Multiplehttp://msj.sagepub.

com/
Sclerosis Journal

Visual dysfunction in multiple sclerosis correlates better with optical coherence tomography derived
estimates of macular ganglion cell layer thickness than peripapillary retinal nerve fiber layer thickness
Shiv Saidha, Stephanie B. Syc, Mary K. Durbin, Christopher Eckstein, Jonathan D. Oakley, Scott A. Meyer, Amy Conger,
Teresa C. Frohman, Scott Newsome, John N. Ratchford, Elliot M. Frohman and Peter A. Calabresi
Mult Scler 2011 17: 1449 originally published online 24 August 2011
DOI: 10.1177/1352458511418630

The online version of this article can be found at:

http://msj.sagepub.com/content/17/12/1449

Published by:

http://www.sagepublications.com

On behalf of:
European Committee for Treatment and Research in Multiple Sclerosis

Americas Committee for Treatment and Research in Multiple Sclerosis

Pan-Asian Committee for Treatment and Research in Multiple Sclerosis

Latin American Committee on Treatment and Research of Multiple Sclerosis

Additional services and information for Multiple Sclerosis Journal can be found at:

Email Alerts: http://msj.sagepub.com/cgi/alerts

Subscriptions: http://msj.sagepub.com/subscriptions

Reprints: http://www.sagepub.com/journalsReprints.nav
Downloaded from msj.sagepub.com at Nat. Taichung Univ. of Sci. & Tech. on April 22, 2014

Permissions: http://www.sagepub.com/journalsPermissions.nav
 >> Version of Record - Nov 30, 2011

OnlineFirst Version of Record - Aug 24, 2011

What is This?

Downloaded from msj.sagepub.com at Nat. Taichung Univ. of Sci. & Tech. on April 22, 2014
Research Paper
Multiple Sclerosis Journal
17(12) 1449–1463
Visual dysfunction in multiple sclerosis Ó The Author(s) 2011
Reprints and permissions:
correlates better with optical coherence sagepub.co.uk/journalsPermissions.nav
DOI: 10.1177/1352458511418630

tomography derived estimates of macular msj.sagepub.com

ganglion cell layer thickness than

peripapillary retinal nerve fiber layer
thickness

Shiv Saidha1, Stephanie B. Syc1, Mary K. Durbin2,

Christopher Eckstein1, Jonathan D. Oakley2, Scott A. Meyer2,
Amy Conger3, Teresa C. Frohman3, Scott Newsome1, John
N. Ratchford1, Elliot M. Frohman3 and Peter A. Calabresi1

Abstract
Background: Post-mortem analyses of multiple sclerosis (MS) eyes demonstrate prominent retinal neuronal ganglion
cell layer (GCL) loss, in addition to related axonal retinal nerve fiber layer (RNFL) loss. Despite this, clinical correlations
of retinal neuronal layers remain largely unexplored in MS.
Objectives: To determine if MS patients exhibit in vivo retinal neuronal GCL loss, deeper retinal neuronal loss, and
investigate correlations between retinal layer thicknesses, MS clinical subtype and validated clinical measures.
Methods: Cirrus HD-optical coherence tomography (OCT), utilizing automated intra-retinal layer segmentation, was
performed in 132 MS patients and 78 healthy controls. MS classification, Expanded Disability Status Scale (EDSS) and
visual function were recorded in study subjects.
Results: GCLþinner plexiform layer (GCIP) was thinner in relapsing–remitting MS (RRMS; n ¼ 96, 71.6 mm), secondary
progressive MS (SPMS; n ¼ 20, 66.4 mm) and primary progressive MS (PPMS; n ¼ 16, 74.1 mm) than in healthy controls
(81.8 mm; p < 0.001 for all). GCIP thickness was most decreased in SPMS, and although GCIP thickness correlated signif-
icantly with disease duration, after adjusting for this, GCIP thickness remained significantly lower in SPMS than RRMS. GCIP
thickness correlated significantly, and better than RNFL thickness, with EDSS, high-contrast, 2.5% low-contrast and 1.25%
low-contrast letter acuity in MS. 13.6% of patients also demonstrated inner or outer nuclear layer thinning.
Conclusions: OCT segmentation demonstrates in vivo GCIP thinning in all MS subtypes. GCIP thickness demonstrates
better structure-function correlations (with vision and disability) in MS than RNFL thickness. In addition to commonly
observed RNFL/GCIP thinning, retinal inner and outer nuclear layer thinning occur in MS.

