Purpose: Choroideremia is a currently incurable X‐linked recessive retinal degeneration that leads to blindness. Gene therapy approaches to date target the outer retinal layers. However, the choroideremia (CHM) gene is expressed in all retinal layers, and a previous study on a small cohort of choroideremia patients suggested possible thinning of the retinal nerve fibre layer (RNFL). The purpose of the study was to examine the RNFL in detail using advanced imaging techniques in a larger cohort of choroideremia patients. Methods: Spectral domain optical coherence tomography of the peripapillary RNFL acquired with the Heidelberg Spectralis HRA circular scan mode were analysed retrospectively in 41 eyes of 21 choroideremia patients aged 39.6 years (±3.7 SEM). As age‐matched controls, 20 eyes from 10 patients with retinitis pigmentosa and 56 eyes from 28 healthy individuals were also assessed. Automated RNFL segmentation was adjusted manually to precisely delineate the RNFL. The data were also compared against an external normative database. Results: Mean peripapillary RNFL thickness in choroideremia was 130 ± 3 μm in the right eye (OD) and 133 ± 3 μm in the left eye (OS). This was 24% and 27% thicker than RNFL thickness in the controls (p < 0.001 for both). Patients with retinitis pigmentosa also showed an increase in RNFL thickness, which was no different to the choroideremia cohort (p > 0.05). Compared with manual analysis, the automated function of the inbuilt software was consistently inaccurate in segmenting the RNFL in choroideremia. Conclusion: The RNFL is significantly thicker in choroideremia compared with age‐matched normal controls, which was similar to what was seen in retinitis pigmentosa.
Keywords: choroideremia; optical coherence tomography; peripapillary retinal nerve fibre layer; retinal nerve fibre layer
Choroideremia is an X‐linked (Xq21) retinal degeneration mainly caused by loss‐of‐function mutations in the choroideremia (CHM) gene, which encodes Rab escort protein‐1 (REP1) (Cremers et al. [
Genead et al. ([
The primary objective of this study was therefore to determine whether or not choroideremia is associated with degeneration of the RNFL, as is observed with the outer retinal layers. Spectral domain OCT can reproducibly measure the thickness of individual retinal layers, including the RNFL (Schuman et al. [
This was a retrospective, noninterventional image analysis of RNFL OCT acquired as per standard operating procedure (see below) in patients with clinical or genetic diagnosis of choroideremia or retinitis pigmentosa at the Oxford Eye Hospital. The RNFL thickness was compared to age‐matched normal controls consisting of healthy volunteers. The study sample comprised: 41 eyes of 21 choroideremia patients; 20 eyes of 10 retinitis pigmentosa patients; and 56 eyes of 28 normal controls (Table ). One eye was excluded due to poor ocular fixation and consequently poor scan acquisition. Due to X‐linked inheritance, only male patients were included in the choroideremia cohort. The study was carried out as part of an ongoing clinical trial (NCT01461213) approved by the National Research Ethics Committee and adhered to the Declaration of Helsinki (2013).
Cohort demographics
Male/Female Eyes Control 28 14/14 56 Choroideremia 21 21/0 41 RP 10 6/4 20
Cohort demographics
p‐value vs. Control p‐value RE vs. LEMean SEM Range Age/years Control 32.6 1.8 21, 55 Choroideremia 39.6 3.7 12, 69 0.07 N/A RP 49.6 5.1 27, 66 <0.001 N/A Refractive error/D (RE/LE) Control −0.6/−0.6 0.3/0.3 −.1,3.5/−4.1,5.8 Choroideremia −1.8/−0.9 0.7/0.8 −10.7,7.5/−10.7,2.9 0.08/0.35 0.65 RP −2.1/−0.6 1.2/1.3 −7.6,7.3/−8.0,2.1 0.08/0.98 0.40 Disc area/mm2 (RE/LE) Control 2.3/2.3 0.1/0.1 1.6,3.7/1.5,3.8 Choroideremia 2.5/2.5 0.1/0.8 1.3,3.9/1.7,3.6 0.08/0.17 0.88 RP 2.7/2.5 0.3/0.2 1.5,4.0/1.6,3.5 0.03/0.27 0.56
1 Demographics of choroideremia (CHM), retinitis pigmentosa (RP) and normal eyes. Control group consisted of volunteers without ocular pathology. The mean, standard error of the mean (SEM), median, range, refractive error and optic disc area of the right (RE) and left eye (LE) were compared. One‐way analysis of variance for comparison of control, choroideremia and RP values was used to calculate p‐values. N/A denotes not applicable.
