Cone photoreceptor dysfunction in retinitis pigmentosa revealed by optoretinography

Abstract

Retinitis pigmentosa (RP) is the most common group of inherited retinal degenerative diseases, whose most debilitating phase is cone photoreceptor death. Perimetric and electroretinographic methods are the gold standards for diagnosing and monitoring RP and assessing cone function. However, these methods lack the spatial resolution and sensitivity to assess disease progression at the level of individual photoreceptor cells, where the disease originates and whose degradation causes vision loss. High-resolution retinal imaging methods permit visualization of human cone cells in vivo but have only recently achieved sufficient sensitivity to observe their function as manifested in the cone optoretinogram. By imaging with phase-sensitive adaptive optics optical coherence tomography, we identify a biomarker in the cone optoretinogram that characterizes individual cone dysfunction by stimulating cone cells with flashes of light and measuring nanometer-scale changes in their outer segments. We find that cone optoretinographic responses decrease with increasing RP severity and that even in areas where cone density appears normal, cones can respond differently than those in controls. Unexpectedly, in the most severely diseased patches examined, we find isolated cones that respond normally. Short-wavelength-sensitive cones are found to be more vulnerable to RP than medium- and long-wavelength-sensitive cones. We find that decreases in cone response and cone outer-segment length arise earlier in RP than changes in cone density but that decreases in response and length are not necessarily correlated within single cones.

Keywords: adaptive optics; optical coherence tomography; optoretinography; photoreceptors; retinitis pigmentosa.

Fig. 1.

It is known that cone OSs briefly increase their optical path length (e.g., elongate as shown) when exposed to a light flash. We hypothesize that this response decreases with severity of the disease, with cones responding less when degenerating than when healthy. a, axon; ELM, external limiting membrane; G, Golgi; l, lysosome; m, mitochondria; md, membranous discs; r, reticulum; and s, soma.

Fig. 2.

Effect of RP disease on cone photoreceptor hyperreflective bands as observed in clinical OCT images. Cross-sectional (B-scan) images obtained with Spectralis OCT reveal the TZ between healthy and severely diseased retina. Shown are one of the controls and the three RP subjects with different stages of RP disease. Foveal center is at the right edge of image; images subtend 10° along the temporal horizontal meridian. The TZ (yellow) is defined as the area between the two retinal eccentricities where the COST and IS/OS hyperreflective bands merge with the underlying RPE band and thus become indistinguishable in the B-scan (12). Superimposed on the OCT images are the retinal locations (beige boxes) we imaged with AO-OCT (2, 4, 6, and 8°). RPE, retinal pigment epithelium.

Fig. 3.

Cone photoreceptor structure and function. AO-OCT en face images of the cone mosaic at 2° eccentricity in one of the controls (A, and B) and the three RP subjects: RP1 (C and D), RP2 (E and F), and RP3 (G and H). Magnified views are shown for cones in red boxes. Cone densities are 31,491 (Control1), 22,750 (RP1), 25,258 (RP2), and 14,041 (RP3) cones/mm². Traces of individual cones’ ΔOPL responses to 637-nm stimulus are shown in B, D, F, and H. Traces are randomly color coded, and histograms of peak responses (average ΔOPL from 0.75 to 1.25 s after stimulus) are shown to the right of each plot.

Fig. 4.

Key steps for processing the cone response traces, illustrated using control subject (Control3). The ∼1,000 cones respond differently depending on the wavelength of the flash stimulus: 637 (A), 528 (B), and 450 nm (C). Flash stimulus occurs at time equal to 0 s. Individual cone traces are randomly colored. The 5-s-long cone traces in A, B, and C are concatenated to form 15-s long traces (D) that are then transformed to PC space to reveal cone clusters (E), one for each of the three cone spectral types (S, M, and L). (F) A kernel density estimate is applied to the PC plot in E to locate the peak of each cluster, whose relative position to the origin [0,0] (green marker) is a measure of each spectral type’s collective response.

Fig. 5.

Cone ΔOPL response decreases as disease severity increases with retinal eccentricity in RP1. The cone response to the three LED stimuli is projected along the first two principal components following the process shown in Fig. 4. The first row is the age-matched control at 2 (A), 4 (B), 6 (C), and 8° (D) retinal eccentricity. (E) All cluster peaks from Control1 overlaid in one plot. Each color corresponds to one location. The second row is RP1 at 2 (F), 4 (G), 6 (H), and 8° (I) retinal eccentricity and a summary plot of the four previous locations (J). In all plots, solid green marker indicates [0,0] location. Black plus markers are the cluster positions of the control (Control1), as computed from the kernel density estimate. Open blue markers are cluster positions of RP1, also computed from the kernel density estimate. In J, black plus markers are the average cluster positions of S, M, and L cones of Control1 over the four retinal locations in A–D.

Fig. 6.

