Photoreceptor Function and Structure in Autosomal Dominant Vitelliform Macular Dystrophy Caused by BEST1 Mutations

Abstract

Purpose: The purpose of this study was to evaluate rod and cone function and outer retinal structure within macular lesions, and surrounding extralesional areas of patients with autosomal dominant Best vitelliform macular dystrophy caused by BEST1 mutations.

Methods: Seventeen patients from seven families were examined with dark- and light-adapted chromatic perimetry and optical coherence tomography. Subsets of patients had long-term follow-up (14–22 years, n = 6) and dark-adaptation kinetics measured (n = 5).

Results: Within central lesions with large serous retinal detachments, rod sensitivity was severely reduced but visual acuity and cone sensitivity were relatively retained. In surrounding extralesional areas, there was a mild but detectable widening of the subretinal space in some patients and some retinal areas. Available evidence was consistent with subretinal widening causing slower dark-adaptation kinetics. Over long-term follow-up, some eyes showed formation of de novo satellite lesions at retinal locations that years previously demonstrated subretinal widening. A subclinical abnormality consisting of a retina-wide mild thickening of the outer nuclear layer was evident in many patients and thickening increased in the subset of patients with long-term follow-up.

Conclusions: Outcome measures for future clinical trials should include evaluations of rod sensitivity within central lesions and quantitative measures of outer retinal structure in normal-appearing regions surrounding the lesions.

Best vitelliform macular dystrophy (BVMD) is an autosomal dominant macular disease with a distinct ophthalmoscopic appearance¹ caused by monoallelic mutations in BEST1.^2,3 Over the last decade, it has become clear that BVMD is the most prevalent member of a spectrum of dominant and recessive inherited retinal pigment epithelium (RPE) diseases collectively named bestrophinopathies caused by both monoallelic as well as biallelic mutations in BEST1.^2,4–7 Patients with BVMD tend to show pathognomonic macular lesions centered on or near the fovea, and disease is thought to progress through stages descriptively named based on their ophthalmoscopic appearances such as vitelliform, pseudohypopyon, or vitelliruptive leading to an atrophic/cicatricial appearance.^7–9 Common to the earlier stages of disease is a serous detachment of the retina from the RPE, which is consistent with the understanding that the molecular defect acts primarily at the interface of photoreceptor outer segments and the RPE apical processes^10,11 despite Bestrophin-1 being localized to the basolateral plasma membrane of the RPE.^12,13 Later stages of BVMD can demonstrate substantial loss of photoreceptors and RPE that is localized to the macular area surrounded by wide expanses of retina that appear to demonstrate normal structure and normal function except for an electrophysiological defect detected by electrooculography (EOG).^8,14 Some patients with monoallelic BEST1 mutations develop one or more extra central lesions and this phenotype is called multifocal vitelliform dystrophy.⁷ 

In patients, BVMD is currently not curable or treatable. In dogs with naturally occurring biallelic BEST1 mutations, gene augmentation therapy has shown promise in treating both central vitelliform lesions as well as peripheral subclinical abnormalities.¹¹ Although gene therapy has not yet been tried in dogs heterozygous for BEST1 mutations, two independent investigations in patient cells have supported the hypothesis that at least a subset of monoallelic BVMD mutations may indeed be amenable to gene augmentation therapy.^15,16 In addition, small molecules have been proposed as BVMD therapies.¹⁷ Such investigational interventions will likely target macular lesions in BVMD as well as surrounding clinically uninvolved retina. In order to plan for appropriate outcome measures, we evaluated rod- and cone-photoreceptor mediated function within macular lesions and investigated subcellular retinal structure of the clinically uninvolved extralesional areas and peripheral retina. 

Methods

Subjects

The study population consisted of 17 patients with BVMD (ages 6–59 at first visit) from 7 families. Patients were heterozygous for BEST1 gene mutations (see the Table). Results in a subset of 10 patients have been published approximately 20 years earlier.¹⁸ Seven of 17 patients (5 of the 10 published earlier¹⁸ and 2 additional patients, see the Table) were evaluated most recently 4 to 22 years (median = 18 years) after their initial visit. Procedures followed the Declaration of Helsinki, and the study was approved by the Institutional Review Board (IRB) of the University of Pennsylvania. Informed consent, assent, and parental permission were obtained, and the work was Health Insurance Portability and Accountability Act (HIPAA)-compliant. 

