Comparative Assessment of Widefield Fundus Autofluorescence Patterns in Retinoschisis and Retinal Detachment

Abstract

Purpose:

This article characterizes widefield fundus autofluorescence (WF-FAF) patterns in retinoschisis (RS), retinal detachment (RD), and combined retinoschisis-detachment (RS/RD), and to correlate them with spectral-domain optical coherence tomography (SD-OCT) findings.

Methods:

A retrospective case series of 13 eyes with senile RS, RD, or RS/RD is presented. One eye underwent imaging of 2 areas within the retina, resulting in 14 data points. Independent, masked graders classified pathology on SD-OCT as RS, RD, or RS/RD and graded WF-FAF images for either hypoautofluorescent areas or mixed autofluorescence (AF) (hyper-AF, iso-AF, hyper-AF with hypo-AF, hyper-AF with iso-AF, or hypo-AF with iso-AF).

Results:

There was no statistically significant correlation between the autofluorescence pattern and the type of retinal abnormality (P = .74).

Conclusions:

High variability was found in the characterization of WF-FAF in patients with RS and RD. SD-OCT remains the criterion-standard imaging modality in distinguishing RS from RD in clinically ambiguous cases.

Introduction

Retinoschisis (RS) is a nonprogressive degenerative splitting of the neurosensory retina, most often at the outer plexiform layer. ¹ It most commonly presents in individuals older than 50 years, is bilateral in approximately 57% to 85% of cases, and is most likely to be located in the inferotemporal quadrant. ^{2 -4} There appears to be no sex predilection and patients are usually asymptomatic. ^2,4 Its prevalence has been cited to be about 7% in those older than 40 years and 3.9% in those older than 60 years. ^2,4 On clinical examination, it typically appears smooth, immobile, and characteristically dome shaped. Subretinal fluid (SRF), hemorrhage, and pigmented cells in the vitreous are generally absent. Pigmented demarcation lines are rarely seen in isolated RS and tend to be more common when RS is associated with an outer retinal layer break or a retinal detachment (RD). ^2,3,5 Spontaneous regression may occur in 2.3% to 8.8% of eyes, and only about 3.2% are reported to extend posteriorly. ^2,4 Despite extension, it rarely causes vision loss in the absence of an RD and most cases are observed. ²

On clinical examination, RS can sometimes prove difficult to distinguish from RD, specifically in patients with chronic detachments that also may present with a smooth surface similar to that seen in RS. In addition, some cases of RS can be complicated by a progressive RD (retinoschisis-detachment [RS/RD]) if there are both inner and outer layer breaks in the retina. ^{2,6 -8} This progression to RD occurs in 0.05% to 2.2% of eyes when fluid within the RS cavity travels through the outer retinal defect and under the subretinal space, leading to detachment of the retina. ^2,4 It is important to differentiate RS from an RD or RS/RD because management differs.

Clinical examination techniques include scleral depression alone or coupled with B-scan ultrasonography. ⁹ These methods are limited because aggressive scleral indentation may be required and chronic or bullous RD may not be collapsible. Agarwal et al proposed high-resolution B-scan ultrasonography as a reliable method of differentiating RS from RD. ¹⁰ With this technique, 3 hyperreflective lines are seen on an ultrasonogram and represent the retinal nerve fiber layer, outer plexiform layer (OPL), and retinal pigment epithelium (RPE)–Bruch complex. An RS showed 2 distinct hyperreflective lines in the outer retina corresponding to the OPL and the RPE. In contrast, RD showed 2 hyperreflective lines in the detached portion, which represent the retinal nerve fiber layer and OPL. This may be useful in patients with significant media opacities that preclude adequate ophthalmoscopy, but technical expertise is required for this test. Other modalities such as laser photocoagulation, indirect ophthalmoscopic perimetry, and automated perimetry have been proposed as well. ^{11 -13}

