Abnormal Foveal Morphology in Ocular Albinism Imaged with Spectral-Domain Optical Coherence Tomography

Home , Ocular albinism, Ocular albinism type 1

CLINICAL SCIENCES Abnormal Foveal Morphology in Ocular Albinism Imaged With Spectral-Domain Optical Coherence Tomography

Gabriel T. Chong, MD; Sina Farsiu, PhD; Sharon F. Freedman, MD; Neeru Sarin, MBBS; Anjum F. Koreishi, BS; Joseph A. Izatt, PhD; Cynthia A. Toth, MD

Objectives: To evaluate the spectrum of foveal archi- Pudlak syndrome: persistence of an abnormal, highly tecture in pediatric albinism and to assess the utility of reflective band across the fovea, multiple inner retinal spectral-domain optical coherence tomography (OCT) layers normally absent at the center of the fovea, and loss in ocular imaging of children with nystagmus. of the normally thickened photoreceptor nuclear layer at the fovea when compared with that in control sub- Methods: Spectral-domain OCT imaging was per- jects. The optic nerve was elevated in multiple eyes formed on study subjects in 3 groups: subjects with of subjects with OA or suspected OA and the subject ocular albinism (OA) or suspected OA with foveal hypo- with oculocutaneous albinism and Hermansky-Pudlak plasia, with nystagmus, and with or without iris trans- syndrome. illumination; a subject with oculocutaneous albinism and Hermansky-Pudlak syndrome; and control subjects. Dense Conclusions: A spectrum of foveal morphological ab- volumetric scans of each fovea were captured using stan- normalities is seen in subjects with OA or suspected OA, dard and handheld spectral-domain OCT devices. Im- which in some cases contrasted with previous studies ages were postprocessed and scored for the presence and using time-domain OCT. These OCT findings clarify configuration of each retinal layer across the fovea. the morphology of foveal hypoplasia seen clinically. Results: High-quality spectral-domain OCT images ob- This imaging modality may be useful in evaluating tained from each subject revealed a range of abnormali- children. ties in subjects with OA or suspected OA and the subject with oculocutaneous albinism and Hermansky- Arch Ophthalmol. 2009;127(1):37-44

CULAR ALBINISM (OA) IS graded, while at the same time with fo- a genetic disorder of veal hypoplasia, central vision should be melanin production that diminished owing to widely spaced cen- occurs with a frequency tral cones.4,8,9 of about 1 in 50 000 per- The nystagmus seen in albinism usu- sonsO in the United States.1 In contrast to ally manifests by age 2 or 3 months and oculocutaneous albinism (OCA) involv- often lessens with age and near vision as ing skin, hair, and eyes, OA is limited to a result of convergence.10 Nystagmus is the eyes.2 Ocular albinism type 1 is an in- probably due to the foveal hypoplasia (ab- herited X-linked disease causing ocular sent foveal pit on light microscopy) and clinical features that may include nystag- the aberrant visual pathways seen in pa- mus, decreased visual acuity, hypopig- tients with albinism.3,10 In addition, astig- mentation of the retina, foveal hypopla- matism and other refractive errors are com- sia, translucency of the iris, macular mon and could lead to some degree of transparency, photophobia, and abnor- amblyopia and low visual acuity.4 Owing mal neuronal wiring (abnormal decussa- to misrouting of the optic pathways, pa- Author Affiliations: tion of nerve fibers at the chiasm).3-5 The tients with OA are at high risk for strabis- Departments of Ophthalmology decreased visual acuity varies, and re- mus and loss of stereovision.4 (Drs Chong, Farsiu, Freedman, ports suggest that visual acuity is re- The genetics involved in OCA type 1 Sarin, Izatt, and Toth and duced to 20/25 to 20/200.6-8 It is possible (OCA1) focus on mutations in the tyrosi- Mr Koreishi) and Biomedical 2 Engineering (Drs Izatt and that with decreased pigmentation, light en- nase gene. Those individuals who lack Toth), Duke University, tering the eye is more prone to scattering pigment owing to complete inactivity of Durham, North Carolina. and the resulting retinal image is de- the tyrosinase gene have OCA1A, whereas

