Can Structural Grading of Foveal Hypoplasia Predict Future Vision in Infantile Nystagmus?

Journal Pre-proof Can structural grading of foveal hypoplasia predict future vision in infantile nystagmus? A longitudinal study Sohaib R. Rufai, BMBS, BMedSc, Mervyn G. Thomas, MBChB, PhD, Ravi Purohit, BSc, MSc), Catey Bunce, MSc, DSc, Helena Lee, PhD, FRCOphth, Frank A. Proudlock, MSc, PhD, Irene Gottlob, MD, FRCOphth PII:

S0161-6420(19)32212-2

DOI:

https://doi.org/10.1016/j.ophtha.2019.10.037

Reference:

OPHTHA 10988

To appear in:

Ophthalmology

Received Date: 23 August 2019 Revised Date:

18 October 2019

Accepted Date: 28 October 2019

Please cite this article as: Rufai SR, Thomas MG, Purohit R, Bunce C, Lee H, Proudlock FA, Gottlob I, Can structural grading of foveal hypoplasia predict future vision in infantile nystagmus? A longitudinal study, Ophthalmology (2019), doi: https://doi.org/10.1016/j.ophtha.2019.10.037. This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of record. This version will undergo additional copyediting, typesetting and review before it is published in its final form, but we are providing this version to give early visibility of the article. Please note that, during the production process, errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain. © 2019 Published by Elsevier Inc. on behalf of the American Academy of Ophthalmology

1

Title Page

2

Full title

3

Can structural grading of foveal hypoplasia predict future vision in infantile nystagmus? A

4

longitudinal study

5

Authors

6

Sohaib R Rufai (BMBS, BMedSc)1, Mervyn G Thomas (MBChB, PhD)1, Ravi Purohit (BSc,

7

MSc)1, Catey Bunce (MSc, DSc)3, Helena Lee (PhD, FRCOphth)2, Frank A Proudlock (MSc,

8

PhD)1, Irene Gottlob (MD, FRCOphth)1*

9

Affiliations

10 11 12 13 14 15

1. The University of Leicester Ulverscroft Eye Unit, Robert Kilpatrick Clinical Sciences Building, Leicester Royal Infirmary, Leicester, UK. 2. Department of Primary Care & Public Health Sciences, King’s College London, London, UK. 3. Clinical and Experimental Sciences, Faculty of Medicine, University of Southampton, Southampton, UK.

16

Corresponding author*

17

Professor Irene Gottlob

18

E-mail: [email protected]

19

Telephone: +44 (0)116 2523268

20

Fax: +44 (0)116 2231996

21

Meeting Presentations: This study was presented in part at the following international

22

meetings: Royal College of Ophthalmologists Annual Congress, May 21-24 2018, Liverpool,

23

UK; Association for Research in Vision and Ophthalmology (ARVO) Annual Meeting, April

24

29-May 3 2018, Honolulu, Hawaii, USA – Awarded ARVO Members in Training Prize for

25

Best Neuro-ophthalmology Poster.

26

Financial Support: Medical Research Council, UK (grant numbers: MR/N004566/1 and

27

MR/J004189/1); Fight for Sight, UK (grant number: 5009/5010); Ulverscroft Foundation,

28

UK; Nystagmus Network, UK. The sponsor or funding organization had no role in the design

29

or conduct of this research

30

Conflict of Interest: No conflicting relationship exists for any author.

31

Running head: Predicting Future Vision in Infantile Nystagmus

32

33 34

Abstract

35

PURPOSE: To evaluate structural grading and quantitative segmentation of foveal

36

hypoplasia using handheld optical coherence tomography (HH-OCT), versus preferential

37

looking, as predictors of future vision in preverbal children with infantile nystagmus.

38

DESIGN: Longitudinal cohort study

39

PARTICIPANTS: 81 eyes from 42 patients with infantile nystagmus (albinism, n=19;

40

idiopathic infantile nystagmus (IIN), n=17; achromatopsia, n=6) were examined.

41

METHODS: A 10×10mm foveal area was sampled using a 3-dimensional raster scan

42

program (500×100, A scans×B scans) with ultra-high resolution spectral-domain HH-OCT

43

(Leica Microsystems Envisu C2300, <4 µm axial resolution) in preverbal children aged up to

44

36 months. Foveal tomograms were graded using our six-point grading system for foveal

45

hypoplasia and segmented for quantitative analysis: photoreceptor (PR) length, outer segment

46

(OS) length and foveal developmental index (FDI, a ratio of inner layers versus total foveal

47

thickness). Patients were followed up until they could perform chart visual acuity (VA)

48

testing. Data was analysed using linear mixed regression models. VA predicted by foveal

49

grading was compared to prediction by preferential looking (PL) – the current gold standard

50

for visual assessment in infants and young children.

51

MAIN OUTCOME MEASURES: Grade of foveal hypoplasia, quantitative parameters (PR

52

length, OS length, FDI) and preferential looking VA were obtained in preverbal children for

53

comparison with future chart VA outcomes.

