Quantitative Fundus Autofluorescence and Optical Coherence Tomography in ABCA4 Carriers

Abstract

Purpose: To assess whether carriers of ABCA4 mutations have increased RPE lipofuscin levels based on quantitative fundus autofluorescence (qAF) and whether spectral-domain optical coherence tomography (SD-OCT) reveals structural abnormalities in this cohort.

Methods: Seventy-five individuals who are heterozygous for ABCA4 mutations (mean age, 47.3 years; range, 9–82 years) were recruited as family members of affected patients from 46 unrelated families. For comparison, 57 affected family members with biallelic ABCA4 mutations (mean age, 23.4 years; range, 6–67 years) and two noncarrier siblings were also enrolled. Autofluorescence images (30°, 488-nm excitation) were acquired with a confocal scanning laser ophthalmoscope equipped with an internal fluorescent reference. The gray levels (GLs) of each image were calibrated to the reference, zero GL, magnification, and normative optical media density to yield qAF. Horizontal SD-OCT scans through the fovea were obtained and the thicknesses of the outer retinal layers were measured.

Results: In 60 of 65 carriers of ABCA4 mutations (age range, 9–60), qAF levels were within normal limits (95% confidence level) observed for healthy noncarrier subjects, while qAF levels of affected family members were significantly increased. Perifoveal fleck-like abnormalities were observed in fundus AF images in four carriers, and corresponding changes were detected in the outer retinal layers in SD-OCT scans. Thicknesses of the outer retinal layers were within the normal range.

Conclusions: With few exceptions, individuals heterozygous for ABCA4 mutations and between the ages of 9 and 60 years do not present with elevated qAF. In a small number of carriers, perifoveal fleck-like changes were visible.

The adenosine triphosphate-binding cassette, subfamily A, member 4 (ABCA4) gene, which is located on the short arm of chromosome 1, encodes for a membrane-associated protein located in outer segment (OS) disc membranes of rod and cone photoreceptors.^1,2 ABCA4, originally identified as rim protein,³ participates in the transfer of retinaldehyde from the interior of the OS disc to the cytosol, after photobleaching of rhodopsin. Failure of this process leads to an increased formation of bisretinoid fluorophores due to condensation reactions of retinaldehyde; these fluorophores subsequently accumulate in the retinal pigment epithelium (RPE) as lipofuscin.^4–6 Numerous toxic effects of bisretinoid have been demonstrated in in vitro studies.⁷ Additionally, in Abca4^−/− mice that accumulate bisretinoids such as A2E in abundance, photoreceptor cells degenerate.⁸ Although accumulation of RPE lipofuscin, albeit at lower levels, is also part of the normal aging process, it is widely accepted that excessive accumulation of RPE lipofuscin is the damaging agent in recessive Stargardt disease (STGD1).^9,10 The role of lipofuscin accumulation in age-related macular degeneration (AMD), a complex multifactorial disease, remains to be determined. 

Homozygous and compound heterozygous ABCA4 mutations are associated with multiple retinal dystrophy phenotypes including autosomal recessive STGD1, cone-rod dystrophy-3, and retinitis pigmentosa-19.^11–14 Although establishing genotype–phenotype correlations has met with difficulty, given the extraordinary allelic heterogeneity in ABCA4 with currently more than 800 known disease-associated genetic variants, ABCA4-related disease is thought to be inversely correlated with the residual ABCA4 activity. For instance, patients with a severe reduction in ABCA4 activity manifest early and severe phenotypic changes. Approximately 5% of individuals of European descent carry a disease-associated ABCA4 allele.^15,16 This high carrier frequency has implications for the extent to which ABCA4 variants contribute to the burden of retinal disease. Carriers of ABCA4 mutations may be at an increased risk of developing AMD,^17–20 may reveal subtle visual dysfunction in psychophysical and electrophysiological tests,²¹ and may demonstrate moderate to severe fundus changes.^17,22 

The increased accumulation of lipofuscin in the RPE of patients with biallelic mutations in ABCA4 has been documented by histology,⁹ by spectrofluorometry,²³ and more recently by quantitative autofluorescence (qAF).²⁴ It is still unknown, however, whether individuals heterozygous for ABCA4 mutations also have elevated lipofuscin levels due to reduced ABCA4 activity. Data from the Abca4 mouse model support this hypothesis. Thus we and others^25,26 have found that in mice heterozygous for a null mutation in Abca4, A2E levels are elevated relative to those in wild-type. 

