Consensus on the Diagnosis and Management of Nonparaneoplastic Autoimmune Retinopathy Using a Modiﬁed Delphi Approach

AUSTIN R. FOX, LYNN K. GORDON, JOHN R. HECKENLIVELY, JANET L. DAVIS, DEBRA A. GOLDSTEIN, CAREEN Y. LOWDER, ROBERT B. NUSSENBLATT, NICHOLAS J. BUTLER, MONICA DALAL, THIRAN JAYASUNDERA, WENDY M. SMITH, RICHARD W. LEE, GRAZYNA ADAMUS, CHI-CHAO CHAN, JOHN J. HOOKS, CATHERINE W. MORGANS, BARBARA DETRICK, AND H. NIDA SEN

PURPOSE: To develop diagnostic criteria for nonpara- second-line treatments, though a consensus agreed that neoplastic autoimmune retinopathy (AIR) through biologics and intravenous immunoglobulin were consid- expert panel consensus and to examine treatment patterns ered appropriate in the treatment of nonparaneoplastic among clinical experts. AIR patients regardless of the stage of disease. Experts DESIGN: Modified Delphi process. agreed that more evidence is needed to treat nonparaneo- METHODS: A survey of uveitis specialists in the Amer- plastic AIR patients with long-term immunomodulatory ican Uveitis Society, a face-to-face meeting (AIR Work- therapy and that there is enough equipoise to justify ran- shop) held at the National Eye Institute, and 2 iterations domized, placebo-controlled trials to determine if nonpar- of expert panel surveys were used in a modified Delphi aneoplastic AIR patients should be treated with long-term process. The expert panel consisted of 17 experts, immunomodulatory therapy. Regarding antiretinal anti- including uveitis specialists and researchers with exper- body detection, consensus agreed that a standardized tise in antiretinal antibody detection. Supermajority assay system is needed to detect serum antiretinal anti- consensus was used and defined as 75% of experts in bodies. Consensus agreed that an ideal assay should agreement. have a 2-tier design and that Western blot and immuno- RESULTS: There was unanimous agreement among ex- histochemistry should be the methods used to identify perts regarding the categorization of autoimmune retinop- antiretinal antibodies. athies as nonparaneoplastic and paraneoplastic, including CONCLUSIONS: Consensus was achieved using a modi- cancer-associated retinopathy and melanoma-associated fied Delphi process to develop diagnostic criteria for retinopathy. Diagnostic criteria and tests essential to nonparaneoplastic AIR. There is enough equipoise to the diagnosis of nonparaneoplastic AIR and multiple sup- justify randomized, placebo-controlled trials to determine portive criteria reached consensus. For treatment, experts whether patients with nonparaneoplastic AIR should be agreed that corticosteroids and conventional immun- treated with long-term immunomodulatory therapy. Ef- osuppressives should be used (prescribed) as first- or forts to develop a standardized 2-tier assay system for the detection of antiretinal antibodies have been initiated as a result of this study. (Am J Ophthalmol 2016;168: Supplemental Material available at AJO.com. 183–190. Published by Elsevier Inc.) Accepted for publication May 12, 2016. From the Laboratory of Immunology, National Eye Institute, National Institutes of Health, Bethesda, Maryland (A.R.F., R.B.N., C.-C.C., J.J.H., H.N.S.); Stein Eye Institute, David Geffen School of Medicine at UTOIMMUNE RETINOPATHIES ARE A GROUP OF the University of California, Los Angeles, Los Angeles, California (L.K.G.); Kellogg Eye Center, University of Michigan, Ann Arbor, inflammatory-mediated diseases characterized by Michigan (J.R.H., T.J.); Bascom Palmer Eye Institute, University of the presence of antiretinal antibodies, visual field Miami Miller School of Medicine, Miami, Florida (J.L.D.); Department A deficits, and photoreceptor dysfunction in the setting of of Ophthalmology, Northwestern University Feinberg School of Medicine, Chicago, Illinois (D.A.G.); Cole Eye Institute, Cleveland progressive, otherwise unexplained vision loss. Autoim- Clinic, Cleveland, Ohio (C.Y.L.); Massachusetts Eye and Ear, Harvard mune retinopathies can be categorized as paraneoplastic Medical School, Boston, Massachusetts (N.J.B.); Veterans Affairs AIR (pAIR), which includes cancer-associated retinopathy Boston Healthcare System, Jamaica Plain, Massachusetts (N.J.B.); Department of Ophthalmology, George Washington University, (CAR) and melanoma-associated retinopathy (MAR), or Washington, DC (M.D.); Department of Ophthalmology, Mayo Clinic, nonparaneoplastic autoimmune retinopathy in the absence of Rochester, Minnesota (W.M.S.); Bristol Eye Hospital, University of malignancy. As autoimmune retinopathy (AIR) is the Bristol, Bristol, United Kingdom (R.W.L.); Ocular Immunology Laboratory, Casey Eye Institute (G.A.), and Department of Physiology preferred term for an acquired and presumed immune- & Pharmacology (C.W.M.), Oregon Health & Science University, mediated retinopathy due to antiretinal autoantibodies in Portland, Oregon; and Department of Pathology, Johns Hopkins the absence of a malignancy, we use AIR to indicate the University, School of Medicine, Baltimore, Maryland (B.D.). Inquiries to H. Nida Sen, 10 Center Drive 10D45, Bethesda, MD 20892- nonparaneoplastic form of autoimmune retinopathy unless 0001; e-mail: [email protected] otherwise indicated.