(REPRINTED) ARCH OPHTHALMOL / VOL 127 (NO. 1), JAN 2009 WWW.ARCHOPHTHALMOL.COM 37

©2009 American Medical Association. All rights reserved. Downloaded From: https://jamanetwork.com/ on 09/28/2021 those with mutations in the tyrosinase gene that allow ing other ophthalmic imaging studies performed at Duke Uni- residual enzyme activity have OCA1B. Individuals have versity Eye Center. Study subjects were examined by a pedi- OCA2, OCA3, and OCA4 when there is a mutation in atric ophthalmologist (S.F.F.). Clinical examination included the P-gene on chromosome 15q, the tyrosine-related pro- best-corrected visual acuity, presence and type of manifest nys- tein 1 gene, or the membrane-associated transporter gene, tagmus, skin color, hair color, iris color, and presence or absence of transillumination defects (assessed in a darkened room respectively.2 Hermansky-Pudlak syndrome (HPS), caused with the subject sitting at a slitlamp biomicroscope unless the by mutations in 1 of 7 HPS genes, and Che´diak-Higashi child could not sit at the table-mounted slitlamp biomicro- syndrome, caused by a mutation in the lysosomal- scope, in which case a portable slitlamp was used). Fundus ex- trafficking regulator gene, are multisystemic disorders as- amination was performed with an indirect ophthalmoscope and sociated with OCA.3 Except for OCA1A, individuals with either a 20-diopter (D) or 28-D condensing lens. Skin biopsy the conditions described earlier may have some pig- and/or genetic testing were not performed in subjects with OA ment in the skin, hair, and irides. or suspected OA or control subjects in this study. The subject Optical coherence tomography (OCT) is the optical with OCA and HPS was diagnosed previously by genetic test- analogue to ultrasonography and measures the echo time ing. Subjects with foveal hypoplasia (determined clinically via delay and magnitude of reflected or backscattered light visualization with indirect ophthalmoscopy using a 20-D lens, using low-coherence interferometry.11 Cross-sectional im- looking for absence of the foveal pit), with nystagmus on clini- ages are obtained by measuring the backscattered light cal examination, and with or without iris transillumination defects were included in the group with OA or suspected OA. Con- while scanning across multiple sites in a transverse fash- trol subjects were white, were aged between 5 and 11 years, ion. The data obtained are displayed as false-color or gray- had normal skin and hair pigmentation, had 20/20 visual acu- scale images. ity in each eye, and lacked nystagmus. The subject with OCA A conventional time-domain (TD) OCT system (Stra- had a diagnosis of HPS. This subject was chosen as a positive tus OCT; Carl Zeiss Meditec, Dublin, California) can pro- control to verify the findings of OCA when compared with the vide 8- to 10-µm axial resolution and 512 axial scans in subjects with suspected OA who did not clearly have albi- 1.3 seconds.11 Clinical limitations of the relatively slow nism. As the subject with OCA and HPS was already sched- standard TD-OCT are partially due to the subject’s eye uled to undergo anesthesia for strabismus surgery, imaging un- motion, which can lead to image artifacts. Such artifacts der anesthesia allowed for the entire posterior pole to be imaged may be reduced by exploiting cross-correlation algo- under controlled conditions to ensure that the area where the fovea might be located would not be inadvertently bypassed rithms that automatically align adjacent axial scans and 11 during SD-OCT imaging. with eye-tracking protocols. Ultrahigh-resolution OCT The retinas of both eyes of the study subjects were then imaging with axial resolutions of 2 to 3 µm has been used scanned using SD-OCT, and the eye with the best images, either to demonstrate retinal morphology with improved de- right or left, was used for analysis. The subject with known OCA tail. Unfortunately, ultrahigh-resolution OCT is even also had SD-OCT imaging of each retina performed while su- slower than the TD systems, where scanning speeds are pine during scheduled anesthesia for strabismus surgery, and typically 150 to 250 axial scans per second.11 Thus, im- SD-OCT was performed using a portable, handheld, noncon- age artifacts due to eye motion are more severe and cov- tact SD-OCT device (Bioptigen Inc, Research Triangle Park, erage is more restricted.11 North Carolina). To overcome such limitations, ophthalmic imaging re- We used an SD-OCT system developed by one of us (J.A.I.) with operating software provided by Bioptigen Inc that was research is now progressively relying on the novel (Fou- 11-13 fined for imaging patients with OA (in collaboration with rier) spectral-domain (SD) OCT systems. The SD- C.A.T.). The SD-OCT light source is a superluminescent di- OCT systems have a higher speed and a sensitivity ode (Superlum Diodes, Ltd, Carrigtwohill, County Cork, Ire- advantage over the conventional OCT systems (50 times land) with a central wavelength of 840 nm and a bandwidth of faster than standard TD-OCT and 100 times faster than 49 nm, yielding a theoretical axial resolution of 6.3 µm in air ultrahigh-resolution OCT imaging systems).11 Owing to and 4.6 µm in tissue. The lateral resolution of each A-scan was the increase in speed, a single cross-sectional scan of 1000 approximately 15 to 20 µm. The power incident on the study A-scans can be captured, processed, and displayed in about subject’s cornea was roughly 500 µW, which is well below the American National Standards Institute’s extended exposure limit 60 milliseconds. This reduces target motion artifacts, re- 14 sulting in a more stable captured image. This is espe- of 750 µW for 8 hours. In each eye, a retinal area of 10ϫ10 mm or 12ϫ12 mm was cially pertinent for the examination of young patients with imaged in 5.8 seconds in a series of 100 horizontal SD-OCT known or suspected albinism, where the time of coop- scans containing 1000 A-scans in each B-scan. This produced erative examination may be limited and nystagmus may 100 B-scans along the horizontal scan and spacing of 80 to 100 affect scans acquired over a longer duration. More pre- µm between scans. In addition to imaging the macula, the se- cise imaging of children with OA or suspected OA may ries of scans included the temporal border of the optic disc and give better insight into the range of morphological macu- the temporal arcades. In several eyes, 2 patterns of additional lar findings in the disease. scans were captured: first, the scans followed the same sam- pling as earlier except that they were oriented 90° to the horizontal; second, a 5ϫ5-mm area was imaged with 200 horizon- METHODS tal B-scans of 500 A-scans each, resulting in 25-µm spacing between scans. Each image set was captured in 5.8 seconds. By collapsing SD-OCT B-scans (averaging all pixels) on an The parents of study subjects consented to the subjects’ par- axial line, a 2-dimensional image analogous to a fundus im- ticipation in a Duke University Institutional Review Board– age, called the summed voxel projection (SVP), is cre- approved study of SD-OCT imaging of the retina. The consent ated.15,16 This technique allows for creating the landmark map comprised investigator access to ophthalmic records, includ- useful to register a cross-sectional OCT image to the fundus