The Spectralis HRA + OCT system's circular scan mode (Heidelberg Engineering GmbH, Heidelberg, Germany) was centred on the optic disc to obtain cross‐sectional images along a 360° path (3.4 mm diameter) of the retinal circumpapillary region. Each image was generated from mean Automatic Real‐Time (ART) alignment of multiple scans. The mean numbers of scans averaged were 28.1, 28.8 and 21.8 for the control, choroideremia and retinitis pigmentosa cohort, respectively. Scan averaging values ranged from 15 to 100, which has shown to be sufficient for identifying retinal layer boundaries in retinal diseases (Sakamoto et al. [
Calculations of retinal layer thickness were performed using Heidelberg Eye Explorer (HEYEX; Heidelberg Engineering). The thickness of the whole retina was defined by manual segmentation of the internal limiting membrane (ILM) and the retinal pigment epithelium (RPE)/Bruch's membrane (BM) complex. RNFL thickness was obtained by manual segmentation of the ILM and ganglion cell layer (GCL). While automated RNFL segmentation within HEYEX works well for normal retina, it could not accurately segment the abnormal retina in choroideremia or retinitis pigmentosa. Therefore, delineation of the ILM, GCL and RPE/BM complex was manually corrected by one grader (DJ) and checked by another grader (KX). Thickness of non‐RNFL retinal layers was calculated by subtracting the RNFL from the whole retinal thickness within each sector.
Choroideremia and retinitis pigmentosa scans were analysed and compared against HEYEX age‐adjusted normative data for peripapillary RNFL thickness. Briefly, the database includes RNFL measurements of 111 males and 90 females with normal intraocular pressure, visual field, optic disc appearance and eyes determined by two ophthalmologists via patient history and clinical examination. All subjects on the normative database were Caucasian with a mean age of 48.2 ± 14.5 (SD) year (range 18–78 year). Age‐adjusted normal distributions were calculated using the following formula: I + S * a + Z *δ ; where I = intercept, S = slope of the regression of RNFL thickness versus age, a = age, Z = inverse of the standard normal cumulative distribution of the percentile and δ = standard deviation. HEYEX was used to compare the RNFL scans of each choroideremia and retinitis pigmentosa eye to the HEYEX normative database in order to determine whether the RNFL thickness of a given sector was below 5th percentile, between 5th and 95th percentile or above 95th percentile.
The data acquired for the right eye (RE) and left eye (LE) were analysed separately. Statistical significance tests were performed using Sigma Plot 12.5 (systat software inc.). Distribution of data was tested for normality and equal variance using Shapiro‐Wilk and Brown‐Forsythe test, respectively. If the data passed both tests, one‐way analysis of variance (anova) was performed. If a statistically significant difference in the mean values among the patient groups was detected, the anova was followed by all pairwise multiple comparison procedures (Bonferroni method). If either the normality or the equal variance test failed, the difference in median values among the patient groups was compared using Kruskal–Wallis one‐way analysis of variance (anova) on ranks, followed by all pairwise multiple comparison procedures (Dunn's Method). Calculated means in text and graphs are expressed with ± error margin corresponding to the 95% confidence interval, unless otherwise specified. Linear regression analyses were conducted using age, laterality of eye, refractive error, area of optic disc and area of AF islands in choroideremia eyes as independent variables. AF area values were not available in two choroideremia eyes due to poor image quality and therefore omitted from analysis. The mean RNFL of each sector in choroideremia eyes were individually interrogated as the dependent variable. A p value <0.05 was considered significant.
Peripapillary OCTs were obtained from a total of 117 eyes of 59 participants consisting of 22 male patients with choroideremia (mean age 39.6 year ± 3.7 SEM), 10 patients with retinitis pigmentosa (mean age 49.6 year ± 5.1 SEM) and 28 normal volunteers of comparable age (mean 32.6 year ± 1.8 SEM) and refractive error as defined by the best vision sphere (Table ). Statistically significant differences were not detected between the mean RNFL thicknesses of male and female normal controls in all sectors (Fig. S1). Therefore, male and female data points were analysed collectively.
