Automatic Measurement of Choroidal Thickness with Swept-Source Optical Coherence Tomography for Clinical Follow-Up in Acute Vogt-Koyanagi-Harada Disease

Background: The course of acute Vogt-Koyanagi-Harada is typically assessed qualitatively using indocyanine green angiography. Swept-source optical coherence tomography may provide a safer, non-invasive, more objective approach to follow up. In this study, we assess the clinical value of the automated measurement capabilities of swept-source tomography to measure choroidal thickness. Design: Prospective, longitudinal case-control study at a tertiary university hospital. Participants: Nine patients with acute Vogt-Koyanagi-Harada disease (18 eyes) and 17 age-matched controls (34 eyes). Methods: Choroidal thickness (subfoveal area and ETDRS grid) was automatically measured with swept-source optical coherence tomography. Changes in thickness were compared to changes in visual acuity and indocyanine green angiography findings to check for correlations. Main outcome measures: Changes in choroidal thickness (micrometersμm) from baseline. Secondary measures included visual acuity and angiography. Results: At baseline, patients presented significantly greater mean (SD) subfoveal choroidal thickness (666.9 μm [258.3] vs. 302.3 [71.4]) and ETDRS grid choroidal thickness (648.7 μm [260.5] vs. 287.5 [69.3]) than controls (p=0.000). Choroidal thinning and improved vision were associated with treatment while increasing thickness and worsening vision were associated with posterior relapse. In 62.5% of recurrences in tomography, no changes in visual acuity were present; however, all recurrences diagnosed with tomography showed signs of inflammation on angiography. Conclusions: Automatic measurement of choroidal thickness with swept-source optical coherence tomography is a rapid, non-invasive manner of detecting posterior segment recurrences and treatment response in acute Harada patients. Swept-source tomography could reduce the need for angiography to monitor patients with Harada disease.


Introduction
Vogt-Koyanagi-Harada (VKH) syndrome is a bilateral granulomatous uveitis that typically presents with distinct clinical features based on the duration and stage of the disease [1,2]. The acute stage of VKH is characterized by diffuse choroiditis, multifocal areas of subretinal fluid and/or bullous serous retinal detachments, with or without neurologic (headaches, meningismus) or auditory (tinnitus, hypoacusia) symptoms. The convalescent stage of the disease develops 12 weeks after onset and is characterized by resolution of retinal detachments with disappearance of cells from the anterior chamber and the vitreous, with characteristic pigmentary changes in the macula and sunset glow fundus. The chronic/recurrent phase is characterized by clinical signs of disease activity in the anterior segment of the eye with anterior granulomatous uveitis and dermatologic signs (vitiligo, alopecia, poliosis).
Choroidal activity is usually evaluated with indocyanine green angiography (ICGA). Herbort et al. systematized the ICGA findings for follow-up, identifying 4 signs: hypofluorescent dark dots (the most important sign, present in the acute and convalescent stage, indicating stromal granulomas), hyperfluorescent choroidal vessels, fuzzy indistinct large choroidal vessels, and disc hyperfluorescence [3]. An important advantage of ICGA is that it can detect choroidal inflammation even when no clinical signs are present; for this reason, ICGA is recommended to diagnose occult choroidal recurrences [3]. Despite the benefits of ICGA-considered the gold standard for the diagnosis of posterior recurrence-this imaging modality presents several important drawbacks. ICGA is an invasive, time-consuming procedure requiring contrast injection. In addition, the results are qualitative (i.e., subjective) rather than quantitative, and thus it is not possible to quantify the degree of inflammation. Moreover, the most important sign of inflammation-hypofluorescent spots-are also commonly observed in choroidal atrophy, making it harder to judge the persistence or recurrence of choroidal inflammation.
The relatively recent development of enhanced-depth imaging optical coherence tomography (EDI-OCT), based on spectral-domain OCT, has enabled in vivo evaluation of the choroid, providing highdefinition cross-sectional images [4,5]. Numerous studies have used EDI-OCT to assess choroidal thickness in both normal and pathologic eyes [6][7][8][9][10]. However, because EDI-OCT is unable to detect the choroid-scleral interface in many cases (from 4-26% of eyes), choroidal margins must be manually identified-a highly cumbersome and subjective (due to inter-observer differences) process. For all these reasons, automatic segmentation of the choroid layer would be preferable [11][12][13]. The recent development of swept-source optical coherence tomography (SS-OCT) may offer an alternative approach to segmentation without the drawbacks of manual labeling. SS-OCT uses a 1 μm band light source that allows for deeper penetration into the retinochoroidal structures, providing increased resolution. SS-OCT is able to simultaneously display a focused image of both the retina and choroid, thus providing better visualization of retinal and choroidal changes, especially of the choroid-scleral interface, thus making it a reliable tool for measuring choroidal thickness [14,15].
Due to time and staffing constraints, manual measurements are typically impractical in daily clinical practice [16][17][18][19][20][21]. For this reason, we use SS-OCT at our centre to automatically measure choroidal thickness. Despite the advantages of automatic segmentation with SS-OCT, to our knowledge, this technique has not been previously used to prospectively monitor the course of disease in patients with acute VKH.
In the present case-control study, we prospectively assessed a series of 9 patients diagnosed with VKH. We used SS-OCT to measure choroidal thickness at baseline and at various time points over the 2year follow-up. Changes in choroidal thickness were compared to changes in visual acuity (VA), ICGA findings, and clinical symptoms to check for significant associations between these variables.