0002-9394/$36.00 PUBLISHED BY ELSEVIER INC. 183 http://dx.doi.org/10.1016/j.ajo.2016.05.013 Despite being described almost 20 years ago,1 AIR re- a structured communication method designed to elicit mains an ill-defined disease. The diagnosis of AIR is typi- and collate the opinions of experts through anonymity, cally made based on the presence of antiretinal controlled feedback, statistical group response, and multi- antibodies and a combination of certain clinical features, ple iterations.10,11 The Delphi method was first developed in the absence of another cause of symptoms. Although by the RAND Corporation in the 1950s to forecast the the prevalence of AIR is unknown, it is thought to be a impact of technology on warfare and has since been used rare entity. However, it is probable that AIR is more prev- throughout numerous healthcare fields including alent than thought and remains undiagnosed in many cases ophthalmology and in uveitis to build consensus among owing to the lack of standardized diagnostic criteria and its experts for the diagnosis and management of disease.10–15 protean clinical features that overlap with other retinal In diseases where clinical evidence is lacking, this degenerative diseases. Nonetheless, it is important to rule method is deemed suitable for the development of out malignant etiologies and treatable conditions when guidelines for diagnosis or management. The goal is to considering the diagnosis of AIR to prevent morbidity narrow the range of responses with each iteration to and treatable vision loss. arrive at an expert consensus. Ultimately, the Delphi While clinical features may vary considerably, process allows experts to work together in a structured commonly recognized manifestations have been identi- manner to gain a better understanding in areas where fied.2–5 The presence of circulating antiretinal antibodies consensus is lacking. is considered essential to the diagnosis of AIR. Although The modified Delphi process used in this study consisted great strides have been made in the detection and of multiple rounds of surveys and a face-to-face meeting, af- measurement of antiretinal antibodies, there is no ter which structured feedback was given for each round. Ex- universally standardized assay for antiretinal antibody perts were then encouraged to reconsider their opinion in testing. Consequently, inconsistent diagnoses among light of the cumulative responses of other experts. This institutions or physicians may result. One study compared allowed experts to remain anonymous while considering the results of antiretinal antibody detection and the responses and opinions of the group and clarifying their measurement between 2 laboratories and found an overall opinions for others. To develop consensus, a survey of uve- concordance rate of any antiretinal antibodies detected itis specialists in the American Uveitis Society (AUS), a to be 64% with a very poor interobserver agreement face-to-face meeting (AIR Workshop) held at the National (kappa ¼0.13). Further, the antiretinal antibody– Eye Institute, and 2 iterations of an expert panel survey specific concordance rate was a mere 36%.6 Currently, were used (Figure 1). This study was in adherence to the te- only 1 center in the United States provides antiretinal anti- nets of the Declaration of Helsinki. body testing commercially through a CLIA (clinical laboratory improvement amendments)-certified laboratory AMERICAN UVEITIS SOCIETY MEMBER SURVEY: (Ocular Immunology Laboratory, Casey Eye Institute, Consensus development began with a 10-question survey Oregon Health & Science University).7 To establish diag- of AUS members to gauge the understanding of autoim- nostic criteria, promote collaboration, and advance our un- mune retinopathies among uveitis specialists. The survey derstanding of AIR, the development of a standardized was developed electronically (Survey Monkey, Palo Alto, assay to detect antiretinal antibodies is essential.8 California, USA), and the survey link was posted in the Criteria have been proposed in the past2,3,9; however, AUS website forum. AUS is a subspecialty society for uve- efforts to establish comprehensive diagnostic criteria, itis specialists with approximately 250 members. Only 1 including clinical criteria and a standardized assay system for response was allowed per computer in an effort to limit antiretinal antibody detection, have not been documented multiple responses from 1 respondent. until now. We believe the establishment of standardized diagnostic criteria and an assay system for antiretinal AUTOIMMUNE RETINOPATHY WORKSHOP: Subse- antibody detection is the first step toward understanding the quently, a meeting of 40 clinicians and researchers in the pathogenesis of AIR. Ultimately, this will improve the field of uveitis and immunology was convened at the Na- management of patients with AIR. The purpose of this tional Institutes of Health (NIH) on September 27, 2013. paper is to describe the consensus process and to report In the AIR Workshop meeting, the diagnosis, manage- results of the consensus process among clinicians and ment, and pathophysiology of AIR were examined through researchers in establishing diagnostic criteria for AIR. presentations, expert panels, and group discussions. In addition, results of the initial AUS survey were reported. The most important questions in the field of AIR were identified through a group effort. METHODS EXPERT SURVEYS: Based on discussions at the AIR TO DEVELOP CONSENSUS FOR THE DIAGNOSIS OF AIR, A Workshop meeting, 2 surveys—1 clinical and 1 basic labo- modified Delphi process was used. The Delphi method is ratory—were developed to further collate opinions of