Keywords
EDSS, ganglion cell layer, multiple sclerosis, optical coherence tomography, retinal pathology, retinal segmentation, visual function

Date received: 14th May 2011; revised: 27th June 2011; accepted: 10th July 2011

3
Department of Neurology and Ophthalmology, University of Texas
Southwestern, USA.
Corresponding author:
Shiv Saidha, 600N Wolfe Street, Pathology 627, Baltimore, MD 21287,
1
Department of Neurology, Johns Hopkins University School of USA. Email: ssaidha2@jhmi.edu; or Peter A. Calabresi, 600N Wolfe
Medicine, Baltimore, USA. Street, Pathology 627, Baltimore, MD 21287, USA.
2
Carl Zeiss Meditec Inc, Dublin, USA. Email: calabresi@jhmi.edu

Downloaded from msj.sagepub.com at Nat. Taichung Univ. of Sci. & Tech. on April 22, 2014
1450
Introduction
Optical coherence tomography (OCT) is an imaging
technique enabling evaluation of retinal structures
with high resolution.1–3 Research using OCT measure-
ments in multiple sclerosis (MS) has primarily focused
on evaluation of the peri-papillary retinal nerve fiber
layer (RNFL), consisting of unmyelinated axons that
originate from ganglion cell neurons (Figure 1A).
RNFL thickness provides estimation of retinal axonal
integrity and correlates with visual function in MS.4–6
Post-mortem analyses demonstrate optic nerve lesions
in 94–99% of MS subjects, irrespective of acute optic
neuritis (AON) history.7,8 Optic nerve demyelination
(due to clinical AON or subclinical optic neuropathy)
is thought to cause either axonal thinning resulting from
loss of neurofilament phosphorylation, or axonal tran-
section with subsequent retrograde degeneration of
constituent axons,9 reflected by RNFL thinning in
OCT results.10 Retinal nerve fiber degeneration is, in
turn, thought to culminate in ganglion cell death.11
While reductions in average macular thickness (AMT)
measured using OCT have been interpreted to reflect
retinal neuronal loss,12 isolated RNFL loss may poten-
tially lead to AMT reduction.13 As such, systematic
quantitative in vivo assessment of the ganglion cell
layer (GCL) remains largely unexplored in MS.
Additional observations highlight the likelihood that
deeper retinal changes may also occur in MS. While
qualitative atrophy of the RNFL and GCL have been
pathologically demonstrated in over 70% of MS
eyes,14,15 a recent ocular post-mortem assessment in
MS identified additional extensive qualitative atrophy
of the inner nuclear layer (INL) in over 40% of eyes,15
as was previously suggested may occur in MS on the
basis of electroretinography findings in advanced MS.16
Outer nuclear layer (ONL) pathology has also been
proposed to occur in MS on the basis of electroretinog-
raphy abnormalities.17–19 While this has not been
illustrated pathologically, quantitative pathologic ultra-
structural evaluations of retinal architecture (including
the RNFL) are limited20 due to challenges posed by
retinal fixation (which alters retinal layer thicknesses),
retinal detachment, retinal layer dehiscence and retinal
protein degradation.21,22 OCT may discern retinal
layers based on differences in tissue reflectivity (deter-
mined by differences in tissue composition)3,23,24 allow-
ing for quantitative assessment of discrete retinal layers
(OCT segmentation).25–29
A recent study utilizing OCT segmentation identified
an MS subset with objective INL or ONL thinning,
with corroborative multifocal electroretinography
abnormalities, and relative preservation of the inner-
most retinal layers (RNFL and GCL), implicating pri-
mary retinal neuronal layer pathology unrelated to
Downloaded from msj.sagepub.com at Nat. Taichung Univ. of Sci. & Tech. on April 22, 2014
Multiple Sclerosis Journal 17(12)
optic nerve pathology (MTP: macular thinning pre-
dominant MS phenotype).30 These findings support
structural (and functional) disruption in the deeper
retina in MS. Extensive retinal evaluation did not
reveal any other retinal disorder and INL/ONL thin-
ning was not observed as a consequence of prior AON.
Previous animal31–33 and electrophysiologic studies34–36
likewise failed to demonstrate retinal atrophy or dys-
function deeper than the GCL following optic nerve
transection, highlighting the possibility that deeper
INL/ONL atrophy may not be the derivative of optic
nerve mediated retrograde transynaptic degeneration.
In this cross-sectional study we used spectral-domain
OCT utilizing retinal segmentation to assess if MS
patients exhibit quantitative in vivo retinal neuronal
loss, in particular ganglion cell loss, and to assess cor-
relations between GCL thickness (and other retinal
neuronal layer thicknesses) with MS subtype and clin-
ical measures. We also sought to determine if MS
patients may exhibit OCT features characteristic of
optic nerve mediated retinal changes (RNFL/GCL
thinning), with concomitant OCT features of MS asso-
ciated primary retinal pathology as recently described
(INL/ONL thinning),30 possibly indicating that retinal
changes in MS may result from mixed retinal pathology
with more than a single site of origin.
Methods
Patients
MS subjects were recruited from the Johns Hopkins
Multiple Sclerosis Center, and represented an unse-
lected convenience sample of patients willing to
undergo evaluation for research purposes. Patients
with MS had their diagnosis confirmed by the treating
neurologist (P.A.C.), based on McDonald criteria.37
Patients with ophthalmologic or neurologic disorders
(in addition to MS) including glaucoma, diabetes and
hypertension, and/or a refractive error of greater than
66 diopters were excluded from the study. MS patients
fulfilling MTP criteria (10% of our historical cohort)
(OCT combination of AMT <5th percentile, with ipsi-
lateral normal average RNFL thicknesses between the
5th and 95th percentiles, in the absence of AON his-
tory)30 were excluded to allow accurate assessment of
MS patients with both deeper macular pathology, as
well as the more typical RNFL/GCL pathology
expected in MS, hereafter referred to as mixed retinal
pathology. MS patients with mixed retinal pathology
(Figure 2) were defined by the presence of RNFL thin-
ning <5th percentile, with concomitant INL or ONL
thinning <5th percentile, as compared to healthy con-
trol participants in this study. Scans acquired within 3
months of AON were excluded, to minimize effects of
Saidha et al. 1451