Table.

 

Demographics and Genetics of Patients with BEST1-BVMD

 

Measures of Rod and Cone Function

Static threshold perimetry was performed in fully dilated eyes under dark- and light-adapted conditions using the modified Humphrey field analyzer (Carl Zeiss Meditec, Dublin, CA, USA), as previously described.^19,20 Dark-adapted test conditions included monochromatic blue (B; 500 nm) and red (R; 650 nm) stimuli (1.7 degrees diameter Goldmann V size, 200 ms duration) used to determine photoreceptor mediation. For loci demonstrating rod mediation with the B stimulus, rod sensitivity loss (RSL) was obtained by subtraction of the B sensitivity from mean normal value at each locus. Light-adapted testing was performed with orange (O; 600 nm) stimuli presented on a 10 cd.m⁻² white background and cone sensitivity loss (CSL) was obtained by subtraction of O sensitivity from mean normal value at each locus. Fixation during visual field testing was monitored via the infrared video imaging. Two profile tests extending 60 degrees along the horizontal and vertical meridia centered at fixation (29 loci separated by intervals of 2 degrees) were recorded. In addition, foveal sensitivity was measured at the center of four fixation lights forming a diamond. 

Retinal Imaging

A confocal scanning laser ophthalmoscope (HRA2 or Spectralis HRA; Heidelberg Engineering, Heidelberg, Germany) was used to obtain near-infrared excited reduced-illuminance autofluorescence imaging (NIR-RAFI).²¹ Wide field image montage was assembled by manually specifying corresponding retinal landmark pairs in overlapping segments using custom-written software (MATLAB version 6.5). 

Optical coherence tomography (OCT) was used to quantify retinal laminar architecture according to previously published methodology.^18,22–26 In brief, the main OCT data originated from 22 eyes of 11 patients recorded with a spectral-domain OCT (SD-OCT) instrument (RTVue-100; Optovue, Fremont, CA, USA). Data included overlapping line scans along the horizontal and vertical meridia extending to 30 degrees eccentricity, and overlapping raster scans with a coverage extending to 30 to 40 degrees eccentricity. In a subset of six eyes of six patients, long-term progression was estimated from overlapping line scans recorded with a time-domain OCT (TD-OCT) instrument (OCT1 or OCT3; Carl Zeiss Meditec, Dublin, CA, USA) along the horizontal meridian crossing the fovea. Post-acquisition processing used custom programs (MATLAB version 7.5; MathWorks, Natick, MA, USA) and raw data of each longitudinal reflectivity profile (LRP). For both SD-OCT and TD-OCT line scans, overlapping segments of LRPs were digitally joined to create 60 degrees wide OCTs centered on the fovea. For TD-OCT, multiple scans taken for each region of interest were considered, and LRPs from the two highest-quality scans were averaged. For both SD-OCT and TD-OCT, segmentation analysis was performed on each LRP to measure the thickness of the outer nuclear layer (ONL). The hyposcattering ONL was defined between the hyperscattering layer attributed to outer plexiform layer (OPL) and the hyperscattering outer limiting membrane (OLM) and included the anatomic layers of both ONL and Henle fiber layer (HFL). Limits of agreement and bias between comparisons of SD-OCT and TD-OCT data obtained decades apart have been published.²⁵ Outer retinal sublaminae were segmented only from SD-OCT data with better axial resolution. Four major sublaminae considered were: (1) a hyper-reflective layer near the junction of inner and outer segments; (2) a hyper-reflective layer near the interface between cone outer segments (COS) and contact cylinder (extrafoveal locations) or apical processes (foveal locations) of the RPE; (3) a hyper-reflective layer near the interface between rod outer segments (ROS) and RPE apical processes; and (4) a hyper-reflective layer near the interface between the basal RPE and Bruch membrane. COS length was defined from (1) to (2), ROS length from (1) to (3), and the combined thickness of the outer segments and subretinal space from (1) to (4). Overlapping three-dimensional SD-OCT raster scans were used for ultra-wide-angle topographic analysis.²² In brief, LRPs making up the OCT scans were aligned by straightening the major RPE reflection. En face images of integrated backscatter intensity were generated, and LRPs were allotted to regularly spaced bins in a rectangular coordinate system centered at the fovea. The waveforms in each bin were aligned, averaged, and segmented. Missing data were interpolated bilinearly; thickness values were mapped to a pseudo-color scale; and fundus landmarks were overlaid for reference. 

