Anti–Retinal Pigment Epithelium Antibodies in Acute Exudative Polymorphous Vitelliform Maculopathy: A New Hypothesis About Disease Pathogenesis

Objective  To investigate the etiology of acute exudative polymorphous vitelliform maculopathy (AEPVM) in a patient with metastatic melanoma, undiagnosed at initial examination, by testing for autoimmune mechanisms.

Methods  Serum samples were obtained from a 50-year-old man with AEPVM and metastatic unknown primary melanoma during the acute stage and 3 years later when subretinal fluid had resolved and melanoma was in remission (AEPVM convalescent stage). Western immunoblots using both serum samples against human donor retinal extract and cultured primary human retinal pigment epithelium (RPE) cell extract were performed to identify antiretinal and anti-RPE antibodies. Serum samples from 5 unaffected participants were tested as controls. Protein identification was performed using 2-dimensional gel electrophoresis and mass spectrometry and was then confirmed by blotting against purified protein.

Results  Western immunoblots of the patient's serum against human donor retinal extract and RPE cell extract demonstrated several antiretinal antibodies, as well as anti-RPE antibodies against a 26-kDa protein that was identified as peroxiredoxin 3 (PRDX3). Serum reactivity against PRDX3 was greatly decreased in the convalescent-stage serum sample compared with the acute-stage serum sample, while results of retinal extract Western immunoblots remained essentially unchanged. Five separate serum samples from participants without AEPVM had no autoantibodies against PRDX3.

Conclusions  Paraneoplastic autoimmune reaction against RPE, with PRDX3 as the putative antigen, may be a cause of AEPVM. This is the first report to date linking a human RPE disease with anti-RPE antibodies against a heretofore undetermined putative protein. Testing for RPE autoantibodies may be useful in exploring the pathogenesis of other presumed RPE-related diseases.

The etiology of acute exudative polymorphous vitelliform maculopathy (AEPVM), a rare disease characterized by serous retinal detachment and subretinal accumulation of hyperautofluorescent yellowish material in the posterior pole, has been unknown.^1-3 The characteristic honeycomb-like pattern of hyperfluorescent whitish spots seen in the posterior pole at the onset of disease is believed to occur at the level of the retinal pigment epithelium (RPE).¹ Visual acuity slowly improves over several months as subretinal fluid resorbs and polymorphous vitelliform deposits accumulate.^1,4,5 Autofluorescence photography has demonstrated that the yellow subretinal deposits are usually brightly autofluorescent.^3,6 Previously, the longest reported follow-up of a patient with AEPVM had been 37 months, and a persistent decrease in photoreceptor function was noted at that patient's final visit using multifocal electroretinogram studies.⁷ The disease is usually accompanied by abnormal electro-oculogram findings that often persist.^1,4

In 2008, a patient with metastatic choroidal melanoma and acquired fundus findings similar to Best disease in the fellow eye was reported not to have a disease-causing mutation in the VMD2 gene responsible for that disease but to have circulating antibodies against an RPE protein, bestrophin 1, which is defective in Best disease.⁸ To better understand the etiology, pathogenesis, and course of disease in AEPVM, as well as to describe an approach for studying autoimmune mechanisms in diseases affecting the RPE, we report long-term clinical follow-up and testing (including for anti-RPE antibodies) of a patient with AEPVM who subsequently was found to have metastatic melanoma.

A 50-year-old man was seen with visual blurring in the central visual field of each eye for about 1 month. Results of laboratory tests performed before referral were normal or negative, including complete blood cell count with a differential cell count, erythrocyte sedimentation rate, serum viscosity, serum protein electrophoresis, antinuclear antibody, dilute Russell viper venom time, and venereal disease research laboratory and fluorescent treponemal antibody absorption. The patient was found to be positive for HLA-B27 antigen. The patient reported 2 sinus infections during the previous 5 months and an ear infection 3 weeks before examination.

His medical history was significant for sclerosing cholangitis and ulcerative colitis with frequent flare-ups, for which he had undergone total colectomy and cholecystectomy 3 years previously. There was no known family history of eye disorders.

The best-corrected Snellen visual acuity was 20/40 OU. Results were normal for external, ocular motility, Amsler grid, confrontation visual field, pupillary, and intraocular pressure examinations. Anterior segments were unremarkable, with no signs of uveitis. In each eye, vitreous was free of cells, the optic disc was pink and flat, and retinal vasculature was normal. Shallow subretinal fluid was noted throughout the posterior pole of each eye, deepest at the macula and extending to just beyond the vascular arcades and slightly nasal to the optic disc. Whitish-yellow round lesions in a honeycomb-like pattern at the level of the RPE were noted in the pericentral macular region and around the optic nerve head bilaterally (Figure 1).

