Quantitative Fundus AutofluorescenceDistinguishes ABCA4-Associated andNoneABCA4-Associated Bull’s-EyeMaculopathy

Tobias Duncker, MD,1 Stephen H. Tsang, MD, PhD,1,2 Winston Lee, MA,1 Jana Zernant, MS,1

Rando Allikmets, PhD,1,2 François C. Delori, PhD,3 Janet R. Sparrow, PhD1,2

Purpose: Quantitative fundus autofluorescence (qAF) and spectral-domain optical coherence tomography(SD OCT) were performed in patients with bull’s-eye maculopathy (BEM) to identify phenotypic markers that canaid in the differentiation of ABCA4-associated and noneABCA4-associated disease.

Design: Prospective cross-sectional study at an academic referral center.Subjects: Thirty-seven BEM patients (age range, 8e60 years) were studied. All patients exhibited a localized

macular lesion exhibiting a smooth contour and qualitatively normal-appearing surrounding retina without flecks.Control values consisted of previously published data from 277 healthy subjects (374 eyes; age range, 5e60years) without a family history of retinal dystrophy.

Methods: Autofluorescence (AF) images (30�, 488-nm excitation) were acquired with a confocal scanninglaser ophthalmoscope equipped with an internal fluorescent reference to account for variable laser power anddetector sensitivity. The grey levels (GLs) from 8 circularly arranged segments positioned at an eccentricity ofapproximately 7� to 9� in each image were calibrated to the reference (0 GL), magnification, and normative opticalmedia density to yield qAF. In addition, horizontal SD OCT images through the fovea were obtained. All patientswere screened for ABCA4 mutations using the ABCR600 microarray, next-generation sequencing, or both.

Main Outcome Measures: Quantitative AF, correlations between AF and SD OCT, and genotyping forABCA4 variants.

Results: ABCA4 mutations were identified in 22 patients, who tended to be younger (mean age, 21.9�8.3years) than patients without ABCA4 mutations (mean age, 42.1�14.9 years). Whereas phenotypic differenceswere not obvious on the basis of qualitative fundus AF and SD OCT imaging, with qAF, the 2 groups of patientswere clearly distinguishable. In the ABCA4-positive group, 37 of 41 eyes (19 of 22 patients) had qAF8 of more thanthe 95% confidence interval for age. Conversely, in the ABCA4-negative group, 22 of 26 eyes (13 of 15 patients)had qAF8 within the normal range.

Conclusions: The qAF method can differentiate between ABCA4-associated and noneABCA4-associatedBEM and may guide clinical diagnosis and genetic testing. Ophthalmology 2015;122:345-355 ª 2015 by theAmerican Academy of Ophthalmology.

Supplemental material is available at www.aaojournal.org.

Varying definitions of bull’s-eye maculopathy (BEM) havebeen reported1e4 since the term was introduced in 1966 todescribe the fundus appearance of chloroquine retinopathy.5

The classical BEM phenotype presents as concentric par-afoveal rings of increased and decreased fundus auto-fluorescence (AF).6e8 However, the rings in BEM can beless obvious because of mottling, bright or reduced fovealAF, diffuse hyperautofluorescence, or a combinationthereof, and the lesion can be round or elliptical.6,8,9 TheBEM phenotype also can progress to extensive macularatrophy.6 Some patients who are homozygous for disease-causing mutations in the ABCA4 gene can present with the

� 2015 by the American Academy of OphthalmologyPublished by Elsevier Inc.

BEM phenotype.2,4 However, BEM is not specific toABCA4 and also can be caused by mutations in photo-transduction-related genes (e.g., GUCA1A, RPGR),10,11

mutations in a gene (PROM1)6 that encodes a proteininvolved in outer segment morphogenesis, and mutations ina gene (RIM1)7 that encodes a photoreceptor synaptic pro-tein. Thus, pinpointing the underlying cause of BEM in anindividual patient can be challenging.

Advances in noninvasive imaging have facilitated thediagnosis and differentiation of retinal dystrophies. Twoimaging methods that have proven especially valuable in thisregard are spectral-domain optical coherence tomography

345http://dx.doi.org/10.1016/j.ophtha.2014.08.017ISSN 0161-6420/14

Ophthalmology Volume 122, Number 2, February 2015

(SD OCT) and fundus AF. By SD OCT, structural changesof the retina can be studied almost at a cellular level. FundusAF emerges from retinal pigment epithelium (RPE) lip-ofuscin,12 a mixture of fluorophores with spectral emissionfeatures that reflect the bisretinoid constituents that have beencharacterized.13 Retinal pigment epithelium lipofuscin accu-mulates with age in healthy eyes,14,15 and its formation isaccentuated in recessive Stargardt disease (STGD1).16e19

Disease-related processes also can alter the distribution of theAF signal. Through image registration, SD OCT and AF canbe correlated.

Using quantitative fundus AF (qAF), an approach wedeveloped to measure fundus AF intensities, we recentlyreported increased qAF levels in STGD1 patients withconfirmed ABCA4 mutations.20 In these patients, qAF levelswere elevated even at young ages and in qualitatively un-affected-appearing fundus areas. Whether qAF levels alsoare increased in retinal dystrophies related to mutationsother than ABCA4 and to what extent lipofuscin is involvedin the pathogenesis of these conditions remains to bedetermined.

The objective of this study was to investigate whetherqAF, a noninvasive approach to measuring RPE lipofuscinin vivo, could aid in distinguishing ABCA4-associated fromnoneABCA4-associated disease in BEM, a phenotype thatis common to some retinal dystrophies. In addition, we wereinterested to determine whether, based on qualitativeassessment of AF and SD OCT images, a distinction be-tween ABCA4-positive and ABCA4-negative patients isfeasible.

