Clinical science

Anatomical and functional impairment of the retina and optic nerve in patients with anorexia nervosa without vision loss
Marilita M Moschos,1 Fragiskos Gonidakis,2 Eleftheria Varsou,2 Ioannis Markopoulos,1 Alexandros Rouvas,1 Ioannis Ladas,1 George N Papadimitriou2
1

1st Department of Ophthalmology, University of Athens, Medical School, Athens, Greece 2 1st Department of Psychiatry, University of Athens, Medical School, Athens, Greece Correspondence to Dr Marilita M Moschos, 6, Ikarias Street, Ekali, 14578, Athens, Greece; moschosmarilita@yahoo.fr Accepted 18 July 2010

ABSTRACT Aim The aim of this cross-sectional study is to evaluate the macular and retinal nerve bre layer (RNFL) thickness, as well as the electrical activity of the macula in female patients suffering from anorexia nervosa (AN) without visual failure. Material and methods 13 female patients (26 eyes) suffering from AN without visual failure and 20 age and sex-matched healthy female controls (40 eyes) were studied. For the measurement of the macula thickness and the electrical activity of the macula, the optical coherence tomography (OCT) and the multifocal electroretinogram were used respectively. Results The visual acuity, as well as the visual elds, the colour vision testing and the dark adaptation test of all patients were normal. However, the mean foveal thickness was 140.04 mm (vs 150.85 in the control group, p0.005), and the RNFL thickness was limited to 116.42 mm in the superior area (vs 123.15 in the control group, p0.372) and 121.08 mm in the inferior area (vs 137.6 in the control group, p<0.001) around the optic nerve. Also, the mean P1 response density amplitude of the foveal area was 159.04 nV/deg2 (vs 292.43 in the control group, p<0.0001), and the perifoveal area was 79.04 nV/deg2 (vs 82.63 in the control group, p0.118). Conclusion The present study shows that in patients with AN, even without visual failure there is a decrease in macular and RNFL thickness, as well as a decrease in the electrical activity of the macula.

and perforation5 and cataract.6 7 One case with rod dysfunction8 and another with central vein occlusion9 were also reported. No other contextual papers on nutrition and retinal impact were found. In our study, we evaluated the thickness of the macula and the retinal nerve bre layer (RNFL) as well as the electrical activity of the macula with optical coherence tomography (OCT) and multifocal electroretinogram (mf-ERG), respectively, in patients with AN. To our knowledge, this is the rst time a study has been performed to evaluate the anatomical and functional damage of the macula and optic nerve in anorectic patients before presenting irreversible visual impairment.

MATERIALS AND METHODS

Consecutive patients with AN followed by the 1st Department of Psychiatry were recruited for our study. On admission to the 1st Department of Ophthalmology, none of the patients suffered from osteoporosis, cardiovascular complications or electrolytic abnormalities. Plasma vitamin A and B12 were normal, and only a subclinical deciency of iron and folic acid was present in three cases. Exclusion criteria were a history of ocular surgery, ocular diseases as well as high refractive errors 66 D. Informed consent for imaging and data collection was obtained from the patients after an explanation of the nature of the study. Patients had no history of ocular disease or eye surgery, and no subjective symptoms, such as itching, redness, photophobia, tearing or low vision, were mentioned by the patients. The best-corrected visual acuity was 6/6 in all eyes, and uorescein angiography did not reveal any leakage or pigmentary lesion of the macula. The study included 66 eyes of 33 female individuals who were divided into two groups. Group A consisted of 26 eyes of 13 female patients suffering from AN. The mean age was 28.62 (SD 6.818) years. Group B consisted of 40 eyes of 20 age and sexmatched volunteers ophthalmologically normal with no ocular or systemic symptoms, who served as normal control subjects. All were female. The mean age was 28.20 (SD 8.118) years. The study was conducted in accordance with the tenets of the Declaration of Helsinki. Two anorectic females refused to participate and were not included in the study. Each individual included in our study underwent a complete ophthalmic examination in the 1st Department of Ophthalmology of Athens
1 of 6