Relative ΔOPL response strength of cones as a function of retinal eccentricity and cone spectral type for RP1 (A) and RP2 (B) subjects. Response strength is relative to the 14-subject control group (100%). The 95% CIs of S, M, and L clusters are for the 14-subject control group and are centered on 100%. Traces are color coded by spectral type: S (blue), M (green), and L (red). For reference, the gray horizontal bars at the bottom denote the location and range of the TZs of the RP subjects.

Fig. 7.

Structural and functional evidence of S cones at 2° eccentricity in subject RP1. En face images are projections over different depths in the photoreceptor layer of the AO-OCT volume: IS/OS band (A), immediately below the IS/OS band (B), and COST band (C). Bands were determined from the projected cross-sectional image of the volume as labeled in D. The three colored rectangles specify the projection ranges and have vertical widths of 7 (green), 5 (blue), and 10 μm (brown). In B, the relatively sparse array of bright punctuated reflections (many indicated by red arrowheads) are consistent with S cones, whose inner segments are known to be 10% longer than M and L cones (36). For this retinal patch, the apparent S-cone IS/OS reflection is shifted about 5 μm deeper than that of M and L cones. The pattern of red arrowheads in B are superimposed in C, revealing that most S cones lack a reflection in the COST band. (E) ΔOPL response of the 11 strongest responding S cones to the 450-nm stimulus are shown, selected from the 45 S cones that were manually identified based on a bright reflection in B and a COST reflection at a depth shallower than the COST band in C (see details in Early-Stage RP Subject). (F) In PC space, the 11 apparent S-cones are colored blue, and those previously identified in Fig. 5F are colored gray. For reference, the black plus markers are the cluster positions of the control (Control1) at 2° eccentricity as computed from the kernel density estimate.

Fig. 8.

Cone ΔOPL response decreases as disease severity increases with retinal eccentricity in RP2. The cone response to the three LED stimuli is projected along the first two principal components following the process shown in Fig. 4. The first row is the age-matched control at 2 (A), 4 (B), and 6° (C) retinal eccentricity. (D) All cluster peaks from Control2 overlaid in one plot. Each color corresponds to one location. The second row is RP2 at 2 (E), 4 (F), and 6° (G) retinal eccentricity and a summary plot of the three previous locations (H). In all plots, solid green marker indicates [0,0] location. Black plus markers are the cluster positions of the control (Control2) as computed from the kernel density estimate. Open blue markers are cluster positions of RP2, also computed from the kernel density estimate. In H, black plus markers are the average cluster positions for S, M, and L cones of Control2 over the three retinal locations in A, B, and C.

Fig. 9.

Cones in RP3 are severely compromised except for a small fraction that respond normally. ΔOPL response in PC space in age-matched control (Control3) (A) and RP3 at 2° retinal eccentricity (B). The cone response to the three LED stimuli is projected along the first two PC components. Solid green marker indicates [0,0] location. Black plus markers are the cluster positions of the control (Control3) as computed from the kernel density estimate. While most cones are clustered around [0,0] in B, 10 cones identified as M (green) and L (red) based on their PC mapping exhibit a strong response (inside the two dashed boxes), some sufficiently so that they are indistinguishable from cones in the control. The sides of the boxes closest to the origin [0,0] are located halfway (50%) between the origin and black plus markers of the control. Traces are shown of all cones’ ΔOPL responses to the 528- (C) and 637-nm stimuli (D); histograms of peak responses (average of ΔOPL from 0.75 to 1.25 s after stimulus) are shown to the right of each plot. In C, the strong response of the seven M cones (green) is distinct from the other cones’ responses (gray), confirming the PC analysis results. Similarly, in D, the strong response of the three L cones (red) is distinct from the other cones’ responses (gray). (E) These 10 cones were then mapped back to the AO-OCT en face image and labeled with green (M cones) and red (L cones) circles. (F) Close-up of a smaller area highlights three of the strongly responsive cones, two identified as M cones and one as an L cone. White arrowheads point to bright cones that did not respond strongly to the stimuli. (Scale bars in E and F, 10 μm.)

Fig. 10.

Cone OS length is a coarse predictor of cone ΔOPL response, a relationship that varies with severity of the RP disease. The magnitude of each cone’s response in PC space is plotted against OS length. Cones are classified as S (blue), M (green), L (red), and undetermined (black). The distribution of cone OS lengths is shown at the top of each plot and color coded by spectral type. Control and RP subjects are in A–C and D–F, respectively. Control1 (A) and RP1 (D) cones were obtained from 6° retinal eccentricity (SI Appendix, Fig. S4 A and D); Control2 (B) and RP2 (E) cones were obtained from 4° retinal eccentricity (SI Appendix, Fig. S4 B and E); and Control3 (C) and RP3 (F) cones were obtained from 2° retinal eccentricity (SI Appendix, Fig. S4 C and F). Black lines are linear regression fits.