54

RESULTS: We imaged 81 eyes from 42 patients with infantile nystagmus of mean age 19.8

55

months (range: 0.9-33.4; S.D. 9.4) at their first HH-OCT scan. Mean follow-up was 44.1

56

months (range: 18.4-63.2; S.D. 12.0). Structural grading was the strongest predictor of future

57

visual acuity (grading: r=0.80, F=67.49, p<0.0001) compared to quantitative measures (FDI:

58

r=0.74, F=28.81, p<0.001; OS length: r=0.65; F=7.94, p=0.008; PR length: r=0.65; F=7.94,

59

p=0.008). PL was inferior to VA prediction by foveal grading (PL: r=0.42; F=3.12, p<0.03).

60

CONCLUSIONS: Infantile nystagmus can be an alarming condition with variable impact on

61

vision. HH-OCT can predict future VA in infantile nystagmus. Structural grading is a better

62

predictor of future VA than quantitative segmentation and PL testing. Predicting future vision

63

could avert parental anxiety and optimise childhood development.

64

FINANCIAL DISCLOSURES: The authors have no proprietary or commercial interest in

65

any materials discussed in this article.

66

Introduction

67 68

Nystagmus is a condition of constant, involuntary to-and-fro movements of the eyes

69

occurring in 24 per 10,000 people.1 Onset is usually in early infancy, which can be alarming

70

for parents and families especially because the pre-verbal child is unable to communicate

71

their level of vision. Parents may initially believe the child is blind or has severely impaired

72

vision and their reaction can include fear, anxiety and uncertainty over their child’s future.2

73

As there is great variation in visual prognosis, predicting future vision could avert anxiety

74

and/or help parents plan adjustments to support the child, particularly to optimise

75

development and educational attainment. Targeted interventions may include accessible

76

learning materials and toys, as well as support with safe mobility and daily living skills.2

77 78

Optical coherence tomography (OCT) is a non-invasive imaging technique that provides

79

ultra-high resolution cross-sectional scans of the retina and optic nerve within seconds.

80

Recently, a handheld spectral-domain OCT (HH-OCT) has been developed for clinical use in

81

the paediatric population. It has overcome the limitations of the conventional adult OCT

82

system and is demonstrably reliable, including in nystagmus.3-5 The normal retinal6 and optic

83

nerve development7 have been recently described in vivo using HH-OCT. Diseases

84

investigated with HH-OCT include retinopathy of prematurity and other retinal disorders,

85

nystagmus, trauma, optic nerve disease, intraocular tumours and central nervous diseases.3,5,8

86 87

Normal development of the human fovea has been traced from 22 weeks gestation to 45

88

months postpartum using ex vivo histological specimens.9 This is characterised by three

89

developmental processes occurring at the fovea: i) centrifugal displacement of inner retinal

90

layers, ii) cone photoreceptor specialisation, and iii) centripetal migration of cone

91

photoreceptors.9 These stages are seen on OCT morphology as the following features of the

92

fovea, respectively: i) formation and deepening of foveal pit with outward displacement,

93

termed “extrusion”, of the plexiform layers, ii) outer segment (OS) lengthening and iii) outer

94

nuclear layer (ONL) widening.10 Failure of any of these processes results in foveal under-

95

development called “foveal hypoplasia”, causing reduced visual acuity (VA). Foveal

96

hypoplasia is often associated with infantile nystagmus.10 Infantile nystagmus is associated

97

with conditions including albinism and aniridia (mutations of the PAX6 gene), or may be

98

idiopathic. “Typical” foveal hypoplasia is characterised by an underdeveloped fovea wherein

99

the lining of individual retinal layers is intact.10 Early onset degeneration of the outer retina as

100

in achromatopsia also causes abnormal persisting inner retinal layers, but since the

101

underlying photoreceptors are disrupted this has been termed “atypical” foveal

102

hypoplasia.10,11

103 104

Structural grading of OCT morphology can enable classification of foveal hypoplasia based

105

on severity. Thomas et al10 proposed a structural grading scheme for foveal hypoplasia based

106

on OCT data from adults and older children, which has become a widely used scheme with

107

Grades 1 to 4 representing the most to least developed fovea, whilst the atypical grade

108

represents disruption of the photoreceptors seen in achromatopsia. Each of the grades

109

corresponded to statistically distinct VA levels. A key recommendation was validation of the

110

scheme using an independent dataset. More recently, Wilk et al12 identified two subsets of

111

Grade 1 hypoplasia and proposed to subdivide these into Grade 1a, in which nearly normal

112

pit metrics (depth, diameter, and/or volume within 2 S.D. of the normal average) are

113

observed, and Grade 1b, in which the pit is only a shallow indent (see Methods: Figure 1 for

114

structural grading scheme employed by the present study and Figure 2 for accompanying

115

grading algorithm).

116 117

An alternative method of evaluating foveal hypoplasia is quantitative segmentation. This

118

involves measuring the thickness of different foveal layers. Mohammad et al13 demonstrated

119

that total photoreceptor thickness and outer segment length are highly correlated with VA in

120

adults with albinism. The ratio of inner layer versus total foveal thickness, termed the foveal

121

developmental index (FDI), correlated with VA in albinism.14

122 123

The gold standard test for VA assessment in older children is logMAR chart VA testing,

124

beginning at four years of age.15 From ages 0-3 years, the gold standard test to assess VA is

125

preferential looking, based on the principle that young children tend to fixate to a pattern

126

stimulus in preference to a plain field.16 However, it can be difficult to assess young children

127

with nystagmus using preferential looking due to constant eye movements17, hence it would

128

be of great value to identify a better tolerated and easier office-based method for VA

129

prediction such as using HH-OCT. Visual evoked potentials (VEP) may represent another

130

option for estimating VA in infantile nystagmus18, but this is time consuming and difficult to

131

perform, requiring specialist diagnostic services.