In this study we used qAF to measure the effect of monoallelic ABCA4 mutations on RPE lipofuscin levels. Essential to the qAF methodology are an internal fluorescent reference installed in the confocal scanning laser ophthalmoscope (cSLO) and a rigorous protocol for image acquisition. Measurement of fundus AF levels is enabled by an internal reference, which is imaged simultaneously with the fundus to account for changes in laser power over time and to adjust for differences in the sensitivity setting of the cSLO.²⁷ In addition, fundus AF gray levels are corrected for magnification and optic media density. We also analyzed spectral-domain optical coherence tomography (SD-OCT) scans acquired from carriers of ABCA4 mutations to assess whether structural alterations could be observed. 

Methods

Genetic Testing and Subjects

Asymptomatic carriers (subjects, S) of ABCA4 mutant alleles were recruited prospectively after disease-causing mutations in ABCA4 were confirmed in their affected family member (probands, P). In one case (S39) the carrier was screened for the purpose of family planning. Screening of clinically diagnosed STGD1 patients involved various versions of the ABCA4 chip, including early chips (∼300 mutations) and more recent versions of the array (>600 variants). When array screening identified only one mutated ABCA4 allele or no ABCA4 mutations, next-generation sequencing (NGS) was carried out. In the latter case, the 50 exons and exon–intron boundaries of the ABCA4 gene were amplified (Illumina TruSeq Custom Amplicon protocol; Illumina, San Diego, CA, USA) and then submitted to NGS on the Illumina MiSeq platform with analysis using the variant discovery software NextGENe (SoftGenetics LLC, State College, PA, USA) and reference genome GRCh37/hg19. Variants were confirmed by Sanger sequencing and analyzed with Alamut software (http://www.interactive-biosoftware.com [in the public domain]). After confirming the mutations in the proband, screening of the identified mutations by direct Sanger sequencing was carried out in the parents and siblings of the proband. 

Seventy-five individuals heterozygous for disease-causing ABCA4 mutations from 46 unrelated families were recruited between December 2011 and December 2014. The mean age of the carrier cohort was 47.3 years; range was 9 to 82 years, and 46 of the subjects were females. Table 1 summarizes demographic and genetic information of the carriers and their familial relationship to the probands (54 parents, 17 siblings, 2 progeny, 1 grandparent, 1 spouse sequenced for family planning). Sixty-five of these carriers were aged 60 and below and thus could be included in the qAF analysis. The remaining 10 carriers were included in the study as part of our examination for fundus changes. 

Table 1

 

Heterozygous ABCA4 Carriers: Summary of Demographic, Clinical, and Genetic Data

 

For comparison, we also included 57 affected family members with biallelic ABCA4 mutations in the study. The mean age of the ABCA4-patient cohort was 23.4 years; range was 7 to 67 years, and 35 of the patients were females. Demographic and genetic information of the affected family members with biallelic ABCA4 mutations is presented in Table 2. For an intrafamily comparison of qAF levels, we also included two noncarrier siblings in the study (S39.5, S27.5). 

Table 2

 

Biallelic ABCA4 Patients: Summary of Demographic, Clinical, and Genetic Data

 

All subjects were examined by a retinal specialist. Snellen visual acuity, converted to logMAR visual acuity, was obtained using the most recent refractive correction. All procedures adhered to the tenets of the Declaration of Helsinki, and written informed consent was obtained from all subjects after a full explanation of the procedures was provided. The protocol was approved by the Institutional Review Board of Columbia University. 