The use of spectral-domain optical coherence tomography (SD-OCT) has become the current criterion standard in distinguishing RS from RD. ^{6,14 -16} Imaging with OCT will clearly show a cleavage plane between retinal layers in RS, whereas an RD will appear as a complete separation of the retina from the underlying RPE. Although OCT has been extremely useful in differentiating between RS and RD in unclear cases, this imaging modality has limitations. High-resolution images are often limited to the posterior pole and may not be able to capture far peripheral abnormalities. Ho and colleagues have suggested widefield infrared imaging as a quick, noninvasive method of diagnosing RS, RD, and RS/RD in which RS appears light, RD appears dark, and RS/RD demonstrates a mixed pattern. Although this may potentially be a helpful adjunct, more experience and further studies are needed to corroborate these results. ¹⁷

One modality that has only recently been evaluated as a possible tool for the differentiation of RS from RD or RS/RD is widefield fundus autofluorescence (WF-FAF). In 1 recent study of 38 eyes that had undergone WF-FAF imaging, 21 showed hyperautofluorescence (hyper-AF) and 17 showed no hyper-AF. ¹⁸ They were able to obtain OCT in 10 of the 21 eyes with hyper-AF and found that 10 of 10 eyes had an RD correlating to the area of hyper-AF. In the 17 eyes without hyper-AF, 11 eyes were successfully imaged with OCT and all 11 eyes demonstrated only RS on OCT. These results have not yet been corroborated in the literature. The purpose of this study is to characterize patterns of WF-FAF images in patients with RS, RD, and combined RS/RD, and to correlate these findings with SD-OCT findings.

Methods

All study procedures adhered to the Declaration of Helsinki. Informed patient consent was not obtained because of the retrospective nature of the study and lack of patient identifiers. After Cleveland Clinic Investigational Review Board approval, a retrospective review was conducted of all patients seen at the Cole Eye Institute between February 2013 and March 2019 who met the following inclusion criteria: diagnosis of “retinal detachment with retinal break” (International Classification of Diseases, Ninth Revision [ICD-9] codes 361.00, 361.01, 361.02; Tenth Revision [ICD-10] codes H33.00, H33.01, H33.02), “unspecified retinal detachment” (ICD-9 code 361.9), and “retinoschisis” (ICD-10 codes H33.10, H33.19) as search terms of the electronic health record system (Epic Inc); age at least 18 years; and history of testing the RS or RD areas with both SD-OCT (Carl Zeiss) and WF-FAF (Optos).

Patients were excluded if they did not have both tests performed the same day or if the time period between the 2 tests was greater than 1 week. Patients were also excluded if they had RS for etiologies other than senile (ie, acquired) RS such as X-linked juvenile RS, had a history of ocular trauma or uveitis, or had a history of RD repair before imaging. Although WF-FAF is not the standard of care when evaluating patients with RS, rhegmatogenous RD (RRD), and RS/RD, several patients underwent imaging to monitor progression that may have been difficult to assess on examination or fundus photographs.

Demographic data were collected for all patients. SD-OCT and WF-FAF images were reviewed by 3 independent graders (J.L.C., T.F.C., and G.L.H.) to ensure reliability, and a fourth expert grader (A.S.B.) was used in cases in which there was less than 100% agreement among the 3 graders. SD-OCT images were first graded for the presence of RD, RS, or combined RS/RD, and the corresponding area scanned using SD-OCT was then identified on each accompanying WF-FAF image. The pattern of AF was characterized as either hypo-AF or mixed AF, which was defined as any of the following: hyper-AF, iso-AF, hypo-AF mixed with hyper-AF, hyper-AF mixed with iso-AF, or hypo-AF mixed with iso-AF (Figure 1). Correlation between SD-OCT results and AF patterns were calculated, and kappa values were determined to evaluate intergrader agreement.

Figure 1.

Autofluorescence (AF) patterns.

Statistical analyses were performed using Microsoft Excel (Microsoft Corp) and R Studio (version 1.1.463). Categorical variables were described using percentages. Continuous variables were described using means, medians, and standard deviations. The Fleiss kappa statistic was used to evaluate intergrader agreement between 3 graders for all patients of each characteristic and a Fisher exact test was used to assess the relationship between retinal abnormality (RD, RS, or RS/RD) and AF pattern. A P value less than .05 was considered statistically significant.

Results

A total of 10 patients and 13 eyes were included in the analysis. One of these eyes had 2 areas within their retina imaged, resulting in a total of 14 data points for evaluation. The median age was 58 years and ranged from 29 to 92 years. Five patients were female and five were male. Eight of 10 (80%) patients identified as Caucasian, 1 (10%) as African American, and 1 (10%) as Asian (Table 1).