132

133

In this longitudinal cohort study, we harness the potential of HH-OCT for the diagnosis and

134

prognosis of foveal hypoplasia in young children. The primary aim was to validate a

135

structural grading system for foveal hypoplasia in infants and young children with infantile

136

nystagmus to predict future vision. Secondary aims were: i) to assess whether structural

137

grading is comparable to quantitative segmentation in predicting future vision; ii) to compare

138

the predictive power of the grading system with preferential looking VA results taken on first

139

examination; iii) to determine whether the grade remains stable over time.

140 141

Materials and Methods

142

Study design and participants

143

This was a prospective longitudinal cohort study of patients with infantile nystagmus

144

recruited in a quaternary referral centre in Leicester, United Kingdom. Longitudinal data

145

were collected between 20th February 2012 and 30th January 2018. This study is reported

146

according to the Strengthening the Reporting of Observational Studies in Epidemiology

147

(STROBE) statement.

148 149

The study adhered to the tenets of the Declaration of Helsinki and ethics committee approval

150

was granted by the East Midlands - Nottingham 2 Research Ethics Committee. Consent was

151

obtained by all parents or guardians of study participants.

152 153

Inclusion criteria were: 1) Diagnosis of infantile nystagmus; 2) Age of 36 months or younger

154

at first HH-OCT scan; 3) Age of at least 42 months or older and ability to participate in chart

155

logMAR visual acuity test on follow up visit. Exclusion criteria included unsuccessful

156

acquisition of HH-OCT scan at first visit, loss to follow-up, scans of inadequate quality for

157

structural grading and co-existing neurological problems. If the HH-OCT scan was only

158

successful in one eye, then the other eye was excluded from analysis. If the HH-OCT scan

159

was unsuccessful in both eyes, then the participant was excluded from analysis. Examination

160

1 involved HH-OCT and full ophthalmic examination including refraction. Examination 2

161

involved full ophthalmic examination and best corrected VA measurement when the

162

participant was old enough to participate in chart logMAR vision testing. This age group was

163

deemed clinically important as it approaches or coincides with primary school education. In

164

addition, longitudinal HH-OCT scans were obtained.

165

166

Diagnosis of albinism was made based on abnormal lateralisation on visual evoked potentials

167

(VEP), iris transillumination defects, skin hypopigmentation and, if clinical uncertainty

168

existed, genetic testing.19,20 Diagnosis of achromatopsia was made based on typical features

169

including light sensitivity, small amplitude nystagmus, inner segment ellipsoid (ISe)

170

disruption identified on OCT, as well as confirmatory electrodiagnostic and genetic testing.10

171

IIN was diagnosed in children with nystagmus and normal ophthalmic examination, normal

172

VEPs and electroretinograms and/or confirmation on genetic testing (i.e. pathogenic FRMD7

173

mutation).

174

HH-OCT image acquisition, grading and segmentation

175 176

A noncontact, spectral domain HH-OCT scanner (ENVISU C class 2300; <4 µm axial

177

resolution; Leica Microsystems, Wetzlar, Germany) was used to image the foveae of all

178

participants in an outpatient setting without sedation. Visual fixation devices were employed

179

to minimise movement during imaging; these included toys, books and cartoons. The

180

acquisition protocol used a 10×10mm scanning window. The 3-dimensional raster scan

181

program for both scan sequences comprised 100 B-scans and 500 A-scans per B-scan line.

182

The acquisition time was short (1.9 seconds) to facilitate successful image acquisition with

183

minimal disruption of quality, thereby avoiding measurement bias.

184 185

The HH-OCT imaging protocol involved scanning the right eye followed by the left eye. The

186

en face view was used to identify the optic nerve as a landmark for identification of the fovea.

187

The region temporal to the optic nerve was navigated frame by frame until the most central

188

foveal B-scan was found. Where a foveal pit was present, the B-scan with the deepest pit was

189

selected for grading and segmentation. Where no pit was present, the B-scan with the highest

190

OS peak or, if absent, that with the highest ONL dome was selected. If no foveal landmarks

191

were present (termed fovea plana), a B-scan was selected within the central foveal area. We

192

repeated scans three times per eye, selecting the highest quality scan where the retinal layers

193

were most easily discernible for analysis. If unsuccessful, scans were repeated until an

194

adequate quality image was achieved or until the child ceased to participate. The total

195

examination time typically took up to 10 minutes per child. All OCT scans were graded by

196

two graders (SRR and MGT) to assess inter-grader reliability. All OCT scans from

197

Examination 1 were segmented for quantitative analysis. Segmentation was performed using

198

ImageJ software (National Institutes of Health, released 2011, Version 1.48, Bethesda,

199

Maryland, U.S.)

200 201

Structural grading was based on the scheme described by Thomas et al10 with Grade 1

202

subdivided into 1a and 1b as per Wilk et al.12 The grading scheme is displayed in Figure 1 -

203

we refer to this as the Leicester Grading System for Foveal Hypoplasia. The accompanying

204

grading algorithm is displayed in Figure 2. Parameters for segmentation are illustrated in

205

Figure 3.