On fluorescein angiography, the white lesions showed early hyperfluorescence with late staining and no apparent leakage (Figure 2). Layering of material in the subretinal space inferiorly blocked background fluorescence throughout the angiogram. Autofluorescence photography demonstrated relative blockage by subretinal fluid, with no areas of hyperautofluorescence (Figure 3). Optical coherence tomography demonstrated subretinal fluid throughout the macular region, with cystoid edema in detached retina, mild accumulation of subretinal debris inferiorly, and folds in the outer retina near the margins of the detachment (Figure 4). Ultrasonographic examination revealed no choroidal or scleral thickening.

At the 1-month follow-up visit, subjective visual function and examination findings were unchanged. The patient was started on an empirical trial of prednisone, 80 mg, daily for 1 week, followed by prednisone, 60 mg daily. Reexamination 3 weeks later showed no apparent response to treatment, and prednisone was tapered to discontinuation over the next 1½ months. Three months after initial examination, ophthalmoscopy and optical coherence tomography showed less subretinal and intraretinal fluid. Small hyperautofluorescent yellow deposits were noted in the areas of resolving subretinal fluid (Figure 5).

At the 5-month follow-up visit, the patient reported that he had been diagnosed 1 month previously as having axillary lymph node and liver biopsy–proven metastatic malignant melanoma with an unknown primary source. Visual acuity had improved to 20/25 OD and 20/30 OS. Clinical evaluation showed further decrease in subretinal fluid and increasing amounts of yellow-brown autofluorescent pigment deposits. Results of Goldmann visual field and Ishihara and Farnsworth D15 color testing were normal. Dark-adapted final thresholds measured using the Goldmann-Weekers dark adaptometer were mildly elevated. Full-field electroretinography (LKC Technologies, Inc, Gaithersburg, Maryland), following the International Society for Clinical Electrophysiology of Vision standards, revealed substantial reduction in rod- and cone-mediated response amplitudes in both eyes (eFigure 1).

The patient was treated orally for metastatic melanoma with the alkylating agent temozolomide. His visual acuity continued to slowly improve as subretinal fluid gradually diminished. At the 1-year follow-up visit, examination revealed 20/20 OU visual acuity and no subretinal fluid bilaterally, with slightly increased vitelliform deposits. Optical coherence tomography confirmed the subretinal location of vitelliform deposits and complete resolution of subretinal fluid (Figure 6).

At the final 41-month follow-up examination, the patient reported completion of his temozolomide regimen 3 months earlier, given an excellent treatment response and a “clear” computed tomographic image. Visual acuity measured 20/15 OU. Ophthalmoscopy and fundus autofluorescence photography revealed slightly increased yellow-brown pigment deposits and no recurrence of subretinal fluid (Figure 7). Dark-adapted thresholds were within normal limits. Electroretinography rod-mediated responses showed marked improvement in each eye. Cone-mediated flash and flicker responses were within the range of normal intervisit variation in the right eye and showed a 50% amplitude increase in the left eye (eFigure 2). The electro-oculographic response in each eye showed a peculiar sharp decrease in the baseline at light onset (eFigure 3). A light peak was present during the light phase, but the baseline did not reach the preadaptation level even after 19 minutes' duration. Arden ratios were computed by shifting the baseline to the dark-phase level. The resulting Arden ratios of light peak to dark trough were subnormal, 1.36 in the right eye and 1.43 in the left eye (Arden ratio >1.8 is normal). Molecular diagnostic testing of the BEST1 gene revealed no significant mutations.

After informed consent was obtained, blood samples were obtained at the 7-month and 41-month follow-up visits to test for melanoma-associated retinopathy (MAR) autoantibodies, other antiretinal autoantibodies, and anti-RPE autoantibodies. A serum sample from the 7-month follow-up was initially analyzed (using commercially obtained purified proteins) for the presence of antiretinal antibodies to the following: aldolase C, aldolase A, carbonic anhydrase II, recoverin, heat shock protein 1, S-arrestin, and alpha-enolase. Serum samples from both time points were subsequently tested for autoantibodies against human donor retinal extract and primary nonimmortalized human RPE cell culture extract using Western immunoblots. Briefly, 20 μg of total protein extract was loaded per lane for sodium dodecyl sulfate–polyacrylamide gel electrophoresis; separated proteins were transferred onto a nitrocellulose membrane. The blots were blocked in 3% milk–phosphate-buffered saline–Tween for 1 hour, incubated with the patient's serum at 1:100 dilution overnight, and then incubated with horseradish peroxidase–conjugated goat antihuman IgG (Chemicon, Temecula, California) at 1:5000 dilution for 90 minutes. Target protein bands were detected using a commercially available kit (SuperSignal West Pico Chemiluminescent Substrate; Thermo Fisher Scientific, Rockford, Illinois). Bands identified by Western immunoblots were further analyzed by 2-dimensional gel electrophoresis and Western immunoblots, followed by matrix-assisted laser desorption ionization time-of-flight (MALDI-TOF) tandem mass spectrometry (MS/MS) analysis. Last, the RPE MS/MS results were confirmed by Western immunoblots with purified protein (Abcam, Cambridge, Massachusetts, and Abnova, Taiwan). Our institutional review board determined that review and oversight of this project were not required.