Study subjects included in the qAF analysis (see below) were compared to our database of 374 healthy eyes of 277 subjects (age 5–60 years) that was reported previously.²⁸ This group of subjects included the following ethnicities and ages (mean age, age range): 87 whites (31.8, 6–58), 79 Hispanics (30.1, 5–60), 47 blacks (36.6, 9–56), 43 Asians (36.5, 22–59), 6 Indians (32.1, 25–39), and 15 others (27.5, 10–57). Carriers included in the quantitative analysis of the SD-OCT scans (see below) were compared to 46 age-similar control subjects (mean age ± standard deviation [SD], 45.3 ± 15.8 years; range, 11–84 years; 24 females). All but six (ages: 62–84 years) of these controls were subjects from our previously published normative qAF database.²⁸ 

Imaging

Short-wavelength (SW) fundus AF images were acquired with a confocal scanning laser ophthalmoscope (Spectralis HRA+OCT; Heidelberg Engineering, Heidelberg, Germany) modified by the insertion of an internal fluorescent reference to account for variable laser power and detector gain.²⁷ Excitation was 488 nm, and the barrier filter in the device transmitted light from 500 to 680 nm. All AF images were recorded for a 30° × 30° field (768 × 768 pixels) in the high-speed mode (8.9 frames/s) as a video consisting of either 9 or 12 frames. Before image acquisition, pupils were dilated to at least 7 mm with topical 1% tropicamide and 2.5% phenylephrine, and room lights were dimmed. The camera was aligned in all three dimensions so that the AF beam was located in the center of the pupil and the fundus image was evenly illuminated. The focus was fine-tuned to the point of maximum signal intensity across the fundus. The detector sensitivity was adjusted to avoid nonlinear effects. While these adjustments took place, the fundus was exposed to the AF light for at least 20 seconds to reduce AF attenuation by rod photopigment to <2%.²⁷ Two or more AF images were recorded, followed by a second imaging session for reproducibility. All videos were reviewed, and those frames without localized or generalized decreased AF signal were aligned, averaged, and saved in “nonnormalized” mode (two images per session). 

In addition, a horizontal 9-mm SD-OCT image through the fovea, registered to a simultaneously acquired AF or near-infrared reflectance (NIR-R) image, was recorded in high-resolution mode as an average of 50 to 100 individual images for each eye. Color fundus photography was performed using a FF450+IR fundus camera (Carl Zeiss Meditec, Jena, Germany). 

AF Image Analysis

Autofluorescence images were analyzed under the control of an experienced operator (TD, WL) with dedicated image analysis software written in IGOR (WaveMetrics, Lake Oswego, OR, USA).²⁷ The software recorded the mean GLs of the internal reference and from eight circularly arranged segments positioned at an eccentricity of approximately 7° to 9° (Fig. 1). Segments were scaled to the distance between the fovea and the temporal edge of the optic disc. An algorithm of the software accounted for the presence of vessels in the sampling area.²⁷ The gray levels (GLs) from the eight circularly arranged segments in each image were calibrated to the reference, zero GL, magnification, and normative optical media density²⁹ to yield qAF₈. Gray levels were also adjusted to a reference calibration factor calculated using a master fluorescent reference.²⁷ 

Figure 1

 

Quantitative fundus autofluorescence image analysis. S22.3. Mean gray levels (GLs) are recorded from the internal reference (white rectangle, top of image) and from 8 circularly arranged segments (red). The segments are scaled to the distance between the temporal edge of the optic disc (white vertical line) and the center of the fovea (white cross). After accounting for the presence of large vessels, qAF values of the 8 segments are averaged to determine qAF₈.