206 207

Visual acuity measurement

208

During Examination 1, preferential looking VA testing using Keeler Acuity Cards (Keeler

209

Ltd, Windsor, UK) was performed where possible; cards were held vertically to more easily

210

determine responses in horizontal nystagmus, thereby avoiding measurement bias.17 During

211

Examination 2, the Keeler LogMAR Crowded Acuity Test, which uses crowded letter

212

optotypes on charts, was used as the gold standard VA test provided the child could

213

cooperate.21 For children who could not cooperate, the Crowded Kay LogMAR Picture test

214

was used.21 This test involves matching easily recognisable pictures (house, car, star, apple,

215

boot and duck) and is easier for children who cannot match letters.21 Where possible,

216

cycloplegic refraction was performed prior to Examinations 1 and 2. Mean spherical

217

equivalent for this cohort was +2.38 (range: -8.75 to +8.38; S.D. +3.22).

218 219

Statistical analysis

220

Our power calculation demonstrated that 38 patients were required to achieve power (α=0.05,

221

β=0.1, r=0.5), based upon the null hypothesis that the correlation is zero. Data was analysed

222

using linear mixed model regression analysis including grade of foveal hypoplasia and eye

223

recorded (right/left) in the model; the dependent variable was chart tested LogMAR VA

224

recorded at Examination 2. Independent variables recorded at Examination 1 included grade

225

of foveal hypoplasia, quantitative segmentation parameters (OS length, ONL length and FDI)

226

and preferential looking VA in LogMAR. Age in months was also included in the model for

227

preferential looking testing because the VA result is expected to improve with age.16 These

228

analyses were computed using SPSS Statistics (IBM Corp. released 2013, Version 22.0.

229

Armonk, NY: IBM Corp.) Agreement between two graders was reported to assess the

230

validity of the grading system. Agreement of grades over time was reported to explore

231

whether grades remain stable over time.22

232

233

Role of the funding source

234

The sponsor or funding organisations had no role in the design or conduct of this research.

235 236

Results

237

Baseline characteristics

238

Out of 47 patients enrolled to this study, data from 42 (89%) were included: two were lost to

239

follow up and three did not cooperate with HH-OCT scanning. During Examination 1, we

240

obtained OCT scans of 81 eyes from 42 patients of mean age 19.8 months (range: 0.9-33.4;

241

S.D. 9.4); 3 right eye tomograms of 3 patients were of inadequate quality for grading and thus

242

excluded. Adequate power was therefore achieved (n=38, α=0.05, β=0.1, r=0.5). Mean

243

follow-up to Examination 2 was 44.1 months (range: 18.4-63.2; S.D. 12.0). 16 patients were

244

female and 26 were male. In Examination 2, LogMAR VA using letter optotypes (Crowded

245

Keeler’s testing) was successfully obtained in 39 patients (93%) with all children being

246

optimally refracted; 3 patients could not read letter optotypes, hence were tested with Kay’s

247

LogMAR Picture Cards. Baseline characteristics of patients are listed in Table 1.

248 249

The cohort included children with albinism (n=19 patients), idiopathic infantile nystagmus

250

(IIN) (n=17) and achromatopsia (n=6).

251 252

Grading of foveal hypoplasia

253

Using the grading scheme alone, the inter-grader correlation coefficient was 0.96 when

254

subdividing Grade 1 into 1a and 1b, suggesting the infant grading system is robust. There was

255

disagreement for 3 eyes from 3 patients during the masked grading by the second grader.

256

Each case of disagreement was over what constituted a nearly normal pit as per Grade 1a

257

versus a shallow indent as per Grade 1b, however there was total agreement following

258

discussion and training, which involved studying the treatment algorithm and discussing

259

examples of 1a versus 1b pits. Agreement between Graders 1 and 2 prior to training is

260

displayed in the Supplemental Table.

261 262

Foveal hypoplasia was identified in 57 eyes according to our structural grading scheme,

263

which included Grade 1a (n=17), Grade 1b (n=10), Grade 2 (n=2), Grade 3 (n=10), Grade 4

264

(n=6) and atypical foveal hypoplasia (n=12). Normal tomograms were seen in 24 eyes.

265

266

Figure 4 displays samples of HH-OCT central B-scans per grade. Figure 5 displays a

267

breakdown of diagnoses per grade.

268 269

Prediction of future visual acuity

270

Figure 6 is a box and whisker plot displaying the grade of foveal hypoplasia during

271

Examination 1 versus VA during Examination 2. Median (M) and interquartile range (IQR)

272

per grade were as follows: no foveal hypoplasia: M=0.26, IQR=0.27; Grade 1a: M=0.41,

273

IQR=0.32; Grade 1b: M=0.65, IQR=0.16; Grade 2: M=0.60, IQR=0.00; Grade 3: M=0.74,

274

IQR=0.22; Grade 4: M=1.01, IQR=0.23; Atypical: M=0.93; IQR=0.28.

275 276

HH-OCT scans of gradable quality were successfully obtained in 81 out of 90 eyes (90.0%).

277

The Leicester Grading System for Foveal Hypoplasia demonstrated strong predictive power

278

for future VA (r=0.80, F=67.49, p<0.0001). This grading system correctly predicted future

279

VA within 0.3 logMAR in 78 out of 81 eyes (96.3%) and within 0.2 logMAR in 60 out of 81

280

eyes (74.1%).

