rod dysfunction Flat ERG

20 PR

=

photoreceptor; YEP

=

Tissue Diagnosis

Retinal Immunohistochemistry

Western Blot (kD)

Oat-cell, lung PR degeneration Oat-cell, lung PR degeneration Small-cell, lung PR degeneration Oat-cell, lung Cervical carcinoma PR degeneration PR degeneration Adenocarcinoma, breast Oat-cell, lung ganglion cell loss Nonsmall cell, lung Oat-cell, lung Oat-cell, lung Oat-cell, lung Melanoma Endometrial carcinoma Oat-cell, lung Oat-cell, lung Oat-cell, lung Oat-cell, lung Oat-cell, lung

Staining large ganglion cells Nonspecific nuclear staining Diffuse staining

Staining inner retina

20/65 23

145/205** 23 23/48

50 Diffuse staining of nuclei in all cell layers

23

23/48 23

Selective staining of outer retina Selective staining of outer retina

Neg

48

visual-evoked potential.

The order of patients is chronological and the relevant references are listed consecutively in the bibliography from 1 to 15. o

Also described as case 1 in Grunwald, 1985, and case 1 in Kornguth, 1986.

t Western blot data found in Kornguth, 1986. t Retinal immunohistochemistry provided in Grunwald,

1985. § Also described as case 1 in Thirkill, 1987. II Case 1 also described as case 1 in Kornguth, 1982. ~ Case 2 also described as case 4 in Thirkill, 1987, and as case 2 in Kornguth, 1986, and the autopsy findings were provided in the 'case report' in Grunwald,1987 . •• Western blot data on same patient also reported in Thirkill, 1987 (case 4), and in the latter situation 23 kD binding was observed instead of 145/ 205 kD binding.

toxins by the tumor. 1.5 Retinotoxic effects of chemotherapeutic agents can cause visual loss but both of our patients experienced visual loss before initiation of chemotherapy. In a recent study of the sera of patients with paraneoplastic visual loss, Polans et al27 found an antibody to a calcium2 + -sensitive photoreceptor protein that regulates an important step in the biochemical recovery of photoreceptors after stimulation by light; how altered regulation of this process leads to cell death is not evident, but the pivotal role played by this protein suggests that alteration of its function could significantly affect the stability of photoreceptors. 28

Diagnosis of this condition is usually predicated on finding an abnormal ERG without typical funduscopic features of the inherited retinal degenerations, absence of family history of retinal disease, and a clinical course consisting of acute or subacute loss of vision-usually in older adults, often with positive visual phenomena. Visual symptoms frequently antedate identification of the tumor and therefore early diagnosis is important. ELISA and Western blot immunologic studies have proven to be useful as markers of the disease. We obtained these tests to assist and confirm the diagnosis in our patients, and we were initially disturbed by the discrepancy between those

1293

Ophthalmology

Volume 99, Number 8, August 1992

test results and our immunohistochemical results-the ELISA and Western blot studies were not diagnostic despite the fact that immunohistochemical staining of the retina was unequivocal, even at high dilutions. In retrospect, the lack of concordance between the immunologic studies should not have been unexpected. Each of the three tests examines characteristics of antibodies with a different strategy: in our patients, the ELISA test was used to quantify the reaction between patient's serum and retina or optic nerve, the Western blot test was used to identify the molecular weight of the antibody from the patient's serum that reacted with either retina or optic nerve, and immunohistochemistry was used to identify the distribution of the retinal antigen(s). It would be reassuring for the tests to provide results that are consistent, but, as in our cases, a disparity may occur. The methods of preparing the material for the ELISA and Western blot studies may alter the conformation of proteins, making them unrecognizable to the antibodies that were directed against them. Immunohistochemistry can be performed while maintaining the overall structure of tissue, thus keeping the proteins in situ. Perhaps this difference reduced the vulnerability of the immunohistochemical tests to conformational alterations that might have affected the other tests. Inconsistencies like those demonstrated in this report emphasize the value of obtaining a battery of immunologic tests. Immunologic testing using ELISA and Western blot methods has certainly contributed to an understanding of paraneoplastic retinopathy. These tests are relatively easy to perform and are the best screening tests. These tests, however, may yield false-negative results, as occurred in our two patients. Administration of prednisone and chemotherapy may have lowered levels of circulating antibodies and contributed to the negative results. It is interesting, however, that immunohistochemistry still uncovered antiretinal antibodies despite these medical therapies. Lack of an identifiable serum antibody in paraneoplastic retinopathy is consistent with a precedent recognized in some cases of myasthenia gravis, the best characterized autoimmune disorder. Myasthenia gravis results from an impairment of neuromuscular transmission mediated by antibodies directed against acetylcholine receptors in muscle. 29 Antibodies are not detectable in all cases of myasthenia gravis, and there is no correlation between the clinical characteristics of the myasthenia and the titer or presence of antibodies. 30 The occurrence of seronegative myasthenia can be explained by a failure of the standard radioimmunoassay to detect the specific antibody causing the syndrome or by the fact that the antibodies could be bound to muscle membrane end plates and consequently removed from the serum. An analogous situation may exist in some cases of the paraneoplastic retinopathy syndrome. Seronegativity also occurs in some cases of another relatively well-characterized syndrome, paraneoplastic cerebellar degeneration. 31,32 The paraneoplastic visual syndrome usually has been referred to as "paraneoplastic photoreceptor retinopathy" despite evidence that visual loss may result from involve-