Figure 1. Panel A illustrates the layers of the retina. Note that the direction of signal transduction is directed from the photore-
ceptors (rods and cones) at the outer aspect of the retina towards the retinal nerve fiber layer (RNFL) at the inner aspect of the
retina. The fibers of the RNFL exit the retina at the optic disc, where they form the optic nerve. The nerve fibers of the RNFL
originate from the ganglion cells in the ganglion cell layer. Optic nerve demyelination results in RNFL degeneration, which in turn leads
to ganglion cell body death. Panel B is a three-dimensional volume cube generated by Cirrus HD-OCT. Note that the individual retinal
layers are readily discernible (due to differences in tissue reflectivity within different retinal layers), except for the ganglion cell layer
(GCL) and inner plexiform layer (IPL) which are virtually indistinguishable. During the segmentation process (performed in three
dimensions) the segmentation software identifies the outer boundaries of the retinal nerve fiber layer (RNFL), inner plexiform layer
(IPL) and outer plexiform layer (OPL), in addition to the inner boundary of the retinal pigment epithelium (RPE) which is identified by
the conventional Cirrus HD-OCT algorithm. The identification of these boundaries facilitates OCT segmentation, enabling the
determination of the combined thicknesses of the GCL and IPL (GCIP), the inner nuclear layer (INL) and OPL, and the outer nuclear
layer (ONL) including the inner and outer photoreceptor segments. Average segmentation thicknesses are measured in an annulus of
inner radius 0.54 mm and outer radius 2.4 mm, centered on the fovea. IS: inner photoreceptor segments, OS: outer photoreceptor
segments, IS/OS: IS/OS junction, ILM: inner limiting membrane, ELM: external limiting membrane.

Downloaded from msj.sagepub.com at Nat. Taichung Univ. of Sci. & Tech. on April 22, 2014
1452 Multiple Sclerosis Journal 17(12)

Figure 2. An optical coherence tomography (OCT) retinal nerve fiber layer (RNFL; panel A) and macular (panel B) report generated
by Cirrus HD-OCT with software 5.0 (Carl Zeiss Meditec, Dublin, California) in a relapsing–remitting multiple sclerosis (MS) patient
without prior history of optic neuritis. Note that the average RNFL thicknesses in the right (OD) and left (OS) eyes (as well as multiple
quadrants and clock-hours) are represented in red indicating reduced RNFL thicknesses bilaterally (red denotes values below 1% of
what would be expected when compared to an age-matched reference population). In the absence of optic neuritis history, the
etiology of such detected changes in MS is thought to relate to subclinical optic neuropathy. The top right section of panel B displays
macular quadrant measurements of retinal thickness in the right eye. Note the thickness reductions in the inner and outer macula
inferred as being indicative of neuronal loss (mostly thought to reflect ganglion cell body loss), while the corresponding reduction in
RNFL thickness in the same eye of this patient (panel A) is thought to represent loss of ganglion cell axons. The bottom right chart
indicates the average macular thickness (cube average thickness) and is also represented in red indicating a reduction in average
macular thickness in that eye. The central area of the macula (top right section of panel B) denoting the parafovea, as well as the
central subfield thickness (bottom right chart in panel B) which also denotes parafoveal thickness are represented in yellow (yellow
denotes values between 1% and 5% of what would be expected compared to an age-matched reference population), indicating
reduction of parafoveal thickness in that eye. Note that the parafovea is essentially devoid of RNFL implicating the presence of retinal
thinning below the RNFL. The macular scan of the left eye in the same patient (not shown) is similar to that of the right eye in panel B.
While the OCT findings described in this figure is from a patient with potential mixed retinal pathology, the identification of such
patients in this study was based upon the definition of RNFL thinning <5th percentile, with concomitant inner or outer nuclear layer
thinning <5th percentile, requiring the application of OCT segmentation to derive quantitative measures not provided by conventional
OCT, as exemplified above.