INTRODUCTION
In developed countries, anorexia nervosa (AN) is a psychiatric disorder affecting 1e3% of middleand upper-class women. It affects predominantly female adolescents, and the female/male sex ratio is approximately 10:1. In teenage girls, AN is the third most common chronic disease with an estimated mortality of 5e10%.1 According to the Diagnostic and Statistical Manual of Mental Disorders (DSMIV), the diagnostic criteria for AN are weight loss leading to a body weight 15% below normal, intense fear of gaining weight and becoming fat, distorted body self-image, amenorrhoea of at least three expected menstrual cycles and absence of other physical disorders causing weight loss.2 AN affects almost every organ system such as the skin, gastrointestinal system and hypothalamicepituitaryeovarian hormonal axis.3 4 The literature investigating the ocular affection in AN is very poor and is limited to the anterior part of the eye with the appearance of corneal ulcer
Moschos MM, Gonidakis F, Varsou E, et al. Br J Ophthalmol (2010). doi:10.1136/bjo.2009.177899

Clinical science
University, including best-corrected visual-acuity assessment (standard Snellen chart), colour vision testing, fundus examination, and intraocular pressure measurement, standardised retinal photography with a wide-angle fundus camera using overlapping elds, uorescein angiography, OCT scan and mf-ERG recording. The recording of mf-ERG and OCT is not inuenced by refractive errors because the subjects during the examination are fully corrected with the appropriate eye contact lens. Also, OCT and mf-ERG recording was performed twice by two experienced ophthalmologists, and the results were similar with little variation. Measurements of RNFL thickness from three scans were averaged to provide a mean measurement of the RNFL thickness average as well as the following retinal regions: temporal (3168 to 458 on a unit circle), superior (468 to 1358 on a unit circle), nasal (1368 to 2258 on a unit circle) and inferior (2268 to 3158 on a unit circle).

Multifocal ERG
For the recording of the mf-ERG, the VERIS III (Visual Evoked Response Imaging System; Tomey, Nagoya, Japan) was used. The stimulus matrix consisted of 61 segments displayed on a CRT colour monitor driven at a frame rate of 72 Hz. These hexagons elicit an approximately equal signal amplitude at all locations on a normal retina. Each hexagon was independently alternated between black and white at a rate of 72 Hz, and the stimulation technique allowed a retinal response from each stimulus, using an M-sequence 1023. The radius of the stimulus array subtended approximately 208 high and 258 wide. The bandwidth of the amplier was 10e300 Hz, and the amplication was 310000. The pupils of the patients were dilated by means of tropicamide 0.5% and phenylephrine 5%, and the eyes were optically corrected for near vision to see clearly the small xation spot in the centre of the stimulus matrix. For signal acquisition,

Optical coherence tomography (OCT)

OCT examination was performed with the OCT model 3000 (Stratus OCT3, Carl Zeiss Meditec, Dublin, CA, USA). The retinal mapping software was used, calculating the averaged retinal thickness of the central ring. All eyes were scanned in a radial spoke pattern centred on the foveola with scan length of 6 mm. To measure RNFL thickness, a light-emitting diode was used, providing low-coherence infrared illumination that generates crosssectional images of the retina with an axial resolution of less than 10 mm. The RNFL thickness (3.4) protocol is designed to acquire three circle scans of diameter of 3.4 mm around the optic disc. Figure 1 (A) Optical coherence tomography (OCT) scans of the fovea of the left eye of a normal control female subject. (B) OCT scans around the left optic nerve. Normalised and aligned scans identify the outer and inner layers of the retinal nerve bre layer. (C) 3D appearance of mf-ERG recording and mf-ERG traces of the left eye.

2 of 6

Moschos MM, Gonidakis F, Varsou E, et al. Br J Ophthalmol (2010). doi:10.1136/bjo.2009.177899

Clinical science
a bipolar contact lens was used in which the active and the reference electrodes were incorporated in the contact lens. The ground electrode was attached to the ear lobe. The fellow eye was closed, and the duration of the data acquisition was 8 min divided into eight sessions of 60 s. The recording procedure was repeated if there were any spurious potentials from eye blinks or if ocular movement was recorded. The response density (amplitude per unit retinal area, nV/ deg2) of each local response was estimated as the dot product between the normalised response template and each local response. The normal ranges for these amplitudes were dened by calculating the median and the 95% CIs in both eyes of 20 normal volunteers (Group B). The mf-ERG stimuli location and anatomical areas corresponded roughly as follows: ring 1 to the fovea, ring 2 to the parafovea, ring 3 to the perifovea, ring 4 to the near periphery and ring 5 to the central part of the middle periphery. The amplitude of each group was scaled to reect the angular size of the stimulus hexagon, which produces the response. These averages give a more accurate view of the relative response densities of each group. The retinal response density decreases with eccentricity, although there is no further decrease from ring 4 to ring 5. measurements. The Z test showed signicant differences in mf-ERG ring 1 (Z3.7, p0.001) mf-ERG ring 2 (Z1.5, p0.02) and RNFL of the inferior area (Z2.3, p0.001). Because of these results, we used a non-parametric test (ManneWhitney U test) to compare the results between the two groups. We have conducted a post hoc power calculation according to the GPower software application for power analysis and the simulations of Olejnik and Algina.10 According to the above approach, we may have a post hoc approach for the sample size needed for a parametric analysis of the data under examination (32 anorectic patients and 32 controls at least). For the same reason, correlations between foveal thickness, P1-response density amplitude and RNFL measurements and factors related to AN, such as body mass index (BMI), min BMI ever measured during adulthood, and duration of the disorder in years were conducted using the Spearman rank test. A generalised linear model test was used to explore further the differences between the two groups.