Methods
We prospectively recruited nine patients with acute VKH and 17 age-matched healthy controls. Inclusion criteria for the VKH patients included an established diagnosis of VKH at our institution according to the revised diagnostic criteria for VKH disease [16]. Exclusion criteria (both cases and controls) included any history of ocular disease or surgery, myopia or hypermetropia greater than ± 3 diopters [9,17,22,23], or any history of systemic disease (other than VKH) with ocular involvement.
Demographic characteristic were recorded before the clinical ophthalmic examination and OCT image acquisition. A complete ophthalmic examination was performed including best-corrected visual acuity (BCVA) using decimal fractions and Snellen equivalent in feet.
Written informed consent was obtained from all subjects in accordance with our institutional guidelines. This study adhered to the tenets of the Declaration of Helsinki. Institutional Review Board/Ethics Committee approval was obtained from the local Research Ethics Committee (CEIC of University Hospital of Bellvitge), reference number PR204/14.
ICGA was used to check for signs of choroidal inflammation and was performed when signs of anterior or posterior recurrence were present, or when an increase of choroidal thickness was detected on the SS-OCT with a simultaneous decrease in VA, or when the increase was ≥ 50 µm without changes in VA. Based on the standard deviation (SD) of measures of choroidal thickness reported in other studies [18][19][20][21], we performed ICGA in all patients who presented an asymptomatic increase of thickness ≥ 50 µm in order to check for signs of inflammation, a finding that would confirm that the increased choroidal thickness was due to relapse.

SS-OCT image acquisition protocol
We scanned the macular area of both the affected and healthy eyes with SS-OCT (Atlantis DRI OCT-1, Topcon, Japan) at a 1,050-nm wavelength and scanning speed of 100,000 A-scans/second using a high-definition set of twelve radial cuts of 12.0 mm each, horizontal scans, and a cube raster scan protocol (12.0 × 9.0 mm). Retinal foveal and choroidal thicknesses were measured automatically. Two choroidal areas were analyzed: the subfoveal choroidal thickness (SFCT) and the area of the Early Treatment Diabetic Retinopathy Study (ETDRS) grid centred on the fovea ( Figure 1). The SFCT and the mean of the 9 measures of the choroidal thickness of the grid (mean grid choroidal thickness: MGCT) were used as reference measures. The SS-OCT 3D scan produces a 12 × 9 mm retinal and choroidal thickness map of the macular area after automated segmentation of the retina and the choroid. The retinal and choroidal thickness maps were overlapped with the modified ETDRS grid (6 × 6 mm), thus obtaining automated measures of SFCT and MGCT. These measures were then compared with SFCT and MGCT measures obtained with SS-OCT in healthy age-matched controls both manually and automatically through the same SS-OCT protocols ( Table 2).
The number of follow-up consultations in VKH patients was individualized according to need. Based on previous experience [9,17,21,22] with EDI-OCT imaging, which suggests that the choroid could change in high myopia or hypermetropia, and that blood pressure could have diurnal variations, we excluded patients with refractive errors ≥ 3 diopters. In addition, blood pressure was measured at each consultation prior to examination to assure that it was within an acceptable range. Finally, all examinations were performed in the morning to assure consistency.