optic disk edema on OCT measurements. Healthy con-
trols without known ocular or neurologic disease were
recruited from among Johns Hopkins and the
University of Texas–Southwestern Medical Center
staff and unaffected family members of patients. Two
healthy control cohorts were recruited to thoroughly
investigate the contribution of age to our study; one
age-matched for the relapsing–remitting MS (RRMS)
cohort and one age-matched for the progressive MS
cohorts. The Johns Hopkins University and
University of Texas–Southwestern Medical Center
Institutional Review Board approval was obtained for
all study protocols. Recruited participants provided
written informed consent. The study was performed
in accordance with Health Insurance Portability and
Accountability Act guidelines.
Clinical data collected
Patient demographics were recorded on study subjects.
MS classification was recorded based on recognized MS
subtypes – RRMS, secondary progressive MS (SPMS),
primary progressive MS (PPMS) and relapsing progres-
sive MS.38 Kurtzke’s Expanded Disability Status Scale

Downloaded from msj.sagepub.com at Nat. Taichung Univ. of Sci. & Tech. on April 22, 2014
Saidha et al.
(EDSS)39 score was determined by a Neurostatus cer-
tified EDSS examiner the same day as OCT examina-
tion. Patient history and medical records were reviewed
to determine MS disease duration, co-morbidities and
history of AON, including their date and side.
Optical coherence tomography
Retinal imaging was performed using Cirrus HD-OCT
(model 4000) with software version 5.0 (Carl Zeiss
Meditec, Dublin, California, USA) as described in
detail elsewhere.30 Briefly, peri-papillary data were
obtained with the Optic Disc Cube 200 3 200 protocol.
Segmentation software determines the location of the
inner limiting membrane (ILM) and the outer bound-
ary (OB) of the RNFL at each A-scan to create a two-
dimensional map of peri-papillary RNFL thickness.
Macular data were obtained using the Macular Cube
512 3 128 protocol (128 horizontal scan lines/B-scans,
each composed of 512 A-scans, and one central vertical
and horizontal scan composed of 1024 A-scans) which
forms a 6 3 6 3 2 mm volume cube. Different segmen-
tation software identifies the ILM and the inner bound-
ary of the retinal pigment epithelium (RPE) in this cube
allowing determination of AMT and average thickness
over the nine macular subfields defined by the Early
Treatment Diabetic Retinopathy Study (ETDRS).40
OCT scanning was performed by three trained techni-
cians, who monitored scans to ensure fixation was reli-
able, as previously described.41 Scans with signal
strength less than 7/10 were excluded from analysis.
Macular cube scans were further analyzed in a
blinded fashion using segmentation software previously
utilized by our group.30 Segmentation was performed in
three dimensions (Figure 1B), utilizing anatomical
information from neighboring A-scans (within exam-
ined and neighboring B-scans) facilitating identification
of anatomical boundaries of interest in each A-scan.
The region of interest (ROI) for the segmentation algo-
rithm includes retinal tissue between the ILM and RPE
(both identified by the regular Cirrus HD-OCT algo-
rithm), and excludes the choroid and vitreous. The
signal within the ROI contains speckle noise analogous
to that of ultrasonagraphy images, since OCT images
are generated by coherent detection. Median filtering
was used as a computationally efficient means of sup-
pressing speckle whilst preserving the edges between
layers.
Images were then filtered to enhance the edges
between layers. Tissue boundaries were characterized
by changes in bulk scattering within tissues, with gra-
dients generally in the axial direction. Post-filtering, the
resulting edge images were sigmoid transformed.
Following this, prior anatomical information was com-
bined with the enhanced images creating cost functions
Downloaded from msj.sagepub.com at Nat. Taichung Univ. of Sci. & Tech. on April 22, 2014
1453
for each boundary, facilitating identification of the
boundary’s of interest: RNFL-OB, inner plexiform
layer (IPL)-OB and outer plexiform layer (OPL)-OB.
Combining this information allowed the algorithm to
simultaneously encode (i) precise edge locations using
local information and (ii) boundary identification infor-
mation using global information related to layer conti-
nuity, fovea location, and relative layer thicknesses and
ordering. A minimum cost graph traversal algorithm
was used, whereby the cost images combined the edge
image and positional cost images based on boundaries
already identified. Each new boundary identified during
the sequential identification of the RNFL-OB, IPL-OB
and OPL-OB provided positional cost information to
the subsequent boundary/boundaries of interest. Using
each boundaries cost functions, the algorithm traversed
the three-dimensional macular cube identifying the
boundary locations that minimized the cumulative
cost for each boundary.
The calculation of metrics followed directly from the
macular layer boundaries identified, yielding combined
thicknesses of the GCL plus IPL (GCIP), the INL plus
OPL, and the ONL (including inner and outer photo-
receptor segments). Average segmentation thicknesses
were measured in an annulus of inner radius 0.54 mm
and outer radius 2.4 mm centered on the fovea, a region
where retinal layers appear thickest on average.
Thickness maps were calculated from the axial distance
between the layer boundaries at each A-scan location
(Figure 3). This segmentation protocol has been shown
to be reproducible in MS patients and healthy controls
(interater ICC: intra-class correlations; 0.91–0.99 for all
OCT segmentation measurements described).30
Visual function testing
Standardized visual function testing was performed
with retro-illuminated eye charts of constant light
source in a darkened room, prior to OCT examination.
High-contrast ETDRS charts (at 4m) and low-contrast
Sloan letter charts (2.5% and 1.25% contrast; at 2 m)
were used. Testing was performed monocularly, with
subjects using their habitual distance spectacles or con-
tact lenses as needed for corrected vision. Charts were
scored letter-by-letter and the number of letters cor-
rectly read was recorded.
Statistical analyses
Statistical analysis was completed on STATA Version
11 (StataCorp, College Station, Texas, USA). Analysis
was completed on right eyes of participants to avoid
bias due to inter-eye correlation. Multivariate linear
regression was used to compare conventional and seg-
mentation OCT measures between the MS and healthy
1454 Multiple Sclerosis Journal 17(12)