SD-OCT Segmentation

Horizontal SD-OCT line scans through the fovea from one eye of the first 58 consecutively recruited carriers of ABCA4 mutations (see Table 1) were segmented in Matlab (MathWorks, Natick, MA, USA) by one of the authors (GES), using a manually corrected automated program. The segmentation technique has been described previously.^30,31 When SD-OCT scans from both eyes were available, one eye was randomly chosen for analysis. Five borders, as indicated in Figure 2, were segmented: (1) the border between the vitreous and inner limiting membrane (ILM); (2) the border between the inner nuclear layer (INL) and the outer plexiform layer (OPL); (3) the proximal border of the band corresponding to the ellipsoid zone (EZ); (4) the proximal border of the retinal pigment epithelium (pRPE); and (5) the border between Bruch's membrane (BM) and the choroid. The thicknesses of two layers, the OS plus layer (OS+, from the BM/choroid border to the proximal border of the EZ band) and the total receptor layer (TRec, from the BM/choroid border to the OPL/INL border), were determined (Fig. 2). 

Figure 2

 

Retinal layer segmentation of SD-OCT scans. Image of horizontal SD-OCT scan from a healthy control subject. Segmented boundaries are indicated as colored lines: red, the border between vitreous and inner limiting membrane (ILM); white, the border between inner nuclear layer (INL) and outer plexiform layer (OPL); green, proximal border of the ellipsoid zone (EZ); pink, the proximal border of the retinal pigment epithelium (RPE); and blue, the border between Bruch's membrane (BM) and choroid. Two layers were derived from these boundaries: OS+, distance between EZ and BM/choroid; TRec, distance between INL/OPL and BM/choroid.

Statistical Analyses

Analyses were performed using Prism 5 (GraphPad Software, La Jolla, CA, USA) and the statistical tests as indicated. We used the Bland-Altman method³² to test the between-session repeatability of qAF measurements and the agreement of qAF measurements between eyes in the carrier population and the ABCA4-affected population. 

Results

Heterozygous ABCA4 mutations were detected in 75 subjects from 46 unrelated families. All mutations were known to be disease causing. The following ABCA4 mutations were frequently present in our cohort: p.G1961E in 11 carriers, p.[L541P; A1038V] in six carriers, p.P1380L in four carriers, and p.L2027F in three carriers. In the group of 57 affected family members, two disease-causing ABCA4 variants were found in 52 patients (91%) while one disease-causing ABCA4 variant was detected in the other five patients (Table 2). 

Quantitative Fundus Autofluorescence

Since age-related changes in ocular media are more pronounced after age 60, and because our normative database was also limited to that age range,²⁸ images from subjects and patients above age 60 were not utilized for qAF. Determination of qAF₈ levels was performed on 107 eyes of 65 carriers of ABCA4 mutations (mean age, 44.3 years; range, 9–60 years). For comparison, we also analyzed the qAF images of 77 eyes of 44 affected family members with biallelic ABCA4 mutations (mean age, 22.7 years; range, 7–52 years); data from 25 of the patients were published previously (Table 2). All subjects and patients had clear media except for some floaters. The qAF data of the carriers of ABCA4 mutations presented in this study were based on AF images of 107 eyes, with 82 of these eyes having a second AF imaging session. The qAF data of the 19 not previously reported affected family members with biallelic ABCA4 mutations presented in this study were based on AF images of 34 eyes, with 26 of these eyes having a second AF imaging session. 

In 102 of 107 eyes of heterozygous subjects (95%), qAF₈ was within normal limits for age and race/ethnicity (95% confidence level; calculated as 2× SD of the residuals of the mixed effects linear regression analysis, with residuals being the deviations of the observed data points from the predicted values that fit the line) (Fig. 3). Three eyes were above the upper limits (S2.3, OD; S20.3, OD; S38.3, OD), and 2 eyes were below the lower limits (S19.4, OD; S26.3, OD) for healthy subjects. As expected, qAF₈ of the subjects heterozygous for ABCA4 mutations increased with age. Of the patients affected with ABCA4-associated disease, 74 of 77 eyes had qAF₈ levels above the normal limits (95% confidence) for age. In agreement with a previous study,²⁴ the highest qAF₈ levels were found for younger ABCA4-affected patients, while the fold difference relative to normal AF levels was less pronounced for older ABCA4-affected patients. 

