Development and Evaluation of Diagnostic Criteria for Vogt-Koyanagi-Harada Disease

Key Points

Question  Can improvements be made in a currently used set of criteria to diagnose Vogt-Koyanagi-Harada (VKH) disease?

Findings  In a case-control study including 1257 patients, the Diagnostic Criteria for VKH Disease (DCV) were developed using latent class analysis on 634 patients with VKH disease and 623 patients with non-VKH uveitis. The DCV were compared with another set of samples containing 537 patients with definite VKH disease and 525 patients with non-VKH uveitis and showed a higher sensitivity and negative predictive value compared with the Revised Diagnostic Criteria for VKH Disease.

Meaning  These results suggest that use of the DCV could result in an improvement in the sensitivity and negative predictive value to diagnose VKH disease.

Importance  To our knowledge, a set of well-defined diagnostic criteria is not yet developed for the diagnosis of Vogt-Koyanagi-Harada (VKH) disease.

Objective  To develop and evaluate a set of diagnostic criteria for VKH disease using data from Chinese patients.

Design, Setting, and Participants  This case-control study reviewed medical records of patients from a tertiary referral center between October 2011 and October 2016. Data from 634 patients with VKH disease and 623 patients with non-VKH uveitis from southern China were used to develop the Diagnostic Criteria for VKH Disease (DCV). Data from an additional group of 537 patients with a definite VKH disease diagnosis and 525 patients with non-VKH uveitis from northern China were used to evaluate the diagnostic criteria.

Main Outcomes and Measures  Sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and receiver operating characteristic.

Results  Of the 1257 patients used to construct the DCV, 665 (52.9%) were male, and the mean (SD) age at disease onset was 38.6 (13.6) years. The 3-class model and 21 clinical findings were selected by latent class analysis. Variables with a high positive rate in the early-phase or late-phase VKH group or high specificity constituted essential parameters. Constellations of these essential parameters constructed the DCV. The sensitivity and NPV of the DCV were higher than those of the Revised Diagnostic Criteria for VKH Disease (RDC) (sensitivity: 94.6% vs 71.9%; difference, 22.7%; 95% CI, 18.5-27.0; NPV: 94.3% vs 76.6%; difference, 17.7%; 95% CI, 13.9-21.5). The specificity and PPV of the DCV were not different from that of the RDC (specificity: 92.2% vs 93.9%; difference, 1.7%; 95% CI, −1.4 to 4.8; PPV: 89.3% vs 92.3%; difference, 3.0%; 95% CI, −1.4 to 4.8). The area under the receiver operating characteristic curve of the DCV and the RDC were 0.934 (95% CI, 0.917-0.951) and 0.829 (95% CI, 0.803-0.855), respectively.

Conclusions and Relevance  The DCV were developed and evaluated using data from Chinese patients with VKH disease and showed a high sensitivity, NPV, and area under the receiver operating characteristic curve in comparison with the RDC. However, they were developed using a retrospective analysis and should be evaluated in prospective studies in other racial/ethnic populations.

Vogt-Koyanagi-Harada (VKH) disease is an immune-mediated disorder affecting both eyes. Although VKH disease does not have a criterion standard for diagnosis, several sets of diagnostic criteria have been proposed, including the American Uveitis Society criteria,¹ Sugiura criteria,^2,3 and the Revised Diagnostic Criteria for VKH Disease (RDC).⁴ The RDC are widely used at present and are highly sensitive and specific compared with the other 2 sets of criteria.^5-8 The RDC divide VKH disease into 3 categories—complete, incomplete, and probable—and do not allow a decision concerning a definite diagnosis for patients without any extraocular manifestation. However, a 2010 study⁹ showed that VKH disease may also manifest itself as an isolated intraocular inflammation. Another problem of the RDC concerns the presence of fundus features in all patients. However, the fundus of many patients with VKH disease in the late phase cannot be clearly seen because of severe media opacities. Additionally, patients whose disease has been appropriately treated early may not present with the typical fundus features. Hence, a diagnosis might be missed when using the RDC at the patient’s first visit. Therefore, we felt that it was necessary to develop another set of diagnostic criteria based on a large sample of Chinese patients with VKH disease.¹⁰