Using the purified MAR protein panel, antibodies to the following proteins were detected in serum obtained at the 7-month follow-up visit: aldolase C, aldolase A, carbonic anhydrase II, recoverin, and heat shock protein 1. Using Western immunoblots with human donor retinal extract, 2-dimensional gel analysis, and MS/MS analysis, there was also evidence of antibodies to retinal isoform 1 of triosephosphate isomerase (approximately 31.1 kDa) and Ran guanidin triphosphatase (approximately 26.7 kDa), which have not previously been associated with MAR to our knowledge (data not shown). Testing of serum from the patient's last visit (the 41-month follow-up) using Western immunoblots with human donor retinal extract revealed similar bands as serum from the 7-month follow-up (Figure 8). A prominent band at approximately 21 kDa was weaker in the older (7-month) serum, which was likely owing to successive use of this serum sample for testing.

Testing of serum from the 7-month follow-up visit with RPE cell extract detected 2 bands on the Western immunoblot at approximately 26 kDa and 75 kDa (confirmed with an additional human RPE cell line [data not shown]). Testing of serum from the final visit detected only the 75-kDa band (Figure 8). Among 5 healthy control patients tested, none had autoantibodies against the RPE cell extract, except for 1 patient who had a single band on Western immunoblot analysis of different size from either of the 2 bands in our patient (data not shown). Subsequent separation of the RPE cell extract on 2-dimensional gel electrophoresis and Western immunoblots with serum from the 7-month follow-up revealed immunoreactivity to 2 spots associated with the 26-kDa band. MALDI-TOF MS/MS protein identification demonstrated high probability of heat shock protein beta-6 (approximately 18.9 kDa) or peroxiredoxin 3 isoform b (PRDX3) (approximately 26.1 kDa) for the first spot and peroxiredoxin 5 (PRDX5) (approximately 17.2 kDa) for the second spot. Because PRDX3 was closest in molecular weight to the bands obtained on the initial Western immunoblot and was in the same protein family as PRDX5, purified PRDX3 and PRDX5 were commercially obtained. Testing with these purified proteins revealed the presence of antibodies against PRDX3 in the 7-month follow-up serum sample in a much higher concentration than in serum from the final visit (Figure 8). No antibody production against purified PRDX5 was detected in either serum sample (data not shown).

Acute exudative polymorphous vitelliform maculopathy is a rare disease of unknown etiology characterized by serous retinal detachment and subretinal accumulation of hyperautofluorescent yellowish material in the posterior pole. We report evidence of autoantibody production against PRDX3, an RPE protein, in a patient with active AEPVM. The autoantibody was undetectable in the convalescent stage of the disease. No other significant difference in retina or RPE autoantibody production in this patient was detected between the active and convalescent stages of disease. PRDX3, a member of a family of peroxidases, is a mitochondrial protein that is important in protection against cellular oxidative damage.^9,10

In addition to residing near the photooxidative retina and in an environment of locally elevated oxygen tension, RPE consumes the shed outer segments of the photoreceptors and is the most phagocytic tissue in the body, elaborating large numbers of free radicals and reactive oxygen species as a function of its high metabolic activity.^11,12 Therefore, proper RPE function is apt to be critically dependent on protective mechanisms against oxidative damage to the cells themselves. Because AEPVM demonstrates lesions at the level of the RPE and is believed to likely be a disease of the RPE that is acquired,^1,2,6,7 we hypothesize based on the evidence presented herein that AEPVM can be caused by autoantibodies to RPE PRDX3. Presumably, the yellowish subretinal deposits seen in our patient and others with AEPVM are due to poorly functioning RPE, demonstrated by abnormal electro-oculographic findings, resulting in buildup of unprocessed cellular debris and possibly lipofuscin, which in turn may be responsible for the autofluorescent findings.⁶

In our patient, given the history of melanoma, generation of anti-RPE antibodies was likely a paraneoplastic event. Although patients with MAR do not typically have exudative vitelliform lesions, clinical findings similar to AEPVM in patients with metastatic melanoma have been reported previously.¹³ In addition to AEPVM, our patient demonstrated negative waveforms in the bright flash dark-adapted electroretinogram (eFigure 1), typical of MAR. He also had bipolar, cone, and RPE staining on indirect immunohistologic examination using fresh donor retina (data reported as case 471 elsewhere¹⁴), as well as antiretinal antibodies to several proteins previously associated with MAR.¹⁴ Patients with metastatic melanoma and vitelliform lesions different from AEPVM have also been described, typically with 1 or more localized yellow serous retinal detachments and without an acute-stage honeycomb-like appearance.^8,15-19 Because AEPVM and other acquired vitelliform retinopathies may be associated with underlying melanoma, patients should be evaluated and monitored accordingly.