Figure 3

 

Quantitative fundus autofluorescence intensities plotted as a function of age. Values are the mean of the 8 segments (qAF₈) shown in Figure 1 and measured in carriers of ABCA4 mutations (red circles), ABCA4-affected patients (blue squares), and subjects with healthy eyes (mean, solid line; upper and lower limits [95% confidence level], dotted lines) of (A) whites (unfilled symbols) and Indians (filled symbols), (B) blacks (filled symbols) and Asians (unfilled symbols), and (C) Hispanics. The values for Indian carriers and probands are plotted with white subjects because the upper 95% CI of whites and Indians is similar.²⁸ Values for both eyes or one eye (23 carriers and 11 affected patients) are plotted.

We previously demonstrated that STGD1 patients carrying the p.G1961E mutation on one allele have relatively lower qAF₈ levels compared to patients with p.[L541P; A1038V], p.P1380L, and p.L2027F mutations (and no p.G1961E mutation on the other allele).²⁴ To determine whether similar differences in the segregation of qAF₈ levels could also be observed for ABCA4 mutations in carriers and whether specific ABCA4 mutations may be associated with higher qAF₈ levels, we plotted qAF values for carriers and affected patients who carried one of the four most common mutations (p.G1961E, p.[L541P;A1038V], p.P1380L, and p.L2027F) (Fig. 4). The qAF₈ levels of the carriers appeared to be relatively evenly distributed regardless of the ABCA4 mutation. Among the carriers, there was no trend for a mutation to be associated with relatively higher or lower qAF₈ levels. 

Figure 4

 

Quantitative autofluorescence intensities associated with 4 common ABCA4 mutations: p.G1961E, p.P1380L, p.[L541P; A1038V], p.L2027F. Values are plotted for carriers and probands (male and female) as indicated by colors and symbols. Other ABCA4 mutations carried by the probands are represented in black. Mean (solid black line) ± 95% confidence intervals (dashed lines) for individuals with healthy eyes are shown. Values are for OD, except in one case where only OS was available. The values (4 carriers and one proband) in the cluster between ages 44 and 52 are replotted in the inset above using expanded scales on both axes (x, y).

In Figure 5, the pedigrees of families 39, 26, 1, and 27 are shown together with the qAF color maps of all corresponding family members included in the study. While age-similar noncarrier and carriers have comparable qAF levels and a normal spatial distribution of the AF signal, the increased qAF levels of affected family members with biallelic ABCA4 mutations are immediately discernable from the qAF color maps. 

Figure 5

 

Color-coded maps of quantitative fundus autofluorescence and inheritance patterns of families carrying ABCA4 mutations p.P1380L (family 39) p.G1961E; p.[L541P, A1038V] (family 26), p.[L541P; A1038V] (family 1), and p.G1961E; p.P1380L (family 27). qAF maps are shown for probands, carriers of ABCA4 mutations, and non-ABCA4 carriers; color-code scale is shown below. Male, square; female, circle. Two generations (I, II) are shown for each family. S, subject; P, patient.

For ABCA4 carriers, the Bland-Altman coefficient of agreement for qAF₈ of right and left eyes (42 subjects) was ±18.7%, and the between-session Bland-Altman coefficient of repeatability was ±10.3% (n = 82). The coefficient of agreement for qAF₈ between the right and left eye in affected family members with biallelic ABCA4 mutations that were not previously reported²⁴ (n = 14) was ±20.5%, and the between-session coefficient of repeatability was ±7.3% (n = 26). 

Qualitative Analysis of Fundus Images

Fundus images of the recruited carriers of ABCA4 mutations were also assessed qualitatively. For most subjects, color fundus photography, conventional fundus AF imaging, and SD-OCT were qualitatively unremarkable. However, as shown in Figure 6, four carriers (S9.3, S7.2, S41.2, and S43.2) exhibited fleck-like changes in the perifoveal region. These changes were also visible in color fundus photographs (not shown) and are similar to the flecks seen in STGD1 disease.³³ The foci of increased AF signal corresponded to hyperreflective deposits traversing photoreceptor-attributable bands in SD-OCT images. A fleck that had faded to darkness on AF (temporal fleck, S7.2) presented with hyporeflectivity at the fleck position on OCT. 