Peripapillary whole retinal thickness (Fig. ), RNFL thickness (Fig. ) and non‐RNFL layers (from the GCL to the RPE) (Fig. S2) were compared between choroideremia, retinitis pigmentosa and controls. In addition, whole retinal thickness and RFNL thickness of the papillomacular bundle region for each group were compared to those in the other radial sectors around the optic disc (Fig. C and Fig. C). In our normal control cohort, the mean whole retinal thickness, RNFL and sub‐RNFL layers (Table ) were similar to previously published normal cohorts (Bowd et al. [
Thickness of retinal layers surrounding the optic nerve in a cohort of normal eyes
Whole retinal RNFL Non‐RNFL RE LE RE LE RE LE Nasal 306 ± 3 302 ± 3 84 ± 2 82 ± 2 222 ± 2 221 ± 2 Infero‐nasal 325 ± 5 328 ± 5 116 ± 4 123 ± 5 209 ± 4 206 ± 3 Infero‐temporal 358 ± 4 352 ± 6 156 ± 3 154 ± 3 202 ± 2 199 ± 6 Supero‐temporal 317 ± 3 320 ± 3 77 ± 2 76 ± 2 240 ± 3 244 ± 3 temporal 358 ± 4 354 ± 4 143 ± 3 140 ± 2 215 ± 3 214 ± 4 Supero‐nasal 328 ± 5 337 ± 4 111 ± 4 121 ± 4 217 ± 4 216 ± 2 Total 328 ± 3 328 ± 3 104 ± 2 105 ± 2 223 ± 2 224 ± 2 Papillomacular 309 ± 3 310 ± 3 56 ± 2 58 ± 1 253 ± 4 252 ± 3
2 Peripapillary retinal layer measurements in 6 disc sectors and the papillomacular bundle in 28 normal volunteers in microns (μm). Mean values of the right eye (RE) and left eye (LE) are presented with ± SE of the mean.
Peripapillary retinal layer measurements in choroideremia
Mean (μm) % change from control p‐value vs. Control p‐value vs. RP RE LE RE LE RE LE RE LE Whole retinal thickness Nasal 205 ± 7 200 ± 8 −33% −34% <0.001 <0.001 NS NS Infero‐nasal 212 ± 11 213 ± 11 −35% −35% <0.001 <0.001 NS NS Infero‐temporal 250 ± 12 258 ± 11 −30% −27% <0.001 <0.001 NS NS Temporal 238 ± 9 234 ± 13 −25% −27% <0.001 <0.001 NS NS Supero‐temporal 252 ± 13 254 ± 12 −30% −28% <0.001 <0.001 NS NS Supero‐nasal 217 ± 12 225 ± 10 −34% −33% <0.001 <0.001 NS NS Total 228 ± 8 230 ± 8 −31% −30% <0.001 <0.001 NS NS Papillomacular 224 ± 10 233 ± 8 −27% −25% <0.001 <0.001 NS NS Retinal nerve fiber layer Nasal 114 ± 3 112 ± 3 36% 37% <0.001 <0.001 NS NS Infero‐nasal 127 ± 5 128 ± 5 10% 5% NS NS NS NS Infero‐temporal 160 ± 6 163 ± 6 3% 6% NS NS NS NS Temporal 127 ± 6 124 ± 5 63% 64% <0.001 <0.001 NS NS Supero‐temporal 151 ± 7 166 ± 6 6% 18% NS <0.001 NS NS Supero‐nasal 122 ± 6 135 ± 4 10% 12% NS NS NS NS Total 130 ± 3 133 ± 3 24% 27% <0.001 <0.001 NS NS Papillomacular 106 ± 6 104 ± 5 89% 81% <0.001 <0.001 NS NS Non‐RNFL retinal layers Nasal 90 ± 9 88 ± 8 −59% −60% <0.001 <0.001 NS NS Infero‐nasal 85 ± 10 85 ± 10 −59% −59% <0.001 <0.001 NS NS Infero‐temporal 90 ± 11 94 ± 8 −56% −53% <0.001 <0.001 NS NS Temporal 112 ± 11 110 ± 16 −54% −55% <0.001 <0.001 NS NS Supero‐temporal 101 ± 14 89 ± 8 −53% −59% <0.001 <0.001 NS NS Supero‐nasal 96 ± 12 90 ± 8 −56% −58% <0.001 <0.001 NS NS Total 98 ± 9 97 ± 8 −56% −57% <0.001 <0.001 NS NS Papillomacular 118 ± 12 129 ± 10 −53% −49% <0.001 <0.001 NS NS
3 Peripapillary retinal layer measurements in the choroideremia cohort of 21 patients. Mean values of the right eye (RE) and left eye (LE) are presented with ± standard error of the mean. Percentage change is calculated in proportion to averages from the control cohort. One‐way analysis of variance for comparison of control, choroideremia and RP values was used to calculate p‐values. p‐values >0.05 were considered not significant (NS). No significant differences were found when comparing any metrics in choroideremia and RP. Whole retinal thickness and non‐RNFL retinal layers in choroideremia are significantly thinned across all sectors. However, the RFNL is comparatively thickened in choroideremia in certain sectors and when the retinal is measured as a whole.