184 AMERICAN JOURNAL OF OPHTHALMOLOGY AUGUST 2016 EXPERT EXPERT AUS AIR SURVEY: SURVEY: SURVEY WORKSHOP ITERATION ITERATION 1 2

FIGURE 1. Consensus development for the diagnosis and management of autoimmune retinopathy (AIR) A modiﬁed Delphi process was used to develop consensus for the diagnosis and management of AIR and consisted of multiple rounds of surveys and the AIR Workshop. AUS [ American Uveitis Society.

experts and define consensus for the diagnosis and manage- were developed using the NICHD/NIH Clinical Trials ment of AIR. The expert surveys mostly consisted of state- Database (Bethesda, Maryland, USA). Unique identifiers ments with 5-point Likert scales (1, strongly disagree; 2, and passwords were distributed to experts via individual disagree; 3, neither agree nor disagree; 4, agree; 5, strongly e-mails. agree). When appropriate, multiple-choice, ranking, and multiple-select questions were used in the modified Delphi DEFINITIONS: Definitions are as follows: process. For all items, experts were encouraged to provide comments and feedback. For the clinical survey, a summary Essential Diagnostic Criteria: essential to the diag- of results from the initial AUS survey was included for ex- nosis of AIR and must be present to make the diag- perts to consider in the initial expert survey. In the subse- nosis of AIR. quent expert surveys, a summary of results from the initial Supportive Diagnostic Criteria: supports the diag- surveys was included for experts to consider as aligned with nosis of AIR but is not necessary to make the diag- our modified Delphi process. nosis of AIR. Supermajority consensus was used and defined as 75% of Core Diagnostic Test: essential to the diagnosis of experts who selected either 4 (agree) or 5 (strongly agree) AIR and should be performed at the initial or first OR 2 (disagree) or 1 (strongly disagree) for items with diagnostic evaluation when AIR is suspected. Likert scales. For questions in which Likert scale was not used, supermajority consensus was defined as 75% of re- spondents who selected a given answer choice. Below in results, we use ‘‘consensus’’ to indicate supermajority RESULTS consensus unless otherwise indicated. Because of the low number (n ¼ 6) of experts to whom the basic laboratory FIFTY-FOUR UVEITIS SPECIALISTS PARTICIPATED IN THE survey was sent (experts with basic science and laboratory initial AUS survey. Seventeen experts participated in the expertise in antiretinal antibody detection), we also report expert surveys (1 person had expertise in both clinical a simple majority consensus (>50%) and indicate when and basic fields). Twelve uveitis specialists participated in doing so. the initial clinical expert survey, and 11 participated in In the subsequent iterations, only questions where at the subsequent expert survey. Six researchers participated least 75% of responses fell within a range of 3 consecutive in the initial and subsequent basic laboratory surveys. values (ie, 3, 4, 5) were included in order to obtain further Here, we report the results of the final rounds of consensus consideration and consensus. Certain questions were refor- development. mulated and reiterated in the subsequent surveys based on comments and feedback from experts. Items reaching su- CLINICAL SURVEY: To gauge the experience of those permajority consensus were not reiterated. For example, surveyed in the final 2 iterations, experts (clinicians) when considering ‘‘response to treatment’’ as one of the were asked to approximate the number of patients with Supportive Diagnostic Criteria in the diagnosis of AIR, 4 the diagnosis of paraneoplastic or nonparaneoplastic AIR experts selected 3 (neither agree nor disagree), 3 experts seen in 1 year’s time: 7 see 3–7 patients, 3 see >7 patients, selected 4 (agree), and 3 experts selected 5 (strongly agree). and 2 see 1–3 patients. This item did not meet supermajority consensus, but As anticipated, there was unanimous agreement among because the response rate was >_75% for consecutive values all clinicians regarding the categorization of autoimmune of 3, 4, 5, the item was reiterated to be considered as Sup- retinopathies as nonparaneoplastic and paraneoplastic, portive Diagnostic Criteria in the subsequent iteration. including CAR and MAR. To disseminate clinical and basic laboratory survey iterations and results, a password-protected site was developed DIAGNOSTIC CRITERIA: All 5 items considered as Essen- using the NIH Clinical Trial Survey System (Bethesda, tial Diagnostic Criteria in the diagnosis of AIR achieved Maryland, USA). The clinical and basic laboratory surveys consensus, including: (1) no apparent cause responsible