Figure 6

 

Fundus changes in a subgroup of heterozygous carriers of ABCA4 mutations. Near-infrared reflectance (NIR-R), short-wavelength fundus autofluorescence (SW-AF), and SD-OCT images of subjects (S) S9.3, S41.2, S7.2, S43.2. The NIR-R and SW-AF images were registered. The axis and horizontal extent of the SD-OCT scan are indicated in the corresponding fundus images. Outer nuclear layer (ONL), external limiting membrane (ELM), photoreceptor ellipsoid zone (EZ), interdigitation zone (IZ), and retinal pigment epithelium/Bruch's membrane (RPE). The nomenclature used for the identification of reflectivity bands in SD-OCT was previously published (Staurenghi et al.).⁴⁴ In all 4 subjects, perifoveal fleck-like changes are visible in SD-OCT images; these changes correspond to hyperreflective foci on NIR-R and have an increased AF signal.

SD-OCT Thickness Measurements

The first 58 consecutively recruited heterozygous subjects (58 eyes of 58 carriers) were included (mean age, 47.1 years; range, 11–82 years) in the thickness measurements of the OS+ and TRec layers (Table 1). In Figures 7 and 8, the individual SD-OCT thickness profiles of these carriers of ABCA4 mutations are shown in gray together with the 95% confidence intervals (CI) for controls (mean ± 1.96× standard error of mean [SEM]; [1.96× SD/√(n-1)] (bold solid and dashed black lines). In Figure 7, the thickness profiles of S43.2, S7.2, S9.3, and S41.2 are indicated. These subjects exhibited qualitative fundus abnormalities that were visible in SW-AF and SD-OCT (Fig. 6). In Figure 8, the segmentation profiles of carriers expressing the most common mutations, p.G1961E, p.[L541P; A1038V], p.P1380L, and p.L2027F, are indicated in color. While some of the thickness values of the carriers fell above or below the CI of controls, there was no clear trend toward thinning or thickening of the segmented retinal layers throughout the cohort. Interestingly, the variation in the thicknesses measured was greater for OS+ than for TRec profiles. 

Figure 7

 

Thickness profiles acquired by segmentation of spectral-domain optical coherence tomography (SD-OCT) images of carriers of ABCA4 mutations. Profiles are shown in color for carriers S43.2, S7.2, S9.3, and S41.2; these carriers presented with qualitative fundus changes in SW-AF and SD-OCT as shown in Figure 6. Thickness profiles of individual carriers are shown as gray lines. Thickness profiles of controls are presented as mean (black solid line) ± 95% confidence intervals (mean ± 1.96× standard error of mean [SEM]; [1.96× SD/√(n-1)]; black dashed lines). Thicknesses of OS+ layer (from EZ to border between Bruch's membrane and choroid) (A, C, E) and TRec (from border between inner nuclear layer and outer plexiform layer to border between Bruch's membrane and choroid) (B, D, F) are presented as a function of distance from the fovea. Right eyes are presented. Subjects are grouped by ages, and numbers of carriers in each group (n) are indicated.

Figure 8

 

Thickness profiles acquired by segmentation of spectral-domain optical coherence tomography (SD-OCT) images of carriers of ABCA4 mutations p.G1961E, p.L541P/A1038V, p.P1380L, and p.L2027F. Thicknesses of OS+ layer (from EZ to border between Bruch's membrane and choroid) (A, C, E) and TRec (from border between inner nuclear layer and outer plexiform layer to border between Bruch's membrane and choroid) (B, D, F) are presented as a function of distance from the fovea. Right eyes are presented. Thickness profiles of individual carriers are shown as gray lines. Thickness profiles of controls are presented as mean (black solid line) ± 95% confidence intervals (mean ± 1.96× standard error of mean [SEM]; [1.96× SD/√(n-1)]; black dashed lines). Subjects are grouped by ages, and numbers of carriers in each group (n) are indicated.