We retrospectively reviewed data from 634 patients who had been diagnosed as having VKH disease based on the RDC at the uveitis center in the First Affiliated Hospital of Chongqing Medical University, Chongqing, People’s Republic of China, between October 2011 and October 2016. Considering the potential influence of geographical factors on the profile of VKH disease, we used the criteria developed based on patients from southern China in the diagnosis of VKH disease in patients from northern China. In the development of the diagnostic criteria, all patients came from southern China, as demarcated by the Yangtze River because the weather, eating habits, and customs vary in the 2 parts of the country divided by the Yangtze River. Most of the patients were referred to us by specialists from other uveitis centers in southern China (eTable 1 in the Supplement). Complete, incomplete, and probable VKH disease defined by the RDC were all considered as VKH disease. They were grouped into an early-phase or late-phase group according to the RDC. A total of 623 patients with other uveitis entities served as controls. The diagnosis of the enrolled patients was initially made by at least 2 specialists in other hospitals and then confirmed by the uveitis specialists from the First Affiliated Hospital of Chongqing Medical University. The study adhered to the tenets of the Declaration of Helsinki. The Ethics Committee of the First Affiliated Hospital of Chongqing Medical University approved this study. Written informed consent was obtained from all included patients.

Clinical findings were recorded at the patients’ initial visit to the uveitis center of the First Affiliated Hospital of Chongqing Medical University, including the previous clinical findings in their medical records. The manifestations listed in the RDC and clinical findings that were not included in the RDC but might be important for diagnosis were also reviewed, including subjective symptoms, abnormal appearances disclosed by indocyanine green fluorescein angiography, standard coherent optical tomography (OCT), and enhanced depth imaging–OCT (EDI-OCT). Both standard OCT and EDI-OCT were performed using the Spectralis OCT (Heidelberg Engineering). Based on the results of a study using a large sample of Chinese people,¹¹ we adopted 370 μm as the upper limit of the normal choroidal thickness. These data were analyzed using a data mining method (latent class analysis [LCA]). Latent class analysis is a statistical method for identifying latent classes among subjects using given variables.¹² In this study, 3 presumptive models that had 2, 3, and 4 classes were used. The 2-class model classified patients as VKH disease and non-VKH disease. The 3-class model could be used to define early-phase VKH disease, late-phase VKH disease, and non-VKH disease. The RDC model, which groups patients into complete VKH disease, incomplete VKH disease, probable VKH disease, and non-VKH disease, is a 4-class model. By comparing the maximal likelihood, the best-fitting model could be selected. In this study, the 3-class model showed the best fit. In a 3-class model, the more important variables should be capable to differentiate early-phase or late-phase VKH disease from the other 2 classes or to differentiate VKH disease from non-VKH disease. By assigning the positive probability of each variable to different classes, important variables were identified.

For each selected variable, those with high specificity or a high positive rate (ie, rate of positive results in the derivation sample) in early-phase or late-phase VKH disease were considered as essential parameters that had to be met in making a diagnosis. We ultimately developed the Diagnostic Criteria for VKH Disease (DCV) based on the data of southern Chinese patients with uveitis whereby we also took the local conditions faced by ophthalmologists at the patients’ first visit to their clinic into account.

To evaluate the diagnostic efficacy of the DCV, another set of samples including 659 patients with VKH disease from northern China were reviewed. The patients were referred to the First Affiliated Hospital of Chongqing Medical University between October 2011 and October 2016 by specialists from uveitis centers or hospitals in northern China (eTable 1 in the Supplement).