Methods

Patients and Genetic Testing

Thirty-seven BEM patients (age range, 8e60 years) from 35families were recruited prospectively at the Department ofOphthalmology, Columbia University. All subjects were examinedby a retinal specialist (S.H.T.) and had clear media except for somefloaters. Patients exhibiting the BEM phenotype were selected onthe basis of fundus AF images. All patients exhibited a localizedmacular lesion exhibiting a smooth contour; outside the macula,the retina was qualitatively normal and without flecks. Of theABCA4-positive patients who have undergone qAF imaging at ourinstitute, approximately 35% exhibited a BEM phenotype withoutflecks. The criteria of BEM without flecks included patients inwhom flecks may have developed at a later stage. Demographic,clinical, and genetic information on the study cohort is presented inTable 1.

The ABCA4 microarray was used for initial screening of moststudy subjects, followed by direct Sanger sequencing to confirmidentified changes, as described previously.21 Because the arrayscreening had been performed over many years, different versionsof the ABCA4 chip had been used, from the least representative(approximately 300 mutations) to the recent version of the array(>600 variants). More recently recruited patients and patients inwhom the array screening identified only 1 mutated ABCA4 alleleor no ABCA4 mutations at all were subjected to next-generationsequencing.

All 50 exons and exoneintron boundaries of the ABCA4 genewere amplified using TruSeq Custom Amplicon protocol (Illumina,San Diego, CA), followed by sequencing on the Illumina MiSeq

346

platform. The next-generation sequencing reads were analyzed andcompared with the reference genome GRCh37/hg19 using thevariant discovery software NextGENe (SoftGenetics LLC, StateCollege, PA). All detected variants possibly associated with dis-ease were confirmed by Sanger sequencing and analyzedwith Alamut software (http://www.interactive-biosoftware.com).Segregation of the new variants with the disease was analyzed infamilies if family members were available.

All procedures adhered to the tenets of the Declaration ofHelsinki, and written informed consent was obtained from allsubjects after a full explanation of the study procedures was pro-vided. The study was approved by the Institutional Review Boardof Columbia University and complied with the Health InsurancePortability and Accountability Act of 1996.

Image Acquisition

Protocols for the acquisition of AF images that meet the qualitystandards necessary for quantification have been described previ-ously.14,20,22,23 Fundus AF images (30�; 488-nm excitation) wereacquired using a confocal scanning laser ophthalmoscope (Spec-tralis HRAþOCT; Heidelberg Engineering, Heidelberg, Germany)modified by the insertion of an internal fluorescent reference toaccount for variations in laser power and detector gain.22 Thebarrier filter in the device transmitted light from 500 to 680 nm.Before image acquisition, pupils were dilated to at least 7 mm withtopical 1% tropicamide and 2.5% phenylephrine. With room lightsturned off, a near-infrared reflectance image was recorded first.After switching to fundus AF mode (488-nm excitation; beampower, <260 mW), the camera was moved slowly toward the pa-tient to allow the patient to adapt to the blue light. Patients wereasked to focus on the central fixation light of the device. Thefundus was exposed for 20 to 30 seconds to bleach rhodopsin,22

whereas focus and alignment were refined to produce a maximumand uniform signal over the entire field. The detector sensitivitywas adjusted so that the grey levels (GLs) did not exceed the linearrange of the detector (GL, <175).22 Two or more images then wererecorded (each of 9 frames, in video format) in the high-speedmode (8.9 frames/second) within a 30��30� field (768 � 768pixels). To assess the repeatability of qAF measurements, a secondsession of AF images was recorded after repositioning the patientand camera. After imaging, all videos were inspected for imagequality and consistency in GLs. For an imaging session, 2 videoswere selected to generate the AF images for analysis. At least 4 of9 frames without localized or generalized decreased AF signal(because of eyelid interference or iris obstruction) and no largemisalignment of frames (causing double images after alignment)were considered for each image. The frames then were aligned andaveraged with the system software and saved in nonnormalizedmode (no histogram stretching). In total, AF imaging was per-formed for 73 eyes, and for 52 of those eyes, a second AF imagingsession was completed. After reviewing the frames of all AFvideos, 18 imaging sessions were excluded because image qualitywas not considered sufficient enough for quantification. Hence,qAF images of 67 eyes, 40 of which had second imaging sessionswere included in this study. Quantitative AF images from all eyesincluded in this study are presented in the Supplemental Appendix(available at www.aaojournal.org).

In addition, a horizontal 9-mm SD OCT image through the fovearegistered to a simultaneously acquired AF or near-infrared reflec-tance (NIR-R) image was recorded in high-resolution mode as anaverage of 100 individual images for each eye. The optical depthresolution in the Spectralis is currently approximately 7 mm. In caseswhere the SD OCT image had been registered to an NIR-R image,i2kRetina software (DualAlign LLC, Clifton Park, NY) was used for

Table 1. Summary of Demographic, Clinical, and Genetic Data

PatientNo. Sex Age (yrs)

Race orEthnicity

Best-Corrected VisualAcuity (Logarithm

of the Minimum Angleof Resolution Units)

Genetic DataAverage of

Quantitative FundusAutofluorescence Values

of the 8 SegmentsABCA4 Mutations

Right Eye Left Eye Allele 1 Allele 2 Right Eye Left Eye

1 M 36 White 0.70 0.70 p.G1961E p.G1961E 282 2792 F 46 White 0.40 0.40y p.G1961E p.R1129C 3913 M 17 Asian Indian 0.70 0.88 p.G1961E c.6729þ4_þ18del 340 3634 M 17 White 0.88 0.88 p.G1961E p.A1773V 340 3665 M 21 White 0.88 0.88 p.G1961E p.W15* 341 3256 F 22 White 0.70 0.48 p.G1961E p.G863A 361 3517 F 20 White 0.70z 0.88z p.G1961E p.L541P 3178 M 12 White 0.80 0.70 p.G1961E p.P1380L 251 2429 F 21 White 0.88 0.88 p.G1961E p.R212C 407 43910 F 26 White 0.40z 0.70z p.G1961E c.5196þ1056A/G 379 34411 F 24 White 0.88z 0.88z p.G1961E p.C2150R 405 39612 F 24 White 0.30z 0.18z p.G1961E p.N96D 513 54913 F 20 White 0.30 0.40 p.G1961E p.N96D 397 35514 M 25 White 0.00y 0.30 p.G1961E p.Q1003* 322 32815 M 8.2 White 0.88 0.88 p.[L541P; A1038V] 438 43216 M 25 White 0.60 0.60 p.S84fs p.R2107H 29417 F 24 Black 0.70 0.88 p.G991R p.L1138P 321 32618 M 26 White 0.00y 0.00y p.R1300* p.R2106C 419 41219 M 11 White 0.40z 0.40z p.W821R p.C2150Y 304 29620 F 16 White 0.70 0.40 p.K1547* p.R2030Q 481 51321 F 13 White 1.30 1.00 pR1108C p.Q1412* 511 52822 F 18 White 0.00 0.00 p.G863A c.5898þ1G/A 465 431