It is possible that several RPE antigens may cause an autoimmune reaction that results in the clinical picture of AEPVM. Headache and upper respiratory tract infection have been reported before manifestation of symptoms^1,2,7 and could implicate a viral trigger for AEPVM, which has been proposed previously.^1,7 Viral antigens present in RPE and autoantibodies against RPE have been detected following viral infection in mice and have been implicated in retinopathy.²⁰ Furthermore, it has been shown that host genetics has a role in the extent of this virus-induced disease.²¹ Inasmuch as we propose that AEPVM likely is an autoimmune retinal pigment epitheliopathy triggered by neoplasm, viral infection, or other autoimmune process, its development is probably dependent on tumor or virus type and interaction with the genetic predisposition of the patient. We speculate that our patient's ocular disease improved as circulating autoantibody levels decreased concomitant with tumor regression induced by immunologic activity and chemotherapy.

Although autoantibodies to peroxiredoxins 1 and 4 have been discovered,²² this is the first report of the existence of autoantibodies to PRDX3,²² a protein that previously has been demonstrated to be present in RPE.^23,24 An earlier study⁸ implicated antibestrophin antibodies in a paraneoplastic Best disease–like condition, but ours is the first report linking a human RPE disease with anti-RPE antibodies against a heretofore undetermined putative protein. Assays for RPE autoantibodies and subsequent identification or confirmation of their antigenic trigger in a manner similar to ours may be useful in exploring the pathogenesis of other presumed RPE-related diseases with unknown etiologies.

Understanding autoimmune disease mechanisms with certainty can be difficult because of issues related to antibody detection,²⁵ antigen concentrations used in testing, and background prevalence of antibodies among normal and diseased populations. Although our patient had cancer and a background of colitis, the anti-PRDX3 antibody was present in high concentrations during active retinopathy and then essentially disappeared when the exudation had resolved. This is consistent with PRDX3 being a potential causative antigen. Although we tested 5 control subjects, it is possible that a larger control group would demonstrate the presence of anti-PRDX3 and other anti-RPE antibodies. Therefore, we cannot be certain that anti-PRDX3 is unique to AEPVM or is not an epiphenomenon. Our study findings suggest that future patients with AEPVM should be tested for RPE antibodies using nonimmortalized early-passage human cultured RPE cells, which likely may exhibit protein expression patterns that are more similar to native eye tissue than those of immortalized human RPE cell lines.^23,26 If they are used, it is important that human postmortem tissues should be fresh so as to avoid the variable gene expression that can result from hypoxia due to delays in handling.²⁴

In summary, we present evidence for an RPE autoimmune etiology of AEPVM and implicate PRDX3 as the putative antigen. The long follow-up of our patient confirms previous reports that AEPVM is unlikely to recur and has a good visual prognosis. Our findings and an overview of the literature show that paraneoplastic processes can be a trigger for AEPVM; clinicians should accordingly be alert for the potential of occult malignant neoplasms in affected patients. Further study of serum samples from other patients with AEPVM needs to be performed to determine the prevalence of anti-PRDX3 antibodies in paraneoplastic and nonparaneoplastic AEPVM. In addition, animal studies need to be undertaken to demonstrate the disease-causing role of anti-PRDX3 antibody. Last, we propose that assays for anti-RPE autoantibodies may prove useful in understanding the pathogenesis of other presumed RPE-related diseases.

Correspondence: Mark W. Johnson, MD, W. K. Kellogg Eye Center, Department of Ophthalmology and Visual Sciences, University of Michigan, 1000 Wall St, Ann Arbor, MI 48105 (markwj@med.umich.edu).

Submitted for Publication: February 10, 2010; final revision received March 30, 2010; accepted April 5, 2010.

Financial Disclosure: None reported.

Author Contributions: Drs Koreen and He contributed equally to this work. Drs Koreen, He, Johnson, and Heckenlively had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Funding/Support: This research was supported by a Foundation Fighting Blindness Center Grant (Dr Heckenlively) and by an unrestricted departmental grant from Research to Prevent Blindness.

Additional Contributions: The Michigan Proteome Consortium performed mass spectrometry analysis.

This article was corrected for errors on February 9, 2011.