VOL. 168 CONSENSUS ON NONPARANEOPLASTIC AUTOIMMUNE RETINOPATHY 185 TABLE 1. Criteria and Tests for the Diagnosis of Autoimmune Retinopathy

Diagnostic Criteria for AIRa

Essential Diagnostic Criteria Supportive Diagnostic Criteria

No apparent cause responsible for visual function abnormalityb Symptoms: photopsias or scotomas or dychromatopsia or nyctalopia or photoaversion ERG abnormality (with or without visual field abnormality) Systemic autoimmune disease: personal or family history Presence of serum antiretinal antibodies Rapidity of onset of vision changee Absence of fundus lesions and retinal degeneration or dystrophy that may explain visual function lossc Absence of overt intraocular inflammationd Core Diagnostic Testsf Malignancy evaluation by appropriate physician Fundus autofluorescence Electroretinogram Optical coherence tomography Serum antiretinal antibody testing Fluorescein angiogram

AIR ¼ autoimmune retinopathy. aAll Essential Diagnostic Criteria must be present and Supportive Diagnostic Criteria are not necessary to make the diagnosis of AIR. bIncluding no evidence of malignancy. cAbsence of chorioretinal lesions (other than incidental/small peripheral benign degenerations such as pavingstone, lattice, etc, or old toxoplasmosis scar) or absence of retinal dystrophy, retinitis pigmentosa, or other hereditary retina vitreal disorders. dLess than 1þ intraocular cells or haze present. eAcute (0–3 months) or subacute (3–6 months). fEssential to the diagnosis of AIR and should be performed at the initial or ﬁrst diagnostic evaluation when AIR is suspected.