Mutations in Other Genes23 F 16 White 0.40 0.48 GUCY2D e p.R838H 152 16524 M 53 Black 0.88 0.88 CNGA3 e p.[R223W; R223Q] 25925 M 28 Hispanic 0.30y 0.30y CRX e p.S150* 346 34926 F 51 Hispanic 0.88 1.30 CRX e p.S150* 277 28427 F 35 White 0.88 0.88 ND 44328 M 48 Black 0.40y 0.18 RPGR e p.N1132fs 293 28329 M 51 Hispanic 1.00 1.00 ND 35930 M 57 White 0.48 0.48y WES, ND 320 36431 M 38 White 0.48 0.48 WES, ND 226 23732 M 33 Black 0.70 0.70 ND 285 30333 F 22 White 0.60 0.48 WES, ND 239 27934 F 60 White 0.30 1.70 ND 486 53635 M 54 White 0.70 0.88 ND 349 43636 M 21 White 0.60 0.88 WES, ND 23937 M 57 White 0.10y 1.18 ND 447 414

F ¼ female; M ¼ male; ND ¼ not determined; WES ¼ whole-exome sequencing.*Indicates stop codon.yFoveal sparing.zOptical empty lesion.

Duncker et al � qAF and SD OCT in Bull’s-Eye Maculopathy

alignment of the AF image of the same field. The assignment ofreflectivity bands was based on the study by Spaide and Curcio.24

Image Analysis

Fundus AF images were analyzed under the control of an experi-enced operator (T.D.) with dedicated image analysis softwarewritten in IGOR (WaveMetrics, Lake Oswego, OR) to determineqAF.22 The software recorded the mean GLs of the internalreference and of 8 circularly arranged segments positioned at aneccentricity of approximately 7� to 9� (Fig 1). Segments werescaled to the horizontal distance between the fovea and the tem-poral edge of the optic disc. If the fovea could not be identified,

then its position was estimated based on an SD OCT scan that wasregistered to an AF or NIR-R image. When a myopic crescentobscured the true edge of the optic disc on AF, an NIR-R imageaided in identifying the disc edge (crescent has high reflectance).The software accounted for the presence of vessels and markedatrophy in the segments and automatically excluded segments thatwere positioned, even partially, at a distance farther than 15� fromthe center of the image. The mean GLs of each segment then weremeasured, and qAF for that segment was calculated, taking intoaccount the reference calibration factor, the internal reference GL(0 GL), magnification, and optical media density from normativedata on lens transmission spectra.22,24 For each eye, the averageqAF from the 8 segments (qAF8) from all available images per eye

347

Figure 1. Fundus autofluorescence image analysis of patient 14 showingmean grey levels recorded from the internal reference (white rectangle, topof the image) and from 8 circularly arranged segments (outlined in white).The horizontal distance, FD, between the temporal edge of the optic disc(white vertical line) and the center of the fovea (white cross) was used todefine inner and outer radii of the ring of segments (0.58 � FD and 0.78 �FD, respectively). The average of quantitative fundus autofluorescencevalues of the 8 segments is defined as qAF8.

Ophthalmology Volume 122, Number 2, February 2015

(2 or 4, depending on whether a second imaging session wasperformed) was generated for comparison with other patients.Control values used in this study consisted of previously publisheddata from 277 healthy subjects (374 eyes; age range, 5e60 years)without a family history of retinal dystrophy.14

During the course of the study, the detection aperture of theSpectralis was changed from 6 to 5 mm in diameter to minimizepupil interference. This modification had no effect on qAF mea-surements because the AF signals of both fundus and referencewere similarly affected by the smaller aperture size. The mea-surements were obtained with different Spectralis devices and in-ternal references; in each case, the reference calibration factor22

was determined using a master fluorescent reference. Correlationsbetween AF and SD OCT images were made by 2 of the authors(T.D., J.R.S.), who reviewed and discussed the images.

Statistical Analyses

Analyses were performed using Prism 5 (GraphPad Software, LaJolla, CA). The qAF8 values were compared with the 95% confi-dence intervals (CIs) of healthy subjects of the same age and raceor ethnicity.14 To evaluate the repeatability of the measurementsbetween sessions, we used the method described by Bland andAltman25 to compute the coefficient of repeatability (CR0, 95% CI)for the differences {log(qAFB) e log(qAFA)},

7 where qAFA andqAFB are the qAF8 in the 2 sessions. The CR expressed as a per-centage of the mean qAF8was calculated as CR¼ 100� (10CR

0e 1).

The coefficient of agreement between the qAF8 of the rightand left eyes was computed similarly. Sensitivity and specificityof qAF as a test to differentiate between ABCA4-associated andnoneABCA4-associated BEM were calculated using MedCalc(http://www.medcalc.org/calc/diagnostic_test.php).

