Early azathioprine in acute Vogt-Koyanagi-Harada disease: a prospective 24-month follow-up study with multimodal imaging and electroretinogram

Marcelo Mendes Lavezzo (  mmlavezzo@gmail.com ) Hospital das Clínicas da Faculdade de Medicina da Universidade de São Paulo Viviane Mayumi Sakata Hospital de Clínicas Universidade Federal do Paraná Fernanda Maria Silveira Souto Hospital das Clínicas da Faculdade de Medicina da Universidade de São Paulo Ruy Felippe Brito Gonçalves Missaka Hospital das Clínicas da Faculdade de Medicina da Universidade de São Paulo Priscilla Figueiredo Campos da Nobrega Hospital das Clínicas da Faculdade de Medicina da Universidade de São Paulo Maria Kiyoko Oyamada Hospital das Clínicas da Faculdade de Medicina da Universidade de São Paulo Carlos Eduardo Hirata Hospital das Clínicas da Faculdade de Medicina da Universidade de São Paulo Joyce Hisae Yamamoto Hospital das Clínicas da Faculdade de Medicina da Universidade de São Paulo

loss as well as imposes a better control of in ammation in CS refractory/intolerant cases (1,17). More recently, several studies propose early IMT associated with high-dose CS as a step towards preventing CS long-term deleterious side effects and avoiding chronicity (18)(19)(20).
In this context, multimodal imaging of posterior segment in ammatory signs, including uorescein angiography (FA), indocyanine green angiography (ICGA) and enhanced depth imaging optical coherence tomography (EDI-OCT), brought light to subclinical choroiditis in non-acute VKHD (4,13,21). There are to date few systematic studies evaluating the relevance of this subclinical choroiditis on visual function outcomes and consequently on tailoring treatment.
Our previous prospective study, including 12 patients with acute onset VKHD treated with initial high-dose CS and slow tapering associated with late IMT, demonstrated an optimal control of acute in ammatory signs with VA ≥20/25 in more than 90% of patients. Nevertheless, 6 months after disease onset and during the 24-month follow-up, anterior uveitis relapse was observed in 33.3% of patients and subclinical choroiditis was present in 91.7% of patients. Full-eld electroretinogram (ffERG) parameters were stable during the follow-up in up to 71% of eyes despite the presence of subclinical choroiditis (13). Based on these previous results and on recent strong indication of early IMT in acute VKHD, the present study aimed to evaluate the outcome of high-dose corticosteroids with early addition of azathioprine (AZA) in patients with acute VKHD followed for 24-month with systematic multimodal and electroretinogram exams.

Subjects And Methods
This prospective study included 15 consecutive patients with acute onset VKHD (22) attended at Uveitis Service, Hospital das Clínicas HCFMUSP, Faculdade de Medicina, Universidade de Sao Paulo, Sao Paulo, SP, Brazil, from September/2015 to February/2018. All patients were followed for at least 24 months.
Patients who had other causes of uveitis, media opacities, refractive errors >5 diopters, and/or systemic contraindication to IMT were excluded. The local institutional review board approved the full protocol, which follows the tenets of the declaration of Helsinki (clinical trials NCT03399175).
All patients were treated with a three-day course of intravenous methylprednisolone (1000 mg/day) pulse therapy, followed by a 1.0 mg/kg/daily oral prednisone with slow taper during at least 9 months (Supplemental Table 1). In case of persistent clinical in ammation, CS tapering was postponed. Oral azathioprine (2 mg/kg/day; maximum 3 mg/kg/day) was added up in the four rst months after pulse therapy. Azathioprine was switched to mycophenolate mofetil (2 g/day; maximum 3 g/day) due to intolerance and/or refractory disease, in spite of optimum azathioprine dosage.