RESULTS
Thirteen anorectic female patients with a disease duration of 10.468.4 years (mean6SD) and 20 age-matched healthy female controls (p0.83 vs age of PD patients) participated in the study (gures 1, 2). Box plots of the clinical measurements are presented for the anorectic group (gure 3) and the control group (gure 4). It is noticeable that the results from the anorectic group show

Statistical analysis
Because of the small sample size, a KolmogoroveSmirnov Z test was conducted in order to detect differences in both the location and the shape of the distribution between the two groups Figure 2 (A) Optical coherence tomography (OCT) scans of the fovea of the right eye of an anorectic female patient. (B) OCT scans around the right optic nerve. Normalised and aligned scans identify the outer and inner layers of the retinal nerve bre layer. (C) 3D appearance of mf-ERG recording and mf-ERG traces of the right eye. Retinal response densities of the fovea are considerably decreased for both eyes.

Moschos MM, Gonidakis F, Varsou E, et al. Br J Ophthalmol (2010). doi:10.1136/bjo.2009.177899

3 of 6

Clinical science
tude of ring 1 (p0.001) and ring 2 (p0.04) were different between the two groups (table 3). The results of the comparison between the two groups regarding the left eyes measurements are presented in table 2. The two groups showed differences in three measurements. First, the foveal thickness measured by OCT in anorectic patients was lower than in controls (p0.0071); second, the analysis of the inferior area in anorectic patients showed a lower RNFL thickness (p0.021); and third, the P1-response density amplitude of ring 1 in anorectic patients was lower than in controls (p0.0003). All other differences did not reach statistical signicance. The generalised linear model analysis showed that the foveal thickness measured by OCT (p0.002), the inferior RNFL thickness (p0.0002) and the P1-response density amplitude of ring 1 (p0.0001) and ring 2 (p0.0006) were different between the two groups (table 4). The Spearman rank test was used to investigate possible correlation between ophthalmological measurements and factors related to AN (BMI, min BMI and duration of the disorder in years). The test showed that BMI of anorectic patients at presentation correlated negatively with RNFL of the inferior area (p0.001), the nasal area (p0.001) and the superior RNFL (p0.001). The duration of the illness was also correlated negatively with the RNFL of the superior area (p0.001), the inferior area (p0.001) and the average RNFL (p0.001). None of the ophthalmological measurements showed a correlation with the reported min BMI (table 5). According to the American Psychiatric Association Diagnostic and Statistical Manual (DSM-IV) 2 classication for mental disorders, there are two types of AN: the restrictive type and the binge/purging type. Restrictive anorectic patients reduce their daily caloric intake, while bingeepurge anorectic patients occasionally succumb to binge episodes that are followed by purging behaviours such as induced vomit and laxative and diuretics abuse. Because of the small size of our sample, the ManneWhitney U test was used to compare the ophthalmological measurements between restrictive-type AN patients (six patients) and bingeepurge-type AN patients (seven patients). The two groups differed only in the left eye OCT measurements (Z2.37, p0.02), with the AN restrictive type having a higher foveal thickness (median142) than the AN bingeepurge type (median134). When the patients (eight patients) who were not inducing vomit were compared with the patients (ve patients) who were inducing vomit, the only marginal statistical difference that was found concerned the P1-response density amplitude of ring 1 of mf-ERG in both the right (z1.8, p0.06) and left eye (z1.7, p0.07), with AN patients who were not vomiting having a higher P1-response density amplitude in both the right (median164) and left eye (median185), compared with the AN patients who were inducing vomit (right eye median107.5, left eye median153). Similarly, the ManneWhitney U test was used to compare the ophthalmological measurements between AN patients who reported to be under any kind of psychiatric medication (ve patients, 10 eyes) and the AN patients who have never received any kind of psychiatric medication (eight patients, 16 eyes). No difference was found between the two groups in any of the measurements.