348

Results

ABCA4 mutations were identified in 22 patients, including 21 pa-tients (95%) with both disease-causing ABCA4 variants (Table 1).One patient was homozygous and 13 patients were compoundheterozygous for the p.G1961E variant. ABCA4 was excluded asthe causal gene in 15 patients because no mutations were detectedafter complete sequencing of the ABCA4 exons and adjacentintronic sequences. ABCA4-positive patients tended to be younger(mean age, 21.9�8.3 years) than ABCA4-negative patients (meanage, 42.1�14.9 years; P<0.0001, unpaired t test). There was nodifference in visual acuities between the 2 groups (right eyes, P ¼0.88, and left eyes, P ¼ 0.13, unpaired t test). International Societyfor Clinical Electrophysiology of Vision standardized full-fieldelectroretinography was performed in 16 of 22 ABCA4-positivepatients. All of these patients were classified as having isolatedmaculopathy (normal full-field electroretinography).3,26,27 Of the15 ABCA4-negative patients, 10 manifested generalized conedysfunction, 3 had normal electroretinography results, and 2 didnot participate in electroretinography testing.

ABCA4-positive and ABCA4-negative patients shared similarAF features (Figs 2 and 3; AF images of all patients are availableat www.aaojournal.org). For instance, we found that in bothgroups of patients, the BEM lesion could be distinctly circular(e.g., ABCA4-positive patients 8 and 9; ABCA4-negative patients26, 30, 31, 33, and 37) or elliptical (e.g., ABCA4-positive patients3, 7, and 16; ABCA4-negative patients 27 and 32). The parafoveacould be mottled (e.g., ABCA4-positive patients 3, 4, 9, 15, and17; ABCA4-negative patients 25, 26, 30, and 33), and thehyperautofluorescent ring could be bright with well-defined bor-ders (e.g., ABCA4-positive patients 5 and 7; ABCA4-negativepatients 25, 26, and 28), relatively faint (e.g., ABCA4-positivepatients 11, 13, 19, and 21; ABCA4-negative patients 27, 29, 30,31, and 37), or diffuse (e.g., ABCA4-positive patients 18 and 22;ABCA4-negative patients 23 and 34). The fovea also could appearbright (e.g., ABCA4-positive patients 2, 6, 9, and 16; ABCA4-negative patients 30, 32, 33, and 35) or dark (e.g., ABCA4-posi-tive patients 14 and 18; ABCA4-negative patients 25, 27, and 28).Lesion size also varied among the patients, whereas there was aremarkable similarity between fellow eyes with respect to lesionsize, shape, and AF pattern.

Quantitative Fundus Autofluorescence

The difficulties inherent in distinguishing patients with and withoutABCA4 mutations based on conventional fundus AF images areillustrated in Figure 4, where age-similar ABCA4-negative patients(left) and ABCA4-positve patients (right) are presented side by side.When comparing the corresponding color-coded qAF maps ofthese patients, however, it becomes immediately notable thatABCA4-positive BEM eyes exhibit higher qAF levels than ABCA4-negative BEM eyes.

This difference in qAF levels also is shown in Figure 5, wherethe qAF8 of each eye is plotted versus age and compared with the95% CIs of healthy subjects.14 ABCA4-positive patients tend tocluster at young ages, whereas ABCA4-negative patients arerepresented at all ages, with a tendency to cluster at older ages. Inthe ABCA4-positive group, 37 of 41 eyes (19 of 22 patients) hadqAF8 of more than the 95% CI for age. Conversely, in theABCA4-negative group, 22 of 26 eyes (13 of 15 patients) hadqAF8 within the normal range. This difference between theABCA4-positive and ABCA4-negative groups was significant(chi-square [2, n ¼ 67], 37.5; P<0.0001). Of the 2 ABCA4-positive patients who had qAF8 within the normal range for age,one was homozygous and the other was heterozygous forp.G1916E. Considering only white ABCA4-positive patients

Figure 2. Fundus autofluorescence images of ABCA4-positve patients: (A) patient 12, (B) patient 7, (C) patient 8, (D) patient 4, (E) patient 17, (F)patient 16, (G) patient 21, and (H) patient 1. The internal autofluorescence reference is visible in the top of the image.

Duncker et al � qAF and SD OCT in Bull’s-Eye Maculopathy

younger than 30 years, we found that mean qAF8 was 366�72qAF units for patients carrying the p.G1961E mutation and416�89 for patients carrying variants other than p.G1961E, butthe difference was not significant (P ¼ 0.2, 2-tailed t test). Therewas no difference in age between ABCA4-positive patients whocarried the p.G1961E variant and those who did not (P ¼ 0.1,unpaired t test).

Figure 3. Fundus autofluorescence images of ABCA4-negative patients: (A) patpatient 35, (G) patient 26, and (H) patient 34. The internal autofluorescence

The sensitivity of qAF measurements in this study was calcu-lated as the percentage of ABCA4-positive BEM patients with aqAF value of more than the 95% CI for age divided by the totalnumber of BEM patients; sensitivity was 86.4% (95% CI, 65.1%e96.94%). The specificity was 86.67 (95% CI, 59.5%e97.9%,calculated as the probability that a qAF value will be within the95% CI for age when the patient is ABCA4 negative).

ient 36, (B) patient 29, (C) patient 28, (D) patient 27, (E) patient 24, (F)reference is visible in the top of the image.

349

Figure 4. Color-coded quantitative fundus autofluorescence (qAF) images shown together with the corresponding fundus autofluorescence (AF) images.Lower qAF levels are coded in blue and higher qAF levels as orange (see color-code scale). ABCA4-negative patients ((A) patient 23, 16 years of age; (B)patient 33, 22 years of age; (C) patient 32, 33 years of age) have lower qAF values than age-similar ABCA4-positive patients ((D) patient 22, 18 years of age;(E) patient 18, 26 years of age; (F) patient 9, 21 years of age). Note that the fundus has similar grey levels in each fundus AF image, but the internal fundusAF reference (top of image) is darker in ABCA4-positive patients, reflecting the higher fundus AF levels.