1294

ment of the photoreceptors, inner retina, or optic nerve. 33 Given the remarkable diversity and specificity of the immune system, it would not be surprising to discover that visual loss might occur from an immune attack directed at various retinal antigens or cell types. For instance, Berson and Lessell 12 reported a patient who developed night blindness secondary to a malignant melanoma, in whom the ERG was interpreted as showing a selective interruptien of intraretinal (i.e., postphotoreceptor) rod signal transmission. It also has become common to refer to paraneoplastic visual loss as the CAR syndrome; while this name simply indicates the association of visual loss with cancer, CAR historically has suggested the presence of an antibody against the CAR antigen. 9 It is reasonable to suspect, however, that the inciting antigen in cases associated with malignant melanoma, and perhaps other heretofore unrecognized clinical syndromes, may not be the CAR antigen (e.g., in our patient 2, an antibody was directed against a 48 kD but not a 23 kD CAR protein). Even if 23 kD antibodies are present, they could be directed against one of several proteins and result in slightly different clinical syndromes, with the degree of cone or rod dysfunction dependent on the specificity of the antibody and the distribution of the antigen. 27 ,28,34 Accordingly, because of the likelihood that several closely related but clinically distinct syndromes may be recognized, we suggest that the syndrome be referred to by the more general term paraneoplastic retinopathy unless laboratory evidence clearly implicates a specific level of retinal dysfunction or there is swelling of the optic nerve. 33 It would be useful to categorize visual loss based on the cell type that had been damaged, since a correlation could exist between cell type and signs/symptoms. 32 Early recognition of a particular visual syndrome might help in the search for occult carcinoma. 9,14 Immunologic studies to date do not prove that paraneoplastic retinopathy is the result of a single, specific antigen detectable by current methods. Our application of immunohistochemistry together with the clinical and pathologic findings in our patients supports the concept of an autoimmune pathogenesis, as originally suggested by Keltner et aV by showing selectivity of the immune response and correlation between the apparent target of the immune response and the clinical and pathologic findings. Immunohistochemical testing of this sort may, however, prove to be neither specific nor sensitive when applied to a larger sample of cases. Given the current uncertainty surrounding laboratory identification of this syndrome, we believe that the diagnosis of paraneoplastic retinopathy remains clinical and pathologic. Acknowledgment. The authors thank Jerome B. Posner, MD, and Eliot L. Berson, MD, for serum used as controls.

References 1. Sawyer RA, Selhorst JB, Zimmerman LE, Hoyt WF. Blindness caused by photoreceptor degeneration as a remote effect of cancer. Am J Ophthalmol 1976;81:606-13.