Figure 3 Box plots of the clinical measurements for the right eye of both groups. CMT, central macular thickness; i, retinal nerve bre layer (RNFL) inferior; n, RNFL nasal; s, RNFL superior; t, RNFL temporal; z1, ring 1; z2, ring 2. a much wider spread from the median than those in the control group. The ManneWhitney U test was used for the comparison of the two groups. (tables 1, 2). The results of the comparison between the two groups regarding the right eyes measurements are presented in table 1. The two groups showed differences in two measurements. First, the analysis of the inferior area in anorectic patients showed a lower RNFL thickness (p0.005), and second, the P1-response density amplitude of ring 1 in anorectic patients was lower than in controls (p0.0003). All other differences did not reach statistical signicance. The generalised linear model analysis showed that the inferior RNFL thickness (p0.0001) and the P1-response density ampli-

Figure 4 Box plots of the clinical measurements for the left eye of both groups. CMT, central macular thickness; i, retinal nerve bre layer (RNFL) inferior; n, RNFL nasal; s, RNFL superior; t, RNFL temporal; z1, ring 1; z2, ring 2.
4 of 6

DISCUSSION
Several physical complications are associated with AN. Many of these problems in behaviour were associated with controlling

Moschos MM, Gonidakis F, Varsou E, et al. Br J Ophthalmol (2010). doi:10.1136/bjo.2009.177899

Clinical science
Table 1 Comparison between anorectic and control groups (ManneWhitney U test) for the right eye
Anorectic group, n[13 Measurements Optical coherence tomography Multifocal electroretinogram (ring 1) Multifocal electroretinogram (ring 2) RNFL T RNFL S RNFL I RNFL N Median 140 131 72 75 114 111 76 Lower quartile 130 96 65 67 103 105 62 Upper quartile 150 199 87 85 124 132 81 Control group, n[20 Median 147 295 81 74 1225 138 79 Lower quartile 138.5 284.5 74 67.5 120 135 71.5 Upper quartile 160 302.5 90.5 79 127.5 140 85.5 ManneWhitney U test Z 1.13 4.18 1.36 0.63 1.79 2.75 0.79 p Value 0.26 0.0003 0.17 0.53 0.073 0.005 0.4

RNFL I, retinal nerve bre layer (RNFL) inferior; RNFL N, RNFL nasal; RNFL S, RNFL superior; RNFL T, temporal.

Table 2

Comparison between anorectic and control groups (ManneWhitney U test) for the left eye
Anorectic group, n[13 Control group, n[20 Upper quartile 138 184 104 78 133 135 85 Median 152 289 83 73 125 139 77 Lower quartile 141 281.5 77 67 118.5 136 70.5 Upper quartile 163 299 89.5 77 128 140.5 83 ManneWhitney U test Z 2.71 4.16 0.81 0.59 0.52 2.3 1.14 p Value 0.0071 0.0003 0.41 0.55 0.61 0.021 0.25

Measurements Optical coherence tomography Multifocal electroretinogram (ring 1) Multifocal electroretinogram (ring 2) RNFL T RNFL S RNFL I RNFL N

Median 134 159 76 70 124 128 72

Lower quartile 132 147 68 65 119 110 63

RNFL I, retinal nerve bre layer (RNFL) inferior; RNFL N, RNFL nasal; RNFL S, RNFL superior; RNFL T, temporal.

body weight in an unhealthy manner, and most of these problems resolved once eating habits and weight loss returned to normal. Vision is frequently affected in AN. Laurence et al11 support that in AN there is impairment in visual discrimination learning. This impairment may be related to decreased appetitive function, possibly resulting from impaired dopaminergic neurotransmission, either as a result of food restriction or, more intriguingly, related to the underlying pathophysiology of AN itself. Dopamine is an important neurotransmitter in the visual pathway, and its presence was documented in the mammalian retina, including the human retina. Previous studies in Parkinsons disease, where there is a reduction in dopamine in the retina and especially in the amacrine cells,12 13 show electrophysiological changes in the retina,14 15 and Palmowski et al16 in a recent study found a reduction in the amplitude of central oscillatory potentials. Also, Inzelberg et al17 suggested that it is possible that impoverished dopaminergic input to a subset of retinal ganglion cells contributes to abnormal production of glutamate and atrophy of these selected optic nerve bres and localised thinning of RNFL. Our results show that the retinal thickness of the macula is higher in restrictive-type anorectic patients than in bingeepurge type patients, which means that the anatomical impairment of the fovea is greater in the AN bingeepurge type.

