Insights Into PROM1-Macular Disease Using Multimodal Imaging

Abstract

Purpose: To describe the features of genetically confirmed PROM1-macular dystrophy in multimodal images.

Methods: Thirty-six (36) eyes of 18 patients (5–66 years; mean age, 42.4 years) were prospectively studied by clinical examination and multimodal imaging. Short-wavelength autofluorescence (SW-AF) and quantitative fundus autofluorescence (qAF) images were acquired with a scanning laser ophthalmoscope (HRA+OCT, Heidelberg Engineering) modified by insertion of an internal autofluorescent reference. Further clinical testing included near-infrared autofluorescence (NIR-AF; HRA2, Heidelberg Engineering) with semiquantitative analysis, spectral domain-optical coherence tomography (HRA+OCT) and full-field electroretinography. All patients were genetically confirmed by exome sequencing.

Results: All 18 patients presented with varying degrees of maculopathy. One family with individuals affected across two generations exhibited granular fleck-like deposits across the posterior pole. Areas of granular deposition in SW-AF and NIR-AF corresponded to intermittent loss of the ellipsoid zone, whereas discrete regions of hypoautofluorescence corresponded with a loss of outer retinal layers in spectral-domain optical coherence tomography scans. For 18 of the 20 eyes, qAF levels within the macula were within the 95% confidence intervals of healthy age-matched individuals; nor was the mean NIR-AF signal increased relative to healthy eyes.

Conclusions: Although PROM1-macular dystrophy (Stargardt disease 4) can exhibit phenotypic overlap with recessive Stargardt disease, significantly increased SW-AF levels were not detected. As such, elevated bisretinoid lipofuscin may not be a feature of the pathophysiology of PROM1 disease. The qAF approach could serve as a method of early differential diagnosis and may help to identify appropriate disease targets as therapeutics become available to treat inherited retinal disease.

Prominin-1 (PROM1; MIM# 604365) is a transmembrane glycoprotein that is widely expressed in human tissues including both rods and cones of retina. Prominin-2 shares 60% amino acid identities with PROM1 and because of the presence of Prominin-2 in most tissues except those of the eye, PROM1-related disease is only observed in retina.¹ 

PROM1 (CD133) is known as a stem cell marker and has been identified in the apical plasma membrane of epithelial cells of several tissues.^2,3 In the retina, PROM1 is present in the photoreceptor outer segments⁴ and is considered to play a role in photoreceptor disk morphogenesis.⁵ Loss of PROM1 function results in cone and rod degeneration and visual impairment.⁴ More recently, knock-out of PROM1 in cultured RPE was reported to increase mTORC1 and mTORC2 signaling and to decrease autophagosome trafficking to the lysosome, whereas overexpression of PROM1 had the opposite effect: mTORC1 and mTORC2 were inhibited.⁶ 

Mutations in PROM1 can present with features of retinitis pigmentosa,^4,7 cone–rod dystrophies,^8–12 and macular dystrophies.^13,14 Autosomal dominant forms have been linked mostly to late-onset disease involving the central retina, whereas recessive forms have been associated with early-onset more severe retinal degeneration.^15,16 PROM1-associated macular dystrophy also known as Stargardt disease 4 (STGD4) presents with a phenotype having some similarities to ABCA4-related Stargardt disease 1 (STGD1).^5,17–19 Specifically, in PROM1 macular dystrophy, retinal imaging reveals mild to severe macular atrophy and hyperautofluorescent flecks can be visible in short wavelength fundus autofluorescence (AF) images. 

The aim of this study was to evaluate short wavelength autofluorescence (SW-AF) and near-infrared autofluorescence (NIR-AF) together with structural correlates evidenced by spectral-domain optical coherence tomography (SD-OCT) in patients with PROM1 macular dystrophy. Because, in phenotypically similar ABCA4-related macular disease, the accumulation of bisretinoid lipofuscin in RPE cells accounts for the increased signal observed in SW-AF images,²⁰ we also measured SW-AF intensities (quantitative fundus autofluorescence [qAF]) in patients diagnosed with PROM1 disease. We found that the phenotypic similarities between ABCA4-associated and PROM1 disease are likely based on photoreceptor cell dysfunction and degeneration that primarily involves the macula, but that proceeds without a background of increased bisretinoid lipofuscin. 