Results

Retained Function Despite Chronic Retinal Detachment

Data from both eyes of 17 patients with BVMD (7 families) were available (see the Table). Ages at first visit ranged from 6 to 61 years, and a subset of 7 patients were followed long term (15.5 ± 6.4 years). The majority (20 of 34, 59%) of eyes at the first visit had best corrected visual acuity (BCVA) of 0.2 logMAR (20/30 Snellen equivalent) or better (Fig. 1A), consistent with previous clinical impressions of relatively retained vision despite substantial foveal lesions.^27,28 Over the long term, BCVA was lost slowly at the average rate of 0.02 logMAR/year (corresponding to loss of 1 Early Treatment Diabetic Retinopathy Study [ETDRS] letter per year) across the 14 eyes with available longitudinal data (see Fig. 1A). Relatively retained acuity that showed slow progression would be counterintuitive considering the severe degenerative consequences of experimental retinal detachment on photoreceptors.²⁹ Therefore, we investigated details of the visual function in eyes that showed subretinal fluid, but retained foveal fixation thus allowing use of perimetric sensitivity. 

Figure 1.

 

Visual function and serous detachment in Best vitelliform macular dystrophy. (A) Best corrected visual acuity as a function of age in patients with monoallelic BEST1 mutations. Connected symbols represent eyes with follow-up. (B) Horizontal OCT from the right eye of patient F1/P2 demonstrating 0 logMAR visual acuity in an eye with a large serous retinal detachment involving the fovea. Inset, NIR-RAFI image showing the location of the OCT in the eye with pseudo-hypopyon stage. (C) Foveal light-adapted cone sensitivity loss in eyes with foveal serous detachment. Eyes with better acuity have better sensitivity. (D) Rod and cone sensitivity losses in parafoveal areas with serous detachment. A, C, D Gray symbols or bars represent eyes with worse than 0.2 logMAR acuity.

A representative example of a BVMD eye with a large vitelliform lesion but normal visual acuity of 0 logMAR and foveal fixation is F1/P2 at age 14 years (Fig. 1B). En face imaging (see Fig. 1B, inset) shows a pseudohypopyon stage with autofluorescent material accumulated at the inferior aspect of the lesion. Cross-sectional imaging along the horizontal meridian shows a large serous detachment extending from approximately 11 degrees temporal to approximately 5 degrees nasal to the fovea. There were six eyes with similar serous detachments and foveal fixation. To better understand the consequence of the chronic retinal detachment on foveal function, we measured light-adapted sensitivities at the fovea. CSL at the fovea ranged from 4 to 5 decibel (dB) in the eyes with 0.2 logMAR or better acuity, to 10 to 16 dB in the eyes with worse than 0.2 logMAR acuity (Fig. 1C). 

Considering there are at least 4 sources of the 11-cis-retinal chromophore required for rod and cone vision,^30–32 we next asked whether rod and cone photoreceptor function are differentially affected in extrafoveal regions of eyes with chronic retinal detachment but retained foveal fixation. In 3 eyes with retained visual acuity, there were 41 extrafoveal locations that corresponded to retinal detachment, whereas in 3 eyes with greater acuity loss, there were 19 such locations. In these regions, RSL were substantially greater than CSL at the great majority of retinal locations (Fig. 1D). Greater rod than cone dysfunction associated with detached locations in BVMD was similar to findings in recessive bestrophinopathies¹¹ and could be explained by the dominant source of visual chromophore originating from the retina for cones as opposed to the visual chromophore originating from the RPE for rods.^30–32 

Microstructural Defects in Clinically Uninvolved Extralesional Retina

To evaluate the retinal microstructure in the clinically uninvolved retinal areas (outside of macular lesions and outside of rare satellite lesions) of patients with BVMD, two key subcellular photoreceptor features were initially assessed: (1) the ONL thickness where rod and cone photoreceptor nuclei reside, and (2) the distance between the junction of inner and outer segments (IS/OS) and the RPE, which would encompass the length of OS plus the subretinal space (Figs. 2A, B, insets). Both features were mapped across an ultra-wide (∼80°) expanse of retina in each eye (Figs. 2A, 2B). As demonstrated in a representative healthy eye, ONL thickness was largest at the fovea with a gradual fall-off with eccentricity (see Fig. 2A). ONL topography in the patient F3/P2 with BVMD was similar to normal but some retinal regions tended to show mild thickening (see Fig. 2B). The normal IS/OS to RPE distance was also largest at the fovea with fast fall-off to parafovea and a nearly homogeneous distribution throughout the rest of the retina (see Fig. 2A). In the representative patient F3/P2 with BVMD, IS/OS to RPE distance was larger than normal throughout the retina (see Fig. 2B). 