Figure 3. Panels A and B illustrate normal GCIP (ganglion cell layer plus inner plexiform layer) thickness maps from the left (OS) and
right (OD) eyes respectively from a healthy control subject. Panels C and D illustrate GCIP thickness maps reflecting GCIP thinning in
both the left and right eyes respectively from a patient. Note the severe extent of GCIP thinning illustrated in panel D (and to a lesser
extent in panel C), by comparing to the normal GCIP thickness maps above (panels A and B).

control cohorts, as well as between the MS subtypes.

Analyses were adjusted for age and sex. In addition,
comparisons between the MS subtypes were also
adjusted for disease duration. Variance ratio testing
was used to compare variance between RNFL and
GCIP thickness measures. Spearman rank correlation
was used to determine correlations between RNFL
thickness, AMT and retinal layer thicknesses (derived
by OCT segmentation), as well as between OCT and
clinical measures of interest. Multivariate linear regres-
sion including age and sex in the model was also used to
adjust for RNFL and GCIP thicknesses when assessing
OCT correlations with EDSS and visual function.
Statistical significance was defined as p < 0.05.
Correction for multiple comparisons was not per-
formed as the variables examined were related and
thus likely to be correlated.
Results
One hundred and thirty-two MS patients were com-
pared to 78 healthy controls (mean age: 36.4 years)
(Table 1). Ninety-six (73%) of the MS patients had
RRMS (mean age: 39.6 years), 20 (15%) had SPMS
(mean age: 51.5 years) and 16 (12%) had PPMS
(mean age: 52.5 years). The SPMS and PPMS cohorts
were further compared to a separate age-matched
healthy control cohort (n ¼ 18, mean age: 48.6).
Average RNFL thickness and AMT were significantly
reduced in all MS subtypes compared to healthy con-
trols (except for in PPMS). GCIP thickness was like-
wise significantly reduced in all MS subtypes (including
PPMS) compared to healthy controls (p < 0.001 in each
subtype). Average RNFL thickness and GCIP thick-
ness were lower in SPMS than RRMS and PPMS.

Downloaded from msj.sagepub.com at Nat. Taichung Univ. of Sci. & Tech. on April 22, 2014
Saidha et al. 1455