for visual function abnormality such as malignancy, inflam- Experts agreed that in order to make the diagnosis of mation, infection, surgery, drug toxicity, trauma, or hered- AIR, all Essential Diagnostic Criteria needed to be present. itary retinal degeneration; (2) electroretinogram (ERG) The significance of supportive criteria in making the diag- abnormality with or without visual field abnormality; (3) nosis was not explored. presence of serum antiretinal antibodies; (4) absence of Consensus was reached on the following Core Diag- fundus lesions and retinal degeneration or dystrophy that nostic Tests to be performed at the initial or first diagnostic may explain visual function loss; (5) absence of overt intra- examination: malignancy evaluation by an appropriate ocular inflammation (Table 1). physician, serum antiretinal antibody testing, ERG, fluores- Absence of fundus lesions and retinal degeneration or cein angiogram, fundus autofluorescence, and optical dystrophy was defined through consensus as the absence coherence tomography (OCT) (Table 1). Consensus could of chorioretinal lesions (other than incidental or small pe- not be achieved for including the following as Core Diag- ripheral benign degenerations such as pavingstone, lattice, nostic Tests: Goldmann visual field (GVF), Humphrey vi- etc, or old toxoplasmosis scar) and the absence of retinal sual field (HVF), dark adaptation testing (DA), or color dystrophy, retinitis pigmentosa, or other hereditary retina vision testing. vitreal disorders. Absence of overt intraocular inflammation was defined through consensus as less than 1þ intraoc- TREATMENT AND MANAGEMENT: To assess preferred ular cells (anterior chamber or vitreous) or vitreous haze. treatment practices among experts, the clinicians were sur- Supportive Diagnostic Criteria for the diagnosis of AIR veyed regarding specific treatments in the management of reaching consensus included: (1) a personal or family history AIR patients. Clinicians unanimously agreed that steroids of systemic autoimmune disease; (2) the presence of photop- (local or systemic) and conventional immunosuppressives sias, scotomas, nyctalopia or photoaversion, or dyschroma- (antimetabolites or T cell inhibitors) should be used as first- topsia; and (3) rapidity of onset of vision change or second-line treatments in the management of AIR pa- (Table 1). Though rapidity of onset of vision change reached tients. Consensus was achieved for the following treatment consensus as supportive criteria, defining rapidity of onset of types to be considered appropriate regardless of the stage of vision change failed to reach consensus: 6 of 11 clinician ex- disease: steroids (local or systemic), conventional immuno- perts selected subacute (3–6 months) and 5 selected acute suppressives (such as antimetabolites or T cell inhibitors), (0–3 months) to define rapidity of onset of vision change biologics (such as monoclonal antibodies), and intravenous in the subsequent expert survey. Age and response to treat- immunoglobulin (IVIG) (Table 2). Consensus was not ment did not reach consensus to be included as Supportive reached for considering plasmapheresis in the management Diagnostic Criteria in the diagnosis of AIR. of AIR patients.

186 AMERICAN JOURNAL OF OPHTHALMOLOGY AUGUST 2016 TABLE 2. Treatments and Follow-up Tests to Consider in Autoimmune Retinopathy

Treatment and Management of AIR Follow-up Tests

Steroids (systemic or local)a Electroretinography Conventional immunosuppressivesa (such as antimetabolites and T cell inhibitors) Humphrey visual ﬁelds, Goldmann visual ﬁelds Biologics (such as monoclonal antibodies) Visual acuity Intravenous immunoglobulin Optical coherence tomography Color vision testing

AIR ¼ autoimmune retinopathy. aAll experts agreed that steroids and conventional immunosuppressives should be considered as ﬁrst- or second-line treatment in the management of AIR patients.