Multimodal analysis
Best-corrected visual acuity (BCVA), measured with the Snellen chart, was converted to LogMAR values. Systematic clinical and multimodal imaging exams were performed at inclusion, 1, 2, 4, 6 months (M), and, thereafter, every 3 months. Multimodal imaging examinations included fundus color photography (Topcon TRC-50DX, Tokyo, Japan), FA, ICGA and EDI-OCT (Spectralis® HRA+OCT, Heidelberg Engineering, Germany). The ffERG recording was performed at inclusion, 30 days and at 6-month intervals thereafter. An additional ffERG recording was carried out based on the presence of subclinical signs, as described below.
Fundus ndings were graded according to a previously described analytical framework (23). The development of SGF was classi ed as either present or absent, based on the fundus color photography.
FA and ICGA reading charts were based on an angiographic scoring system for uveitis (24). For the followup, we also evaluated dark dots (DD) alone, since they are the most constant and easily recordable ICGA sign (25). The total DD score varied from 0 to 8 (24). The uctuation with worsening was de ned as a change in DD score ≥0.5 between three consecutive time points (24). ffERG recordings were performed following the ISCEV guidelines with the RETI-port system (Roland Consult®; Electrophysiological Diagnostic Systems, Brandenburg, Germany) using ERG-jet electrodes (Universe SA, La Chaux-de-Fons, Switzerland)(26). The recording sequence was scotopic rod response, scotopic cone-rod combined response, scotopic oscillatory potentials, photopic cone response and 30-Hz icker cone response. The normal value limits for each speci c ERG response were established at 95% con dence intervals. Amplitude of a and b waves of eight averaged stimulus responses for each phase were analyzed. The results were compared with a normative database, consisting of 30 healthy subjects (60 eyes) matched for age and sex. During the follow-up, value reduction ≥30% (con rmed in two sequential exams) (27) in any ffERG parameter in consecutive exams indicated an increase of IMT. A subnormal ffERG recording was de ned as the presence of at least one parameter below the 5th percentile of healthy age, gender-matched controls. ffERG measurements stability or worsening obtained at 12 and 24 months were compared; worsening was de ned as at least 31% change in any of the parameters, except for icker, in which at least a 44% change was considered(28, 29) For this study implicit time was not included for the analysis (30).
After the acute in ammatory phase (6 months following disease onset), the disease activity was systematically evaluated based on both clinical and subclinical signs. Clinical signs included anterior chamber cells (ACC) of at least 1+ cells(31); exudative retinal detachment; choroidal neovascularization (CNV) revealed by fundus examination and/or OCT and/or FA, and macular edema con rmed on FA and/or OCT. Subclinical signs included optic disc leakage/hyper uorescence and/or perivascular leakage on FA; DD persistence or score uctuation with worsening on ICGA and subfoveal choroidal thickness (CT) increase (de ned as an increase ≥30%) on EDI-OCT.
In this prospective study protocol, the need for systemic IMT increment was based solely on the presence of clinical signs and/or worsening of ffERG exam. Isolated subclinical signs of activity did not infer a change in treatment schedule, but the patient was directed to the ffERG exam. If ffERG parameters worsened as described above, IMT was increased. The study ow chart is demonstrated in Supplemental Figure 1. Based on the prede ned protocol, the term " are" was used to describe the appearance or worsening of in ammatory signs observed at least 6 months after disease onset.
During the 24-month follow-up period, the presence of ocular complications was recorded. CNV was diagnosed on FA and/or on OCT and was treated with an intravitreal bevacizumab injection and IMT increment(32).
The primary outcomes were clinical and subclinical ares, BCVA and ffERG results at 24 months of follow-up. The frequency of ocular complications was observed, as a secondary outcome as well as CSE, de ned as a prednisone dose ≤10 mg/day while maintaining clinically inactive uveitis.