Table 1. Summary of demographics and clinical characteristics

Parameter All MS RRMS SPMS PPMS Young HC Older HC

Subjects, n 132 96 20 16 78 18
Age at OCT scan, years (SD) 42.9 (9.9) 39.6 (9.1) 51.5 (6.0) 52.5 (5.2) 36.4 (9.7) 48.6 (7.2)
Sex, n (%)
Male 42 (31.8) 27 (28.1) 6 (30.0) 9 (56.2) 26 (33.3) 4 (22.2)
Female 90 (68.2) 69 (71.9) 14 (70.0) 7 (43.8) 52 (66.7) 14 (77.8)
Median EDSS (range) 2.5 (0–8) 2 (0–6.5) 6 (2–8) 5.5 (2–6.5) – –
Mean disease duration, years (SD) 10.4 (7.7) 8.6 (6.0) 20.4 (8.4) 10.1 (6.3) – –
Eyes with history of AON, n (%) 41 (31.0%) 37 (38.5%) 4 (20.0%) 0 (0%) – –
Letters read correctly at 100% (SD)* 56.5 (13.1) 57.2 (12.7) 55.9 (15.2) 55.8 (9.1) 62.3 (6.6) 62.0 (6.7)
Letters read correctly at 2.5% (SD)* 25.6 (13.7) 25.6 (14.2) 26.8 (12.1) 25.7 (12.3) 32.3 (8.3) 31.0 (10.7)
Letters read correctly at 1.25% (SD)* 12.2 (11.6) 13.2 (12.0) 9.4 (9.3) 10.2 (11.7) 20.7 (10.3) 10.7 (9.9)
MS: multiple sclerosis, RRMS: relapsing–remitting MS, SPMS: secondary progressive MS, PPMS: primary progressive MS, young HC: healthy control
cohort age-matched to RRMS, older HC: healthy control cohort age-matched to SPMS and PPMS, OCT: Optical coherence tomography SD: standard
deviation, EDSS: expanded disability status scale, AON: acute optic neuritis.
*Visual acuity data was not available for all participants in the study – all MS: n ¼ 120, RRMS: n ¼ 86, SPMS: n ¼ 18, PPMS: n ¼ 16, young HC: n ¼ 67,
older HC: n ¼ 16

While average RNFL thickness and GCIP thickness
correlated significantly with disease duration (RNFL:
r ¼ 0.25, p ¼ 0.004, GCIP: r ¼ 0.20, p ¼ 0.02), after
adjusting for this (as well as age and sex) GCIP thick-
ness remained significantly lower in SPMS than RRMS
(p ¼ 0.04). OCT measurements between the two healthy
control cohorts were similar (Table 2). While the mag-
nitude of significance changed for some OCT measure-
ment comparisons after the SPMS and PPMS cohorts
were compared to their separate age-matched healthy
control cohort, the direction of significance remained
the same as when these progressive MS cohorts were
compared to the larger healthy control cohort (age-
matched for the majority RRMS cohort included in
the study). Specifically, following comparison to this
older age-matched healthy control cohort, GCIP thick-
ness remained significantly lower in SPMS (p < 0.001)
and PPMS (p ¼ 0.003), and INL and ONL thicknesses
still did not differ significantly.
Average GCIP thickness was significantly lower in
eyes with AON history (n ¼ 41; 66.8 mm) and without
AON history (n ¼ 91; 73 mm) compared to healthy con-
trols (p < 0.001 for both), and was also significantly
lower in eyes with AON history compared to non-
AON eyes (p ¼ 0.001). ONL thickness was significantly
lower in RRMS compared to healthy controls
(p ¼ 0.04), but otherwise average INL and ONL thick-
nesses did not differ according to MS subtype or history
of AON. Summary of conventional OCT and OCT seg-
mentation findings are found in Table 2 and Figure 4.
Variance ratio testing demonstrated significantly
smaller standard deviations with GCIP thickness mea-
sures than RNFL thickness measures (p ¼ 0.001),
consistent with less variance of OCT derived measures
of GCIP than RNFL thickness. Average RNFL thick-
ness correlated strongly with GCIP thickness in MS
(and healthy controls), but not with INL or ONL thick-
nesses. AMT correlated strongly with average RNFL,
GCIP, INL and ONL thicknesses in MS and healthy
controls (Table 3). Multivariate linear regression
including age and sex in the model was also used to
adjust for RNFL and GCIP thicknesses when assessing
OCT correlations with EDSS and visual function.
EDSS correlated inversely and significantly with
GCIP thickness in RRMS (Table 4), and remained sig-
nificantly correlated after adjusting for RNFL thick-
ness (r ¼ 0.25, p ¼ 0.01), while RNFL thickness did
not achieve significant correlation with EDSS in the
RRMS cohort after adjusting for GCIP thickness
(r ¼ 0.17, p ¼ 0.11). Although GCIP thickness did not
correlate with EDSS amongst the entire MS cohort
(inclusive of progressive MS patients), following adjust-
ment for RNFL thickness, GCIP thickness did achieve
significant correlation with EDSS (r ¼ 0.20, p ¼ 0.02),
while after adjusting RNFL thickness for GCIP thick-
ness significant correlation was not attained between
RNFL thickness and EDSS (r ¼ 0.08, p ¼ 0.33).
Although average RNFL thickness correlated signif-
icantly with 100% high-contrast, 2.5% and 1.25% low-
contrast letter acuity in the entire MS cohort,
Spearman correlation coefficients were higher between
these and GCIP thickness (Table 4, Figure 5). When
GCIP thickness was adjusted for RNFL thickness it
retained its significant correlations with 100% high-
contrast (r ¼ 0.22, p ¼ 0.02), 2.5% (r ¼ 0.27, p ¼ 0.003)
and 1.25% (r ¼ 0.29, p ¼ 0.001) low-contrast letter