(REPRINTED) ARCH OPHTHALMOL / VOL 127 (NO. 1), JAN 2009 WWW.ARCHOPHTHALMOL.COM 38

©2009 American Medical Association. All rights reserved. Downloaded From: https://jamanetwork.com/ on 09/28/2021 Table 1. Demographic Data for Control Subjects, Subjects With Ocular Albinism, and a Subject With Oculocutaneous Albinism

Color Foveal Patient No./ Depression Transillumination Sex/Age, y Diagnosis BCVA logMAR Nystagmus Fundus Eye Hair Skin or Pita Defectsb 1/F/5 Control 20/20 0.00 None Normal Normal Normal Normal ϩ − 2/F/5 Control 20/20 0.00 None Normal Normal Normal Normal ϩ − 3/M/6 Control 20/20 0.00 None Normal Normal Normal Normal ϩ − 4/M/11 Control 20/20 0.00 None Normal Normal Normal Normal ϩ − 5/F/4 OA 20/60 ϩ 3 0.48 Horizontal, perpendicular Blond Blue Blond Normal − ϩ/− 6/M/10 Suspected OA 20/40 0.30 Nystagmus Blond Blue Blond Pale ϩ/− − 7/F/8 Suspected OA 20/70 ϩ 1 0.54 Irregular horizontal, Blond Blue Light brown Normal − − perpendicular 8/M/8 Suspected OA 20/40 0.30 Horizontal, perpendicular Blond Hazel Light brown Normal ϩ/− − 9/F/10 Suspected OA 20/40 0.30 Horizontal, perpendicular Blond Light green Light brown Normal − − 10/M/4 OCA and HPS 20/200 1.00 Horizontal, perpendicular Blond Blue Light brown Pale − ϩ 11/M/13 OA 20/80 ϩ 1 0.60 Horizontal, perpendicular Blond Blue Light Pale ϩ/− ϩ 12/M/4 Suspected OA 20/70 0.54 Horizontal, perpendicular Normal Hazel/green Scant dark hair Very pale ϩ/− −

Abbreviations: BCVA, best-corrected visual acuity; HPS, Hermansky-Pudlak syndrome; OA, ocular albinism; OCA, oculocutaneous albinism. a ϩ Indicates present; −, entirely absent; and ϩ/−, indistinct. b ϩ Indicates present; −, not present; and ϩ/−, trace.

image and to orient the location of the foveal scan. The SVPs nystagmus. Additionally, the study subject aged 4 years show features such as the macula, optic disc, and blood ves- with known OCA and HPS, 20/200 visual acuity (1.00 log- sels. While the SVPs created with stationary SD-OCT systems MAR), little to no skin melanin pigment present, foveal are usually of high quality, the quality of the SVP images in the hypoplasia, blue irides with prominent transillumination pediatric SD-OCT (with handheld SD-OCT systems) is poorer. defects, and nystagmus was imaged. Demographics of study Therefore, the quality of the SVP images was enhanced by ap- plying adaptive image processing algorithms. First, the raw SD- subjects are included in Table 1. OCT images were denoised using an iterative maximum a pos- Table 2 summarizes the results of analysis of SD- teriori–based algorithm. In our maximum a posteriori OCT images of each subject. As shown in Figures 1, framework, we used the least squares likelihood penalty term 2, 3, 4, and 5, imaging revealed a range of changes in and a variation of the Tikhonov prior as the regularization term.17 the foveal architecture of eyes with OCA and those with Then, the contrast of the SVP image was enhanced by compar- OA when compared with control subjects. These in- ing and normalizing the intensity of each horizontal line (pro- cluded variable loss of normal foveal architecture pre- jected B-scan) on the SVP with respect to the global intensity sent on SD-OCT images of the fovea and increased sig- of the SVP. Therefore, the erroneously dark or bright outlier nal from the choroid and sclera. Optic nerve thickening frames are adjusted with respect to the intensity of their neigh- boring projected B-scans.12 was also observed on SD-OCT scans of 4 subjects with To create summed, less noisy images from a sequence of re- OA or suspected OA and the subject with OCA and HPS. peated B-scans from a unique azimuthal position, large rota- Normal SD-OCT morphology included a sloping foveal tional motions were corrected using the ImageJ (National In- depression, thickening of the hyporeflective photorecep- stitutes of Health, Bethesda, Maryland) “StackReg” registration tor nuclear layer at the fovea with bulging upward cen- plug-in.18 Then, the resulting images were registered using an trally within the fovea, and centripetal displacement of optical flow–based technique with subpixel accuracy19 and fused the nerve fiber layer, ganglion cell layer, inner plexi- 20 using the second-order classic kernel regression algorithm. form layer, inner nuclear layer, and outer plexiform layer with a broadening Henle fiber layer extending from the RESULTS foveola (Figure 1). In control subjects, the choroidal vessels were minimally visible and the sclera was not im- Seven white study subjects (aged 4-13 years), 2 with OA aged with SD-OCT. and 5 with suspected OA, had retinal imaging of both eyes The normal foveal depression was a clear hallmark of by SD-OCT. These study subjects had foveal hypoplasia the fovea in the eyes of the control subjects (Figure 1). Al- on clinical examination, best-corrected visual acuities from though the control subjects showed subtle variation in the 20/40 to 20/80 (0.30-0.60 logMAR; median, 0.48 log- thickness of the inner retinal layers surrounding the fo- MAR), mild skin pigmentation, blue, green, or hazel iri- vea, they all demonstrated the normal centripetal distri- des, and nystagmus (Table 1). Five of the 7 subjects with bution of cells extending out of the fovea. The normal dis- OA or suspected OA did not have definite transillumina- placement of these layers and the broadening of the Henle tion of the iris. Transillumination defects were recorded fiber layer out of the fovea were absent in every one of the as absent only after careful examination using slitlamp bio- eyes of the subjects with OA or suspected OA and the sub- microscopy in a darkened room before pupil dilation. Iris ject with OCA and HPS. In 1 eye from a subject with OA, transillumination was not graded. Four white control sub- a subtle depression of the retina was visible at the foveal jects of similar age without OA were imaged for compari- site even though reflective bands consistent with inner reti- son. These study subjects had normal skin and hair pig- nal layers—including the nerve fiber layer—persisted across mentation, pigmented irides, normal visual acuity, and no this site (subject 5) (Table 2). In another subject with sus-