Experts agreed that more evidence is needed when asked antiretinal antibody subtypes should have differential whether AIR patients should be treated with long-term weight toward the diagnosis of AIR, a simple majority immunomodulatory therapy. Subsequently, consensus consensus agreed that there is NOT enough evidence agreed that there is enough equipoise to justify randomized, for or against defining this as a criterion. For example, placebo-controlled trials to determine if AIR patients although the presence of 3 antibodies to different antigens should be treated with long-term immunomodulatory would be more meaningful than detecting an antibody to therapy. 1 antigen, we do not have enough evidence to determine Lastly, the follow-up testing and monitoring of AIR pa- whether the presence of 1 type of antibody (enolase) tients was surveyed. Consensus was reached for the would be more meaningful than another (carbonic anhy- following tests to be used to regularly follow AIR patients: drase) for the diagnosis of AIR. Consensus was not ERG, HVF, GVF, visual acuity, OCT, and color vision reached for whether an assay system should provide testing (Table 2). Experts agreed that follow-up time in- weight to the presence of antibodies against ‘‘yet un- tervals (for these tests) for AIR patients should be 3- known’’ antigens. Though multiple items regarding a diag- month intervals in general; however, it was acknowledged nostic assay system met consensus (simple or that monitoring may differ among patients depending on supermajority) (Table 3), multiple experts stated that the treatment types and clinical characteristics of AIR. more evidence and studies are needed before developing Consensus was not reached for repeated serum antiretinal a more complex, weighted assay system. antibody testing or DA testing in the follow-up of AIR Consensus was not reached on the ideal tissue type to patients. Utility of other potentially useful tests, such as use for fixation and detection of serum antiretinal anti- automated kinetic visual fields, was not investigated in bodies: 3 experts selected human tissue and 3 selected this survey, and although ERG was considered essential monkey tissue as the ideal tissue type. Experts suggested to the diagnosis and follow-up of AIR, consensus was that fresh normal human tissue would be ideal, but the not sought on the specific type of ERG to use for the diag- limited availability of human tissue may restrict its use nosis of AIR. in a standardized assay system. In addition, it was suggested that monkey tissue would allow for standardized BASIC LABORATORY SURVEY: Experts agreed that a fixation conditions and reproducible dissection for a stan- diagnostic assay system to detect antiretinal antibodies dardized assay system. Consensus was also not reached should have a 2-tier design. For example, experts suggested regarding optimal dilution of patient serum; however, it that Western blot (WB) or immunohistochemistry (IHC) was suggested that assays should be conducted for a range can be performed initially and subsequently followed by a of dilutions owing to multiple variables affecting antireti- different diagnostic method, including WB, IHC, or nal antibody titers. Experts acknowledged that standardi- enzyme-linked immunosorbent assay (ELISA), to maxi- zation of a diagnostic assay system to detect antiretinal mize sensitivity and specificity. Consensus was also reached antibodies is essential to better understand the role of on the methods to identify antiretinal antibodies; a antiretinal antibodies in the pathogenesis of AIR. consensus of experts agreed that IHC and WB should be The percentage of items achieving consensus between used to detect retinal proteins in the diagnosis of AIR, the first and second iterations of the expert surveys while a simple majority consensus agreed that ELISA remained stable and marginally increased overall (6.7%) should be used. (Figure 2). Percent agreement and the median including A simple majority consensus agreed that the number of interquartile range are included for Likert scale items positive antiretinal antibody subtypes should have more used in our study and can be found in the Supplemental weight toward the diagnosis of AIR. When asked whether Table (Supplemental Material available at AJO.com).

VOL. 168 CONSENSUS ON NONPARANEOPLASTIC AUTOIMMUNE RETINOPATHY 187 TABLE 3. Basic Laboratory Survey Summary: Detection of Antiretinal Antibodies for Diagnosis of Autoimmune Retinopathy

Basic Laboratory Survey Consensus Itemsa

Standardization of a diagnostic assay system is essential to understanding the role of antiretinal antibodies in the pathogenesis of AIR A 2-tier diagnostic assay system should be used to detect antiretinal antibodies Methods to identify antiretinal antibodies should include: Western blot, immunohistochemistry, or enzyme-linked immunosorbent assayb Number of positive antiretinal antibody subtypes identiﬁed should have more weight toward the diagnosis of AIRb There is not enough evidence for or against determining whether antiretinal antibody subtypes should have differential weight toward the diagnosis of AIRb

AIR ¼ autoimmune retinopathy. aIncludes items meeting simple or supermajority consensus. bItems meeting simple majority consensus.

Clinical Survey Basic Laboratory Survey Overall 100% 90% 80% 70% 60% 50% 40% 30% 20% 10%

Proportion of Survey Survey Items Proportion of 0% Iteration 1 Iteration 2 Iteration 1 Iteration 2 Iteration 1 Iteration 2

Consensus No Consensus

FIGURE 2. Percentage of expert survey items achieving consensus for the diagnosis and management of autoimmune retinopathy (AIR). Percentage of items achieving consensus between iteration 1 and 2 of the expert surveys remained stable and marginally increased overall (6.7%). (Supermajority consensus was used for the clinical survey, while simple majority consensus was used for the basic laboratory survey owing to the smaller number of items and experts surveyed.)