Methods

Subjects

This prospective cross-sectional study included patients who presented to the Edward S. Harkness Eye Institute at Columbia University Irving Medical Center. Institutional Review Board (IRB)/Ethics Committee approval was obtained under protocol IRB-AAAI9906. Informed written consent was obtained before study enrollment. Informed written assent was obtained from the individual younger than the age of 18 years (patient 13) under supervision from a parent. All study-related procedures adhered to the tenets established in the Declaration of Helsinki. After study enrollment, retinal images used for qAF were acquired and whole blood was obtained by venipuncture for DNA isolation and genetic screening. Additional anonymized clinical data from each study participant was obtained including the clinical notes and standard-of-care testing results when available. Patients underwent a comprehensive ophthalmic examination by a retina specialist (S.H.T.). The fundus was examined under pupillary dilation and best corrected visual acuity was determined. Full-field ERG was recorded according to the International Society for Clinical Electrophysiology of Vision standards.²¹ 

Image Acquisition and Analysis

SW-AF (488-nm excitation, 500–680 nm emission) images (30° × 30° fields) were acquired with a modified confocal scanning laser ophthalmoscope (Spectralis HRA+OCT; Heidelberg Engineering, Heidelberg, Germany) in automatic real-time tracking capture mode. qAF images were acquired in a non-normalized mode to avoid histogram stretching, and signal intensities were measured at an eccentricity of 7° to 9° in eight predetermined circular segments. The qAF values were calculated using a published formula that adjusts for variation in laser power, detector sensitivity, refractive error, and anterior media transmission. The method has been described in more detail in our previous work.²² Comparison was made to a normative database of healthy eyes (374 eyes; age range, 5–60 years). NIR-AF (787-nm excitation, >830 nm emission) images (30° × 30° fields) were acquired with a confocal scanning laser ophthalmoscope (Spectralis HRA+OCT; Heidelberg Engineering). Semiquantitative NIR-AF (qNIR-AF) analysis has been described in our earlier work.²³ We used the i2kRetina software (DualAlign LLC, Clifton Park, NY, USA) for the alignment of SW-AF, NIR-AF, and IR-R images. 

Color fundus photographs were obtained with a FF 450plus Fundus Camera (Carl Zeiss Meditec, Jena, Germany). Ultrawidefield (2008) pseudocolor images were captured with a Zeiss Clarus 700 (Carl Zeiss Meditec) and Optos Optomap (Optos Daytona; Optos, Inc., Marlborough, MA, USA) was operated in the composite pseudocolor mode. SD-OCT line and volume scans in high-resolution mode and a corresponding infrared reflectance fundus image were acquired with a confocal scanning laser ophthalmoscope (Spectralis HRA+OCT; Heidelberg Engineering). 

Retinal layer thickness measurements were performed using horizontal 9-mm (high-resolution mode) SD-OCT scans centered at the fovea in both eyes of all study patients and one eye of 25 age-matched healthy controls (mean age = 47.5 years; age range, 13–84 years) (Supplemental Table S2). The boundaries of total receptor+ (TREC+) were delineated using the automated segmentation tool in the HEYEX software (Heidelberg Engineering). TREC+ was defined as the distance between the inner nuclear layer/outer plexiform layer boundary and the Bruch's membrane/choroid interface. The segmentation protocol was performed in both patients and controls in an unmasked manner. The inner nuclear layer/outer plexiform layer and Bruch's membrane/choroid interface borders were automatically segmented and, when necessary, subsequent manual corrections were made by a single grader (W.L.). For each patient, individual thickness profiles spanning 6 mm (3 mm in the temporal and nasal directions from the fovea) were compared with the thickness profiles of control subjects. Significant thinning was defined as values lower than 2 standard deviations from the control mean. Statistical analyses and thickness plots were analyzed using R (https://www.r-project.org).²⁴ For qNIR-AF and TREC+ SD-OCT analyses, measurements were confined to the perifoveal area (2.5 mm eccentricity, spatially matching the qAF eccentricity).²² 

Genetic Testing

Genomic DNA was extracted from peripheral blood lymphocytes from each patient and exome capture kits v.5 and v.7 from Agilent Technologies (Santa Clara, CA, USA) and Integrated DNA Technologies’ (Coralville, IA, USA) Exome Research Panel v.1 were performed. Sequence reads were aligned to the hg19 reference genome using a Burrows–Wheeler Transform, and processed with GATK according to best practices recommendations. 