(REPRINTED) ARCH OPHTHALMOL / VOL 127 (NO. 1), JAN 2009 WWW.ARCHOPHTHALMOL.COM 39

©2009 American Medical Association. All rights reserved. Downloaded From: https://jamanetwork.com/ on 09/28/2021 Table 2. Clinical Findings and Features of Retinal Imaging by Spectral-Domain Optical Coherence Tomography in Control Subjects, Subjects With Ocular Albinism, and a Subject With Oculocutaneous Albinism

Patient No. Diagnosis FDa NFLb GCLc IPLd INLd OPLd PRe RPEf Choroidg Sclerah ONHi 1 Control 2 2 2 2 2 2 2 2 2 2 . . . 2 Control 2 2 2 2 2 2 2 2 2 2 . . . 3 Control 1 1 1 1 1 1 2 4 3 3 . . . 4 Control 1 1 1 1 1 1 2 2 2 2 . . . 5OA100000123 33 6 Suspected OA 0 1 0 0 0 0 1 3 3 3 3 7 Suspected OA 0 0 0 0 0 0 1 2 3 3 2 8 Suspected OA 0 0 0 0 0 0 0 3 3 3 3 9 Suspected OA 0 0 0 0 0 0 0 3 2 2 2 10 OCA and HPS 0 0 0 0 0 0 0 2 3 3 4 11 OA 0 0 0 0 0 0 1 2 2 2 3 12 Suspected OA 0 0 0 0 0 0 1 2 3 3 . . .

Abbreviations: FD, foveal depression; GCL, ganglion cell layer; HPS, Hermansky-Pudlak syndrome; INL, inner nuclear layer; IPL, inner plexiform layer; NFL, nerve fiber layer; OA, ocular albinism; OCA, oculocutaneous albinism; ONH, optic nerve head; OPL, outer plexiform layer; PR, photoreceptor; RPE, retinal pigment epithelium. a A grade of 2 indicates normal; 1, nearly absent; and 0, none. b A grade of 2 indicates that the NFL does not persist across the FD; 1, trace NFL across the FD; and 0, the NFL persists across the FD. c A grade of 2 indicates absent in the fovea (normal); 1, somewhat thinned across the fovea; and 0, no thinning within the fovea. d A grade of 2 indicates absent in the fovea; 1, thinned across the fovea; and 0, no thinning within the fovea (same thickness throughout). e A grade of 2 indicates a normal PR contour, thickened at the fovea; 1, a diminished PR contour; and 0, an absent PR contour. f A grade of 2 indicates normal; 3, increased reflectivity; and 4, hyperreflectivity. g A grade of 2 indicates minimally visible; 3, increased visualization. h A grade of 2 indicates not visible; 3, increased definition. i A grade of 4 indicates bulging; 3, elevated; 2, a normal contour without elevation; and ellipses, absent data.

