Promises and pitfalls of evaluating photoreceptor-based retinal disease with adaptive optics scanning light ophthalmoscopy (AOSLO)

Abstract

Adaptive optics scanning light ophthalmoscopy (AOSLO) allows visualization of the living human retina with exquisite single-cell resolution. This technology has improved our understanding of normal retinal structure and revealed pathophysiological details of a number of retinal diseases. Despite the remarkable capabilities of AOSLO, it has not seen the widespread commercial adoption and mainstream clinical success of other modalities developed in a similar time frame. Nevertheless, continued advancements in AOSLO hardware and software have expanded use to a broader range of patients. Current devices enable imaging of a number of different retinal cell types, with recent improvements in stimulus and detection schemes enabling monitoring of retinal function, microscopic structural changes, and even subcellular activity. This has positioned AOSLO for use in clinical trials, primarily as exploratory outcome measures or biomarkers that can be used to monitor disease progression or therapeutic response. AOSLO metrics could facilitate patient selection for such trials, to refine inclusion criteria or to guide the choice of therapy, depending on the presence, absence, or functional viability of specific cell types. Here we explore the potential of AOSLO retinal imaging by reviewing clinical applications as well as some of the pitfalls and barriers to more widespread clinical adoption.

Copyright © 2020 Elsevier Ltd. All rights reserved.

Figures

Fig. 1.

Representative resolution of various retinal imaging modalities relative to human retinal anatomy, inspired by Miller et al. (2011). Drawing of a sagittal section of a human eye created by Teresa Patitucci, PhD, Medical College of Wisconsin (not to scale). Histological cross section provided by Dr. Christine Curcio, University of Alabama at Birmingham, from a larger image originally published in Tian et al. (2015). The width and length of the shapes approximate the resolution of each modality (lateral and axial, respectively). None of these are absolute, as a number of variables can influence the actual resolution – including differences in imaging light source (wavelength, bandwidth), confocal pinhole diameter, pupil size, and axial length of the eye. Retinal layer labels: RNFL = retinal nerve fiber layer; GCL = ganglion cell layer; IPL = inner plexiform layer; INL = inner nuclear layer; OPL = outer plexiform layer; HFL = Henle fiber layer; ONL = outer nuclear layer; IS = photoreceptor inner segments; OS = photoreceptor outer segments; RPE = retinal pigment epithelium; ChC = choriocapillaris; Chor. = choroid. Scale bars = 50 μm.

Fig. 2.

Example confocal AOSLO montage of the living human foveal cone mosaic from a 26-year-old female with normal vision. Each bright reflective structure is an individual cone photoreceptor. The location of peak cone density at the fovea is marked with an ‘x’ at the left side of the image. Moving away from the fovea, cone density declines and cone spacing increases precipitously, with smaller rods beginning to appear. Scale bar = 100 μm.

Fig. 3.

AOSLO images of the parafoveal cone mosaic. A, C: Confocal AOSLO images of the rod and cone mosaic. Cones appear as a dark ring with a central reflective core, with the smaller rods filling the space between cones. B, D: Corresponding split-detection AOSLO images at the exact same retinal location as the confocal images. The large circular structures are cone inner segments, with the smaller rods not typically visible due to the lower lateral resolution of this modality. Scale bar = 50 μm.

Fig. 4.

False-colored image of the human trichromatic cone mosaic. The reflectance of each cone in an AOSLO image was evaluated following various selective bleaching protocols. Based on the relative change in reflectance under each bleaching condition, the spectral identity of the photopigment within each cell could be inferred. Cones interpreted as long-wavelength sensitive are colored red, those that were middle-wavelength sensitive are colored green, and the short-wavelength sensitive cones are colored blue. Scale bar = 2 arcmin. Reproduced fromSabesan et al. (2015).

Fig. 5.

Images of retinal pigment epithelial (RPE) cells. A: RPE cells visualized in confocal adaptive optics scanning laser ophthalmoscopy (AOSLO) images of a patient with cone loss due to cone rod dystrophy (Roorda et al., 2007). B: AOSLO with short wavelength autofluorescence (AO-SWAF) capability shows RPE images in a monkey (Morgan et al., 2009a). C: AOSLO dark-field images show hexagonal RPE cells in a normal subject (Scoles et al., 2013). D: AO optical coherence tomography (AO-OCT) images of RPE cells (Liu et al., 2016). E: AO-indocyanine green (AO ICG) image of RPE cells (Tam et al., 2016). F: AO infrared autofluorescence (AO IRAF) images of RPE cells (Liu et al., 2017b). Scale bars, 50 μm. Modified fromRoorda et al. (2007) and Morgan et al. (2009a), copyright by the Association for Research in Vision and Ophthalmology, Liu et al. (2016), Tam et al. (2016), Scoles et al. (2013)reprinted/adapted with permission from Drew Scoles, Yusufu N. Sulai, and Alfredo Dubra andLiu et al. (2017b), © The Optical Society.

Fig. 6.