Figure 2.

 

Ultra-wide field topography of outer retinal structure in BVMD. (A, B) Topographical maps of ONL thickness (gray map, center) and IS/OS to RPE distance (pseudocolor map, right) obtained from OCT scans (left) in a normal subject and in patient F3/P2. ONL is the major hyposcattering layer on the OCT, IS/OS and RPE peaks are highlighted with orange and brick red, respectively. Purple represents indeterminate regions. (C) Quantitation of the ONL thickness and IS/OS to RPE distance along the two cardinal meridians, horizontal and vertical, crossing the fovea. Black traces are data from individual BVMD eyes, green lines delimit the normal range. N, I, S, and T represent nasal, inferior, superior, and temporal retina, respectively.

ONL thickness was quantified in 22 eyes of 11 patients along the horizontal and vertical meridia crossing the fovea (Fig. 2C). When measurable, ONL overlying macroscopically obvious lesions tended to be thinner than normal suggesting photoreceptor degeneration. In extralesional regions, on the other hand, ONL was either normal or mildly thickened. IS/OS to RPE distance is also shown along the two meridians (see Fig. 2C). Majority of the BVMD eyes across the clinically uninvolved retinal areas showed expansion of the distance from IS/OS to the RPE suggesting either elongation of the OS or widening of the subretinal space or both. 

Outer Segment Length and Subretinal Space

To attempt to differentiate among elongation of cone OS, rod OS, or expansion of the subretinal space with interdigitating OS and microvilli of the RPE, we measured corresponding markers from OCT scans at 2 locations at 16 degrees eccentricity in the superior and inferior retina (Fig. 3). Designation of the backscatter peaks on LRPs corresponding to subcellular photoreceptor and RPE components are shown for a representative normal, and 2 patients F4/P1 and F3/P2 with BVMD (see Figs. 3A, 3C). In most BVMD eyes, there was normal or near normal cone OS length in the superior (see Fig. 3B) and inferior (see Fig. 3D) retina. Rod OS length could be normal or elongated, and the distance to the RPE could be normal or extended both in the superior (see Fig. 3B) and inferior (see Fig. 3D) retina. 

Figure 3.

 

Detailed structure of the photoreceptor-RPE interface in BVMD. (A, C) OCT scan and the longitudinal reflectivity profile of a representative normal and two patients with BVMD in the superior and inferior retina locations. Schematic shows the 16 degrees eccentric location (black square) sampled. (B, D) Distances measured from the IS/OS peak to the cone OS tips (COST) representing COS length, to the rod OS tips (ROST) representing ROS length and to the RPE peak representing combination of outer segments, subretinal space, and the RPE thickness in the superior and inferior retina locations. Left panels show absolute thickness, and the right panels show the thickness difference from mean normal values. Gray rectangles delimit the 99th percentile (±2.33 SD) of the normal values for each parameter.