Downloaded from msj.sagepub.com at Nat. Taichung Univ. of Sci. & Tech. on April 22, 2014
 1456

Table 2. Cirrus HD-OCT comparisons according to MS subtype

All MS RRMS SPMS PPMS Young HC Older HC All MS vs. RRMS vs. SPMS vs. SPMS vs. PPMS vs. PPMS vs.
OCT (SD) (SD) (SD) (SD) (SD) (SD) young HC young HC young HC older HC young HC older HC

Unsegmented OCT
ARNFLT, mm 84.3 (13.5) 85.2 (13.6) 77.9 (13.0) 87.0 (10.3) 93.7 (10.5) 92.4 (8.3) p < 0.001 p < 0.001 p < 0.001 p ¼ 0.001 p ¼ 0.08 p ¼ 0.11
AMT, mm 270.4 (15.7) 269.8 (15.7) 265.9 (16.6) 280.1 (11.9) 284.1 (13.6) 285.8 (13.1) p < 0.001 p < 0.001 p < 0.001 p < 0.001 p ¼ 0.14 p ¼ 0.09
Segmented macular OCT
AT GCIP, mm 71.2 (9.8) 71.6 (9.8) 66.4 (10.2) 74.1 (7.1) 81.8 (6.3) 81.9 (6.6) p < 0.001 p < 0.001 p < 0.001 p < 0.001 p < 0.001 p ¼ 0.003
AT INLþOPL, mm 65.5 (4.7) 65.7 (4.6) 65.1 (4.9) 65.0 (5.6) 64.7 (4.4) 65.9 (4.2) p ¼ 0.17 p ¼ 0.16 p ¼ 0.90 p ¼ 0.48 p ¼ 0.93 p ¼ 0.65
AT ONL, mm 119.7 (7.2) 119.2 (7.0) 119.0 (5.8) 123.2 (8.3) 121.3 (7.8) 121.0 (7.7) p ¼ 0.04 p ¼ 0.04 p ¼ 0.15 p ¼ 0.15 p ¼ 0.93 p ¼ 0.90
Multivariate linear regression was used to adjust for age and sex. Results additionally adjusted for disease duration are presented in the results section where stated.
OCT: optical coherence tomography, MS: multiple sclerosis, SD: standard deviation, RRMS: relapsing–remitting MS, SPMS: secondary progressive MS, PPMS: primary progressive MS, young HC: healthy
control cohort age-matched to RRMS, older HC: healthy control cohort age-matched to SPMS and PPMS, ARNFLT: average peri-papillary retinal nerve fiber layer thickness, AMT: average macular thickness,
AT: average thickness, GCIP: ganglion cell layer þ inner plexiform layer, INL: inner nuclear layer, OPL: outer plexiform layer, ONL: outer nuclear layer (including inner and outer photoreceptor segments).

Downloaded from msj.sagepub.com at Nat. Taichung Univ. of Sci. & Tech. on April 22, 2014
Multiple Sclerosis Journal 17(12)
Saidha et al. 1457

Figure 4. Panels A to E represent box plots illustrating the inter-quartile range for peripapillary retinal nerve fiber layer (RNFL)
thickness (panel A), average macular thickness (AMT; panel B), GCIP: ganglion cell layer (GCL) and inner plexiform layer (IPL)
thickness (panel C), inner nuclear layer (INL) (plus outer plexiform layer) thickness (panel D) and outer nuclear layer (ONL) thickness
(including the inner and outer photoreceptor segments) (panel E) in healthy controls and multiple sclerosis (MS) subjects according to
MS subtype. The healthy control cohort represented in this figure is the young healthy control cohort (more appropriately age-
matched to the majority relapsing remitting MS cohort). Significantly reduced thicknesses in comparison to this healthy control cohort
(after adjusting for age and sex) are denoted by an asterisk (*). Separate comparison of the secondary progressive MS (SPMS) and
primary progressive MS (PPMS) cohorts to an older age-matched healthy control cohort revealed the same findings as when these
were compared to the younger healthy control cohort (as in this figure). GCIP thickness was significantly lower in SPMS compared to
relapsing–remitting MS (RRMS) and PPMS and remained significantly lower after adjusting for age and sex. Following additional
adjustment for disease duration, GCIP thickness remained significantly lower in SPMS compared to RRMS.