DISCUSSION This study indicates that consensus among experts can be used to establish clinical criteria for the diagnosis of AS AIR IS A RARE, COMPLEX DISEASE WITH PROTEAN CLIN- AIR (Table 1). We recommend these criteria be used for ical features and eventual vision loss, the natural history the diagnosis of AIR to allow for meaningful comparisons and response to treatment is variable. There are no practice of data between institutions, for creating a registry, and guidelines for diagnosis or management of AIR. In addi- for future prospective multicenter trials. As previously tion, the detection of antiretinal antibodies, whose pres- recognized, a developed standardized assay is important to ence is considered essential to the diagnosis by most, understanding the role of antiretinal antibodies in the lacks standardization. Both the diagnosis and management pathogenesis of AIR and correlation with clinical out- rely heavily on the individual physician’s opinion. Here, we comes.7,8 Such a standardized assay system for the used a modiﬁed Delphi process to facilitate communication detection of antiretinal antibodies has not yet been among experts to develop consensus and establish criteria established; however, efforts to develop a standardized 2- for the diagnosis of AIR. The lack of standardized criteria tier assay system for the detection of antiretinal antibodies for the diagnosis of AIR hinders our progress in understand- have been initiated as a result of this study. ing the pathophysiology of AIR, and developing such stan- Consensus was reached for core tests needed to make the dardized criteria is an imperative step toward gaining a diagnosis, treatment options, and follow-up guidelines. It is better understanding of the pathophysiology, diagnosis, important to note that the panel agreed that steroids, bio- and management of AIR. logic and nonbiologic immunosuppressives, and IVIG

188 AMERICAN JOURNAL OF OPHTHALMOLOGY AUGUST 2016 should be considered in the treatment of AIR. However, the taken to conduct the Delphi process in a rigorous manner. supermajority agreed that more evidence was needed for Anonymity, a core feature of the Delphi method, was main- considering long-term immunomodulatory therapy and tained throughout the process to limit interaction and po- that there is currently enough equipoise to justify random- tential biases. In addition, to ensure that the structure and ized, placebo-controlled trials to determine if AIR patients content of the survey did not impose preconceptions, ex- should be treated with long-term immunomodulatory ther- perts were encouraged to comment and provide feedback apy. As more cases of AIR are identified, future studies and at every survey item. An additional limitation in the collaborations will be needed to validate the proposed diag- Delphi process is the limit of time and low response rates. nostic criteria. Further, the diagnostic criteria and the assay Because we could not exhaustively survey all aspects in systems for detection of antiretinal antibodies should be the diagnosis and management of AIR, we carefully formu- improved as the pathogenesis and natural history of AIR lated each iteration to include the most necessary items to are further characterized in the future. gauge consensus while maximizing the response rate. To do Though the Delphi process has great utility to develop so, we initially conducted the AUS survey of 54 uveitis spe- guidelines for diseases where clinical evidence is lacking, cialists and held the AIR Workshop to identify the most there are limitations to such a study. It is a possibility necessary issues, and as a result, our response rate for all sur- that experts erroneously reinforce incorrect concepts. For veys was very high (97%), allowing us to maximize the this reason, it is essential to select qualified experts with Delphi process. utmost experience and to conduct the process in a rigorous Though steps may be taken to conduct a rigorous and manner. We selected experts from diverse institutions who efficient Delphi study, we recognize that incorrect concepts we considered highly qualified in uveitis and experienced may inevitably be propagated. We acknowledge that the in AIR. For the panel, an invitation was initially extended criteria for the diagnosis of AIR are not definitive, but we to over 50 clinicians and researchers from over 20 different encourage all to use these criteria as a standard to allow institutions, and only those with significant exposure to collaboration and gain a better understanding of AIR. It this rare disease who attended the AIR Workshop were is our hope that a larger and more diverse group of experts considered. In addition, to limit the potential to reinforce will participate in similar consensus studies in the future to incorrect concepts throughout the process, great care was ultimately improve the diagnosis and management of AIR.