A total of 537 of the 659 patients (81.5%) had a definite VKH disease diagnosis, as evidenced either by exudative retinal detachment, choroidal thickening, and early punctate staining and late subretinal dye pooling disclosed by fluorescein angiography (FFA) in the early phase or by definite sunset glow fundus and bilateral recurrent granulomatous anterior uveitis in the late phase. A total of 525 patients with a diagnosis of other non-VKH entities were included in the control group. These 537 patients with VKH disease and 525 patients with non-VKH uveitis were used to compare the sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) between the RDC and the DCV. In calculations of sensitivity, specificity, PPV, NPV, and receiver operating characteristic, complete, incomplete, and probable VKH disease defined by the RDC were all considered as a positive diagnosis, and non-VKH disease was considered a negative diagnosis in the DCV.

Demographic characteristics of the patients with VKH disease and the control patients with non-VKH uveitis were compared using Mann-Whitney U test tests and χ² test using SPSS version 20.0 (IBM). Normal distribution was tested using Kolmogorov-Smirnov test. R version 3.2.3 (The R Foundation) was used for running the LCA. The sensitivity, specificity, PPV, and NPV between the RDC and the DCV were compared using McNemar test. The receiver operating characteristic and all the aforementioned tests except the LCA were analyzed using SPSS. All the hypothesis tests were 2-sided, and a P value less than .05 was considered significant.

Of the 634 patients with VKH disease from southern China, 261 (41.2%) were grouped into the early-phase group and 373 (58.8%) into the late-phase group. The time duration from disease onset to the first visit to the First Affiliated Hospital of Chongqing Medical University ranged from 2 days to 41 years. The fundus of 88 patients (23.6%) with late-phase VKH disease could not be clearly seen because of severe cataract. Of the patients with VKH disease with a clearly visible fundus, fundus photography was performed in 546 of 546 patients (100%); OCT, 537 (98.4%); FFA, 475 (87.0%); and indocyanine green fluorescein angiography, 320 (58.6%). Extraocular manifestations were observed in 561 of 634 patients (88.5%) with VKH disease. The most frequent extraocular findings were neurologic manifestations (342 patients [53.9%]), followed by poliosis (227 [35.8%]), tinnitus (225 [35.5%]), alopecia (223 [35.2%]), hearing loss (219 [34.6%]), and vitiligo (99 [15.6%]).

The demographic characteristics of patients with VKH disease and control patients are summarized in Table 1. There was no difference between the 2 groups. The components of the control group are listed in Table 2.

A total of 37 variables were subjected to LCA analysis. The 3-class model was found to show the best fit. Twenty-one variables were selected as the important variables. Of the selected clinical findings, the positive rates in the early-phase VKH, late-phase VKH, and non-VKH uveitis groups as well as the sensitivity, specificity, PPV, and NPV are presented in Table 3.

Because of the very similar clinical appearances and laboratory findings between VKH disease and sympathetic ophthalmia, the criteria had an absolute requirement for the exclusion of a history of penetrating ocular trauma or intraocular surgery preceding the onset of uveitis. The criteria also excluded infectious uveitis or accompanying systemic rheumatic diseases or other ocular disease entities and the presence of bilateral ocular involvement within 2 weeks.^13-15 The clinical findings with high positive rates in early-phase or late-phase VKH disease or with high specificity were considered as essential parameters to be included in the diagnostic criteria. To make a diagnosis of typical early-phase VKH disease, diffuse choroiditis and exudative retinal detachment should be present. In patients without clinically apparent exudative retinal detachment, at least 1 of the 2 variants should be met. One of them was serous retinal detachment found by OCT or B-scan ultrasonography, coinciding with choroidal thickening as identified by EDI-OCT. The other was the presence of early punctate staining and late subretinal dye pooling, as disclosed by FFA. In patients who had been treated with systemic corticosteroids or with immunosuppressive agents elsewhere, hyperfluorescence of the optic disc disclosed by FFA was considered as an essential parameter. To make a diagnosis of late-phase VKH disease, 1 of the 3 following variants should be met. The first was the presence of definite sunset glow fundus or retinal pigment epithelium clumping/migration in combination with bilateral recurrent granulomatous anterior uveitis. The second variant was the presence of bilateral recurrent granulomatous anterior uveitis plus Dalen-Fuchs nodules or multifocal chorioretinal atrophy or window defects/moth-eaten fluorescence disclosed by FFA in patients without sunset glow fundus or visible pigment alternations owing to prompt and appropriate treatment soon after disease onset. The third variant was the presence of clinical findings capable of diagnosing early-phase VKH disease from their prior records combined with bilateral recurrent granulomatous anterior uveitis in patients without visible fundus due to significant media opacity. The extraocular manifestations were not considered essential parameters. The DCV are summarized in the Box, in a form of constellations of the essential parameters.