It is also interesting that these ndings correlated negatively with the duration of the disease and the BMI on admission to the Department of Ophthalmology, but they never show any correlation with the patients min BMI. However, as indicated in the box plot gures, there is a high spread of the RNFL measures and mf-ERG responses of ring 1, which is much greater in anorectic patients than in controls. This is difcult to explain, because a similar high spread of the measurements was not observed in the OCT and mf-ERG. A larger population study would probably provide a reason for this nding. There is a dearth of literature investigating the ocular impact of AN. Except from two studies concerning the anterior part of the eye,5 6 only one case of an anorectic patient is reported with retinal lesions. Berthout et al8 described a case with affection of the cones and rods of the retina resulting in a decrease in central vision and visual-eld constriction bilaterally. The scotopic electroretinogram was almost absent, and the photopic ERG was abnormal with a decrease in a- and b-wave amplitude and absence of oscillatory potentials. Fluorescein angiography did not reveal any lesion of the pigmentary epithelium of the macula and the perimacular area, and the affection of the retina was attributed to a lack of vitamin A. In our study, the electrical activity of the macula and the thickness of the macular area in AN without visual failure have been studied, and the ndings are very interesting. Indeed, there is

Table 3
Variables

Generalised linear model for the right eye

Estimate 0.69 0.0013 0.0058 0.0050 0.008484 Standard 0.47 0.0022 0.00056 0.0024 0.0022 Wald 2.166 0.35 107.57 4.36 14.43 p Value 0.1 0.6 0.001 0.04 0.0001

Table 4
Variables

Generalised linear model for the left eye

Estimate 0.74 0.0074 0.0055 0.00096 0.19 Standard 0.41 0.0023 0.00051 0.0023 0.026 Wald 3.35 10.04 113.29 11.89 13.86 p Value 0.07 0.002 0.0001 0.0006 0.0002

Intercept Optical coherence tomography Multifocal electroretinogram z1 Multifocal electroretinogram z2 Retinal nerve bre layer inferior
z1, ring 1; z2, ring 2.

Intercept Optical coherence tomography Multifocal electroretinogram z1 Multifocal electroretinogram z2 Retinal nerve bre layer inferior
z1, ring 1; z2, ring 2.

Moschos MM, Gonidakis F, Varsou E, et al. Br J Ophthalmol (2010). doi:10.1136/bjo.2009.177899

5 of 6

Clinical science
Table 5 Spearman correlation matrix between factors related to anorexia nervosa (BMI, BMI min and duration of anorexia nervosa) and ophthalmological measurements
Variables Optical coherence Multifocal Multifocal tomography electroretinogram z1 electroretinogram z2 RNFL T RNFL S RNFL I RNFL Av BMI 0.36 1 0.64y 0.23 0.38 0.04 0.24 0.17 0.19 0.64y 1 0.32 0.23 0.01 0.18 0.19 0.19 0.23 0.32 1 0.14 0.23 0.4* 0.05 0.41* 0.16 0.39 0.38 0. 04 0.17 0.23 0. 01 0.19 0. 14 0. 23 0.05 1 0.56y 0.79y 0.56y 1 0.8y 0.34 0.63y 0.64y 0.79y 0.8y 1 Duration of the BMI min disorder in years 0.33 0. 25 0.09 0. 06 0.47* 0. 57y 0.35 0.58y

Optical coherence tomography 1 Multifocal electroretinogram z1 0.36 Multifocal electroretinogram z2 0.19 RNFL T 0.19 RNFL S 0.41* RNFL I 0.16 RNFL N 0.07 RNFL Av 0.39

0. 38 0.32 0. 24 0.23 0. 03 0.29 0.25 0.19 0.6y 0.15 0.63y 0. 13 0.53y 0.12 0.71 0.05

*Correlation is signicant at the 0.05 level (two-tailed). yCorrelation is signicant at the 0.001 level (two-tailed). BMI, body mass index; min, minimum; retinal nerve bre layer (RNFL) Av, average; RNFL I, inferior; RNFL N, nasal; RNFL S, superior; RNFL T, temporal; z1, ring 1; z2, ring 2.