Functional Consequences of an Expanded Subretinal Space

Rod sensitivities were normal across clinically uninvolved extralesional regions of BVMD retinas; this implied that minor changes in the subretinal space do not result in detectable changes to the primary function of rods, which is to signal dim lights under fully dark-adapted conditions. However, because the bulk of the chromophore for rod cells is thought to originate from RPE cells, we hypothesized that abnormalities at the interdigitation of the outer segments and RPE microvilli may slow down the kinetics of the resupply of chromophore and thus the rate of dark adaptation. In a subset of 4 patients with BVMD , at a macular region 10 degrees nasal to the fovea, we recorded dark-adaptation kinetics of the rod system (Fig. 4). All four patients had near-normal ONL thickness and dark-adapted sensitivities ruling out retinal degeneration. Three of the four patients (F4/P1, F1/P4, and F7/P1), had normal or near normal rod and cone OS thicknesses and normal subretinal space (see Fig. 4A) and they demonstrated normal kinetics of dark-adaptation recovery (see Fig. 4B). One patient (F3/P2) with elongated cone OS, rod OS, and wider subretinal space (see Fig. 4A), demonstrated a substantial slowing of the dark-adaptation recovery rate (see Fig. 4B). These data appear to support a relation between rod dark-adaptation recovery rate and the abnormality of the subretinal space. However, it is important to note the results from a fifth patient (F3/P1) demonstrating substantial thinning of the ONL (see Fig. 4C) but retaining apparently normal distances of IS/OS to COS, ROS, and RPE. Dark-adapted sensitivities of F3/P1 were elevated by more than a log unit, and there was substantial slowing of the dark-adaptation rate (see Fig. 4D). This retinal-degeneration-associated abnormality in dark-adaptation rate was similar to an effect previously described in patients with ABCA4-associated Stargardt disease³³ and BEST1-associated autosomal recessive bestrophinopathy.¹¹ Thus, it is important to note that dark-adaptation kinetic abnormalities could have multiple causes but the combination of normal ONL thickness, normal dark-adapted sensitivities, and slowed kinetics likely places the physiological defect measured in F3/P2 at the abnormal RPE-PR interface. 

Figure 4.

 

Relation of retinal ultrastructure and dark-adaptation kinetics. (A) Outer retinal ultrastructure at 10 degrees nasal to the fovea (schematic, black square) shows the thickness difference from mean normal for COS and ROS length and IS/OS to RPE distance similar to Figures 3B and 3D. Four patients had normal ONL thickness whereas one patient (F3/P2) had abnormally thinned ONL. (B) Dark-adaptation kinetics following a transient adapting light presented at time zero. Data from four patients (unfilled circles) are shown in comparison to normal data (filled gray circles). Symbols identify the individual patients plotted in panel A. (C) OCT scan of patient F3/P1-OS at the locus evaluated with dark-adaptation kinetics (white rectangle) demonstrating substantially diminished ONL thickness compared to a representative normal. (D) Dark-adaptation kinetics showing elevated thresholds before the adapting light, and exceedingly slow recovery timecourse after the adapting light.

Long-Term Natural History of Retinal Structure

Among the outer retinal abnormalities in clinically uninvolved retina, was minor ONL thickening observed in some retinal locations of some patients with BVMD (see Fig. 2). To better understand the temporal development of this finding, we took advantage of the long-term natural history of retinal structure in a subset of 6 patients with BVMD who were followed with OCT over nearly 2 decades (Fig. 5). Qualitatively, in five of six patients, there were relatively small changes; central lesions showed minor changes, and clinically uninvolved extralesional areas appeared to be unchanged (see Fig. 5). An important exception was F1/P3, who showed formation of an extra-macular satellite lesion temporal to the fovea. Next, we quantified the ONL thickness across the horizontal meridian crossing the fovea. Overlying the central lesions, ONL thickness was reduced implying a progressive photoreceptor degeneration. Altered visibility of the HFL secondary to lesion changes could possibly contribute to the measured ONL but does not fully account for large magnitude of ONL thinning seen in F4/P1-OD and F1/P4-OD. ONL thinning also occurred over the newly formed satellite lesion in F1/P3-OS. Changes in extralesional regions were mostly within the variability observed with similar ultra-long-term evaluations^25,26; however, it was notable that there was a tendency toward ONL thickening over time. In some patients and some retinal locations (F1/P4-OD and F1/P5-OS), ONL thickness moved from being near the upper limit of normal to being significantly hyperthick (see double arrows in Fig. 5); in other patients and other retinal locations, there was evidence for relative thickening of ONL albeit remaining within normal limits (see single arrows in Fig. 5). 

Figure 5.

 

Long-term changes in retinal structure. (A) Long-term (17–21 years) follow-up of younger (20–23 years old at first visit) patients with BVMD. (B) Long-term (10–20 years) follow-up of older (34–47 years old at first visit) patients with BVMD. In each panel, OCT scans along the horizontal meridian crossing the fovea are shown at two ages. ONL thickness is quantified (gray trace first visit and black trace last visit) and compared to the normal range. Bar graphs show ONL thickness change (last minus first) sampled across the retina. Double arrows mark regions of thickening beyond the normal range and single arrows mark regions of thickening within the normal range.