After variant calling, all identified variants were functionally annotated with the ANNOVAR program²⁵ using the dbnsfp version 4.2a dataset.^26,27 Pathogenicity scores for PolyPhen2 (HVAR), SIFT4G, M-CAP, REVEL, Eigen, CADD version 1.6, DANN, and ClinPred were considered. The mutational intolerance landscape and tolerance scores (dn/ds) of PROM1 (NM_006017) were derived using the MetaDome web server (https://stuart.radboudumc.nl/metadome/dashboard).²⁸ Putative splicing defects were assessed for intronic variants using SpliceAI²⁹ and splicing module in the Alamut software version 2.2 (Interactive Biosoftware, Rouen, France). Allele frequencies of all variants were obtained from the gnomAD database (https://gnomad.broadinstitute.org/gene/ENSG00000007062?dataset=gnomad_r2_1) (accessed July 2022). 

Results

Clinical Data

Eighteen subjects (36 eyes) were analyzed in this study with the mean age being 42.4 years (range, 5–66 years). A summary of the clinical and demographic characteristics of the cohort is presented in Table 1; the patients’ ages at the time of the imaging are presented. Nine patients reported a family history consistent with autosomal dominant inheritance (Fig. 1). The family history and related family members within this study are shown in Table 1 and Figure 1. No patients reported using any medication toxic to the retina. 

Table 1.

 

Demographic, Clinical, and Genetic Characteristics of Patients With PROM1-associated Maculopathy

 

Figure 1.

 

PROM1 macular dystrophy. Four pedigrees with dominant retinal disease inheritance patterns. Subjects with PROM1 macular dystrophy are shown in black. “–” is for pathogenic variants and “+” denotes a wild-type allele. Male, square; female, circle. Diagonal line denotes a deceased individual. Filled symbols are for subjects presenting with PROM1 disease and open symbols for a subject without known PROM1-disease. SW-AF images of the patients are shown next to each pedigree.

Fifteen patients presented with central visual disturbances starting in the second or third decade of life; three were asymptomatic relatives (P8, P10, and P13) or otherwise had preserved best-corrected visual acuity (20/20) (Table 1). Five patients reported difficulty with night vision or photopsias. 

Full-field ERG recordings were unremarkable in 6 of the 11 subjects tested. Moderate attenuation of the 30-Hz flicker and single flash cone responses were detected in the remaining five patients (P1, P5, P16, P17, and P18). 

Retinal Imaging

Fundus examination revealed that, in 15 of the 18 patients, disease was confined to the macular area. In one family in which individuals (P7 and P8) were affected across two generations, the fundus exhibited granular fleck-like lesions across the posterior pole (Fig. 1). In another patient (P11) with more advanced disease, peripapillary atrophy was present and central atrophy reached the major vascular arcade (Fig. 1). None of the patients presented with bone spicules. Color fundus images are shown in Supplementary Figure S1. Although in most cases the disease was confined to central macula, in four cases the disease reached the peripapillary region (P7, P8, P11, and P17). Only one of those four patients (P7) demonstrated peripapillary sparing. 

AF images revealed macular atrophy and RPE mottling in the SW-AF and NIR-AF modalities (Fig. 2). Granular hypoautofluorescence was present in almost all eyes (36/38). Small hyperautofluorescent foci surrounding central atrophy were seen in SW-AF images in 30 eyes (Fig. 2A, red arrow; P3, P6, P9, and P15) and a hyperautofluorescent halo was seen in 20 eyes (20/38) (Fig. 2A, blue arrow; P1, P9, P15, and P17). 

Figure 2.

 

SW-AF (A), NIR-AF (B), and SD-OCT (C) in patients with PROM1. Macular dystrophy presents as granular alterations in SW-AF and NIR-AF images. Loss of the ellipsoid zone (EZ) band is evident in the SD-OCT scans. Hyperautofluorescent foci are marked with a red arrow in SW-AF (A: P3, P6, P9, and P15) and NIR-AF (B: P6) images. Hyperautofluorescent halos are indicated with blue arrows in SW-AF (A: P1, P9, P15, and P17) and in NIR-AF (B: P9, P15, and P17) images. Advanced outer retinal disruption or atrophy is seen as hypertransmission into the choroid in SD-OCT images (C: P1, P9, P15, and P17; yellow stars).