a statistically signicant decrease in macular and RNFL thickness as well as a decrease in electrical activity in the macula and especially of the foveal area. Although we cannot pinpoint the cause of the photoreceptors abnormality, we presume that the substantial evidence against vitamin A deciency in these cases may support the idea that other nutritional abnormalities or impaired dopaminergic neurotransmission play a more substantial role in the function of the photoreceptors. To the best of our knowledge, there are no publications concerning the relationship between the affection of photoreceptors and the lack of dopamine in AN. Nevertheless, it is stated that patients with AN have an impaired dopamine function, as shown by a reduced concentration of dopamine metabolite homovenillic acid and altered growth hormone response to apomorphine stimulation.18 The limited number of patients included in our study prevents denitive conclusions. Whether these decits are preliminary signs, which will conduct to visual failure, or will be normalised on recovery is an important issue for further research. Despite the limitations of the study, a signicant anatomical and functional impairment has been proven for the rst time in the literature.
Competing interests None. Patient consent Obtained. Ethics approval Ethics approval was provided by the Athens University. Provenance and peer review Not commissioned; externally peer reviewed.

2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18.

REFERENCES
1. Mitchell JE. Anorexia nervosa: medical and physiological aspects. In: Brownell Kd, ed. Handbook of Eating DisordersdPhysiology Pathology and Treatment Obesity, Anorexia and Bulimia. New York: The Press, 1986:666e8.

American Psychiatric Association. Diagnostic and Statistical Manual of Mental Disorders. 4th edition text revision (DSM-IV-TR). Washington, DC: American Psychiatric Association, 2000:65e8. Langan SM, Farell PM. Vitamin E, vitamin A and essential fatty acid status of patients hospitalized for anorexia nervosa. Am J Clin Nutr 1985;41:1054e60. Sheniker IR, Shaw H. Diagnosis and early management of anorexia nervosa and bulimia. Semin Adolesc Med 1966;2:21e5. Gilbert JM, Weiss JS, Sattler AL, et al. Ocular manifestations and impression cytology of anorexia nervosa. Ophthalmology 1990;97:1001e7. Stigmar G. Anorexia nervosa associated with cataract (report of a case). Acta Ophthalmol 1965;43:787e9. Miller DA. A case of anorexia nervosa in a young woman with development of subcapsular cataract. Trans Ophthalmol Soc U K 1958;78:217e22. Berthout A, Sellam M, Denimal F, et al. Oeil et anorexie. A propos dun cas (in French). J Fr Ophtalmol 2007;30:15. Shibuya Y, Hayasaka S. Central retinal vein occlusion in a patient with anorexia nervosa. Am J Ophthalmol 1995;119:109e10. Olejnik S, Algina J. Measures of effect size for comparative studies: Applications, Interpretations and limitations. Contemp Educ Psychol 2000;25:241e86. Laurence AD, Dowson J, Foxal GL, et al. Impaired visual discrimination learning in anorexia nervosa. Appetite 2003;40:85e9. Haggendal J, Malmfors T. Evidence of dopamine concerning neurons in the retina of the rabbit. Acta Physiol Scand 1963;64:58e66. Harnois C, Di Polo T. Decrease dopamine in the retina of patients with Parkinson disease. Invest Ophthalmol Vis Sci 1990;31:2473e5. Ikeda H, Head GM, Ellis CJ. Electrophysiological signs in dopamine deciency in recently diagnosed Parkinsons disease and a follow-up study. Vision Res 1994;34:2629e38. Gottlob I, Weghaupt H, Vass C, et al. Effect of levedopa on the human pattern electroretinogram and pattern visual evoked potentials. Graefes Arch Clin Exp Ophthalmol 1989;227:421e7. Palmowski-Wolfe AM, Perez MT, Behnke S, et al. Inuence of dopamine deciency in early Parkinsons disease on the slow stimulation multifocal ERG. Doc Ophthalmol 2006;112:209e15. Inzelberg R, Ramirez JA, Nisipeanu P, et al. Retinal nerve ber layer thinning in Parkinsons disease. Vision Res 2004;44:2793e7. Brambilla F, Bellodi L, Arancio C, et al. Central dopaminergic function in anorexia and bulimia nervosa. Psychoneuroendocrinology 2001;26:393e408.

6 of 6

Moschos MM, Gonidakis F, Varsou E, et al. Br J Ophthalmol (2010). doi:10.1136/bjo.2009.177899