pected OA, the nerve fiber layer thinned at the fovea, and with loss of foveal contour,4,21 the high-resolution and the photoreceptor layer also showed a slight upward bulge 3-dimensional imaging achieved with SD-OCT handheld even though there was no visible foveal depression and the scanning of a subject during examination under anesthe- remaining inner retinal layers were prominent across the sia allows assured definition of the absence of foveal mor- fovea (subject 6) (Table 2 and Figure 2). The classic up- phology. From 3-dimensional scan stacks, the inner ward bulge of the thicker photoreceptor nuclear layer on retinal layers including the ganglion cell and nerve fiber OCT at the fovea was always present in the control eyes. layers are shown to definitely extend intact across the fo- In contrast, the photoreceptor layer showed no change in vea without thinning. Similar unusual foveal abnormali- thickness on SD-OCT scans across the abnormal site of the ties were found on examination of eyes of children with fovea in 2 subjects with suspected OA and in the subject possible OA or OCA. These children, with visual acuities with OCA and HPS (subjects 8, 9, and 10) (Table 2 and of 20/40 to 20/80 and mildly hypopigmented irides, dem- Figure 3) and only very mild prominence in 5 of the sub- onstrated loss of foveal depression and persistence of in- jects with OA or suspected OA. ner retinal layers across the fovea. The normal centripetal We did not plan to evaluate the optic nerve head with displacement of inner retinal elements with the Henle fi- SD-OCT in this study; however, because of the field of cap- ber layer radiating out from the foveola was absent in all ture and the nystagmus in the subjects with OA or sus- of these eyes. A few of these eyes in subjects with OA or pected OA and the subject with OCA and HPS, we were suspected OA showed a slight suggestion of 1 or 2 elements able to view images of the optic nerve in one and some- of foveal development (Table 2); however, no eyes showed times both eyes of all of these subjects. On identifying op- inner retinal layer displacement. tic nerve head elevation in 1 subject, we went back and re- Comparing between the 2 affected groups, we found viewed images of the optic nerve in all of the subjects. The persisting ganglion cell nuclei and plexiform and inner optic nerve head appeared very prominently elevated in both nuclear layers in all of the subjects (in contrast to the con- eyes of the single subject with OCA and HPS (subject 10) trol group). We also found that although the scale of fo- (Table 2 and Figures 4 and 5) and moderately elevated in veal depression loss, nerve fiber layer thinning, and pho- at least 1 eye of an additional 4 subjects with OA or sus- toreceptor nuclear layer thinning varied among the pected OA (subjects 5, 6, 8, and 11) (Table 2). The optic subjects with OA or suspected OA, when the microar- nerve was not imaged in the control subjects. chitecture of the entire inner retina was considered, all of the eyes of subjects with suspected OA and the eye of COMMENT the subject with OCA and HPS were clearly abnormal. In this small study group, we were unable to correlate In this study, we demonstrate that foveal hypoplasia in OCA visual acuity with the level of foveal development. is associated with absence of all of the morphologic hall- The difficulty in using OCT technology to image chil- marks of the fovea on SD-OCT scanning of the macula. dren with known and suspected OA is movement on the Although conventional TD-OCT has shown single scans part of the child, whether due to the nystagmus or short-

(REPRINTED) ARCH OPHTHALMOL / VOL 127 (NO. 1), JAN 2009 WWW.ARCHOPHTHALMOL.COM 40

58 Scan No. 53 48 48

Scan No. 43 38

Figure 1. Summed voxel projection image (A) and multiple cross-sectional scans corresponding to the red dotted lines in part A (B) across the center of the fovea and at several steps away from the fovea in a control subject (subject 1) (Table 2). There is an obvious foveal pit with normal retinal architecture.

ened attention typical of young children. The rapid image acquisition of SD-OCT does not completely elimi- nate this challenge but improves resolution of each scan and allows more precise imaging even when faced with nystagmus. The rapid scanning also allows repeated imaging at a single location and results in higher- resolution summed scans (Figure 2). However, in children with nystagmus, it still may be difficult to be cer- Figure 2. A spectral-domain optical coherence tomographic summed B-scan tain that the fovea is captured during an SD-OCT scan. cross-section of the retina at the fovea of a subject with suspected ocular Use of a handheld SD-OCT system allowed us to bring albinism (subject 6) (Table 2). There is high reflectivity across the fovea, SD-OCT imaging into the examination under anesthe- suggesting a persisting nerve fiber layer. Multiple inner retinal layers, normally absent at the center of the fovea, persist across the fovea. The sia and to capture definitive 3-dimensional images across photoreceptor nuclear layer bulges upward at the fovea; the external limiting the macula in a child with nystagmus (Figure 3). membrane, photoreceptor inner segment layer, and photoreceptor outer The analysis in this study is based on SD-OCT imaging segment layer appear normal. Note the prominent view to vessels within the choroid (compared with Figure 1), consistent with minimal reflection or and is compared with previous TD-OCT studies, expand- absorption of signal from pigmentation of the retinal pigment epithelium and ing on those findings. Some of our findings support many choroid in this eye. of those in previously published articles,2,21,22 including the persisting retinal layers across the fovea, although using SD- OCT allows better determination of which layers actually reflectivity of the fovea, lack of foveal depression (which persist and affect overall image quality. Seo et al4 describe we also noted), and increased reflectivity from the choroidal the characteristic TD-OCT findings of albinism as hypo- layer. The hyporeflectivity of the photoreceptor layer of the

(REPRINTED) ARCH OPHTHALMOL / VOL 127 (NO. 1), JAN 2009 WWW.ARCHOPHTHALMOL.COM 41

49 Scan No. 44 39

Scan No. 34 29

Figure 3. Summed voxel projection of a spectral-domain optical coherence tomographic 3-dimensional scan of a subject with oculocutaneous albinism (subject 10) (Table 2) (A) and the cross-sectional scans corresponding to the red dotted lines in part A (B) across the center of the presumed fovea and at several steps away from the fovea. There is high reflectivity across the fovea, suggesting a persisting nerve fiber layer. Multiple inner retinal layers, normally absent at the center of the fovea, persist across the fovea. The photoreceptor nuclear layer bulges upward at the fovea; the external limiting membrane, photoreceptor inner segment layer, and photoreceptor outer segment layer appear normal. The white dotted line in part A illustrates the position of the B-scan in Figure 4.