281 282

Quantitative segmentation analysis

283

All quantitative segmentation parameters demonstrated statistically significant associations

284

with future VA in typical foveal hypoplasia, but none as strongly as structural grading (OS

285

length: r=0.65; F=7.94, p=0.008; PR length: r=0.65, F=6.90, p=0.01; FDI: r=0.74, F=28.81,

286

p<0.001). Scatter plots to show quantitative segmentation analyses for Examination 1 OCT

287

scans versus VA in Examination 2 are shown in eFigure 1.

288 289

Preferential looking

290

Preferential looking testing at Examination 1 was successfully assessed in 49 out of 72 eyes

291

(68.1%). Preferential looking was a poorer predictor of future VA (r=0.42; F=3.12; p=0.03)

292

compared to HH-OCT. Age did not have a statistically significant effect (p=0.09).

293

Preferential looking testing correctly predicted future VA within 0.3 logMAR in 19 out of 49

294

eyes (38.8%) and within 0.2 logMAR in 17 out of 49 eyes (34.7%).

295 296

Longitudinal grading

297

Longitudinal OCT scans were obtained in 80 eyes from 40 patients with mean follow-up 40.5

298

months (range: 10.4-62.3, S.D. 15.7). The longitudinal intra-grade correlation co-efficient

299

was

1.0,

as

all

grades

remained

the

same

over

time.

300 301 302

Discussion

303

To our knowledge, this is the first longitudinal study in infants and young children using HH-

304

OCT to provide a visual prognosis. In foveal hypoplasia the grading scheme10,12 has been

305

validated and demonstrates the strongest prediction of vision compared to quantitative

306

measures and preferential looking in this cohort. The inter-grader reliability testing

307

demonstrates that this scheme is robust. All grades remained stable throughout the study

308

period. In infants and young children, all typical grades achieved VA approximately 2 lines

309

worse than that predicted by grading in adults and older children10: in whom VA per grade

310

was as follows: Grade 1: median 0.20, IQR=0.12; Grade 2: median 0.44, IQR=0.18; Grade 3:

311

median 0.60, IQR=0.0; Grade 4: median=0.78, IQR=0.11; Atypical: median=1.0,

312

IQR=0.08.10 In infants and young children they were as follows: No foveal hypoplasia: 0.26;

313

Grade 1: 0.41; Grade 2: 0.60; Grade 3: 0.74; Grade 4: 1.01; Atypical: 0.93. As only one

314

patient with grade 2 foveal hypoplasia was recruited in this study, results for grade 2 cannot

315

be generalized. These differences could be due to visual acuity still undergoing development

316

and cooperation/ability for VA testing improving with increasing age.23 The slightly better

317

predicted VA in young children with atypical foveal hypoplasia in our study compared to the

318

study in adults and older children is likely because achromatopsia may be a progressive

319

disease.24

320 321

Grading versus quantitative segmentation

322

Our study revealed that grading was a stronger predictor of vision compared to quantitative

323

segmentation. An explanation could be that the grading system incorporates all key elements

324

of foveal development, while measurement of FDI, OS length and PR length only represent

325

individual developmental landmarks. However, quantitative segmentation may give insight

326

into subtle differences within a single grade, which may be particularly useful in

327

achromatopsia, represented solely in one atypical grade. Longitudinal quantitative

328

measurements may also help in the monitoring of patients undergoing emerging therapies.

329

For example, measurements of the ONL in achromatopsia, which changes with increasing

330

age24, are particularly important to determine the degree of cone degeneration in view of gene

331

therapy.

332 333

Grading can be performed rapidly in clinic whereas quantitative segmentation requires

334

special training and takes approximately 20 minutes per patient. Furthermore, grading can

335

help clinicians make appropriate decisions regarding management and investigation. For

336

example, if VA is poorer than expected according to grading, there should be raised suspicion

337

of other pathology limiting the VA. This may include other retinal pathology such as retinal

338

dystrophy, cortical/neurological disorders, amblyopia, conditions affecting the anterior

339

segment or sub-optimally corrected refractive errors. Similarly, if a patient has poor vision

340

consistent with a high degree of foveal hypoplasia, it is likely that the limiting factor for poor

341

VA is accounted at the retinal level.

342 343

Grading versus preferential looking

344

Preferential looking was a poor predictor of future VA, tending to underestimate visual

345

potential in nystagmus. Constant eye movement may increase difficulties of assessing

346

whether the child has correctly identified test cards.17 Suboptimal refraction at Examination 1

347

may have contributed to the poor correlation between preferential looking and chart VA.

348

Mean chart VA predicted by preferential looking in our cohort was 1.0 logMAR, better than a

349

previously published figure in another young cohort with infantile nystagmus of mean age 43

350

months (1.7 logMAR), but still poor.25 Moreover, the success rate in preferential looking was

351

poor (68.1%) compared to HH-OCT (90.0%). Dubowitz et al26 achieved a similar success

352

rate of 70% (n=96) for preferential looking in children with retinopathy of prematurity, also

353

associated with foveal hypoplasia, of a similar age to our study. In our study, preferential

354

looking testing correctly predicted future VA within 3 lines in only 38.8% compared to

355

96.3% for HH-OCT.