Onset of Extra-Macular Satellite Lesions

In nine eyes of five patients, there were extra-macular satellite lesions at one or more visits. Unexpectedly, in both eyes of one patient, onset of several de novo extra-macular satellite lesions was observed in retinal regions where there was no evidence of a lesion at an earlier visit. In F1/P4-OD, NIR-RAFI at ages 29 and 37 years showed only a macular lesion (Fig. 6A). At age 43 years, however, several small lesions had formed along the superior arcade and OCTs were consistent with development of local outer retinal deposits where none existed previously (Fig. 6B). IS/OS-RPE thickness topography performed across an ultra-wide extent of retina at age 37 years showed the existence of an arcuate region of subclinical abnormality (IS/OS-RPE distance >10 um greater than normal) at the superior vascular arcade. Several serous detachments developed in this area 6 years later (Fig. 6C). The region around the newly developed serous detachments showed greater abnormality of the IS/OS-RPE distance (see Fig. 6C). In F1/P4-OS, NIR-RAFI at age 29 years showed only a macular lesion (Fig. 6D). At age 37 years, a satellite lesion had formed superior to the optic nerve head near the eccentricity of the vascular arcades. At age 43 years, a second satellite lesion formed superotemporal to the macula. OCTs were consistent with the development of a local serous detachment where none existed previously (Fig. 6E). IS/OS-RPE thickness topography showed the existence of an arcuate region of subclinical abnormality (IS/OS-RPE distance >20 um greater than normal) along a band at the superior vascular arcades, which appear to precede the onset of the lesion (Fig. 6F). These rare results appear to support the hypothesis that clinically obvious outer retinal deposits and serous detachments may be preceded by subclinical widening of the IS/OS-RPE distance. 

Figure 6.

 

Late development of satellite lesions in both eyes of patient F1/P4. (A, D) NIR-RAFI results at ages 29, 37, and 43 years show development of de novo lesions (pink outlines) at age 43 years in Panel A and at ages 37 and 43 years in Panel D. (B, E) Representative OCTs demonstrate serous detachments in retinal locations (arrows in A and D) that were previously attached. (C, F) Ultrawide-field topography of the IS/OS to RPE distance shows arcuate regions of hyper-thickening in the superior paramacular retina at age 37 years preceding the development of de novo serous detachments 6 years later. Purple represents regions with indeterminate IS/OS to RPE distance.

Discussion

Visual Function in BVMD

Clinical studies in BVDM during the pre-molecular era,^34,35 as well as large series performed more recently in molecularly confirmed patients,^36–38 have shown many patients over a substantial range of ages to have at least one eye with BCVA of 20/40 or better. This level of BCVA would be consistent with retained cone photoreceptor function in many patients with clinically obvious and often chronic serous detachments of the fovea.²⁸ In our cohort, there were BVMD eyes with 20/20 vision and fixation at the fovea which was chronically detached from the underlying RPE. Taking the previously published results together with the current findings, the only conclusion that can be reached is that in BVMD, survival and function of foveal cone photoreceptors are not dependent on their attachment to the RPE cells. 

BCVA is a coarse measure of spatial vision and does not scale linearly with photoreceptor degeneration at least for the loss of the first approximately 50%.^39–41 Perimetric light sensitivity provides another measure of visual function and allows sampling of distinct retinal locations surrounding the fovea. Previously, fundus-controlled microperimetry has shown foveal and parafoveal sensitivity losses with retinal lesions in patients with BVMD^38,42–44; however, tests were performed under mesopic conditions and the relative contribution of rod versus cone photoreceptors losses were not known. In a subset of eyes retaining stable foveal fixation, we performed light- and dark-adapted perimetry and determined that at retinal loci with chronic serous detachments rod sensitivity losses tended to be substantially higher than cone sensitivity losses. Relatively greater involvement of rod sensitivity or low luminance visual acuity has previously been demonstrated in central serous retinopathy (CSR)^45,46 but we are not aware of investigations of direct colocalized comparison of rod and cone function in BVMD or CSR. Publications reporting histopathology from BVMD eye donors have not quantitatively evaluated the comparative involvement of rod and cone photoreceptors.^47–53 It remains to be determined whether the observed loss of rod function in central lesion areas of BVMD corresponds to degeneration or dysfunction of rod photoreceptors, or to remodeling of postreceptoral connections.⁵⁴ 