NIR-AF imaging was available for nine patients (18 eyes) and showed decreased granular signal in the central macular area (Fig. 2B). Hyperautofluorescent foci were observed in only one subject (2/18 eyes) (Fig. 2B, red arrow; P6) and a hyperautofluorescent halo was seen in 3 subjects (6/18 eyes) (Fig. 2B, blue arrow; P9, P15, and P17). 

In SD-OCT images, all patients presented with thinning of outer retina. These changes ranged from discontinuities in the ellipsoid zone band (Fig. 2C; P3 and P6) to total outer retinal atrophy (Fig. 2C; P1, P9, P15, and P17). This presentation fulfilled the criteria for defining the lesion as complete RPE and outer retinal atrophy, the classification system used to describe progression of geographic atrophy.³⁰ In the macula, increased transmission of SD-OCT signal into the choroid was commonly observed, the extent of which varied (Fig. 2C, yellow stars; P1, P9, P15, and P17). Areas of granular hypoautofluorescence in SW-AF and NIR-AF images corresponded to partial outer retinal degeneration (Figs. 2A, C; P3 and P6), whereas discrete regions of hypoautofluorescence corresponded to complete loss of outer retinal layers and RPE in SD-OCT scans (Figs. 2A, C; P9 and P15). Foveal sparing was evaluated in SD-OCT, SW-AF, and NIR-AF images and was observed in 31 of 38 eyes. 

Videos of aligned SW-AF, NIR-AF, and IR-R images are included as Supplementary Videos S1 through S7 for those patients for whom these aforementioned images were available (P3, P4, P6, P9, P12, P13, and P17). For P11, the image quality was not sufficient for the software to complete the alignment. 

qAF analysis in 12 patients (20 eyes) indicated that in 18 eyes the qAF values were within the 95% confidence intervals calculated for healthy eyes. In one eye (P17 OD), the qAF values were higher and in one eye (P8 OS) lower than the 95% confidence interval (Fig. 3A). For comparison. we included previously published data from Burke et al.²⁰ (2014) (Fig. 3A); these data show that patients with STGD1 presented with qAF values above the 95% confidence interval. Comparison with color-coded qAF maps of healthy age-matched individuals (Fig. 3B) also demonstrated that there was no generalized increase of qAF signal in patients with PROM1 (Fig. 3C; P3 and P9). Centrally decreased qAF signal corresponded to areas of outer retinal atrophy (Fig. 3C). 

Figure 3.

 

Quantitative fundus autofluorescence (qAF8) intensities recorded from patients with PROM1 are presented with previously published data acquired from patients with STGD1 and reported in Burke et al. 2014.²⁰ qAF values acquired from patients with PROM1 (red circles) are plotted as a function of age together with mean (solid black line) and 95% confidence interval (CI) (dashed lines) of the healthy control group. Values acquired from patients with STGD1 are represented by blue circles. (A) Color-coded maps of qAF distribution in healthy age-matched individuals (B) and patients with PROM1 (C). Only 2 of the 17 qAF8 values acquired from patients with PROM1 were outside the 95% confidence intervals of the healthy controls; one eye was above and the other eye was below the range. In the color-coded images, decreased SW-AF signal in the central macula corresponded with atrophic changes.

Semiquantitative analysis of NIR-AF signal (8 patients,16 eyes) (Fig. 4A) along a horizontal axis through the fovea revealed reduced mean qNIR-AF in central macula at an eccentricity from 0 to 2 mm relative to the fovea in both nasal and temporal retina. This area also presented with outer retinal atrophy in SD-OCT scans (Fig. 2C; P9). Outside this central area (2–4 mm eccentricity), however, the mean qNIR-AF intensity in patients with PROM1 was comparable with that of healthy subjects (Fig. 4B). 

Figure 4.