Discussion

Homozygous and compound heterozygous mutations in the ABCA4 gene are associated with macular dystrophies that include STGD1 and cone-rod dystrophy.^12,13,18 The inheritance pattern of ABCA4-associated disease is exclusively autosomal recessive, and the age of onset of disease is variable. Elevated RPE lipofuscin^9,23 that can be measured as increased qAF²⁴ is typical of disease linked to ABCA4 mutations. In our recent study of qAF in STGD1 patients,²⁴ limited genotype–phenotype correlations were possible. Nevertheless, we concluded that based on qAF values (measured at an eccentricity of 7°–9°), the mutations p.L2027F and p.P1380L and the complex allele p.[L541P; A1038V] conferred a faster rate of lipofuscin accumulation, whereas accumulation in the presence of the p.G1961E and p.G851D mutations was slower. Here we report that carriers of mutations in ABCA4, recruited as family members of affected patients, do not exhibit elevated qAF intensities when compared to controls. In a carrier of a mutant allele, half of the protein is defective. Alternatively, there could be posttranscriptional mechanisms that adjust ABCA4 protein to required levels irrespective of whether one or two functional copies of the gene are present.³⁴ 

In the Abca4^−/− mouse, fundus autofluorescence intensities, measured as qAF²⁶ or corrected gray levels,³⁵ were found to be 2- to 2.6-fold greater than in wild-type mice, and A2E levels, measured chromatographically, were 4-fold higher.²⁶ In the heterozygous mice, qAF intensities were increased by approximately 15% while A2E levels were amplified 2-fold. In another study,²⁵ A2E accumulation was determined to be several-fold higher in Abca4^± than in wild-type mice. We also found that ONL thinning, a measure of photoreceptor cell demise, was a feature of both Abca4^−/−and Abca4^± mice, but the thinning was less pronounced in the latter. With absence of one copy of a gene, as in the Abca4^± heterozygous mouse, the amount of expressed protein is expected to be 50% of that in the wild-type. We do not have corresponding information in human ABCA4-associated disease. Several years ago a case–control study of unrelated subjects with AMD identified heterozygous ABCA4 mutations in a subgroup of AMD cases; six of these patients harbored the p.G1961E mutation.¹⁸ A follow-up study detected the p.G1961E variant statistically significantly more frequently in AMD cases than in matched controls.³⁶ The p.G1961E mutation is exceptional in that STGD1 patients homozygous for the mutation or compound heterozygous for p.G1961E and another disease-associated allele exhibit qAF levels (measured 7°–9° from fovea) that are either within the normal range or modestly higher (Fig. 4).²⁴ These observations with respect to G1961E are consistent with an earlier study wherein most patients with this mutation did not present with a dark choroid during fundus angiography.³⁷ The dark choroid is thought to be conferred by high lipofuscin levels. Thus since STGD1 patients expressing the G1961E mutation have relatively normal qAF intensities, the finding that carriers of a G1961E mutation also do not exhibit elevated qAF is not informative with respect to the burden of disease. 

An association between AMD and heterozygosity for ABCA4 mutations has not been replicated in all studies,^38,39 yet an analysis of 23 families revealed that carriers of ABCA4 disease-causing mutations, that is, relatives of STGD1 probands, were significantly more likely than by chance to be affected by AMD.⁴⁰ Additionally a subgroup of patients diagnosed with AMD are reported to have geographic atrophy and a SW-AF phenotype that overlaps with STGD1 (fine granular pattern with peripheral punctate spots, GPS[+]). Fritsche et al.¹⁷ demonstrated that patients with the GPS[+] phenotype were often heterozygous for ABCA4 mutations. This group of patients did not possess the second ABCA4 mutant allele required for STGD1; therefore they did not represent late-onset STGD1, which is sometimes phenotypically confused with AMD. The GPS[+] phenotype was most strongly associated with the p.A1038V ABCA4 allele (5/20 patients). In our cohort, none of the carriers presented with geographic atrophy and the GPS[+] phenotype. Nevertheless, our carrier subjects were significantly younger (range, 11–82 years; mean, 47.3 ± 14.2 years) than the GPS[+] phenotype cohort reported by Fritsche et al. (range, 50–84 years; mean, 63.6 ± 10.4 years).¹⁷ Importantly, Fritsche et al.¹⁷ also noted that only a small fraction of carriers of ABCA4 mutations develop this particular phenotype (1/60). The latter finding may be indicative of a role for other unknown modifiers that contribute to this specific phenotype in the presence of a heterozygous mutant ABCA4 allele. 