Disrupted cone mosaics in patients with inherited red-green color vision deficiency. A: A deuteranope with the LIAVA haplotype expressed by his OPN1MW gene. B, C: Both patients are protanopes as a result of the LIAVA haplotype being expressed by their OPN1LW genes. The normally waveguiding cones are thought to be S-cones and L-cones in the patient in panel A and S-cones and M-cones in the patients in panels B & C. The dark gaps in all three mosaics are the location of non-functional cones expressing the LIAVA haplotype. The variable number of cones with altered waveguiding is thought to be due to the variable stochastic expression of the OPN1LW and OPN1MW genes across different patients. Scale bar = 25 μm.

Fig. 7.

S-cone free zone of 16-year-old subject with BCM caused by C203R L/M interchange mutation. A: OCT image shows a focal disruption of the EZ band at the fovea, with the arrows indicating the location of the corresponding confocal AOSLO montage in panel B. A large dark region can be seen, surrounded by sparsely distributed reflective structures, presumably S cones. Scale bar for A = 200 μm. Scale bar for B = 100 μm.

Fig. 8.

Confocal and corresponding split detection AOSLO images from one 43-year old male with CNGB3-associated ACHM. Dark areas on confocal imaging (left panels) correspond to cone inner segments on split-detection (right panels). The top and bottom image pairs are from 2.6 degrees and 4.6 degrees temporal to the fovea, respectively. Scale bar = 50 μm.

Fig. 9.

Variability in the foveal cone mosaic in patients with ACHM. (Top) Foveal montages obtained using split-detector AOSLO for two subjects with sparse foveal mosaics—PCI-008 with a peak density of 7,273 cones/mm2 and PCI-007 with 12,231 cones/mm2. (Bottom) Foveal montages for two subjects with relatively contiguous mosaics—PCI-009 with a peak density of 19,835 cones/mm2 and PCI-021 with 44,959 cones/mm2. Scale bar = 50 μm. Reproduced fromLanglo et al. (2016).

Fig. 10.

Adaptive optics microperimetry (AOMP) revealed reduced sensitivity per cone density in eyes with mutations in RPGR, which is expressed in rod and cone photoreceptors, compared to normal eyes and eyes with mutations in RHO, which is expressed exclusively in rods. AOSLO split detector (top) and confocal (center) images with test locations and sensitivities are shown as colored circles above a spectral domain OCT scan (bottom) from the same retinal location. Retinal sensitivities are displayed using a color scale ranging from green (normal) to red (not seen). Center panels show AOSLO images at 200% magnification; black scale bars in all AOSLO images are 1 degree. A: Sensitivities from a normal subject are normal, B: sensitivities from a patient with RHO c.810C>A, p.Ser270Arg are normal near the fovea but reduced (yellow) beginning at 4 degrees and not measurable (red) at 6 degrees temporal to the fovea, and C: sensitivities from a patient with RPGR c.1243_1244delAG, p. Arg415Glyfs*37 are normal near the fovea but more severely reduced (orange) beginning at 4 degrees and not measurable (red) at 6 degrees temporal to the fovea. D: Retinal sensitivity per cone density was significantly lower in eyes with RPGR mutations than normal eyes and eyes with RHO mutations. E: Outer segment (OS) thickness per cone density was significantly lower in eyes with RPGR mutations than eyes with RHO mutations and also significantly lower than normal. F: Retinal sensitivity was not significantly different among eyes with RHO and RPGR mutations or different from normal when normalized for OS thickness. Modified fromFoote et al. (2020).

Fig. 11.

AOSLO images show changes in cone spacing over time in eyes with rod-cone degeneration. Images from 3 patients who received sham surgery (A and C, red outlines) in one eye and sustained-release ciliary neurotrophic factor (CNTF) (B, D and F, blue outlines) were imaged at baseline and 31–35 months later. Rectangles outline regions of interest where cone density was measured. E) Cone density decreased over time in all but 1 sham-treated region (red lines, n = 9), while all regions in the CNTF-treated eyes (n=12) remained within the measurement error (± 6.3%, gray shaded bar). Reproduced fromTalcott et al. (2011), copyright by the Association for Research in Vision and Ophthalmology.

Fig. 12.

Imaging rod photoreceptors in RP. Shown is a split detector image of the photoreceptor mosaic in a 49-year-old subject with autosomal recessive RP due to mutations in the USH2A gene. The panel on the right is the same image, with white arrows indicating the location of presumed remnant rod photoreceptors, which are much smaller in size than the larger cone inner segments. Scale bar = 50 μm.

Fig. 13.

AOSLO imaging and microperimetry demonstrates cone structure and function in eyes with choroideremia. AOMP demonstrates sensitivity near the margin of atrophy (yellow and orange spots), with dense scotomas (red spots) corresponding to outer retinal tubulations in confocal (A) or split detector (B) AOSLO images. C) Fundus-guided microperimetry shows retinal sensitivity in decibels ranging from good (yellow) to not seen (black) in an eye with choroideremia. The margin of retinal atrophy as determined from SW-AF images (black line) differs from the margin as defined from swept-source OCT images (blue line). Retinal sensitivity is measurable beyond the margin of retinal atrophy defined using both methods (orange spots), although the accuracy of the microperimetry stimulus delivery likely accounts for some of the difference. Other regions within the margin of retinal atrophy as defined from swept-source OCT images (blue line) show dense scotomas (black spots, Modified from (Foote et al., 2019b), copyright by the Association for Research in Vision and Ophthalmology and reprinted fromTuten et al. (2019)Copyright 2019, with permission from Elsevier.