Box Section Ref ID

Records of another set of samples in which the patients were from hospitals from northern China were included for an evaluation of the DCV. Of the 537 patients with VKH disease who had a definite diagnosis, 216 (40.2%) were grouped into the early-phase group and 321 (59.8%) into the late-phase group. The time duration from disease onset to the first visit to the First Affiliated Hospital of Chongqing Medical University ranged from 3 days to 40 years. The fundus of all these patients could be clearly observed since they did not show media opacities.

The 537 patients with definite VKH disease and the 525 patients non-VKH uveitis were assessed against the RDC and DCV. The specificity and the PPV of the DCV were not different from those of the RDC (specificity, 92.2% vs 93.9%; difference, 1.7%; 95% CI, −1.4 to 4.8; P = .23; PPV, 89.3% vs 92.3%; difference, 3.0%; 95% CI, −1.4 to 4.8; P = .72). However, the sensitivity and the NPV of the DCV were higher than that of the RDC (sensitivity, 94.6% vs 71.9%; difference, 22.7%; 95% CI, 18.5-27.0; P < .001; NPV, 94.3% vs 76.6%; difference, 17.7%; 95% CI, 13.9-21.5; P = .02). The receiver operating characteristic of the 2 sets of criteria is shown in the Figure. The area under the receiver operating characteristic curve of the DCV was 0.934 (95% CI, 0.917-0.951) and of the RDC was 0.829 (95% CI, 0.803-0.855).

In this study, we developed and evaluated the DCV based on data from patients from southern and northern China. Higher sensitivity, NPV, and area under the receiver operating characteristic curve were observed in the DCV compared with the RDC. These results suggest that the DCV might have a lower rate of missed diagnosis and higher diagnostic efficiency in comparison with the RDC. eTable 2 in the Supplement shows the comparisons of the 2 sets of criteria. Compared with the RDC, the DCV do not use extraocular manifestations and cerebrospinal fluid findings as essential parameters and show similar specificity but a significantly higher sensitivity. Our findings are in agreement with an earlier study by Rao et al,⁹ which showed that exudative retinal detachment and sunset glow fundus are highly specific for VKH disease irrespective of the racial/ethnic background of the patients. Our study was carried out in the largest uveitis center in China (the First Affiliated Hospital of Chongqing Medical University), and referred patients had been diagnosed by at least 2 uveitis specialists in other hospitals and then reconfirmed by the uveitis specialists in Chongqing. To assure that the patients included in the evaluation step had a definite VKH disease diagnosis, patients without characteristic examination results were not included. Additionally, we included control patients with various uveitis entities resembling VKH disease either in inflammatory nature, the affected tissues, or the clinical manifestations in different phases. Since latent class modeling was introduced in medical research in 1982,¹⁶ data mining tools have been widely used to extract information from a data set to transform it into an understandable structure for further use.¹⁷ The LCA and the analogous algorithms were used in developing or validating classification criteria for systemic lupus erythematosus¹⁸ and psoriatic arthritis.¹⁹ As clinical manifestations vary in the course of VKH disease, the 3-class model of classifying the patients as having early-phase VKH disease, late-phase VKH disease, and non-VKH uveitis was adopted.