Heterozygous mutations in ABCA4 (p.V2050L) have also been reported to contribute to an exacerbation of the phenotype conferred by a monoallelic mutation in PRPH2 (p.R172W).⁴¹ In another study, 12 nonsymptomatic mutation-carrying relatives of STGD1 patients were found to have normal visual acuity but impaired contrast sensitivity and reduced multifocal ERG amplitudes.²¹ More recently, a subset of ABCA4 carriers were reported to have reduced visual acuity, fundus abnormalities that included pigmentary changes (8/18) and flecks, and multifocal ERGs of reduced amplitude and delayed implicit times.²² Some missense mutations, including the complex allele p.[L541P;A1038V], have been shown to be associated with ABCA4 mislocalization; the p.L541P mutation in particular prevents correct localization of ABCA4 in OS and thus retention in inner segments.⁴² It is expected that under these conditions the pathogenesis of STGD1 could be exacerbated by endoplasmic reticulum (ER) stress and the unfolded-protein response (UPR).⁴³ Since the onset of the UPR may be dependent on mutation type and gene dosage, a monoallelic p.[L541P; A1038V] mutation may leave photoreceptor cells with a limited and latent capacity to deal with other sources of ER stress. 

Only four carriers had fundus abnormalities in the form of central fleck-like changes. The foci of increased and decreased AF signals corresponded to hyper- and hyporeflective deposits traversing the photoreceptor-attributable bands in the SD-OCT images. The pathognomonic significance of this observation remains to be elucidated. In addition, quantitative analysis of the SD-OCT images showed that for the majority of carriers there was no clear trend toward thinning or thickening of the segmented retinal layers. 

Limitations of our study included age restrictions. For instance, subjects older than age 60 were not included in the qAF analysis because of reduced ocular media transmission. Nevertheless, these older subjects may reveal changes that are not detectable at earlier ages. Imaging of pseudophakic subjects could enable the study of these older age groups. Perhaps if we had used psychophysical and electrophysiological tests,²¹ other changes would have been revealed. All abnormal fleck-like changes we observed (S7.2, S9.3, S41.2, S43.2) were situated perifoveally, while qAF₈ measurements were taken at an eccentricity of 7° to 9° (see Fig. 1), a location outside the area of fundus change. Another limitation of this study is that the control group was not genotyped for ABCA4. Given the estimated carrier frequency of ∼5% in the general population, we cannot exclude that some of the healthy control subjects were actually carrying a disease-associated ABCA4 allele. 

In summary, we observed somewhat unexpectedly that most individuals between the ages of 9 and 60 who are heterozygous for disease-causing mutations in ABCA4 do not present with qAF levels higher than the normal range. In four carriers, central fleck-like changes were visible in SW-AF and SD-OCT images. Otherwise, carriers had normal retinal structure on SD-OCT. 

Acknowledgments

Supported in part by grants from the National Eye Institute/National Institutes of Health EY024091, EY021163, EY019007; Arnold and Mabel Beckman Foundation; Foundation Fighting Blindness; a grant from Research to Prevent Blindness to the Department of Ophthalmology, Columbia University; Robert L. Burch III Fund, Columbia University, New York, New York, United States; New York Community Trust–Fredrick J. and Theresa Dow Wallace Fund, Columbia University, New York, New York, United States. The authors alone are responsible for the content and writing of the paper. 

Disclosure: T. Duncker, None; G.E. Stein, None; W. Lee, None; S.H. Tsang, None; J. Zernant, None; S. Bearelly, None; D.C. Hood, None; V.C. Greenstein, None; F.C. Delori, None; R. Allikmets, None; J.R. Sparrow, None