Statistical analysis
For the descriptive data analysis, medians, means and standard deviations were calculated for numerical variables and absolute/relative frequencies were presented for qualitative variables (Table 1). Patients had both eyes included for analysis, therefore, generalized estimation equations (GEE) were used to evaluate the difference between repeated measurements using the appropriate correction for inter-eye dependency (Tables 2 and 3, Supplemental Table 2). GEE with Poisson's distribution and identity link function, likelihood ratio, Kruskal-Wallis test and Fisher exact test were used when comparing different intervals to start treatment. (Table 3 and Supplemental Table 2). P values ≤0.05 were considered signi cant. Data analysis and statistical tests were conducted using SPSS20.0 (SPSS Science, Chicago, IL, USA).
Oral prednisone was used for a median time of 12 months (range: 9-22 months). A daily prednisone dose ≤10 mg was achieved in all 15 cases within a median of 8 months (range: 4-17 months). Four patients (26.7%) had to switch from azathioprine to mycophenolate mofetil due to refractoriness (2 cases) and intolerance (gastrointestinal symptoms and pharmacodermia, one case each). The median interval to this switch was 223.5 days (range: 124-252 days). All patients were still under IMT after 24 months.
During the follow-up, 5 patients (9 eyes, 30%) had anterior uveitis are and/or macular edema. All these patients had in ammation controlled with IMT increment, except one patient (2 eyes, 6.7%) who had refractory anterior uveitis are with CNV at month 18 of follow-up in spite of IMT increment. All eyes persisted with subclinical in ammation during the follow-up: DD were still present in all eyes at M24 and DD score uctuation with worsening was present in 24 eyes (80%), CT increase was detected in 15 eyes (50%) and perivascular leakage on FA was present in 17 eyes (56.7%). Choroidal folds were observed in the acute phase in ve eyes (16.7%). An illustrative case is demonstrated in Figure 1.
Regarding ocular complications, during the follow-up, cataract was observed in seven eyes (23.3%); ocular hypertension in 14 eyes (46.7%); subretinal brosis in 11 eyes (36.7%); and, CNV in four eyes of 2 patients (13.3%). One patient with CNV was under azathioprine and another patient under mycophenolate mofetil therapy at the time of CNV diagnosis. At the 24-month evaluation, SGF was observed in 18 eyes (60%). No patients had cataract extraction or other ocular surgeries during the follow-up.
Concerning ffERG evaluation, all parameters improved within 12 months of follow-up but remained subnormal ffERG in 23 eyes (76.7%) at the M24 follow-up point. Comparing ffERG parameters of M12 and M24 points, they remained stable or improved in 20 eyes (66.7%) and they worsened in at least one parameter in 10 eyes (33.3%). (Table 2 and Supplemental Table 2) Some correlations between clinical and functional data were observed. Among the 18 eyes with SGF, 17 eyes (94.4%) had subnormal ffERG at the M24 evaluation (p=0.022). Fibrosis was more frequently observed in eyes that had anterior uveitis are (p=0.007). Higher number of cells in CSF was observed in the worsening ffERG group than in the stable ffERG group (median 262.5 cells versus 32 cells, respectively, p=0.011). The presence of anterior uveitis are and/or macular edema was associated with a higher FA score at M12 (p=0.022) and M24 (p=0.002) and a higher DD score at M6 (p=0.002).
Evaluating the impact of very early treatment (CS <14 days and therapeutic dose IMT <4 months), it was observed that an earlier treatment was related with a lower ICGA score at M1 (p=0.012) and a lower DD score at M1 (p=0.047). Besides that, an earlier treatment was related to a faster period to reach CSsparing effect (p=0.018). (Table 3)