perreflectivity of the underlying sclera. Reflectivity of a tissue layer on SD-OCT is a function of both the reflectivity of the structure and the shadowing or loss of signal from reflectivity of overlying structures. The absence of pigment in the retinal pigment epithelium and choroid in the subjects with OA or suspected OA and the subject with OCA and HPS results in a pronounced increase in signal (pertinent at the 830-nm central wavelength) extending into Figure 4. Cross-sectional scan at the optic nerve corresponding to the white the choroid and sclera. Because of increased depth of scan- dotted line in Figure 3A from the left eye of the subject with oculocutaneous ning with this system compared with the TD system, what albinism and Hermansky-Pudlak syndrome. Note the elevation of the optic appeared as a tram track in the previous study is clearly nerve, which was seen in both eyes in this subject. hyperreflectivity from the retinal pigment epithelium on one side and the sclera beneath the hyporeflective choroi- fovea may have been due to limited image quality from the dal vessels. In their observational case report of a 10-year- old girl with OCA who was imaged using OCT, Meyer et TD system or from the hyperreflectivity of the nerve fiber 23,24 layer across the fovea. We did not find a difference in pho- al also showed a lack of foveal depression. They re- toreceptor reflectivity in the subjects with OA or sus- ported increased foveal thickness greater than 300 µm in pected OA or the subject with OCA and HPS. Seo and col- their patient with OCA compared with a 150-µm-thick fo- leagues also describe a “tram track sign” caused by increased vea in a control eye. They speculated that the fovea was transillumination of the choroidal layer and increased re- filled with hyperreflective tissue, possibly including mul- flectivity of the choroidal space. This was not seen in our tiple ganglion cell layers. In our study, we believe that the study. We found increased visualization of the choroidal increased foveal thickness in subjects with OA or sus- vessels and sclera in subjects with OA or suspected OA and pected OA and the subject with OCA and HPS is due to the subject with OCA and HPS, with hyporeflectivity of the the persisting nerve fiber layer, ganglion cell layer, inner large choroidal vessels contrasting with the prominent hy- plexiform layer, inner nuclear layer, and outer plexiform

(REPRINTED) ARCH OPHTHALMOL / VOL 127 (NO. 1), JAN 2009 WWW.ARCHOPHTHALMOL.COM 42

©2009 American Medical Association. All rights reserved. Downloaded From: https://jamanetwork.com/ on 09/28/2021 layer. In their observational case report of a 79-year-old man with foveal hypoplasia who was imaged using OCT, A McGuire et al21 confirm some of our findings as they also found continuation of the outer nuclear layer, inner nuclear layer, and ganglion cell layer across the fovea. However, they describe a prominent photoreceptor layer centrally. In contrast, we found that the photoreceptor layer showed mild loss of thickness in SD-OCT scans across the abnormal site of the fovea in the subjects with OA or suspected OA and in the subject with OCA and HPS when compared with the normal thickening of the photoreceptor layer in the fovea of control eyes. In using SD-OCT technology, the inner retinal layers persisting across the fovea and also the choroid and scleral reflectivities can be better defined in subjects with OA, suspected OA, and OCA. Evaluating how each layer contrib- utes to the overall reflectivity and hyporeflectivity seen on high-resolution scanning provides a better overall picture of the anatomy of the hypoplastic fovea seen in OCA and OA or suspected OA. Albinism as a disease can vary from a total lack of melanin in all tissues and poor visual out- comes seen in some forms of OCA to mild hypopigmentation (or normal pigmentation in most tissues) and un- B impaired visual acuity seen in albinoidism.7,25 For example, choroidal pigmentation was comparable to that in control subjects in 1 subject with OCA (subject 10) (Table 2). By comparing findings in control subjects with the extreme findings in OCA, a better understanding of changes at the retinal microarchitecture can be obtained. This imaging study based on tissue reflectance demonstrates morphology but cannot demonstrate the function of cellular elements in the retina. Of note, we unexpectedly observed abnormal eleva- Figure 5. Summed voxel projection of the right eye of the subject in Figure 3 tion of the optic nerve head in 4 of the 6 subjects with OA, (A) and a B-scan cross-section corresponding to the yellow line in part A possible OA, or OCA. We did not design the study to evalu- through the prominently elevated optic nerve (B). ate the optic nerve head and did not have images of the optic nerve in any of the control subjects. However, pre- scribe abnormal melanocytes in ocular tissues, poste- vious reports of OCT imaging of the nerve fiber layer at rior embryotoxon, ciliary body hyalinization, foveal the optic nerve in healthy pediatric subjects26 and in pe- hypoplasia, marked hypopigmentation, and ocular hem- diatric subjects with amblyopia27 showed no such eleva- orrhages. The study did not include any images of the tion. The optic nerve thickening in subjects with OA or fovea or any descriptions of the retinal layers at the pre- suspected OA and the subject with OCA and HPS could sumed site of the fovea. They did note a lack of differ- be a coincidental finding; however, this is an intriguing entiation at the fovea and a retinal pigment epithelium association in a disease with abnormal redirection of reti- with few melanin granules. In another article, Mietz et nal axons further along the optic pathway at the chiasm. al25 describe an eye from a 99-year-old patient with com- It is possible that in a very blond fundus where the optic plete OCA in which they were unable to find an area of nerve appears gray and its borders are more difficult to de- foveal differentiation after serial sections through the pos- lineate, optic nerve head elevation could be present with- terior pole. They noted partial atrophy of the optic nerve out being clinically appreciated. The optic nerve thick- head in a vertical hourglass configuration. ness will be evaluated prospectively in a future study. Imaging with SD-OCT gives the clinician much more It is important to correlate the SD-OCT findings in this detailed information regarding the retina in patients with study with known histopathological findings in eyes of OCA and in those with known or suspected OA. A grad- subjects with OCA. Foveal hypoplasia was noted to be a ing schema similar to the one proposed by Seo et al4 could characteristic in OCA in several histopathological stud- be implemented, one with greater definition based on the ies in the early 20th century.28-32 Zhou et al33 describe the more exact details provided by high-speed, high- ocular findings in a patient with type 1 HPS. Hermansky- resolution SD-OCT imaging. Diagnostic signs of OA should Pudlak syndrome has a triad of findings including OCA, be searched for in an adequate retinal OCT image set. It is platelet dysfunction, and ceroid lipofuscin accumula- very important for the clinician to recognize that analysis tion. Zhou and colleagues describe a patient who had of an OCT image from outside the fovea could be mislead- macular transparency, absence of foveal pits, and foveal ing as the image could appear to represent foveal hypopla- light reflexes suggesting foveal hypoplasia, which was con- sia in an eye with a normal fovea (Figure 1B, scan 38 or firmed with TD-OCT. On pathology specimens, they de- 58). The following cardinal signs should be evaluated with