To further interrogate whether the RNFL thickness in choroideremia also deviates from the normative database, we compared each RNFL measurement to age‐adjusted values from the HEYEX normative database of 201 normal participants (Heidelberg Eye Explorer software; Heidelberg Engineering GmbH) (Fig. S3A). Here, sectors were considered thickened when surpassing the 95th percentile of a given age‐adjusted range. Thickening was detected in at least two quadrants for all choroideremia eyes studied and at least three quadrants in 83% (34/41) of eyes; the majority of which were the nasal, temporal and papillomacular bundle sectors (all >80% of eyes). RNFL thinning (measurements that fell below 5th percentile) was only detected in 10% (4/41; RE: 5%, LE: 15%) of eyes (Fig. S3B). Hence the increased RNFL seen in our choroideremia group was validated against both our internal controls and the external normative database.
Global peripapillary RNFL thickening has also been described in retinitis pigmentosa using OCT (Walia & Fishman [
Peripapillary retinal layer measurements in retinitis pigmentosa
Mean (μm) % change from control p value vs. Control RE LE RE LE RE LE Whole retinal thickness Nasal 211 ± 9 215 ± 10 −31% −29% <0.001 <0.001 Infero‐nasal 225 ± 13 221 ± 14 −31% −33% <0.001 <0.001 Infero‐temporal 275 ± 12 263 ± 13 −23% −25% <0.001 <0.001 Temporal 251 ± 8 256 ± 7 −21% −20% <0.001 <0.001 Supero‐temporal 267 ± 12 268 ± 17 −25% −24% <0.001 <0.001 Supero‐nasal 233 ± 11 226 ± 13 −29% −33% <0.001 <0.001 Total 240 ± 8 240 ± 10 −27% −27% <0.001 <0.001 Papillomacular 235 ± 19 239 ± 8 −24% −23% <0.001 <0.001 Retinal nerve fibre layer Nasal 105 ± 6 103 ± 7 25% 26% <0.001 <0.001 Infero‐nasal 117 ± 7 122 ± 10 −1% 1% NS NS Infero‐temporal 155 ± 11 154 ± 9 0% 0% NS NS Temporal 119 ± 10 114 ± 7 53% 51% <0.001 <0.001 Supero‐temporal 145 ± 10 153 ± 13 1% 9% NS NS Supero‐nasal 126 ± 8 121 ± 10 14% 0% NS NS Total 125 ± 6 123 ± 6 19% 18% 0.01 0.01 Papillomacular 99 ± 20 90 ± 7 77% 57% <0.001 <0.001 Non‐RNFL retinal layers Nasal 106 ± 8 112 ± 9 −52% −49% <0.001 <0.001 Infero‐nasal 107 ± 12 99 ± 9 −49% −52% <0.001 <0.001 Infero‐temporal 120 ± 12 110 ± 12 −40% −45% <0.001 <0.001 Temporal 133 ± 12 142 ± 11 −45% −42% <0.001 <0.001 Supero‐temporal 122 ± 12 115 ± 6 −43% −46% <0.001 <0.001 Supero‐nasal 107 ± 9 105 ± 5 −51% −46% <0.001 <0.001 Total 116 ± 9 117 ± 7 −48% −46% <0.001 <0.001 Papillomacular 136 ± 25 149 ± 14 −49% −46% <0.001 <0.001
4 Summary of peripapillary retinal layer measurements in a cohort of 10 patients with retinitis pigmentosa (RP). Mean values of the right eye (RE) and left eye (LE) are presented with ± standard error of the mean. Percentage change is calculated in proportion to averages from control cohort. One‐way analysis of variance for comparison of control, choroideremia and RP values was used to calculate p‐values. p‐values >0.05 were considered not significant (NS). Whole retinal thickness and non‐RNFL retinal layers in RP are significantly thinned across all sectors. However, the RFNL is comparatively thickened in RP when the retinal is measured as a whole and in specific sectors (nasal, temporal and papillomacular bundle sectors).