356 357

Can the fovea be graded from birth?

358

Our grading system involves identifying the presence or absence of intact ISe, foveal pit, OS

359

lengthening and ONL widening. Histological and OCT studies have demonstrated that all of

360

these landmarks can be identified from birth except OS elongation, which continues up to 5

361

months postnatally.6,27 Hence, it is possible to determine all grades apart from Grade 2 and

362

Grade 3 from birth, as an infant may be inappropriately assigned Grade 3 before OS

363

elongation has occurred. In this study, no patient under 5 months with Grade 2 or 3 foveal

364

hypoplasia was identified. However, we successfully obtained scans from patients aged 1

365

month and 3 months with Grade 4 and atypical foveal hypoplasia, respectively. Longitudinal

366

OCT scans revealed that all grades remained the same until the end of the study period, hence

367

adding age into the regression model would not impact the correlation between grading and

368

future VA.

369 370

Study strengths and limitations

371

To our knowledge, this is the first longitudinal cohort study using HH-OCT to predict future

372

vision in infantile nystagmus. This helps to reassure and/or enable patients and families to

373

optimise the development and educational attainment of the child during this crucial age, as

374

well as helping clinicians in their decision-making process for investigation and management.

375

A powered cohort of infants and young children has been achieved with adequate follow up

376

until the child could participate with chart VA, approaching or coinciding with

377

commencement of primary school education. Our grading system demonstrated an inter-

378

grader coefficient of 0.96, suggesting it is robust. Our grading system shows lower prediction

379

of VA than previously found in older patients.10 This is likely due to less cooperation with

380

logMAR VA testing and an immature visual system.23 Nystagmus characteristics and head

381

posture may influence VA, however, even without taking these factors into account, we

382

found very good correlation between foveal structure and later VA. In the future, a

383

multivariate model including foveal structure and nystagmus could be calculated to

384

investigate whether VA prediction could be refined.

385 386

A limitation of our study is that we are currently only able to predict VA in childhood rather

387

than adulthood due to the recent availability HH-OCT. However, it is still of great value to

388

predict an infant’s VA at the time they commence primary education. It is likely that VA will

389

continue to improve based on adult data.10 Another limitation is that only three aetiologies of

390

foveal hypoplasia were represented within this cohort; the grading system was not applied to

391

other conditions associated with foveal hypoplasia such as PAX-6 mutation28, retinopathy of

392

prematurity29, nanophthalmos30 and other photoreceptor dystrophies. There were only two

393

eyes from one patient that fulfilled the criteria for Grade 2, therefore, median VA predicted

394

from this single patient (0.60) cannot be generalized. With respect to image acquisition, if

395

three scans per eye were unsuccessful, patients could not be included. However, most eyes

396

(90%) were successfully scanned.

397 398 399

Author Contributions

400

SRR: Concept, literature search, figures, study design, data collection, data analysis, data

401

interpretation, writing.

402

MGT: Concept, literature search, figures, study design, data analysis, data interpretation,

403

writing, critical revision.

404

RP: Data collection, data interpretation, critical revision.

405

CB: Study design, data analysis, data interpretation, writing, critical revision.

406

HL: Literature search, data collection, data interpretation, writing, critical revision.

407

FAP: Figures, study design, data analysis, data interpretation, writing, critical revision.

408

IG: Concept, study design, data collection, data interpretation, writing, critical revision,

409

overall supervision.

410 411

Acknowledgements

412

The authors would like to thank vision scientists Dr Rebecca McLean and Dr Zhanhan Tu,

413

research orthoptists Helen Kuht and Michael Hisaund, and research optometrist Sonal Shah,

414

for their support in data collection for this study. The authors would like to thank all the

415

patients and families who took part in this study.

416 417

SRR’s post is funded by a National Institute for Health Research (NIHR) Academic Clinical

418

Fellowship. MGT’s and HL’s posts are funded by National Institute for Health Research

419

(NIHR) Academic Clinical Lectureships. CB’s post is part funded/supported by the National

420

Institute for Health Research (NIHR) Biomedical Research Centre based at Guy's and St

421

Thomas' NHS Foundation Trust and King's College London. The views expressed are those

422

of the authors and not necessarily those of the NHS, the NIHR or the Department of Health

423

Figure 3 is reprinted from Ophthalmology, 118(8), Thomas MG, Kumar A, Mohammad S,

424

Proudlock FA, Engle EC, Andrews C, Chan WM, Thomas S, Gottlob I. Structural grading of

425

foveal hypoplasia using spectral-domain optical coherence tomography a predictor of visual

426

acuity? 1653-1660, Copyright (2011), with permission from Elsevier.

427

428

References

429

1. Sarvananthan N, Surendran M, Roberts EO, Jain S, Thomas S, Shah N, Proudlock

430

FA, Thompson JR, McLean RJ, Degg C, Woodruff G, Gottlob I. The prevalence of

431

nystagmus: the Leicestershire nystagmus survey. Invest Ophthalmol Vis Sci. 2009

432

Nov;50(11):5201-6.

433

2. Nystagmus

Network.

Parents

booklet.

2019.

434

https://nystagmusnetwork.org/documents/parents-booklet/ Accessed 31 January 2019.