Definition of relapse
Changes in choroidal thickness were classified as a posterior recurrence when these changes were associated with diminished VA (excluding other causes) and/or with signs of inflammation on ICGA.

Statistical analysis
A descriptive analysis was performed. Measures of the central value (mean and median) and dispersion (SD, interquartile range [IQR]) were determined. For quantitative variables, the Student t test or the Mann-Whitney U test was used, as appropriate, to compare means. All statistical tests were two-tailed, and P values of <0.05 were deemed significant. All analyses were performed using Minitab 17.    Visual acuity  There were no statistically significant differences between the MGCT and the SFCT values in the study group (p=0.9114). Given that the SFCT is measured in the area of maximum vision, and that previously-reported studies have used this measure, we elected to use the SFCT as the reference value to monitor choroidal thickness during follow up.

Relapses
A total of 16 recurrences (increased choroidal thickness with or without loss of VA) were observed during the 2-year follow up period. In all cases, choroidal thickness was measured automatically by SS-OCT. All recurrences were confirmed by ICGA (signs of inflammation) ( Table 3). In 6 cases (37.5%), the relapse was accompanied by a worsening VA and in 10 eyes presented with an unchanged vision (62.5%); consequently, 62.5% of recurrences were diagnosed based on the SS-OCT findings.

Results
We prospectively recruited 9 patients (3 women and 6 men) with acute VKH (18 eyes) and 17 age-matched healthy controls (34 eyes). All subjects were followed for up to 2 years (range, 6 to 24 months; mean, 12.5 months). Follow up in the patient group was as follows: <1 year (four patients), 1 year (two patients), and two years (three patients).
All VKH patients presented with headaches and blurred vision in addition to other common clinical features of the disease (Table 1). Median age was 40 years (range, 37-53) ( Table 1). The control group consisted of 10 women (59%) and 7 men (41%), with a median age of 49 (range, 24-76 years). There were no significant differences between age in the two groups at baseline (p=0.328). Overall analysis of the study variables are shown in Table 2 (Table 4). Thus, 14 eyes experienced an improvement in vision at study completion while vision remained unchanged (maximum vision) in 2 eyes, and 2 eyes experienced a decrease in vision (from VA 1 at baseline to 0.7-0.8) at final follow up.
Using the choroidal thickness of the control group (220-360 µm) as a reference value, 17 eyes in the VKH group presented choroidal thickening at baseline. By study end, 10 eyes had recovered normal choroidal thickness, 6 showed a slight thickening, and 2 showed a notable thinning but without affecting VA (Table 4).