There are similar characteristics between the DCV and the RDC. The different manifestations in the early phase and late phase of VKH disease are taken into consideration by both criteria. Apart from these similarities, advanced techniques were emphasized in the DCV. Ophthalmologists can now readily recognize choroidal thickening and subtle serous retinal detachment using high-resolution OCT and EDI-OCT.^20-22 This technique was not yet available when the RDC were developed. Like the study by Rao et al,⁹ our previous study¹³ and the present study revealed that some patients presenting with a sunset glow fundus and nummular chorioretinal scars had a typical evolutionary process of VKH disease but did not develop extraocular manifestations during the disease course. The extraocular findings were not included in the final DCV.

It is worthwhile to point out that the RDC requires typical FFA manifestations for early-phase VKH disease. This is true in most cases without treatment. In fact, a large number of patients are already treated with systemic corticosteroids at local hospitals despite the unclear diagnosis before being referred to a tertiary uveitis center. Early use of systemic corticosteroids may have pronounced effects on the FFA appearances, especially with respect to the leakage and pooling.^9,23 However, the finding of optic disc staining is less influenced by a short period of treatment.²⁴ In this study, 15 patients without typical FFA appearance but with bilateral optic disc staining were diagnosed as having VKH disease according to the typical appearances during follow-up visits. Choroidal thickening and exudative retinal detachment also diminished rapidly following high-dose systemic corticosteroids.^25,26 Therefore, optic disc staining combined with early typical appearances of exudative retinal detachment are important indexes for the diagnosis of early-phase VKH disease after treatment. For the patients with subtle exudate retinal detachment that could be missed with an ophthalmoscope, the DCV provide 2 options. These include the characteristic features in FFA. Another choice is the use of OCT or B-scan ultrasonography in patients when they are not amenable to FFA. For the diagnosis of late-phase VKH disease, the depigmentation or pigment migration/clumping of the ocular fundus may not be detectable owing to complicated cataract in patients with recurrent anterior uveitis. In these patients, bilateral recurrent granulomatous anterior uveitis combined with records of clinical findings corresponding to early-phase VKH disease diagnosis may allow them to be diagnosed. Sunset glow fundus and characteristic pigment alternations also may not appear in some patients because of an early and appropriate treatment. In our recent study,²³ a normal fundus without sunset glow appearance or signs of pigment alternations was found in 46.3% of the patients who were treated immediately after disease onset. Therefore, the presence of bilateral recurrent granulomatous anterior uveitis plus window defects/moth-eaten fluorescence disclosed by FFA or plus Dalen-Fuchs nodule/multifocal chorioretinal atrophy are criteria enabling an ophthalmologist to make a diagnosis of late-phase VKH disease.

There are some limitations concerning the criteria presented in this study. The developed criteria were based on retrospective data of patients, and prospective studies are needed to evaluate the results. Although indocyanine green fluorescein angiography has been found to be valuable in the detection of some choroidal inflammatory vasculopathies,^27,28 it is not routinely used in China, presumably owing to accessibility and costs.

In conclusion, another set of criteria was developed using data from a large group of patients with VKH disease and non-VKH uveitis from southern China and evaluated using another patient data set from northern China. High diagnostic efficiency was found in the DCV. More studies are needed to ascertain whether these criteria can also be used in patients with VKH disease from other racial/ethnic populations.

Accepted for Publication: May 4, 2018.

Corresponding Author: Peizeng Yang, MD, PhD, the First Affiliated Hospital of Chongqing Medical University, Chongqing Key Laboratory of Ophthalmology and Chongqing Eye Institute, Youyi Road 1, Chongqing 400016, People’s Republic of China (peizengycmu@126.com).

Published Online: July 5, 2018. doi:10.1001/jamaophthalmol.2018.2664

Conflict of Interest Disclosures: All authors have completed and submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest and none were reported.

Funding/Support: This work was supported by grants 2016YFC0904000, 2016YFC0904001, 2016YFC0904002, and 2016YFC0904003 from the National Key R&D Program of China, CSTC grant 2008CA5003 from the Chongqing Key Laboratory of Ophthalmology, grant cstc2014pt-sy10002 from the Chongqing Science & Technology Platform and Base Construction Program, and grant cstc2017shmsA0503 from the Natural Science Foundation Project of Chongqing.

Role of the Funder/Sponsor: The funders had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.