Age, ethnicity, refractive error and optic disc area have each been shown independently to predict RNFL thickness in normal eyes (Budenz et al. [
The evidence presented here demonstrates that the peripapillary RNFL in choroideremia patients is thickened when compared to a control cohort matched in age, refractive error and optic disc area. This was largely accounted for by the nasal and temporal quadrants. These data suggest that the absence of REP1 does not lead to thinning of the RNFL, as is seen with outer retinal layers following degeneration. Conversely, choroideremia consistently features marked thickening of the nasal and temporal quadrants of the peripapillary regions. Temporal thickening of RNFL may be related to the previous findings of preservation of the central macula region until the very terminal stages of the disease (Jolly et al. [
Our RNFL findings do however contrast with observations from a group of 16 choroideremia patients previously reported by Genead et al. ([
Quantification of retinal layers surrounding the optic nerve head in our retinitis pigmentosa cohort was consistent with previously reported mean values (Walia & Fishman [
It is important to understand the mechanism that gives rise to thickening of the RNFL as it would contribute to our understanding of choroideremia pathogenesis with possible insights into the prognosis of therapies in development. Our data suggest that RNFL thickening in choroideremia is independent of outer retinal changes and present in early disease. Age, refractive error, optic disc area and FAF are not significant predictive factors of peripapillary RNFL thickness in the multiple linear regression models of our measurements in choroideremia. This implies that the RNFL thickening seen in choroideremia is not directly linked to the severity of outer retinal degeneration. If the RNFL thickening were secondary to disease progression, one would expect correlation with age and the area of residual retina. This is also supported by the observation that – in comparison to the control cohort – choroideremia patients featured thinning of the outer retinal layers and whole retinal thickness (the sum of all retinal layers). Critically, this was the case across all circumpapillary sectors including the sectors of prominent RNFL thickening (nasal, temporal and papillomacular bundle sectors).
In light of the results, one could postulate that the thickening phenotype may be due to a developmental mechanism. The retinal pigment epithelium has a key role in ganglion cell neurogenesis through mechanisms such as the Wnt (Liu et al. [
An alternative explanation for the RNFL thickening may be a homoeostatic mechanism, in which inner retinal thickening occurs as a compensatory response to outer retinal thinning. In support of this hypothesis is our observation that RNFL thickening was greater in the temporal than in the nasal quadrant. The fact that RNFL thickening also occurs in retinitis pigmentosa caused by a variety of gene defects implies that the changes may be a general response to outer retinal degeneration. Metabolic signalling may play a role, as it has been shown that the peripapillary RNFL thickness in retinitis pigmentosa correlates with metabolic function and blood vessel architecture of the inner retina (Bojinova et al. [
In considering the mechanisms underlying RNFL thickening in choroideremia, it is important to recognise that OCT imaging offers in vivo morphometric assessment but does not directly measure cell mass, number or type. Expansion of the RNFL could therefore be due to hypertrophy, proliferation or oedema of ganglion cell nerve fibres. Other cells may also contribute to the observed thickening. This could be Müller cells, as they span the inner retina and undergo reactive gliosis after retinal injury (MacLaren [
Here, we provide evidence to suggest that the RNFL appears relatively spared in choroideremia and somewhat thickened compared with age‐matched controls. This suggests that the integrity of the optic nerve is preserved and that the absence of REP1 in the ganglion cells may not hinder therapies aimed at restoring outer retinal function, such as gene therapy and electronic photosensitive microchip implantation (Zrenner et al. [
The research was funded by the National Institute for Health Research (NIHR) Oxford Biomedical Research Centre based at Oxford University Hospitals NHS Trust and University of Oxford, The Health Innovation Challenge Fund, Fight for Sight. The views expressed are those of the authors and not necessarily those of the NHS, the NIHR or the Department of Health. The sponsor and funding organization had no role in the design or conduct of this research.
GRAPH: Figure S1. Comparison of segmental RNFL thicknesses in male and female normal controls revealed no statistically significant difference.
GRAPH: Figure S2. Differences in peripapillary sub‐RNFL (retinal nerve fibre layer) thickness between choroideremia (CHM), retinitis pigmentosa (RP) and normal (Control) eyes. (A) Mean sub‐RNFL thickness (ganglion cell layer to retinal pigment epithelium) and error bars representing standard error of the mean was plotted against location around the optic disc and comparison made between normal (black), CHM (blue) and retinitis RP (red). (B) The sub‐RNFL thickness along the papillomacular bundle (PMB) was compared against global (360 degree) retinal thickness in normal, CHM and RP eyes. Error bars represent 95% confidence intervals of the means. One‐way analysis of variance for comparison of control, CHM, and RP values was used to calculate P‐values. *** indicates P ≤ 0.001 when compared with control.
By Dun J. Fu; Kanmin Xue; Jasleen K. Jolly and Robert E. MacLaren
Reported by Author; Author; Author; Author