 

Quantitative NIR-AF analysis in patients with PROM1. NIR-AF intensity profiles acquired from patients with PROM1 are plotted as mean (red solid line) and upper and lower 95% confidence intervals (CI) (dashed red lines) together with mean NIR-AF values of healthy subjects (green solid line) and corresponding 95% CI (dotted green line) (A). In patients with PROM1, the NIR-AF values are decreased in the central macula due to outer retinal disruption or atrophy seen in PROM1 macular dystrophy; however, the NIR-AF values in the adjacent macular area are comparable with those of healthy subjects (A). NIR-AF image of a healthy aged-matched eye (left) and PROM1 patient (P9) (right) (B). The horizontal axes along which NIR-AF values were recorded are shown as green and red lines (B).

TREC+ thickness measurements using SD-OCT scans were acquired from 17 patients (34 eyes). Subnormal values were observed in 15 patients (28 eyes) and only two subjects (P3 and P6) presented with values within the normal range (Table 2). 

Table 2.

 

Comparative Assessment Across Modalities of Patients With PROM1-associated Maculoapthy

 

The results of qAF, qNIR-AF values, and TREC+ OCT thickness measurements were compared with each other. For spatial correspondence, the values in the qAF8 ring (eccentricity of 7°–9°; approximately 2.1–2.7 mm) were compared with corresponding regions examined by qNIR-AF and SD-OCT TREC+. Although TREC+ values were normal in four patients, the qAF values were within the normal range in all but two eyes. We evaluated individual qNIR-AF values at an eccentricity of 7° to 9° outside the fovea and found that intensities were normal in four eyes, elevated in six eyes, and subnormal in six eyes (Table 2). 

Retinal Findings in Consecutive Family Members

An analysis of patients presenting with PROM1 disease more than two generations showed that the retinal findings were mostly similar within a family with the older subjects presenting with more advanced retinal degeneration (Fig. 1). However in a two-generation family (P7 and P8), the daughter presented with a more widespread retinal degeneration and more advanced disease compared with the mother. SW-AF demonstrated uneven signal and small hyperautofluorescent foci reaching from central macula to the posterior pole and nasal to the optic disc (Fig. 1). Despite being younger, the daughter (P7) also presented with loss of SW-AF signal in a well delineated central area corresponding to chorioretinal atrophy in SD-OCT scans, whereas in the mother relative sparing of the fovea was seen in SD-OCT. Peripapillary sparing was observed in the SW-AF images in the daughter but not in the mother (Fig. 1). This family has been described in greater detail in our previous work.³¹ 

Genetic Analysis

Molecular screening identified 19 likely pathogenic alleles composed of 11 missense, 1 nonsense (stop-gain), 1 frameshift, and 6 intronic variants. All variants have either been previously associated with disease such as the c.1117C>T (p.Arg373Cys) variant (CADD = 24.8, REVEL = 0.66) identified in seven patients—P9, P10, P11, P12, P12, P13, P14, and P15—or are predicted to be deleterious by computational algorithms (Supplementary Table S1). All patients were heterozygous for variants except P6, who was homozygous for c.314A>G (p.Tyr105Cys). This variant occurs in a highly conserved nucleotide position (phyloP: 7.40 [−20.0; 10.0]) that is intolerant to variation (dn/ds = 0.44) and, consequently, unanimously predicted to be pathogenic across all algorithms considered (Supplementary Table S1). The noncanonical splice site variant 2373+5G>T in P16 abolishes the upstream donor site (SpliceAI Δscore = 0.84) (Supplementary Fig. S2), which affects the exon 23/intron 23 splice junction. 

Discussion

In all but three patients, PROM1 macular dystrophy was confined to the central macula (Figs. 1, 2). P7, P8, and P11 (ages 34, 63, and 64 years) presented with a more widespread retinal degeneration affecting the entire macula (Fig. 1; P7, P8, and P11). PROM1-related recessive disease has been associated with a more severe retinal degeneration^15,16; however, in our cohort P6, who had a biallelic mutation in PROM1, presented with a mild maculopathy with no peripheral retinal changes and the ERG recording was within normal limits. 