435

3. Lee H, Proudlock FA, Gottlob I. Pediatric Optical Coherence Tomography in Clinical

436

Practice—Recent Progress. Invest Ophthalmol Vis Sci. July 2016, Vol.57, OCT69-

437

OCT79.

438

4. Maldonado RS, Izatt JA, Sarin N, et al. Optimizing hand-held spectral domain optical

439

coherence tomography imaging for neonates, infants, and children. Invest Ophthalmol

440

Vis Sci. 2010; 51: 2678–2685.

441

5. Lee H, Proudlock F, Gottlob I. Is handheld optical coherence tomography reliable in

442

infants and young children with and without nystagmus? Invest Ophthalmol Vis Sci.

443

2013; 54: 8152–8159.

444

6. Lee H, Purohit R, Patel A, Papageorgiou E, Sheth V, Maconachie G, Pilat A, McLean

445

RJ, Proudlock FA, Gottlob I. In Vivo Foveal Development Using Optical Coherence

446

Tomography. Invest Ophthalmol Vis Sci. July 2015, Vol.56, 4537-4545.

447

7. Patel A, Purohit R, Lee H, Sheth V, Maconachie G, Papageorgiou E, McLean RJ,

448

Gottlob I, Proudlock FA. Optic Nerve Head Development in Healthy Infants and

449

Children Using Handheld Spectral-Domain Optical Coherence Tomography.

450

Ophthalmology. 2016 Oct; 123(10): 2147–2157.

451

8. Papageorgiou E, Pilat A, Proudlock F, Lee H, Purohit R, Sheth V, Vasudevan P,

452

Gottlob I. Retinal and optic nerve changes in microcephaly: An optical coherence

453

tomography study. Neurology. 2018 Aug 7;91(6):e571-e585.

454 455

9. Hendrickson AE, Yuodelis C. The morphological development of the human fovea. Ophthalmology. 1984;91:603–12.

456

10. Thomas MG, Kumar A, Mohammad S, Proudlock FA, Engle EC, Andrews C, Chan

457

WM, Thomas S, Gottlob I. Structural Grading of Foveal Hypoplasia. Ophthalmology.

458

2011;118(8):1653–1660.

459 460

11. Thomas MG, Kumar A, Kohl S, Proudlock FA, Gottlob I. High-resolution in vivo imaging in achromatopsia. Ophthalmology. 2011 May;118(5):882-7.

461

12. Wilk MA, McAllister JT, Cooper RF, Dubis AM, Patitucci TN, Summerfelt P,

462

Anderson JL, Stepien KE, Costakos DM, Connor TB, Wirostko WJ, Chiang PW,

463

Dubra A, Curcio CA, Brillian MH, Summers CG, Carroll J. Relationship Between

464

Foveal Cone Specialization and Pit Morphology in Albinism. Invest Ophthalmol Vis

465

Sci. July 2014, Vol.55, 4186-4198. : s.n.

466

13. Mohammad S Gottlob I Kumar A. The functional significance of foveal abnormalities

467

in albinism measured using spectral-domain optical coherence tomography.

468

Ophthalmology. 2011; 118: 1645–1652.

469 470 471 472 473 474

14. Ather S. Imaging the visual pathway in human albinism. PhD thesis. University of Leicester, June 2016. 15. Huang RZ. Design of visual acuity chart for preschool children. Anhui Medical Journal. 1983;4(3):184–185. 16. Huurneman B, Boonstra, FN. Assessment of near visual acuity in 0–13 year olds with normal and low vision: a systematic review. BMC Ophthalmol. 2016; 16: 215.

475

17. Getz LM, Dobson V, Luna B, Mash C. Interobserver Reliability of the Teller Acuity

476

Card Procedure in Pediatric Patients. Invest Ophthalmol Vis Sci. 1996;37:180-187.

477

18. Yasmin S. Bradfield; Thomas D. France; James Verhoeve; Ronald E. Gangnon.

478

Sweep Visual Evoked Potential Testing as a Predictor of Recognition Acuity in

479

Albinism. Arch Ophthalmol. 2007;125(5):628-633.

480

19. Clinical and oculomotor characteristics of albinism compared to FRMD7 associated

481

infantile nystagmus. Invest Ophthalmol Vis Sci. 2011 Apr 8;52(5):2306-13. Kumar A,

482

Gottlob I, McLean RJ, Thomas S, Thomas MG, Proudlock FA.

483

20. Thomas MG, Maconachie G, Sheth V, McLean RJ, Gottlob I. Development and

484

clinical utility of a novel diagnostic nystagmus gene panel using targeted next-

485

generation sequencing. Eur J Hum Genet. 2017 Jun;25(6):725-734.

486

21. O'Boyle, Chen SI, Little J. Crowded letter and crowded picture logMAR acuity in

487

children with amblyopia: a quantitative comparison. Br J Ophthalmol. 2017 Apr;

488

101(4): 457–461.

489 490

22. Cohen, J. (1960) "A coefficient for agreement for nominal scales" in Education and Psychological Measurement. Vol. 20, pp. 37–46.

491

23. Fu VLN, Bilonick RA, Felius J, Hertle RW, Birch EE. Visual Acuity Development of

492

Children with Infantile Nystagmus Syndrome. Invest Ophthalmol Vis Sci. 2011 Mar;

493

52(3): 1404–1411.