Table 1.

Demographic Characteristics and Data.

Patient	Age, y	Sex	Ethnicity	Laterality	SD- OCT	WF-FAF Pattern
1	69	M	Caucasian	OD	RD	Hypo-iso
2	92	F	Asian	OD	RD	Hypo
3	56	F	Caucasian	OS	RD	Hyper-iso
4	41	M	Caucasian	OS	RD	Hypo-iso
5	29	M	Caucasian	OD	RD	Hyper
6	73	F	Caucasian	OD	RS/RD	Hyper-hypo
				OS	RS/RD	Hyper-hypo
7	83	F	AA	OS	RS/RD	Hyper-hypo
8	60	M	Caucasian	OS	RS	Hypo
9	41	M	Caucasian	OD	RS	Iso
				OS	RS	Hyper-iso
10	51	F	Caucasian	OD temporal	RS	Hyper-iso
				OD inferior	RS	Hypo
				OS	RS	Hypo

Abbreviations: AA, African American; F, female; Hyper, hyperautofluorescent; Hypo, hypoautofluorescent; Iso, iso-autofluorescent; M, male; OD, right eye; OS, left eye; RD, retinal detachment; RS, retinoschisis; RS/RD, retinoschisis-detachment; SD-OCT, spectral-domain optical coherence tomography; WF-FAF, widefield fundus autofluorescence.

Of the 14 data points, 5 (35.7%) were confirmed to have an RD on SD-OCT, 6 (42.9%) had RS, and 3 (21.4%) had combined RS/RD. Of the 5 data points with RD on SD-OCT, 4 (80%) were graded as having a mixed AF pattern on WF-FAF and 1 (20%) was graded as having hypo-AF on WF-FAF. Of the mixed patterns, 2 were hypo-iso, 1 was hyper-iso, and 1 was hyper-AF. Of the 6 data points with RS on SD-OCT, 3 (50%) had mixed patterns, and 3 (50%) had hypo-AF only. Of the mixed patterns, 1 was iso-AF, 1 was hyper-hypo, and 1 was hyper-iso. Two of these data points came from the same eye, and although both SD-OCT cuts showed RS only, 1 had a mixed pattern (hyper-iso) whereas the other had hypo-AF only. Of the 3 eyes with mixed RS/RD, all had mixed patterns and were all graded as hyper-hypo.

The Fleiss kappa statistic was calculated to confirm intergrader agreement and validate the characteristic choice for presence of RD, RS, or both and the AF pattern (Table 2). For each characteristic, the P value was less than .05 and statistically significant, suggesting that agreement between the 3 graders was not due to random chance.

Table 2.

Fleiss Kappa Statistic for Intergrader Agreement.

Characteristic	Kappa Value	Level of Agreement	P
SRF presence	0.89	Near perfect	<.000
Retinoschisis presence	0.89	Near perfect	<.000
Combined RS/RD	0.65	Substantial	<.000
AF pattern	0.71	Substantial	<.000
Hyperautofluorescence	0.90	Near perfect	<.000

Abbreviations: AF, autofluorescence; RS/RD, retinoschisis-detachment; SRF, subretinal fluid.

There was less than 100% agreement on the image grading for 3 eyes. In these eyes, a fourth grader independently evaluated the images and the final classification of each image depended on the majority opinion of the 4 graders. A Fisher exact test was performed based on intergrader agreed-on choices to determine whether there was a statistically significant correlation between the AF pattern and the presence of RS, RD, or RS/RD. Calculations based on a 2-by-3 contingency table revealed that there was no statistically significant correlation between the retinal finding on SD-OCT and the AF pattern (P = .74) or presence of hyper-AF (P = .99).

Conclusions

It is often difficult in the clinical setting to discernably identify RS vs RD because a chronic RD may have a smooth surface similar to that of RS. Although other modalities exist for differentiating these 2 entities, they can be cumbersome or require technical expertise. Having multiple modalities to noninvasively and quickly identify the correct pathology would be beneficial to clinicians. Use of FAF in patients with RD has been studied and the proposal of possible mechanisms for different AF patterns has been hypothesized.