(REPRINTED) ARCH OPHTHALMOL / VOL 127 (NO. 1), JAN 2009 WWW.ARCHOPHTHALMOL.COM 43

©2009 American Medical Association. All rights reserved. Downloaded From: https://jamanetwork.com/ on 09/28/2021 an adequate OCT image set: (1) absence of central foveal 3. Biswas S, Lloyd IC. Oculocutaneous albinism. Arch Dis Child. 1999;80(6):565- depression; (2) persistence of ganglion and plexiform (re- 569. 4. Seo JH, Yu YS, Kim JH, Choung HK, Heo JW, Kim SJ. Correlation of visual acu- flective) retinal layers across the fovea; and (3) persis- ity with foveal hypoplasia grading by optical coherence tomography in albinism. tence of the nerve fiber layer (although this is slightly less Ophthalmology. 2007;114(8):1547-1551. common than the other signs). A very prominent image 5. Shen B, Samaraweera P, Rosenberg B, Orlow SJ. Ocular albinism type 1: more of the entire choroid is also common in these subjects, al- than meets the eye. Pigment Cell Res. 2001;14(4):243-248. though this may vary slightly depending on the level of 6. King RA, Summers CG. Albinism. Dermatol Clin. 1988;6(2):217-228. 7. Kinnear PE, Jay B, Witkop CJ Jr. Albinism. Surv Ophthalmol. 1985;30(2):75-101. choroidal pigmentation. Our study was limited by the in- 8. Summers CG. Vision in albinism. Trans Am Ophthalmol Soc. 1996;94:1095-1155. clusion of only 1 patient with known OCA. Study of SD- 9. Abadi RV, Bjerre A. Motor and sensory characteristics of infantile nystagmus. OCT in additional eyes of children with OCA could pro- Br J Ophthalmol. 2002;86(10):1152-1160. vide data corroborating our current findings. With the new 10. Rosenberg T, Schwartz M. Ocular albinism, X-linked. GeneReviews Web site. http: data provided by this imaging technique, new questions //www.ncbi.nlm.nih.gov/bookshelf/picrender.fcgi?book=gene&&partid =1343&blobtype=pdf. Accessed August 16, 2008. arise regarding the pathophysiology of OA. Is it solely the 11. Srinivasan VJ, Wojtkowski M, Witkin AJ, et al. High-definition and 3-dimen- lack of melanin production involved in this disease or is sional imaging of macular pathologies with high-speed ultrahigh-resolution op- there another process causing the persistence of multiple tical coherence tomography. Ophthalmology. 2006;113(11):2054.e1-2054.e14. retinal layers normally absent in the fovea? Is it the chias- doi: 10.1016/j.ophtha.2006.05.046. mal misrouting seen in OA and OCA that leads to a retro- 12. Farsiu S, Chiu SJ, Izatt JA, Toth CA, inventors. Fast detection and segmentation 22 of drusen in retinal optical coherence tomography images. US provisional patent grade effect on foveal development? As a new imaging application 61/021,821. January 17, 2008. modality easily applied to children even in the presence 13. Wojtkowski M, Srinivasan V, Fujimoto JG, et al. Three-dimensional retinal imaging of nystagmus, SD-OCT may provide valuable structural with high-speed ultrahigh-resolution optical coherence tomography. detail in children with suspected foveal hypoplasia and Ophthalmology. 2005;112(10):1734-1746. albinism. 14. American National Standards Institute. American National Standard for Safe Use of Lasers, ANSI Z 136.1-1993. Orlando, FL: Laser Institute of America; 1993. 15. Jiao S, Knighton R, Huang X, Gregori G, Puliafito C. Simultaneous acquisition of Submitted for Publication: April 11, 2008; final revi- sectional and fundus ophthalmic images with spectral-domain optical coher- sion received August 20, 2008; accepted September 3, ence tomography. Opt Express. 2005;13(2):444-452. 2008. 16. Stopa M, Bower BA, Davies E, Izatt JA, Toth CA. Correlation of pathologic fea- Correspondence: Cynthia A. Toth, MD, Departments of tures in spectral domain optical coherence tomography with conventional retinal studies. Retina. 2008;28(2):298-308. Ophthalmology and Biomedical Engineering, Box 3802, 17. Farsiu S, Robinson MD, Elad M, Milanfar P. Fast and robust multiframe super Duke University Medical Center, Durham, NC 27710 resolution. IEEE Trans Image Process. 2004;13(10):1327-1344. ([email protected]). 