Thickening of the Outer Nuclear Layer

Human IRDs and their animal models involve progressive degeneration of the photoreceptors, which is reflected by the thinning of the ONL readily measurable on histology.^55,56 Soon after development of the OCT,⁵⁷ our group was the first to recognize that ONL thickness could be measured noninvasively in humans and animals,^58,59 and progression rates based on loss of photoreceptors could be quantitatively inferred in vivo from serial measurements of the ONL thickness over time.^26,60 It is less appreciated that counterintuitive thickening can sometimes precede the ONL thinning.⁶¹ A mild ONL thickening can be chronic or a transient stage in a continuum and it has been detected at the foveas of patients with choroideremia,⁶² periphery of patients with NPHP5- or CEP290-associated Leber congenital amaurosis (LCA),²² and surrounding the macular drusen in age-related macular degeneration (AMD).⁶³ Thickened ONL has also been observed in animals as a consequence of interventions or mild disease states.^11,64,65 More recently, thickening of ONL upon gene augmentation therapy was seen.⁶⁶ Our current work showed that mild ONL thickening is detectable across the clinically normal-appearing retinas of many patients with BVMD, and this feature can slowly be accentuated over decades in otherwise normal retinal areas. Our work is consistent with “apparent thickening” of the ONL outside lesion areas and in previtelliform disease stages previously observed in some patients with BVMD.⁶⁷ It is unlikely that thickened ONL observed across diverse IRDs and AMD represent the same exact underlying pathophysiological cause; however, involvement of subclinical edema due to abnormalities at the RPE-PR interface or sub-threshold signaling of photoreceptor stress to glial cells can be postulated in BVMD. 

Outer Segment and RPE Interface in Clinically Uninvolved Extralesional Areas

Clinically obvious serous detachments of the retina from the RPE have been well described in BVMD.^18,68 This grossly detectable pathology is consistent with the recent understanding that the primary molecular defect acts at the RPE-PR interface in the canine model^10,11 despite Bestrophin-1 being normally localized to the basolateral plasma membrane of the RPE.^12,13 What remains less clear is the state of the photoreceptor outer segments and RPE across wide expanses of clinically uninvolved extralesional areas of the retina in patients with BVMD. Previous investigations to evaluate the earliest disease features in BVMD have resulted in some controversy. Some qualitative studies have suggested subclinical abnormalities at the outer retina,^69–71 but quantitative studies have come to opposing conclusions. One study measured “photoreceptor equivalent thickness” or “outer segment equivalent length” between the “IS/OS junction” and the “inner surface of the RPE” and found that distance to be on average 6.5 um wider in patients with BVMD compared to controls.^52,72 Another study measured the “length of outer segments” between the “IS/OS junction” and the “RPE1, outer segment/RPE interface” and found no difference between BVMD and controls⁴³ and controversy ensued.^73,74 Using our interpretation of OCT peaks in human eyes,²³ measurements of the earlier study^52,72 would closely relate to what we have labeled ROS length in the current study, and that of the latter study⁴³ would correspond to what we have labeled COS length. At the superior and inferior perimacular regions, we found greater numbers of eyes with lengthening of the ROS than those with lengthening of the COS in patients with BVMD (see Fig. 3). Thus, our results confirm both previous conclusions and help clarify the source of the apparent controversy in the literature. 

Understanding of the microscopic details of the minor structural abnormalities in some BVMD eyes well away from central lesions in clinically normal-appearing retina remains elusive. It has been suggested that RPE-PR interface is filled with extremely long and/or unphagocytized photoreceptor outer segments⁵² and that may be true. Extrapolation from the results in dogs with biallelic BEST1 mutations would suggest a developmental abnormality of the RPE microvilli preceding detectable disease onset.¹¹ Studies in dogs heterozygous for BEST1 mutations could potentially distinguish between these hypotheses in the future. More pragmatically, do these micron-level subclinical abnormalities have any consequence to the progression of disease or loss of vision that BVMD patients may experience? To answer this question is challenging due to the slow progression of disease and limited longitudinal studies with higher resolution cross-sectional techniques. However, in both eyes of one patient, we showed de novo formation of clinically obvious lesions in the exact area of the retina that 6 years earlier showed the largest subclinical RPE-PR abnormality but no lesions. In another patient, we showed dark-adaptation kinetic abnormalities corresponding to the location with a subclinical RPE-PR abnormality. If these anecdotal findings are representative of the progression and vision loss experienced in BVMD, it would be justifiable to treat such retinal areas with subclinical disease assuming a safe and effective treatment becomes available in the future. 