494

24. Thomas MG, McLean RJ, Kohl S, Sheth V, Gottlob I. Early signs of longitudinal

495

progressive cone photoreceptor degeneration in achromatopsia. Br J Ophthalmol.

496

2012 Sep;96(9):1232-6.

497

25. Katsumi O, Chedid SG, Kronheim JK, Henry RK, Denno S, Hirose T. Correlating

498

preferential looking visual acuity and visual behavior in severely visually

499

handicapped children. Acta Ophthalmol Scand. 1995 Oct;73(5):407-13.

500

26. Dubowitz LMS, Mishin J, Morante A, Placzek M. The maturation of visual acuity in

501

neurologically normal and abnormal newborn infants. Behav Brain Res. 1983; 10: 39-

502

45.

503

27. Gupta T, Kapoor K, Sahni D, Singh B. Mapping the Time Line of Development in

504

Each Layer of Human Foetal Retina. J Clin Diagn Res. 2016 Mar; 10(3): AC04–

505

AC07.

506

28. Hingorani M1, Williamson KA, Moore AT, van Heyningen V. Detailed

507

ophthalmologic evaluation of 43 individuals with PAX6 mutations. Invest Ophthalmol

508

Vis Sci. 2009 Jun;50(6):2581-90.

509

29. Hammer DX, Iftimia NV, Ferguson RD, et al. Foveal fine structure in retinopathy of

510

prematurity: an adaptive optics Fourier domain optical coherence tomography study.

511

Invest Ophthalmol Vis Sci. 2008;49:2061–70.

512 513

30. Bijlsma WR, van Schooneveld MJ, Van der Lelij A. Optical coherence tomography findings for nanophthalmic eyes. Retina. 2008;28:1002–7.

514 515

Figure Legends

516 517 518 519 520 521 522 523 524 525 526 527 528 529

Figure 1: The Leicester Grading System for Foveal Hypoplasia. Schematic demonstrating the features of a normal fovea detectable using optical coherence tomography (OCT), followed by the features of typical and atypical grades of foveal hypoplasia. The normal fovea features outward displacement, termed “extrusion”, of the plexiform layers. All grades of foveal hypoplasia feature continuation of the plexiform layers. Grade 1a foveal hypoplasia is associated with a nearly normal pit resembling a 'v' shape, OS lengthening and ONL widening relative to the parafoveal OS and ONL lengths, respectively. Grade 1b foveal hypoplasia is associated with a shallow indent, OS lengthening and ONL widening relative to the parafoveal OS and ONL lengths, respectively In Grade 2 foveal hypoplasia, all features of Grade 1 are present except there is no pit. Grade 3 foveal hypoplasia represents all features of Grade 2 except there is no lengthening of the OS segment. Grade 4 foveal hypoplasia represents grade 3 except there is no ONL widening at the fovea (termed fovea plana). Atypical foveal hypoplasia is characterised by a shallow foveal pit and disruption of the inner segment ellipsoid (ISe) band forming a hyporeflective zone. ILM = internal limiting

530 531 532 533 534 535

membrane; RNFL = retinal nerve fibre layer; GCL = ganglion cell layer; IPL = inner plexiform layer; INL = inner nuclear layer; OPL = outer plexiform layer; ONL = outer nuclear layer; ELM = external limiting membrane; ISe = inner segment ellipsoid; OS = outer segment; RPE = retinal pigment epithelium. Reprinted and adapted from Ophthalmology, 118(8), Thomas MG et al, 1653-1660, Copyright (2011), with permission from Elsevier.

536 537

Figure 2: Algorithm for Leicester Grading Scheme for Foveal Hypoplasia. Key: ISe = inner segment ellipsoid; OS = outer segment; ONL = outer nuclear layer.

538 539 540 541 542 543

Figure 3: Parameters for quantitative segmentation analysis. 1A: central foveal B-scan; 1B: En face fundal view. Three metrics were included in the segmentation analysis: (1) outer segment (OS) length: distance between Point I (cone outer segment tips: COST) to Point II (OS); (2) photoreceptor (PR) length: distance between Point I to Point III (outer nuclear layer: ONL) (3) Foveal Developmental Index (FDI): ratio of distance between Point I to Point III divided by distance between point I to point IV (foveal pit).

544

Figure 4: Examples of OCT scans for each grade of foveal hypoplasia.

545

Figure 5: Breakdown of grades by condition

546 547 548

Figure 6: Box and whisker plot to show typical grade of foveal hypoplasia in Examination 1 versus VA in Examination 2. VA = visual acuity, FH = foveal hypoplasia, M = median; IQR = interquartile range.

Variable n Mean (range; S.D.) Number of eyes (patients) 81 (42) Albinism 38 (19) IIN 31 (17) Achromatopsia 12 (6) Males 26 Females 16 Age at Examination 1 in months 19.8 (0.9-33.4; 9.4) Age at Examination 2 in months 63.9 (44.7-88.8; 11.6) Follow-up in months 44.1 (18.4-63.2; 12.0) Table 1: Baseline characteristics of cohort. Key: S.D. = standard deviation; IIN = idiopathic infantile nystagmus.

Handheld optical coherence tomography can predict future vision in infantile nystagmus. The Leicester Grading System for Foveal Hypoplasia provides better prediction of future vision than preferential looking – the current gold standard.