Sekiryu et al investigated the AF of SRF surgically collected from eyes with RRD and proposed that macrophages in the SRF generate AF after accumulating photoreceptor outer segments. ¹⁹ They compared macrophage AF with that of lipofuscin and concluded that they were spectroscopically similar. Thus, it may be possible that these macrophages contribute to increased AF in the case of an RRD. Others suggest that outer retinal disruption in the case of an RRD causes increased AF because of a window defect. ²⁰ It is possible that photopigment loss reduces optical pigment density, increasing the AF signal. Whereas these 2 papers seem to suggest that RDs would be hyperautofluorescent on imaging, another study showed that macular hole RD was associated with hypo-AF. ²¹ In that study, they proposed that hyperviscous SRF may block the AF signal in an RD, leading to hypo-AF. Eyes with myopic foveoschisis, however, had more variable AF patterns.

AF patterns have also been found to be variable in central serous chorioretinopathy. Han and colleagues looked at 126 eyes, characterized FAF patterns into 5 categories, found that the pattern type was associated with chronicity and visual prognosis. ²² Eyes with a “blocked AF” pattern had more acute pathology and better visual prognosis. These eyes had either no changes or uniform changes in decreased AF at the location of SRF. Eyes with a “descending tract” pattern, defined as a downward swath of hypo-AF extending below the inferior arcade, were associated with poorer prognosis. Similar studies have verified the use of AF patterns in determining chronicity and prognosis. ^23,24 For example, Imamura et al found that both confluent and granular hypo-AF of the macula were predictors of decreased visual acuity. ²³

Although a previous study of a small number of patients did show a positive correlation between WF-FAF findings and the presence of RS from RD and RS/RD, our study did not corroborate these results (Figures 2 –4). ¹⁸ It is worth noting that in contrast to their study, which evaluated WF-FAF imaging before evaluating OCT imaging, we chose to grade OCT images first because this modality is considered the current criterion standard. The only consistent finding in our study was in the 3 patients with mixed RS/RD, who all demonstrated mixed AF patterns and were graded as hyper-hypo. That said, a hyper-hypo AF pattern was also seen in 1 data point with RS. It is possible the lack of a positive correlation between WF-FAF findings and the presence of RS, RD, and RS/RD may be due to the small sample size in our study; a higher-powered study may have revealed a significant outcome. In addition, all but 1 eye had only a single SD-OCT cut through the area of pathology, which limits our study by potentially missing other areas of pathology. Collectively, our results suggest that WF-FAF cannot be reliably used to differentiate RS from RD or RS/RD.

Figure 2.

Varying patterns of retinoschisis. OD, right eye; OS, left eye.

Figure 3.

Varying patterns of retinal detachment.

Figure 4.

Varying patterns of schisis-detachment. OS, left eye.

Theories on AF patterns in RD may not always hold true. Variability in patterns may exist when varying lengths of time abnormalities are present. For example, if several prior theories hold true, perhaps SRF in an RD blocks the AF signal, causing a hypo-AF pattern early on. Because products of the photoreceptor outer segments are processed within RPE cells or macrophages within the SRF, it might be possible that this causes hyper-AF later on. As previously mentioned, this study is limited by its small sample size, which may have hampered the ability to detect a statistically significant correlation.

Another major limitation is that WF-FAF is not currently the standard of care when evaluating patients with RS and/or RD, which further limits the sample size because few patients underwent testing with both imaging modalities. In addition, the study is limited by its retrospective nature, the subjective grading of interpreting images, the use of multiple ophthalmic imaging technicians with variable levels of experience, and the somewhat poorer quality of some of the images. Poorer-quality images may be affected by patient cooperation during image accumulation and/or technician experience. Future investigations with an increased sample size and imaging obtained by a single, skilled, ophthalmic imaging technician are warranted to validate our results.

In our limited sample of patients, we did not find a correlation between WF-FAF patterns and the presence of RS, RD, or RS/RD. Further studies focused on the characterization of WF-FAF patterns in patients demonstrating RS, RD, and combined RS/RD are needed to define the utility of this imaging modality. For now, we agree with continuing to use OCT as the criterion-standard imaging modality for distinguishing RS from RD and RS/RD and would defer to larger studies for further information.