The outer retinal degeneration in central macula was seen as an aberrant granular signal in SW-AF and NIR-AF images. Because SW-AF signal is absorbed by macular pigments in the fovea, atrophic changes in central macula were better visualized with NIR-AF imaging (Fig. 2B). Small hyperfluorescent foci surrounding the central retinal atrophy were observed in 30 of the 38 eyes; these hyperautofluorescent foci were more visible in SW-AF images than NIR-AF images. There was also a trend toward subnormal values of outer retinal thickness in almost all patients. The PROM1-associated disease process seems to affect the perifoveal region first and then advances to peripheral macular areas, while leaving the fovea intact until late-stage disease; foveal sparing was observed in 31 of the 38 eyes and was not detected in 7 eyes (P7 OU, P11 OU, P16 OU, and P18 OS). Thinning of the RPE/Bruch's membrane and increased sub-RPE transmission of OCT signal were observed; both of these changes are indicative of RPE cell loss. Unexplained is why the dysfunction of a protein (PROM1) expressed in photoreceptor cells results in RPE degeneration. 

PROM1 maculopathy is often compared to ABCA4-associated disease. We also observed that the current PROM1 cohort presented with clinical findings similar to STGD1. These features include macular atrophy that in some cases is surrounded by a hyperautofluorescent band, autofluorescence mottling, and hyperautofluorescent foci. Conversely, the involvement of the peripapillary area in P8, P11, and P17 was unlike the peripapillary sparing observed in STGD1.³² In STGD1, a decrease in or loss of the ABCA4 protein leads to mishandling of retinaldehyde and unabated random reactivity of the aldehyde moiety with amine-bearing lipids.^20,33–36 As a result, a mixture of di-retinal adducts (bisretinoids) are produced in photoreceptor cells at levels that exceed their formation in healthy retina. These fluorophores subsequently accumulate in RPE, are the source of SW-AF and account for the downstream cellular damage, leading to disease processes in ABCA4 disease.³⁷ 

The measurement of SW-AF signal strength by qAF is an indirect measure of retinal bisretinoids. An analysis of cone–rod and macular dystrophies by Gliem et al.³⁸ previously revealed increased levels of qAF in two patients diagnosed with PROM1-related disease. A greater SW-AF signal has been demonstrated in ABCA4-related disease relative to age-matched healthy subjects,^20,36 with the increase in qAF signal observed even in healthy appearing retina in SD-OCT scans. However, the increased qAF that is a hallmark of ABCA4-associated disease^36,39 was not observed in PROM1-positive cases in our analysis of 12 subjects. As is the case for non–ABCA4-related bull's eye maculopathy,⁴⁰ qAF may be effective in distinguishing between ABCA4 and PROM1-related macular dystrophy. 

We have observed previously that, in ABCA4 disease, high levels of lipofuscin not only present as elevated SW-AF signal, but an increase in NIR-AF signal is also observed.²³ To determine whether PROM1 macular dystrophy presents with distinctive findings regarding NIR-AF, we also measured the NIR-AF signal intensities in the patients. Although the NIR-AF intensities in the foveal area were slightly decreased, in the perifoveal area the NIR-AF signal was comparable with healthy eyes. 

A limitation of this study was that several participants were lacking some of the imaging modalities analyzed here as well as some ocular parameters. Nevertheless, we believe the number of patients was enough to support our conclusions or observed trends. We also acknowledge that the lack of longitudinal data does not allow us to follow the disease progression. Additionally in this study of PROM1 patients, atrophic lesions were confined to a central zone of less than 7° eccentricity. This distribution may be indicative of early and severe cone-driven disease.¹⁵ Because our measurements of qAF are acquired in segments situated 7° to 9° outside the fovea, we cannot exclude the possibility that the SW-AF fluorophores derived from cones were initially elevated. 

Our findings do not indicate that excessive accumulation of bisretinoid lipofuscin is a component of the PROM1-related disease process. That knowledge could be useful in developing future therapeutic targets for treating PROM1-retinal disease and also as a diagnostic tool in distinguishing STGD1 and STGD4 before genetic testing. 

Acknowledgments

Supported by grants from the National Eye Institute EY012951, P30EY019007, and EY009076; Foundation Fighting Blindness; and a grant from Research to Prevent Blindness to the Department of Ophthalmology, Columbia University. Jonas Children's Vision Care is supported by R01EY018213 and R01EY024698. 

Disclosure: M. Paavo, None; W. Lee, None; R. Parmann, None; J.R. Lima de Carvalho, None; J. Zernant, None; S.H. Tsang, Abeona Therapeutics, Inc (F), Emendo Bio, Inc. (F), Rejuvitas (O), Nanoscope Therapeutics (S); R. Allikmets, None; J.R. Sparrow, None