18. Thevenaz P, Ruttimann UE, Unser M. A pyramid approach to subpixel registra- Financial Disclosure: Dr Farsiu has received research sup- tion based on intensity. IEEE Trans Image Process. 1998;7(1):27-41. port from North Carolina Biotechnology Center and Bi- 19. Farsiu S, Elad M, Milanfar P. Constrained, globally optimal, multi-frame motion estimation. In: 2005 IEEE/SP 13th Workshop on Statistical Signal Processing. optigen, Inc. Dr Izatt is an investor and employee of Bi- 2005:1396-1401. doi:10.1109/SSP.2005.1628814. optigen, Inc and has patents with Bioptigen, Inc. Dr Toth 20. Takeda H, Farsiu S, Milanfar P. Kernel regression for image processing and has received research support and has been a consultant reconstruction. IEEE Trans Image Process. 2007;16(2):349-366. for Genentech, has received research support from Sirion 21. McGuire DE, Weinreb RN, Goldbaum MH. Foveal hypoplasia demonstrated in Therapeutics, has received research support and a speaker vivo with optical coherence tomography. Am J Ophthalmol. 2003;135(1):112- 114. honorarium from Bioptigen, Inc, has received research 22. van Genderen MM, Riemslag FC, Schuil J, Hoeben FP, Stilma JS, Meire FM. support and patent royalties from Alcon Laboratories, Inc, Chiasmal misrouting and foveal hypoplasia without albinism. Br J Ophthalmol. and has received research support from North Carolina 2006;90(9):1098-1102. Biotechnology Center. 23. Meyer CH, Lapolice DJ, Freedman SF. Foveal hypoplasia in oculocutaneous albinism demonstrated by optical coherence tomography. Am J Ophthalmol. 2002; Funding/Support: This study was supported by An- 133(3):409-410. gelica and Euan Baird, by a software donation from 24. Meyer CH, Lapolice DJ, Freedman SF. Foveal hypoplasia demonstrated in vivo Bioptigen, Inc, and in part by the North Carolina Bio- with optical coherence tomography. Am J Ophthalmol. 2003;136(2):397-398. technology Center Collaborative Funding Grant 2007- 25. Mietz H, Green WR, Wolff SM, Abundo GP. Foveal hypoplasia in complete ocu- CFG-8005 entitled “Improving Imaging of Phenotypes locutaneous albinism: a histopathologic study. Retina. 1992;12(3):254-260. 26. Hess DB, Asrani SG, Bhide MG, Enyedi LB, Stinnett SS, Freedman SF. Macular in Age-Related Macular Degeneration.” and retinal nerve fiber layer analysis of normal and glaucomatous eyes in chil- Additional Information: Through their university, Drs dren using optical coherence tomography. Am J Ophthalmol. 2005;139(3): Farsiu, Izatt, and Toth have filed a provisional patent ap- 509-517. plication for fast detection and segmentation of drusen 27. Repka MX, Goldenberg-Cohen N, Edwards AR. Retinal nerve fiber layer thick- in retinal optical coherence tomography images, and Dr ness in amblyopic eyes. Am J Ophthalmol. 2006;142(2):247-251. 28. Naumann GO, Lerche W, Schroeder W. Foveolar aplasia in tyrosinase-positive Toth has filed a patent application for a method and sys- oculocutaneous albinism (author’s transl) [in German]. Albrecht Von Graefes Arch tem of coregistrating optical coherence tomography with Klin Exp Ophthalmol. 1976;200(1):39-50. other clinical tests. 29. Fulton AB, Albert DM, Craft JL. Human albinism: light and electron microscopy study. Arch Ophthalmol. 1978;96(2):305-310. 30. Velhagen C. Ueber Albinismu. Muenchen Med Wchnschr. 1917;64:845. REFERENCES 31. Usher C. Histological examination of an adult human albino’s eyeball with a note on mesoblastic pigmentation in foetal eyes. Biometrika. 1920;13(1):46-56. 1. King RA, Hearing VJ, Creel D, Oetting WS. Albinism. In: Scriver CR, Beaudet AL, 32. Elschnig A. Zur Anatomie des menschlichen Albinoauges. Graefes Arch Ophthalmol. Sly WS, Valle D, eds. The Metabolic and Molecular Basis of Inherited Disease. 1913;84:401-419. 7th ed. New York, NY: McGraw-Hill; 1995:4353-4392. 33. Zhou M, Gradstein L, Gonzales JA, Tsilou ET, Gahl WA, Chan CC. Ocular patho- 2. Harvey PS, King RA, Summers CG. Spectrum of foveal development in albinism logic features of Hermansky-Pudlak syndrome type 1 in an adult. Arch Ophthalmol. detected with optical coherence tomography. J AAPOS. 2006;10(3):237-242. 2006;124(7):1048-1051.

(REPRINTED) ARCH OPHTHALMOL / VOL 127 (NO. 1), JAN 2009 WWW.ARCHOPHTHALMOL.COM 44