Ophthalmology Volume 122, Number 2, February 2015

For qAF8 of right and left eyes (n ¼ 30), the Bland-Altmancoefficient of agreement was 15.6%. The between-session Bland-Altman CR was �8.8% (40 eyes of 24 patients). The study cohorthad a coefficient of agreement between eyes and a between-sessionCR that were similar to those of healthy subjects (�15.3% and�9.4%, respectively),14 STGD1 patients (�13% and �10.3%,respectively),20 and Best vitelliform macular dystrophy patients(�19.2% and �7.0%, respectively).23

Spectral-Domain Optical Coherence Tomography

Correlations between AF and SD OCT images were made for 73eyes. Severe nystagmus did not permit acquisition of an AF image

350

for the right eye of patient 25. Fellow eyes of each patient generallyhad very similar SD OCT findings. In Figure 6, representative SDOCT images of (A) a healthy subject, (B) patient 10, (C) patient 15,(D) patient 14, and (E) patient 25 are shown. The high AF ringsurrounding the low AF lesion area correlated in all eyes with theouter limits of a central area of inner segment ellipsoid loss orbreak-up. The SD OCT features of the low AF lesion area includedobvious RPE thinning in 53 eyes (27 patients, 15 of whom wereABCA4 positive), increased choroidal reflectivity in 61 eyes (32patients, 19 of whom were ABCA4 positive), and variable amountsof hyperreflective debris above the RPEeBruch’s membranecomplex in 53 eyes (28 patients, 15 of whom were ABCA4 posi-tive). An optical empty lesion (OEL) with localized inner segment

Figure 5. Scatterplots showing quantitative fundus autofluorescence values of the 8 segments (qAF8) from each eye of bull’s-eye maculopathy (BEM)patients, plotted as a function of age. ABCA4-positive BEM patients are shown as black circles. ABCA4-negative BEM patients are shown as white squares.For comparison, qAF8 from healthy subjects14 (grey circles) are plotted together with mean (solid line) and 95% confidence intervals (CIs; dashed lines).Because qAF8 varies with race and ethnicity, white, black, and Hispanic patients are shown in separate plots. The eyes of an Indian patient (patient 3;marked with white X) are included in the plot of white subjects because the upper 95% CI of whites and Indians is similar.14

Duncker et al � qAF and SD OCT in Bull’s-Eye Maculopathy

ellipsoid loss at the fovea and an intact external limiting membrane(ELM) throughout the SD OCT scan were present in 10 eyes (5patients, all ABCA4 positive; Fig 6B). Relative foveal sparing, asshown in Figure 6D, E, was present in 9 eyes (7 patients, 3 ofwhom were ABCA4 positive). In all eyes except for those of patient36, the outer nuclear layer (ONL) was thinned in the central part ofthe lesion and in an adjacent transition zone. In 32 eyes (16 pa-tients, 14 of whom were ABCA4 positive), the ONL was disruptedby hyperreflective dots in the transition zone bordering the centrallesion. Interestingly, this feature was most pronounced in patientswith an OEL (Fig 6B). In some patients, the ELM appeared morepronounced than in healthy subjects (Fig 6C). At the eccentricitieswhere qAF measurements were obtained (double-headed arrows inFig 6), abnormalities were observed in at least some eyes in theform of hyperreflective dots at the level of the ONL (Fig 6B), ELMthickening (Fig 6C), or ONL thinning (Fig 6E).

Discussion

In this study, we assessed whether qAF, an indirect measureof RPE lipofuscin, could aid in differentiating ABCA4-positve from ABCA4-negative cases of BEM. QuantitativeAF clearly distinguished the 2 groups. Specifically, qAFanalysis indicated that ABCA4-positive BEM patients haveincreased lipofuscin levels throughout the posterior pole,whereas patients with BEM resulting from mutations inother genes have qAF levels within normal limits for age.These data reinforce the clinical usefulness of qAF. Becauseof the association between ABCA4-related disease and adark choroid in fluorescein angiography, the latter imagingmethod also can benefit diagnosis. However, qAF has theadvantage of being noninvasive. It is also worth noting thatin the case of G1961E mutations in ABCA4, neither fluo-rescein angiography nor qAF may be helpful imagingmethods; a dark choroid is absent in these individuals,27 and

in the presence of the G1961E mutation, qAF values can bewithin the normal range for age (Fig 5).

The ABCA4 disease pathway is known. FunctioningABCA4 protein28,29 is needed to remove N-retinylidene-phosphatidylethanolamine, formed by the binding of reti-naldehyde to phosphatidylethanolamine, from the outersegment disc membranes of rods and cones. Failure of thisprocess, such as in STGD1, permits N-retinylidene-phos-phatidylethanolamine to accumulate in excess such that asecond molecule of retinaldehyde can react with N-reti-nylidene-phosphatidylethanolamine, leading to the forma-tion of A2E and related bisretinoids. These molecules arephagocytosed and stored in RPE cells as lipofuscin andhave potential to cause cellular damage.13 The diseasepathways in ABCA4-negative BEM patients are less clear,and the phenotypic similarities between ABCA4-positiveand ABCA4-negative patients currently are difficult toexplain. Moreover, although lipofuscin accumulation likelyis involved in the primary pathogenesis of ABCA4-relateddisease, whether this is the case for noneABCA4-disease isnot known.

Although retinal changes were most obvious in the centralmacula, we also observed ONL thinning in a transition zoneat the lesion border in all but 1 patient. In a subgroup ofpatients, the ONL in the transition zone was altered by thepresence of hyperreflective material. The ELM also waspronounced outside the central part of the lesion in someABCA4-positive patients. A case of ELM thickening in ayoung STGD1 patient without any obvious RPE changes hasbeen reported before,30 and it was suggested that ELMthickening could be an early disease marker. Although in ourstudy cohort all patients with an OEL were ABCA4 positive,other studies have shown that OEL also can be associatedwith mutations in other genes, for example, in achroma-topsia.31 We identified several quantifiable features by SD

351

Figure 6. Fundus autofluorescence (AF) images (left) and corresponding spectral-domain optical coherence tomography (SD OCT) images (right) of (A) a21-year-old healthy subject, (B) patient 10, (C) patient 15, (D) patient 14, and (E) patient 25. Horizontal axis and extent of SD OCT image are indicatedby the white arrow in the AF images. The SD OCT images were stretched vertically for presentation purposes. The retinal layers corresponding to the outernuclear layer (ONL), external limiting membrane (ELM), inner segment ellipsoid (ISe), ensheathed outer segments (eOS), and retinal pigment epithelium(RPE)/Bruch’s membrane are indicated in (A). The double-headed white arrows in the SD OCT images indicate the position where the quantitative fundusAF measurements were obtained (Fig 1).