Discussion
In this prospective study, 15 patients with acute onset VKHD were treated with early AZA in association with high-dose corticosteroid. Acute in ammatory signs rapidly receded with BCVA recovery of ≥20/40 in 90% of eyes at one month of follow-up and, at nal follow-up, all eyes had BCVA ≥20/25. Chronic refractory anterior uveitis was observed in one patient (2 eyes, 6.7%) in spite of IMT increment. ffERG parameters remained stable in 66.7% of the eyes despite persistence of subclinical signs of in ammation in all included eyes.
Immunosuppression therapy as rst-line treatment in acute VKHD has been proposed to achieve faster control of the uveitis and to facilitate earlier tapering of the corticosteroids(18). However, there are very few studies that systematically evaluated visual and in ammatory outcomes in acute VKHD with early IMT. In 2012, Abu El-Asrar et al. described the effectiveness of early IMT in acute VKHD(33). These authors used mycophenolate mofetil with high-dose corticosteroid as a rst-line therapy in 19 acute VKHD patients and compared their results with a historical group of 68 acute VKHD patients treated with steroid monotherapy with a mean follow-up of 27 months (range, 16-54 months)(33). The early IMT group had better results than the CS group: visual acuity of 20/20 (74% versus 38%); recurrent in ammation (3% versus 18%); complications such as SGF (0% versus 100%)(33). In 2017, the same group con rmed their results by evaluating the long-term effectiveness (mean follow-up period of 37 months; range, 9-120 months) of mycophenolate mofetil (2 g/day) as a rst-line therapy associated with CS in 38 prospectively followed patients(33, 34). Other authors did not present such a clear-cut difference between rst-line IMT with CS and CS alone or with late IMT. Chee et al, in a retrospective study, compared outcomes of patients with IMT before 6 weeks (n=15) and after 6 weeks of disease onset (n=14) and did not nd a difference in SGF prevalence and in the proportion of eyes with resolved or chronic recurrent uveitis (19). Visual acuity was signi cantly better in the early IMT patients at the 4 th year of follow-up. Chee et al. also compared data of patients with IMT within 3 months from disease onset (n=29 patients) and patients with CS only or with late IMT (n=60 patients) and did not nd differences in SGF prevalence; the proportion of eyes with resolved (41.4% versus 33.3%), with chronic (51.7% versus 28.3%) or chronic recurrent (6.9% versus 38.4%) indicated signi cantly more eyes became chronic but fewer chronic recurrent among those with IMT within 3 months. These authors also found signi cantly better visual acuity in patients with IMT within 3 months at the 3 rd year of follow-up (19). Recently a prospective, multicenter, randomized and non-inferiority trial was conducted in Japan comparing the e cacy and safety of a combination therapy of prednisolone and cyclosporine (n=34 patients) and corticosteroid pulse therapy (n=36 patients) in VKHD with a 1 year follow-up. These authors observed lower recurrence/worsening risk, sunset glow fundus grade, and cataract rate in the combination group than in the corticosteroid group (35).
The following considerations could be made, when comparing our data with those previous studies.
Firstly, the timing of introduction of IMT may vary from a very early treatment, de ned as within 2 weeks from disease onset(36) to an interval up to 6 months (19,35,37,38). The rapid control of acute in ammatory signs depends primarily on the early use of high-dose systemic corticosteroid followed by its slow tapering(37). Conventional IMT takes at least 2 to 3 months to reach its ideal therapeutic action (17); therefore, the IMT introduction interval should not impact signi cantly on the outcomes. In our study, 13 patients (86.7%) achieved the AZA therapeutic dose within 4 months. Indeed, in the present study the comparison between patients treated with very early treatment (CS within 14 days and therapeutic dose of AZA within 4 months) and those patients treated otherwise (CS>14 days and/or therapeutic dose of AZA after 4 months) pointed out that very early group had a lower DD score at M1 than the other group and reached CSE in a shorter median time. We did not nd differences in other outcomes including in ammatory or functional parameters. In the literature, early IMT has already shown its bene ts on CSE(33, 34, 37, 40).
Secondly, besides azathioprine and mycophenolate mofetil, other drugs have been used, e.g., methotrexate, cyclosporine, cyclophosphamide and biologics, with slight differences in CSE and frequency of ocular complications (18, 37, 40-45). The most frequent IMT used in VKHD patients by Chee et al. was azathioprine, followed by methotrexate and mycophenolate mofetil (19). Concerning the differences in antimetabolites prescribed, azathioprine is approved by the Public Brazilian Healthcare System to be used in non-infectious uveitis, while mycophenolate mofetil is not. Both immunosuppressive drugs are effective in controlling non-infectious uveitis (17,40,44).