Discussion
Given the drawbacks of ICGA and EDI-OCT, it would be useful to have an alternative approach to assessing the course of VKH disease. In our center, we have been using the automated measurement feature of SS-OCT for the past several years to measure choroidal thickness in patients with VHK. SS-OCT offers a rapid, non-invasive, and objective method of detecting posterior segment recurrences and treatment response. The main aim of our study was to determine the clinical value of measuring choroidal thickness with the automatic segmentation capabilities of SS-OCT. Consistent with previous reports [7,8,18,19,[23][24][25][26], we found that patients with VKH had significantly thicker SFCT (666.9 μm) than healthy controls (302.3 μm). During follow up, increasing choroidal thickness and worsening VA were associated with posterior relapse. Importantly, in nearly two-thirds (62.5%) of eyes in which relapse was diagnosed by SS-OCT, no changes in VA were present. By contrast, all relapsed eyes showed signs of inflammation on ICGA. These findings suggest that SS-OCT may be a valuable adjunct to ICGA to rapidly, objectively, and non-invasively The minimum increase in SFCT associated with loss of VA was 31 µm (the maximum was 660 µm). In patients with a recurrence, the mean increase in SFCT was 167.25 µm. Interestingly, loss of VA was not directly correlated with the amount of increase in SFCT: for example, one patient experienced an increase of 660 µm in SFCT but only a 20% decrease in VA (from 1 to 0.8), whereas another patient had an increase of 31 µm that resulted in a 10% decrease in VA (from 1 to 0.9). Relapses occurred as soon as 1 month after disease onset to as late as 17 months after onset. The mean time to relapse was 8.81 months from onset.
At study completion, no recurrences were observed in patients with <1 year from diagnosis. VA and SFCT at baseline and at study completion (Table 4).
detect the likely presence of posterior segment recurrence and to evaluate treatment response in patients with acute VKH.
In the acute phase of VKH disease, bilateral granulomatous panuveitis is present with diffuse choroiditis and multifocal exudative retinal detachments. While EDI-OCT allows for visualization and measurement of the choroidal thickness, it may be difficult to delineate the outer edge of the choroid with this technique. For this reason, the thickness must be measured manually-a time-consuming task that makes this approach prohibitive in many case [18,23,24,27,28]. The benefit of these studies is that they have confirmed the presence of choroidal thickening-which may be related not only to inflammatory infiltration but also increased exudation-in acute VKH. Importantly, however, those studies did not evaluate choroidal thickness over time as a follow-up measure [18]. Although the manual approach to measuring choroidal thickness provide valuable data-particularly with regards to the association between choroidal thickness and acute VKH-it is not practical for clinical use due to the time required to perform the measurements. The discovery of the presence of choroidal thickening in acute VKH opens up the possibility of alternative approaches to diagnosing and monitoring the course of VKH. The emergence of SS-OCT-with its automatic measurement capabilities-provides clinicians with a new tool to quickly evaluate choroidal thickness to monitor the clinical course of VKH. In this study, we hypothesized that we could use SS-OCT to measure changes in the thickness of the choroid to both assess treatment response and to detect posterior relapse. Our results appear to confirm this hypothesis. We found that, after treatment, patients with acute VKH experienced a significant decrease in choroidal thickness and a gain in VA, a finding that seems to support the use of automatic measurement of choroidal thickness to evaluate treatment response. In addition, in symptomatic patients, we found that an increased choroidal thickness was associated with partial vision loss. By contrast, in asymptomatic patients with recurrent disease, the first sign of relapse was an increase in choroidal thickness (detected by SS-OCT) without loss of VA (all relapses were confirmed with ICGA). Thus, although ICGA is necessary to confirm the relapse, SS-OCT offers the possibility of detecting recurrent disease without the need for such an invasive, time-consuming procedure. Moreover, although ICGA gives a qualitative diagnosis of choroidal inflammation, SS-OCT provides an objective, quantitative diagnosis of that inflammation (evidence by the increase in choroidal thickness). Notably, personalized follow-up with SS-OCT scanning allowed us to diagnose pathologic increases in choroidal thickness in all eyes with posterior recurrences; it was also useful for the diagnosis of 10 posterior recurrences without loss of VA that were later confirmed with ICGA.  The findings in our study are consistent with previous reports describing a significant increase in choroidal thickness in patients with acute VKH [18,19,23,[26][27][28][29][30]. To verify the representativeness of the SFCT values measured automatically with SS-OCT in our control group, we compared these with the manual measurements (also obtained with SS-OCT) in 276 healthy volunteers in another study [14]. Those authors reported a mean SFCT of 301.89 μm (SD=80), a finding that was similar to our control group (mean, 302.3 μm). We performed a t-test to check for differences between these two groups, finding a mean difference of only 11.89 µm (95% CI, 14.27-38.05), which was not statistically significant (p=0.3718). This comparison confirmed the representativeness of our control group. Perhaps more importantly, our SFCT data was obtained using the automated measurement feature of SS-OCT. This is an enormous advantage because it does not require a trained professional to make the measurements and it can be done much faster and easier, making it an excellent method for use in daily clinical practice. Nakayama et al. evaluated 8 patients recently diagnosed with VKH. In that study, the authors measured choroidal thickness manually with EDI-OCT, defining a recurrence as an increase in choroidal thickness >100 µm from measurement [24]. In our study, we defined the relapse as the minimum increase in thickness required to induce a decrease in VA (not otherwise attributable to other causes) and with the presence of inflammatory signs on ICGA, or without loss of vision but with an increase of ≥ 50 µm in thickness combined with the presence of inflammatory signs on ICGA. The 50 µm cut off was selected because this was the mean SD among published studies [7,8,19,21]. As in our study, Nakayama et al. found that VKH patients had an increased SFCT at baseline (mean, 578 μm). Likewise, those authors found, as we did, that choroidal thickness decreased with treatment. However, they also found a rebound in choroidal thickening in 3 patients (5 eyes) during corticosteroid tapering but without evidence of increased inflammation (1 year follow-up and manual measurements). By comparison, in our study, 5 patients (10 eyes) presented asymptomatic relapses. We suspect that the higher relapse rate observed in our study versus Nakayama et al. is due to our longer follow-up (12 months vs.