Ophthalmology Volume 122, Number 2, February 2015

OCT; however, none of these clearly distinguished ABCA4-positive and ABCA4-negative cases.

Because all photoreceptors of ABCA4-positive caseshave a defective visual cycle, it is perhaps not surprising thatretinal changes also extended into areas that appeared un-affected on AF imaging. It is interesting to consider, how-ever, why the central macula seems to be more susceptibleto the damaging effects of the ABCA4 mutations than therest of the retina. One possibility is that certain ABCA4mutations, for instance, the p.G1961E variant, may elicit astronger effect on cones than rods. Although p.G1961E isthe most frequent ABCA4 variant in STGD1 (allele fre-quency, approximately 10%),32,33 it is still remarkablethat 14 of 22 ABCA4-positive BEM patients (64%) carriedthis variant. It has been noted before, however, that the

352

p.G1961E variant seems to be associated with the BEMphenotype and a milder disease spectrum.2,20,32,34 Thenotion that STGD1 patients with a BEM phenotype haveless widespread disease is supported further by Figure 7,which shows the qAF8 levels of the white ABCA4-positveBEM patients in comparison with previously published20

results from white STGD1 patients with phenotypes otherthan BEM, including patients with extramacular funduschanges. ABCA4-positve BEM patients have higher qAFlevels than normal subjects (and ABCA4-negative BEMpatients) but can have relatively low qAF levels whencompared with other STGD1 patients.

It is perhaps not a coincidence that most ABCA4-positivepatients in our study clustered at young ages. In STGD1, theBEM phenotype in some may cases be a transient stage, and

Figure 7. Scatterplot showing quantitative fundus autofluorescence inpatients homozygous or compound heterozygous for disease-causing muta-tions in ABCA4. The average of quantitative fundus autofluorescencevalues of the 8 segments (qAF8) of both eyes of white recessive Stargardtdisease (STGD1) patients with a phenotype other than bull’s-eye macul-opathy (BEM; white circles)20 are shown together with qAF8 of whiteABCA4-positive BEM patients (black circles). For comparison, qAF8 fromwhite healthy subjects14 (grey circles) together with their 95% confidenceintervals (CIs; dashed lines) and means (solid line) are shown.

Duncker et al � qAF and SD OCT in Bull’s-Eye Maculopathy

eventually more extensive fundus changes will develop inpatients. This is illustrated in Figure 8, in which diseaseprogression over 2.5 years is shown for the brother of

Figure 8. Fundus autofluorescence images from 3 visits (AeC) are shown for thbull’s-eye maculopathy (BEM) phenotype can develop extramacular disease andof presence of peripheral flecks), but he also carries the complex allele L541P/A

patient 15, who also carries the complex allele L541P/A1038V. At the initial visit (Fig 8A), this patient readilycould have been classified as having BEM. However, wedecided not to include the patient in this study because someperipheral flecks were visible already. At subsequent visits1.2 (Fig 8B) and 2.5 (Fig 8C) years later, the flecks weremore pronounced and numerous.

A question that is not only of interest for the interpreta-tion of the AF signal in BEM, but also for other retinaldystrophies, is what may be the basis for the high AF ringthat surrounds the low AF part of the central lesion. Wefound that, similar to retinitis pigmentosa,35 the innersegment ellipsoid was discontinued at the outer border of orwithin the high AF ring. Thus, the high AF ring may reflectan accelerated production of bisretinoids within degenerat-ing photoreceptors.36 As an alternative explanation, Freundet al37 recently suggested that the high AF ring may resultfrom what they termed a window defect. They proposed thatloss of the outer retina may cause less absorption of the AFsignal originating from the RPE. However, photopigment,the major absorber of the fundus AF exciting and emittedlight, was bleached before image acquisition in our protocol.

A limitation of this study is that retinal layer thicknesseson SD OCT were not analyzed quantitatively. By seg-menting retinal layers, one could confirm whether the ONLwas thinned throughout the retina or whether the thinningwas limited to the central part of the lesion and the adjacenttransition zone.38

Although finding the definitive causal genes in theABCA4-negative patients is beyond the scope of this study,we investigated the genetic causality in most ABCA4-negative cases by whole-exome sequencing (7/15 patients)or targeted candidate gene sequencing (4/15 patients).Although these studies have not yielded the causal gene inmost cases, we found that (1) the genetic causality varieswidely in the ABCA4-negative group; (2) except for onepatient with an RPGR mutation, in all cases we excludedknown genes that previously have been associated with theBEM phenotype (PROM1, GUCA1A, etc.); and (3) in somecases, we unexpectedly found disease-causing mutations in

e brother of patient 15 at the indicated ages, illustrating that patients with aflecks. This ABCA4-positive patient was not included in the study (because1038V.

353

Ophthalmology Volume 122, Number 2, February 2015

known retinal disease genes such as CRX39 and CNGA3,which have not been previously associated with the BEMphenotype, suggesting phenotypic expansion (BEM phenotypein genes known to cause a very different disease phenotype).Because almost all of our patients exhibited sporadic diseasewith no available family members, we were not able toidentify the causal gene even by whole-exome sequencing,in most cases because unless one finds clearly pathogenicvariants in known disease genes it is impossible to distin-guish between many candidate genes that result from whole-exome sequencing studies of a single patient. In summary,we found a very likely causal gene in 5 of 15 cases,excluded known BEM-associated genes in most cases, andidentified multiple possible new causal genes in 4 of 15cases that need detailed follow-up analyses.