FUNDING/SUPPORT: THIS WORK IS SUPPORTED BY THE NEI INTRAMURAL RESEARCH PROGRAM, THE NIH OFFICE OF RARE DIS- eases Research (ORDR), and the NIH Medical Research Scholars Program. The NIH Medical Research Scholars Program is a public-private partnership supported jointly by the NIH and generous contributions to the Foundation for the NIH from Pﬁzer Inc, The Doris Duke Charitable Foundation, The Newport Foundation, The American Association for Dental Research, The Howard Hughes Medical Institute, and the Colgate-Palmolive Company, as well as other private donors. For a complete list, please visit the Foundation website at: http://fnih.org/work/education-training-0/medical-research- scholars-program. Financial disclosures: Janet L. Davis, Consultant: AbbVie Inc (Chicago, IL); Debra A. Goldstein, Consultant: Bausch & Lomb (Bridge- water, NJ), Xoma (Berkeley, CA), Clearside Biomedical (Alpharetta, GA), Independent Data Monitoring Committee: AbbVie Inc (Chicago, IL); Careen Y. Lowder, Data and safety monitoring committee: Allergan (Dublin, Ireland); Richard W. Lee, Consultant: Roche-Genentech(South San Francisco, CA), EMD Serono (Darmstadt, Germany); Barbara Detrick, Consultant: Siemens Healthcare Diagnostics Inc (Tarrytown, NY), Abbott Laboratories (Chicago, IL). The following authors have no ﬁnancial disclosures: Austin R. Fox, Lynn K. Gordon, John R. Heckenlively, Robert B. Nussenblatt, Nicholas J. Butler, Monica Dalal, Thiran Jayasundera, Wendy M. Smith, Grazyna Adamus, Chi-Chao Chan, John J. Hooks, Catherine W. Morgans, and H. Nida Sen. All authors attest that they meet the current ICMJE criteria for authorship.

REFERENCES 6. Faez S, Loewenstein J, Sobrin L. Concordance of antiretinal antibody testing results between laboratories in autoim- 1. Mizener JB, Kimura AE, Adamus G, Thirkill CE, Goeken JA, mune retinopathy. JAMA Ophthalmol 2013;131(1): Kardon RH. Autoimmune retinopathy in the absence of can- 113–115. cer. Am J Ophthalmol 1997;123(5):607–618. 7. Adamus G, Wilson DJ. The need for standardization of anti- 2. Ferreyra HA, Jayasundera T, Khan NW, He S, Lu Y, retinal antibody detection and measurement. Am J Ophthal- Heckenlively JR. Management of autoimmune retinopathies mol 2009;147(3):557. author reply 557-558. with immunosuppression. Arch Ophthalmol 2009;127(4): 8. Forooghian F, Macdonald IM, Heckenlively JR, et al. 390–397. The need for standardization of antiretinal antibody 3. Sen HN, Nussenblatt RB. Autoimmune retinopathies. In: detection and measurement. Am J Ophthalmol 2008; Ryan, ed. Retina. 5th ed. London: Elsevier; 2013:1381–1389. 146(4):489–495. 4. Braithwaite T, Holder GE, Lee RW, Plant GT, Tufail A. 9. Grange L, Dalal M, Nussenblatt RB, Sen HN. Autoimmune Diagnostic features of the autoimmune retinopathies. Autoim- retinopathy. Am J Ophthalmol 2014;157(2):266–272.e1. mun Rev 2014;13(4-5):534–538. 10. Jones J, Hunter D. Consensus methods for medical and health 5. Larson TA, Gottlieb CC, Zein WM, et al. Autoimmune reti- services research. BMJ 1995;311(7001):376–380. nopathy: prognosis and treatment. Invest Ophthalmol Vis Sci 11. Dalkey N. The Delphi Method: an experimental study of 2010;51. ARVO E-Abstract 6375. group opinion. Rand Corporation 1969. RM-5888-PR.

VOL. 168 CONSENSUS ON NONPARANEOPLASTIC AUTOIMMUNE RETINOPATHY 189 Available at http://www.rand.org/content/dam/rand/pubs/ of a clinical evidence base utilizing informatics tools research_memoranda/2005/RM5888.pdf. Accessed March 21, and techniques. Methods Inf Med 2013;52(3):259–265. 2015. S1-6. 12. Lee PP, Sultan MB, Grunden JW, Cioffi GA. IOP Consensus 14. Ferris FL 3rd, Wilkinson CP, Bird A, et al. Clinical classifica- Panel. Assessing the importance of IOP variables in glaucoma tion of age-related macular degeneration. Ophthalmology using a modified Delphi process. JGlaucoma2010;19(5): 2013;120(4):844–851. 281–287. 15. Behrens A, Doyle JJ, Stern L, et al. Dysfunctional tear syn- 13. Trusko B, Thorne J, Jabs D, et al. The Standardization drome: a Delphi approach to treatment recommendations. of Uveitis Nomenclature (SUN) Project. Development Cornea 2006;25(8):900–907.

190 AMERICAN JOURNAL OF OPHTHALMOLOGY AUGUST 2016