Visual Evoked Potential Detection of Amblyopia in the Clinic Sina Salehi Omran, Casey M. Howards, Sean P. Donahue

Amblyopia is thought to affect around 2% of all children1. Preverbal children with risk factors for amblyopia can be challenging to manage, as they may have a fixation preference but not amblyopia. By measuring visual evoked potentials (VEPs) in response to predictable patterns, we may have an objective tool for diagnosing and managing amblyopia in these preverbal children.

Abstract

Objective

Introduction: Preverbal children with amblyopia risk factors can be challenging to manage, as they may have a fixation preference but may or may not have amblyopia. We developed a user- and patient-friendly test using visual evoked potentials to aide in management of such children.

Methods: We tested 33 normal children and 43 amblyopic children (age 3-12 years) using the Diopsys NOVA-TR system. A checkerboard pattern reversal visual evoked response was produced at 2 Hz for each of 5 spatial frequencies. Stimuli were presented for 10 seconds (20 reversals) and the entire test typically took less than five minutes. The relative diminution of P100 amplitude between the eyes, and the absolute prolongation of P100 latency were compared for each spatial frequency.

Results: A 9 millisecond difference in P100 latency for at least two spatial frequencies, or a 25% diminution in P100 amplitude between the eyes for at least three spatial frequencies correctly identified 94% of normal subjects as being normal, and 81% of amblyopic subjects as being amblyopic, including 72% of subjects with mild amblyopia (2-3 lines of inter-ocular difference), and 88% of subjects with moderate or severe amblyopia (>3 lines difference).

Discussion: A pattern reversal visual evoked response at five spatial frequencies can correctly identify children with amblyopia, without misclassifying too many normal individuals. Thus, we have an objective test for detecting amblyopia in preverbal children.

Conclusion: The development of a clinically useful evoked potential test will help pediatric ophthalmologists manage amblyopia in preverbal children.

Subjects Age: 3-12 years old 33 normal subjects

•  20/40 vision or better and less than 2 lines of inter-ocular difference (IOD)

43 amblyopic subjects •  Amblyopia: 2 lines or more of IOD •  18 have mild amblyopia (2-3 lines IOD) •  25 have moderate/severe amblyopia (>3 lines IOD)

5 spatial frequencies at 10 seconds each (20 reversals) Test took ~5 minutes, setup time was longer.

Data Analysis

OD 32x32 OD 64x64 OD 128x128 OD 256x256 OD Combined

OS 32x32 OS 64x64 OS 128x128 OS 256x256 OS Combined

A pattern reversal VEP at 5 spatial frequencies can be used to correctly identify children with amblyopia, without misclassifying many normal individuals. This is important since fixation preference testing can be unreliable in detecting amblyopia in preverbal children. The test takes less than five minutes to complete using commercially available software. Automation of the test is under development.

The development of a clinically useful evoked potential test will help pediatric ophthalmologists manage amblyopia in preverbal children.

Conclusion

Thank you to Niraj Nathan and Diana Tamboli. This poster is supported by the Emphasis program at Vanderbilt University School of Medicine. Many thanks to various individuals at the Vanderbilt Eye Institute for help in recruiting patients. The support of the many patients and parents who consented for this research was critical for its successful completion.

Acknowledgements

Dr. Donahue is a consultant for Diopsys Inc. and has also served as a consultant to several other companies.

Conflicts

Discussion

Rel

ativ

e D

imin

utio

n of

Am

plitu

de (%

) •  (

Larg

est –

Sm

alle

st) /

Lar

gest

in n

orm

als

•  (Fe

llow

– A

mb)

/ Fe

llow

in a

mbl

yopi

cs

Normal Subjects Amblyopic Subjects

16x16 32x32 64x64 128x128 256x256 16x16 32x32 64x64 128x128 256x256 Spatial Frequency Spatial Frequency

Del

ta A

mpl

itude

The following VEP amplitude vs. time graphs are all from one amblyopic subject tested across 5 spatial frequencies. The fellow (OD, HOTV 20/30) eye appears on the top row, while the amblyopic (OS, HOTV 20/150) eye appears on the bottom row. The final graph is a combination of all 5 spatial frequencies.

In the graphs below, each dot in each spatial frequency represents one subject. The top two graphs show the relative diminution of P100 amplitude, while the bottom two graphs show absolute prolongation of P100 latency, between the eyes at each spatial frequency. The red lines represent the threshold values.

Del

ta A

mpl

itude

25% 25%

P10

0 La

tenc

y (m

s)

9

P10

0 La

tenc

y (m

s)

9

Abs

. Pro

long

atio

n of

P10

0 La

tenc

y (m

s)

•  Abs

olut

e O

D –

OS

P10

0 La

tenc

y in

nor

mal

s •  A

bsol

ute

Am

b –

Fel P

100

Late

ncy

in a

mbl

yopi

cs

We can use the following two criteria to detect amblyopia:

With this protocol: •  93.9% of normal subjects do not meet these criteria, i.e. are correctly identified as normal •  81.4% of amblyopic subjects meet these criteria (13/18 of the subjects with mild amblyopia, 22/25 of subjects with moderate to severe amblyopia)

Furthermore, by setting a higher threshold for the diminution in amplitude, subjects with moderate/severe amblyopia can be separated from those with mild amblyopia. For example, by adding the condition of “60% or greater diminution in N75-P100 amplitude in at least 2 spatial frequencies”, we have:

•  3/18 (16.7%) of mild amblyopes are misclassified •  20/25 (80%) of moderate/severe amblyopes are correctly identified.

9 ms or greater prolongation of P100 latency in at least 2 spatial frequencies, same direction

OR 25% or greater diminution in N75-P100

amplitude in at least 3 spatial frequencies

1 Chou R, Dana T, Bougatsos C. Screening for visual impairment in children ages 1-5 years: update for the USPSTF. Pediatrics 2011;127(2):e442-e479.

References