A common challenge in the diagnosis of retinal dystro-phies is that mutations in different genes can result in verysimilar phenotypes. Identifying the causal gene has impli-cations with respect to the patient’s prognosis and will becritical when personalized treatment options such as genetherapy become available. Quantitative AF cannot circum-vent the need for genetic testing of causal genes such asABCA4. However, clinical criteria such as qAF that providesome indication of how likely a specific gene is the cause ofa patient’s phenotype can be a valuable clinical tool. Forinstance, in patients with a clinical diagnosis of STGD1,complete sequencing of exons and adjacent intronic se-quences in the ABCA4 gene may identify only 1 disease-causing mutation in 15% to 20% of individuals and nomutations in 10% to 15% of individuals.32 In these groupsof patients, qAF analysis can be particularly valuable.Although qAF levels in the normal range do not completelyexclude the possibility of ABCA4 mutations, they substan-tially increase the likelihood of finding causal mutations ingenes other than ABCA4.

Quantitative AF is clinically useful to recognize BEMpatients with qAF levels higher than the normal limitsfor age; in these patients, one may expect to confirm thepresence of ABCA4 mutations with genotyping. Thus, theqAF method may guide genetic testing and help to counselpatients at a stage where genetic results are not yet avail-able. Ongoing longitudinal qAF studies involving STGD1patients hold promise in providing valuable informationregarding the natural course of the disease. QuantitativeAF also may help to ascertain the role of lipofuscin indifferent disorders. Whether the qAF approach will beuseful as an outcome measure in clinical trials remains tobe determined.

References

1. Kellner U, Renner AB, Tillack H. Fundus autofluorescenceand mfERG for early detection of retinal alterations in patientsusing chloroquine/hydroxychloroquine. Invest Ophthalmol VisSci 2006;47:3531–8.

2. Cella W, Greenstein VC, Zernant-Rajang J, et al. G1961Emutant allele in the Stargardt disease gene ABCA4 causesbull’s eye maculopathy. Exp Eye Res 2009;89:16–24.

354

3. Kurz-Levin MM, Halfyard AS, Bunce C, et al. Clinical vari-ations in assessment of bull’s-eye maculopathy. Arch Oph-thalmol 2002;120:567–75.

4. Michaelides M, Chen LL, Brantley MA Jr, et al. ABCA4mutations and discordant ABCA4 alleles in patients and sib-lings with bull’s-eye maculopathy. Br J Ophthalmol 2007;91:1650–5.

5. Kearns TP, Hollenhorst RW. Chloroquine retinopathy. Eval-uation by fluorescein fundus angiography. Arch Ophthalmol1966;76:378–84.

6. Michaelides M, Gaillard MC, Escher P, et al. The PROM1mutation p.R373C causes an autosomal dominant bull’s eyemaculopathy associated with rod, rod-cone, and macular dys-trophy. Invest Ophthalmol Vis Sci 2010;51:4771–80.

7. Michaelides M, Holder GE, Hunt DM, et al. A detailed studyof the phenotype of an autosomal dominant cone-rod dystro-phy (CORD7) associated with mutation in the gene for RIM1.Br J Ophthalmol 2005;89:198–206.

8. Michaelides M, Johnson S, Poulson A, et al. An autosomaldominant bull’s-eye macular dystrophy (MCDR2) that maps tothe short arm of chromosome 4. Invest Ophthalmol Vis Sci2003;44:1657–62.

9. Robson A, Michaelides M, Saihan Z, et al. Functional charac-teristics of patients with retinal dystrophy that manifest abnormalparafoveal annuli of high density fundus autofluorescence; areview and update. Doc Ophthalmol 2008;116:79–89.

10. Robson AG, Michaelides M, Luong VA, et al. Functionalcorrelates of fundus autofluorescence abnormalities in patientswith RPGR or RIMS1 mutations causing cone or cone roddystrophy. Br J Ophthalmol 2008;92:95–102.

11. Michaelides M, Wilkie SE, Jenkins S, et al. Mutation in thegene GUCA1A, encoding guanylate cyclase-activating protein1, causes cone, cone-rod, and macular dystrophy. Ophthal-mology 2005;112:1442–7.

12. Delori FC, Dorey CK, Staurenghi G, et al. In vivo fluorescenceof the ocular fundus exhibits retinal pigment epithelium lip-ofuscin characteristics. Invest Ophthalmol Vis Sci 1995;36:718–29.

13. Sparrow JR, Gregory-Roberts E, Yamamoto K, et al. Thebisretinoids of retinal pigment epithelium. Prog Retin Eye Res2012;31:121–35.

14. Greenberg JP, Duncker T, Woods RL, et al. Quantitativefundus autofluorescence in healthy eyes. Invest OphthalmolVis Sci 2013;54:5684–93.

15. Delori FC, Goger DG, Dorey CK. Age-related accumulationand spatial distribution of lipofuscin in RPE of normal sub-jects. Invest Ophthalmol Vis Sci 2001;42:1855–66.

16. Delori FC, Staurenghi G, Arend O, et al. In vivo measurementof lipofuscin in Stargardt’s diseasedfundus flavimaculatus.Invest Ophthalmol Vis Sci 1995;36:2327–31.

17. Birnbach CD, Jarvelainen M, Possin DE, Milam AH. Histo-pathology and immunocytochemistry of the neurosensoryretina in fundus flavimaculatus. Ophthalmology 1994;101:1211–9.

18. Cideciyan AV, Aleman TS, Swider M, et al. Mutations inABCA4 result in accumulation of lipofuscin before slowing ofthe retinoid cycle: a reappraisal of the human diseasesequence. Hum Mol Genet 2004;13:525–34.

19. Eagle RC Jr, Lucier AC, Bernardino VB Jr, Yanoff M. Retinalpigment epithelial abnormalities in fundus flavimaculatus: alight and electron microscopic study. Ophthalmology 1980;87:1189–200.