Thirdly, study designs differ. The unique characteristics of the present study are as follows: prospective with all included patients with the same 24-month follow-up period; clinical, multimodal and electroretinogram exams at prede ned intervals and treatment management based on clinical in ammation and on electroretinogram parameters. On the other hand, most studies analyzed, as main outcomes, visual acuity, recurrences based on clinical signs and ocular complications. Few studies included systematic multimodal imaging, but the results are not described (33, 34, 37, 38). An exception is Chee´s study in which choroidal thickness decreased over time and average ICGA score decreased throughout the follow-up period and, at the 5-year follow-up, 2 patients had persistent DD (19). Most studies had varied follow-up periods after the minimum initial six months.
Concerning subclinical signs in the non-acute phase of VKHD, they may indicate choroidal in ammation; however, in our study, even after 24-month of follow-up, we could observe that they did not impact in visual function when evaluated by VA, and, in at least two thirds of our patients, when evaluated by ffERG. It should be noted that subclinical signs, more importantly those observed on ICGA, are di cult to quantify and interpret. Herbort (47). On the other hand, Herbort et al. strongly suggest that monitorization of VKHD through ICGA is essential to avoid chronicity and sunset glow fundus(36). Therefore, subclinical signs may be important to a global understanding of the in ammatory status of the disease but, if presented isolated, may be not enough to dictate treatment intervention. Nevertheless, our study pointed out that a portion of the 33% of patients with worsening ERG could deserve a more aggressive treatment. Thus, biomarkers of disease severity and the understanding of timing for being more aggressive should be pursued. Our study is the rst prospective study with systematic multimodal and ffERG evaluation and early IMT; therefore, further studies are compelling to still clarify these points.
Despite the prospective study design with a systematic follow-up, some limitations should be considered. Firstly, a larger cohort sample size and a longer follow-up could add statistical power to detect differences. Secondly, ffERG was used as an objective method to evaluate retinal function and ffERG parameters were de ned to compare groups. Nevertheless, the de nition of retinal dysfunction progression was based on other in ammatory and non-in ammatory retinal/choroidal diseases (27)(28)(29) due to very scarce literature using ERG in VKHD. Furthermore, ffERG is a complex exam, which is not always available and the de nition of worsening or not should be validated in future studies (Supplemental Table 2). Thus, besides ffERG, other means of evaluating the impact of in ammation on visual function may play a role in VKHD appraisal, e.g., visual eld, microperimetry, contrast sensibility and self-reported quality of life questionnaires (47)(48)(49). It is important to highlight that all of these methods are subjective and their interpretation should be done cautiously. Thirdly, no control group using late IMT was presented to compare with the early IMT group. Nevertheless, our group recently published a study with a similar prospective design using corticosteroid only or late IMT in acute VKHD (13). Lastly, even though patients were recruited from early disease onset, there is heterogeneity in the interval to diagnosis and in disease severity.
In conclusion, early azathioprine with high-dose corticosteroid was effective in improving BCVA with control of clinical in ammation in more than 90% of patients. Isolate subclinical choroidal in ammation persistence may not be enough to indicate treatment increment, since VA was 20/25 in 100% of patients and visual function evaluated by ffERG remained stable in two thirds of patients, even after 24 months of follow-up. However, selected patients, particularly those with worsening ffERG, may deserve further treatment.

Declarations Ethics approval
All procedures performed in this study were in accordance with the ethical standards of the institutional research committee (CAPPesq 0496/11) and the 1964 Helsinki declaration and its later amendments (clinical trials NCT03399175).

Consent to participate
Informed consent was obtained from all individual participants included in the study.

Consent for publication
Not applicable.

Availability of data and material
The authors con rm that the data supporting the ndings of this study are available within the article and/or its supplementary materials.

Code availability
The authors declare that data analysis and statistical tests were conducted using SPSS20.0 (SPSS Science, Chicago, IL, USA).

Competing interests
The authors declare that they have no con ict of interest.