There has been increasing evidence that measurable structural changes occur in the outer retina with light exposure.^52,75–83 For the current work, all OCT imaging was performed in dark-adapted eyes. It remains to be determined whether previously hypothesized light dependent changes to BVMD outer segment lengths^52,73 can be confirmed. 

Outcome Measures for Clinical Trials of BEST1 BVMD

Our results support choices of outcome measures for clinical trials attempting to improve retinal structure and visual function. We showed relatively retained BCVA and relatively large losses of scotopic light sensitivity. In AMD with a scotopic deficit,⁸⁴ low-luminance visual acuity (LLVA)⁸⁵ has been found to be a simple but informative outcome.^86,87 In the current study, we did not record LLVA, and we are not aware of literature measuring LLVA in BVMD. However, patients with CSR have shown large reductions in LLVA despite retaining good BCVA.⁴⁶ Assuming comparability of dysfunction resulting from serous retinal detachments in CSR and BVMD, future studies should evaluate LLVA in BVMD as a prerequisite for its potential use as an outcome measure in addition to standard BCVA. 

Perimetric methods provide topographic distribution of light sensitivity and can allow comparison of treated areas to neighboring untreated regions in localized interventions, such as subretinal gene therapy.⁶⁰ Use of light-adapted or dark-adapted conditions allows direct comparison of cone and rod function as was done in the current study. However, standard perimetric methods require foveal fixation which may not be attainable in some patients with BVMD. Thus “microperimetry” methods performed with real time tracking of the retina is required. Measurement of cone function with microperimetry is challenging and we are aware of only one instrument that provides testing with the standard photopic background.⁸⁸ Measurement of rod function with microperimetry is also challenging due to the limited dynamic range of stimuli available in all devices.^89–91 Previous studies in BVMD have used microperimetry under mesopic conditions which do not allow distinction of rod and cone function.^38,42–44 For a future clinical trial, it would be important to use a microperimetric method, the results of which can be interpreted confidently in terms of the function of the underlying photoreceptor system. 

En face imaging methods provide convenient analytics to measure changes in lesion appearance and extent as part of the natural history of disease or interventions. Historically, color fundus photographs were used to describe the evolution of BVMD lesions^14,35 but advent of imaging with short-wavelength autofluorescence (SW-AF) showed a specific hyperintense signal within vitelliform lesions.^92–95 However, conventional SW-AF is performed with a high intensity excitation light which could, at least in principle, accelerate retinal disease.⁹⁶ Therefore, we developed an RAFI method using NIR excitation.²¹ NIR-RAFI can provide en face information regarding RPE health with some similarities to the standard SW-AF results.^97,98 In the case of BVMD, NIR-RAFI can be more sensitive to earliest disease features,⁹⁹ but unlike the ABCA4 form of macular degeneration,³³ extralesional areas in BVMD do not show abnormal increases in SW-AF and NIR-RAFI signals.^67,99 In the current work, we used NIR-RAFI to delineate not only the gross macular lesions but also de novo development of small satellite lesions. Similarly, NIR-RAFI can form an important outcome measure to provide information regarding RPE health with comfortable lights and a short examination time without undue light hazard potential. 

Cross sectional imaging with OCT is a key outcome for BVMD clinical trials. The extent of the vitelliform lesions can be quantified. Additionally, current results strongly suggest that hyperthickening of the ONL and widening of the distance between the IS/OS and the RPE may be the only subclinical abnormalities in paralesional retinal areas that can be evaluated quantitatively for efficacy and safety experimental interventions in clinical trials. 

Acknowledgments

Supported by IVERIC Bio, Inc., NEI R01EY06855, Foundation Fighting Blindness, unrestricted funds from Research to Prevent Blindness, and the Sanford and Susan Greenberg End Blindness Outstanding Achievement Prize. 

A.V.C., S.G.J., K.E.G., G.D.A., and W.A.B. are listed as co-inventors on two US Patent Applications related to treating bestrophinopathies. 

Disclosure: A.V. Cideciyan, IVERICBIO (F, P); S.G. Jacobson, IVERICBIO (F, P); M. Swider, None; A. Sumaroka, None; R. Sheplock, None; A.K. Krishnan, None; A.V. Garafalo, None; K.E. Guziewicz, IVERICBIO (F, P); G.D. Aguirre, IVERICBIO (F, P); W.A. Beltran, IVERICBIO (F, P); E. Heon, None