20. Burke TR, Duncker T, Woods RL, et al. Quantitative fundusautofluorescence in recessive Stargardt disease. Invest Oph-thalmol Vis Sci 2014;55:2841–52.

Duncker et al � qAF and SD OCT in Bull’s-Eye Maculopathy

21. Jaakson K, Zernant J, Kulm M, et al. Genotyping microarray(gene chip) for the ABCR (ABCA4) gene. Hum Mutat 2003;22:395–403.

22. Delori F, Greenberg JP, Woods RL, et al. Quantitative mea-surements of autofluorescence with the scanning laser ophthal-moscope. Invest Ophthalmol Vis Sci 2011;52:9379–90.

23. Duncker T, Greenberg JP, Ramachandran R, et al. Quantitativefundus autofluorescence and optical coherence tomography inBest vitelliform macular dystrophy. Invest Ophthalmol Vis Sci2014;55:1471–82.

24. Spaide RF, Curcio CA. Anatomical correlates to the bandsseen in the outer retina by optical coherence tomography.Literature review and model. Retina 2011;31:1609–19.

25. van de Kraats J, van Norren D. Optical density of the aginghuman ocular media in the visible and the UV. J Opt Soc AmA Opt Image Sci Vis 2007;24:1842–57.

26. Bland JM, Altman DG. Statistical method for assessingagreement between two methods of clinical measurement.Lancet 1986;1:307–10.

27. Lois N, Holder GE, Bunce C, et al. Phenotypic subtypes ofStargardt macular dystrophydfundus flavimaculatus. ArchOphthalmol 2001;119:359–69.

28. Fishman GA, Stone EM, Grover S, et al. Variation of clinicalexpression in patients with Stargardt dystrophy and sequencevariations in ABCR gene. Arch Ophthalmol 1999;117:504–10.

29. Quazi F, Lenevich S, Molday RS. ABCA4 is an N-retinylidene-phosphatidylethanolamine and phosphatidyletha-nolamine importer. Nat Commun 2012;3:925.

30. Burke TR, Yzer S, Zernant J, et al. Abnormality in the externallimiting membrane in early Stargardt disease. OphthalmicGenet 2013;34:75–7.

31. Greenberg JP, Sherman J, Zweifel SA, et al. Spectral-domainoptical coherence tomography staging and autofluorescenceimaging in achromatopsia. JAMA Ophthalmol 2014;132:437–45.

32. Burke TR, Fishman GA, Zernant J, et al. Retinal phenotypes inpatients homozygous for the G1961E mutation in the ABCA4gene. Invest Ophthalmol Vis Sci 2012;53:4458–67.

33. Zernant J, Schubert C, Im KM, et al. Analysis of the ABCA4gene by next-generation sequencing. Invest Ophthalmol VisSci 2011;52:8479–87.

34. Fujinami K, Sergouniotis PI, Davidson AE, et al. The clinicaleffect of homozygous ABCA4 alleles in 18 patients. Ophthal-mology 2013;120:2324–31.

35. Duncker T, Tabacaru MR, Lee W, et al. Comparison ofnear-infrared and short-wavelength autofluorescence inretinitis pigmentosa. Invest Ophthalmol Vis Sci 2013;54:585–91.

36. Sparrow JR, Yoon KD, Wu Y, Yamamoto K. Interpretationsof fundus autofluorescence from studies of the bisretinoids ofthe retina. Invest Ophthalmol Vis Sci 2010;51:4351–7.

37. Freund KB, Mrejen S, Jung J, et al. Increased fundus auto-fluorescence related to outer retinal disruption. JAMA Oph-thalmol 2013;131:1645–9.

38. Lazow MA, Hood DC, Ramachandran R, et al. Transitionzones between healthy and diseased retina in choroideremia(CHM) and Stargardt disease (STGD) as compared to retinitispigmentosa (RP). Invest Ophthalmol Vis Sci 2011;52:9581–90.

39. Yamamoto SJ, Jaiswal M, Charng WL, et al. A Drosophilagenetic resource of mutants to study mechanisms underlyinghuman genetic diseases. Cell 2014;159:200–14.

Footnotes and Financial Disclosures

Originally received: May 14, 2014.Final revision: June 26, 2014.Accepted: August 6, 2014.Available online: October 3, 2014. Manuscript no. 2014-739.1 Department of Ophthalmology, Columbia University, New York, NewYork.2 Department of Pathology and Cell Biology, Columbia University, NewYork, New York.3 Schepens Eye Research Institute and Department of Ophthalmology,Harvard Medical School, Boston, Massachusetts.

Financial Disclosure(s):The author(s) have no proprietary or commercial interest in any materialsdiscussed in this article.

Supported in part by the National Eye Institute, National Institutes ofHealth, Bethesda, Maryland (grant nos. EY024091, EY021163, EY019861,and P30EY019007); Foundation Fighting Blindness (Owings Mills, MD;grant no. C-CL-0710); and a grant from Research to Prevent Blindness,

New York, New York, to the Department of Ophthalmology, ColumbiaUniversity. During initial stages of the work, Dr. Delori was supportedpartially by the National Eye Institute (grant no. EY015520).

Abbreviations and Acronyms:AF ¼ autofluorescence; BEM ¼ bull’s-eye maculopathy; CI ¼ confidenceinterval; CR ¼ coefficient of repeatability; ELM ¼ external limiting mem-brane; GL ¼ grey level; NIR-R ¼ near-infrared reflectance;qAF ¼ quantitative fundus autofluorescence; OEL ¼ optical empty lesion;qAF8 ¼ mean quantitative fundus autofluorescence from the 8 segments;ONL ¼ outer nuclear layer; RPE ¼ retinal pigment epithelium; SDOCT¼ spectral-domain optical coherence tomography; STGD1 ¼ recessiveStargardt disease.

Correspondence:Janet R. Sparrow, PhD, Department of Ophthalmology, ColumbiaUniversity, 630 West 168th Street, New York, NY 10032. E-mail:jrs88@columbia.edu.

355