Fig. 14.

Multimodal imaging in albinism. Shown are images depicting the features of foveal hypoplasia in a patient with oculocutaneous albinism. A: OCT B scan through the location of the incipient fovea (arrows delineate the region imaged with AOSLO in C). B: OCT-A image of the inner retinal vasculature showing the absence of a foveal avascular zone. C: AOSLO montage extending about 3 degrees temporal from the location of peak foveal cone density (asterisk). The extent of the montage is marked on the OCT image in panel A with small arrows under the RPE. Note the gradual decrease in packing density of cones moving away from the fovea. Scale bars for A and B = 200 μm. Scale bar for C = 25μm.

Fig. 15.

Multimodal imaging in closed-globe blunt ocular trauma subject WW_0923. The location of the en face OCT shown in (D) is outlined on the fundus photograph (A). En face imaging in this subject revealed a tripetaloid EZ disruption centered at the fovea (D). Dashed lines on (D) indicate locations of the horizontal and vertical B-scans (B and C), whereas the square represents the area shown in (E–G). Confocal AOSLO imaging revealed a similarly shaped, although enlarged, region of nonwaveguiding photoreceptors (F). Split-detector imaging revealed a nearly contiguous mosaic of photoreceptors, which change dramatically in size within a small area (G). There is only one small region at the bottom right corner of the disruption, where there seems to be a complete absence of photoreceptor structure (G). B–D. Scale bars = 200 μm. E–G. Scale bars = 100 μm. Reproduced fromScoles et al. (2016)with license from Wolters Kluewer Health Inc. Please note the Creative Commons license does not apply to this content. Use of the material in any format is prohibited without written permission from the publisher, Wolters Kluwer Health, Inc. Please contactpermissions@lww.comfor further information.

Fig. 16.

A schematic of cone mosaic illustrating the basis of some commonly used spatial cone metrics. Each cone within the rectangular region of interest (ROI) is represented by a circle, light gray circles with unfilled surrounding areas are cells with a Voronoi domain that is unbounded. In contrast, the black circles with a shaded surrounding area are cells with a bound Voronoi domain (with the color representing the number of sides of the Voronoi polygon). Bound density is estimated by taking the number of bound cells and dividing by the total area of the bound Voronoi cells. Unbound density is simply the total number of cells in the ROI divided by the area of the ROI. For each bound cell in the ROI, the mean intercell distance (ICD) to its immediate neighbors can be calculated, along with the nearest immediate neighbor and the farthest immediate neighbor (NND and FND, respectively). In a perfectly triangular mosaic, all bound Voronoi cells will have six sides (i.e, a hexagon). The percentage of bound cells with six sides can be used to assess mosaic packing geometry. Finally, regularity metrics (M) look at the mean value of a metric across all bound cells (μM) divided by the standard deviation of the metric across those same cells (σM).

Fig. 17.

The optoretinogram images biophysical changes in photoreceptor outer segments (OS) in response to light. A) Line scanning ophthalmoscope retinal image with stimulus (528nm ± 20nm) shown as horizontal red bars, B) OPL before stimulus shows minimal baseline activity, C) OPL 1.05 seconds after stimulus increases in stimulated regions. D) Stimulus areas of different size (yellow, 0.27 degrees2; gray, 0.20 degrees2; red, 0.14 degrees2; violet, 0.07 degrees2 corresponding to about 10 cones) elicit mean (solid) of 6 single OPL responses (dotted) at 0.27 degrees2 (E); stimuli of different sizes shown in (D) elicit OPL changes of similar magnitude (F). (G) AO image shows individual cones, and (H) shows greater OPL changes in L/M cones in response to 29.7% average L and M bleach, than in S cones exposed to 0.3% bleach; magnified inset from area outlined in dashed square reveals early reduction in OPL immediately after stimulus onset in L and M cones (orange), but negligible early response in putative S-cones (blue). I) Light (yellow shaded area) causes changes in the electrical potential and surface tension of the OS disc membrane; this causes the OS to change shape, flattening the disks immediately after stimulus onset (early response). This is followed by a slower increase in optical path length (OPL) (late response) caused by osmotic changes during phototransduction. The OS length changes can be measured as a change in OPL by interferometry using phase-resolved OCT; with adaptive optics, resolution improves to study individual cells. The magnitude of the OPL change increases (J) and the time to peak response decreases (K) with stimulus intensity. Modified with permission from (Pandiyan et al., 2020b). Available athttps://advances.sciencemag.org/content/6/37/eabc1124. It is made available under a CC-BY-NC 4.0 International license:https://creativecommons.org/licenses/by-nc/4.0/legalcode. Copyright the authors.