months).
In VKH, the reported relapse rate ranges from 25% to 54% with anterior recurrences accounting for 50% of all recurrences [3,25,31]. Posterior recurrences are usually associated with poor VA at diagnosis or with a rapid tapering of the corticosteroid dose [25]. In our study, recurrent inflammation was found in 5 out of 9 patients (56%), all of which were posterior recurrences. In general, baseline VA was not particularly low (two-thirds of patients had a VA>0.63 and 3 of the 18 eyes had a decimal VA of 1. In addition, corticosteroids were tapered slowly in all cases (minimum of 1 year of treatment with prednisone). Consequently, given that VA in our sample was relatively acceptable and that corticosteroids were tapered slowly, the high relapse rate must be due to other factors. Based on our data, which showed that recurrences were associated with the duration of follow-up from diagnosis (all cases of recurrence were observed only in patients with ≥ 1 year from diagnosis), it appears that the likelihood of recurrence depends on the duration of the disease.
It is well known that the presence of a hypopigmented fundus (sunset glow) could be attributable to unidentified (and thus, untreated) posterior recurrences. For this reason, the routine use of ICGA during follow up is recommended in VKH patients [3]. However, given the aforementioned drawbacks of ICGA, the use of this modality should be minimized to the extent possible. In this sense, monitoring SFCT with SS-OCT may offer a valuable complement to ICGA. We used ICGA to confirm the presence of inflammation in all relapses but we found that SS-OCT was quicker, easier and safer than angiography. We observed sunset glow fundus in only 1 patient (both eyes), but this finding was not unexpected given that this patient experienced three asymptomatic relapses and it seems likely that there is an association between asymptomatic relapse and sunset glow fundus.

Study strengths and limitations
To our knowledge, the present study is the first to use the automated measurement features of SS-OCT to measure choroidal thickness in patients with acute VKH. Moreover, our prospective study includes the largest sample to Caucasian, Western European patients with VKH. We are aware of other studies that measured choroidal thickness in acute VKH; however, those studies were conducted in Japanese patients using EDI-OCT or SS-OCT with manual measurement [18,19,24]. Ours is the first study to present data on automated-and therefore objective-measurement of choroidal thickness obtained during the course of this disease.
This study has several limitations, primarily the small sample size and the fact that-despite the recurrent nature of this disease-we did not perform standardized monthly follow-up appointments for all patients. Rather, examinations were scheduled according to need depending on the severity of each individual case. Another limitation is that the decision to use a cut-off value of ≥ 50 µm before performing the ICGA, as this may have led to an under diagnosis of recurrences; however, this was a conscious decision to limit ICGA choroidal assessments to avoid performing unnecessary angiographies.

Conclusions
The results of this study suggest that the automatic measurement of choroidal thickness using SS-OCT can be a valuable tool to both evaluate treatment response and to help diagnose posterior recurrences. If confirmed, the findings presented here may reduce the need to use ICGA for routine follow-up, as angiography may only be necessary in asymptomatic patients who develop an increase in choroidal thickness. However, this finding must be first confirmed in prospective studies with larger patient samples.