Rizzo and Gittinger . Paraneoplastic Retinopathy Syndrome 2. Kornguth SE, Klein R, Appen R, Choate J. Occurrence of anti-retinal ganglion cell antibodies in patients with small cell carcinoma of the lung. Cancer 1982;50: 1289-93. 3. Kornguth SE, Kalinke T, Grunwald GB, et al. Anti-neurofilament antibodies in the sera of patients with small cell carcinoma of the lung and with visual paraneoplastic syndrome. Cancer Res 1986;46:2588-95. 4. Keltner JL, Roth AM, Chang RS. Photoreceptor degeneration: possible autoimmune disorder. Arch Ophthalmo1 1983;101 :564-9. 5. Buchanan TAS, Gardiner TA, Archer DB. An ultrastructural study of retinal photoreceptor degeneration associated with bronchial carcinoma. Am J Ophthalmol 1984;97:277-87. 6. Klingele TG, Burde RM, Rappazzo JA, et al. Paraneoplastic retinopathy. J Clin Neuro Ophthalmol 1984;4:239-45. 7. Grunwald GB, Klein R, Simmonds MA, Kornguth SE. Autoimmune basis for visual paraneoplastic syndrome in patients with small-cell lung carcinoma. Lancet 1985; 1:65861. 8. Grunwald GB, Kornguth SE, Towfighi J, et al. Autoimmune basis for visual paraneoplastic syndrome in patients with small cell lung carcinoma. Retinal immune deposits and ablation of retinal ganglion cells. Cancer 1987;60:780-6. 9. Thirkill CE, Roth AM, Keltner JL. Cancer-associated retinopathy. Arch Ophthalmol 1987; 105:372-5. 10. van der Pol BAE, Planten JTh. A non-metastatic remote effect of lung carcinoma. Doc Ophthalmol 1987;67:89-94. 11. Vargas Nunez JA, Bonilla Velasco F, Cases Fernandez-Tejerina J, Vaquero Ruano M. Degeneraci6n retiniana paraneoplasica. Med Clin 1987;88:431. 12. Berson EL, Lessell S. Paraneoplastic night blindness with malignant melanoma. Am J Ophthalmol 1988;106:307-11. 13. Crofts JW, Bachynski BN, Odel JG. Visual paraneoplastic syndrome associated with undifferentiated endometrial carcinoma. Can J Ophthalmol 1988;23:128-32. 14. Thirkill CE, Fitzgerald P, Sergott RC, et al. Cancer-associated retinopathy (CAR syndrome) with antibodies reacting with retinal, optic-nerve, and cancer cells. N Engl J Med 1989;321: 1589-94. 15. Jacobson DM, Thirkill CE, Tipping SJ. A clinical triad to diagnose paraneoplastic retinopathy. Ann Neurol 1990;28: 162-7. 16. Brain WR, Norris FH, eds. The Remote Effects of Cancer on the Nervous System. New York: Grune & Stratton, 1965. 17. Henson RA, Urich H. Cancer and the Nervous System: the Neurological Manifestations of Systemic Malignant Disease. Oxford: Blackwell, 1982. 18. Anderson NE, Cunningham JM, Posner JB. Autoimmune pathogenesis of paraneoplastic neurological syndromes. Crit Rev Neurobiol 1987;3:245-99.

19. Wacker WB, Donoso LA, Kalsow CM, et al. Experimental allergic uveitis. Isolation, characterization, and localization of a soluble uveitopathogenic antigen from bovine retina. J Immunol 1977;119:1949-58. 20. Donoso LA, Yamaki K, Merryman CF, et al. Human Santigen: characterization of uveitopathogenic sites. Curr Eye Res 1988;47: 1077-85. 21 . Pfister C, Chabre M, Plouet J, et al. Retinal S antigen identified as the 48K protein regulating light-dependent phosphodiesterase in rods. Science 1985;228:891-3. 22. Stefansson K, Marton LS, Dieperink ME, et al. Circulating autoantibodies to the 200,OOO-daiton protein of neurofilaments in the serum of healthy individuals. Science 1985;228: 1117-19. 23. Christian CL, Elkon KB. Autoantibodies to intracellular proteins. Clinical and biologic significance. Am J Med 1986;80:53-61. 24. Thirkill CE, Roth AM, Takemoto DJ, et al. Antibody indications of secondary and superimposed retinal hypersensitivity in retinitis pigmentosa. Am J OphthalmoI1991 ;112: 132-7. 25. Koroshetz WJ, McKee AC. A 76-year-old man with confusion, agitation, and a gait disorder. Case Records of the Massachusetts General Hospital Case 39-1988. N Engl J Med 1988;319:849-60. 26. Tyler KL. Host and viral factors that influence viral neurotropism. I. Viral cell attachment proteins and target cell receptors. Trends Neurosci 1987; 10:455-60. 27. Polans AS, Buczylko J, Crabb J, Palczewski K. A photoreceptor calcium binding protein is recognized by autoantibodies obtained from patients with cancer-associated retinopathy. J Cell Bioi 1991 ;112:981-9. 28. Dizhoor AM, Ray S, Kumar S, et al. Recoverin: a calcium sensitive activator of retinal rod guanylate cyclase. Science 1991;251 :915-18. 29. Lindstrom J, Shelton D, Fujii Y. Myasthenia gravis. Adv Immunol 1988;42:233-84. 30. Soliven BC, Lange DJ, Penn AS, et al. Seronegative myasthenia gravis. Neurology 1988;38:514-17. 31 . Hammack JE, Kimmel DW, O'Neill BP, Lennon VA. Paraneoplastic cerebellar degeneration: a clinical comparison of patients with and without Purkinje cell cytoplasmic antibodies. Mayo Clin Proc 1990;65: 1423-31. 32. Anderson NE, Rosenblum MK, Posner JB. Paraneoplastic cerebellar degeneration: clinical-immunological correlations. Ann Neurol 1988;24:559-67. 33. Boghen D, Sebag M, Michaud J. Paraneoplastic optic neuritis and encephalomyelitis. Report of a case. Arch Neurol 1988;45:353-6. 34. Yamagata K, Goto K, Kuo C-H, et al. Visinin: a novel . calcium binding protein expressed in retinal cone cells. Neuron 1990;4:469-76.

1295