Downloaded from msj.sagepub.com at Nat. Taichung Univ. of Sci. & Tech. on April 22, 2014
1458 Multiple Sclerosis Journal 17(12)

acuity, while when RNFL thickness was adjusted for subtypes (temporal quadrant RNFL thickness data
GCIP thickness it lost its significant correlations with only shown for entire MS cohort). Spearman correla-
100% high-contrast (r ¼ 0.02, p ¼ 0.80), 2.5% (r ¼ 0.05, tion coefficients between the three RNFL temporal
p ¼ 0.56) and 1.25% (r ¼ 0.08, p ¼ 0.37) low-contrast clock hour thicknesses and the three RNFL nasal
letter acuity. Likewise, Spearman correlation coeffi- clock hour thicknesses with letter acuity (high and
cients were higher between GCIP thickness and letter low-contrast) were even lower than those detected
acuity (both high and low-contrast) than RNFL thick- between RNFL or temporal quadrant RNFL thick-
ness or temporal quadrant RNFL thickness in all MS nesses with letter acuity (clock-hour data not shown).
INL and ONL thicknesses did not correlate with EDSS
or visual function.
The mixed retinal pathology phenotype (RNFL
Table 3. Cirrus HD-OCT layer correlations in MS and healthy
controls
thinning <5th percentile, with INL or ONL thinning
<5th percentile) was observed in 7.6% (n ¼ 10; female:
ALL MS Young HC seven, male: three) of MS patients (RRMS: n ¼ 7,
Layer (p-value) (p-value) SPMS: n ¼ 2, PPMS: n ¼ 1), of whom four had INL
RNFL and GCIP 0.69 (<0.001) 0.71 (<0.001) and six had ONL thinning. Eight additional MS par-
ticipants (6.0%) demonstrated INL (n ¼ 4) or ONL
RNFL and INLþOPL 0.16 (0.06) 0.24 (0.04)
(n ¼ 4) thinning <5th percentile, in the absence of
RNFL and ONL 0.10 (0.27) 0.06 (0.61)
RNFL thinning <5th percentile, despite having
AMT and GCIP 0.80 (<0.001) 0.70 (<0.001) normal AMT between the 5th and 95th percentiles (six
AMT and INLþOPL 0.27 (0.002) 0.46 (<0.001) female, two male; six RRMS, one SPMS, one PPMS).
AMT and ONL 0.62 (<0.001) 0.48 (<0.001) In total, INL/ONL pathology was observed in 13.6%
AMT and RNFL 0.60 (<0.001) 0.55 (<0.001) of our MS cohort, distinct from the 10% of MTP
Entries represent R values.
patients with normal RNFL thickness, reduced AMT
RNFL, peri-papillary retinal nerve fiber layer. Other abbreviations as in and detectable INL/ONL pathology on OCT segmen-
Table 2. tation, which we previously described.

Table 4. Cirrus HD-OCT clinical correlations in MS

EDSS (p-value) 100% LA* (p-value) 2.5% LA* (p-value) 1.25% LA* (p-value)

ALL MS
RNFL 0.12 (0.16) 0.25 (0.005) 0.38 (<0.001) 0.39 (<0.001)
Temporal quadrant RNFL 0.06 (0.50) 0.27 (0.003) 0.32 (<0.001) 0.41 (<0.001)
AMT 0.03 (0.73) 0.27 (0.003) 0.31 (<0.001) 0.29 (0.002)
GCIP 0.14 (0.12) 0.41 (<0.001) 0.49 (<0.001) 0.52 (<0.001)
RRMS
RNFL 0.11 (0.27) 0.31 (0.004) 0.43 (<0.001) 0.44 (<0.001)
AMT 0.14 (0.17) 0.31 (0.003) 0.32 (0.003) 0.30 (0.006)
GCIP 0.20 (0.048) 0.52 (<0.001) 0.58 (<0.001) 0.59 (<0.001)
SPMS
RNFL 0.13 (0.60) 0.20 (0.42) 0.06 (0.82) 0.16 (0.54)
AMT 0.06 (0.79) 0.38 (0.12) 0.44 (0.07) 0.44 (0.06)
GCIP 0.03 (0.90) 0.27 (0.28) 0.39 (0.11) 0.50 (0.04)
PPMS
RNFL 0.14 (0.59) 0.43 (0.01) 0.49 (0.06) 0.36 (0.17)
AMT 0.15 (0.58) 0.06 (0.82) 0.07 (0.81) 0.15 (0.57)
GCIP 0.13 (0.62) 0.66 (0.006) 0.57 (0.02) 0.70 (0.002)
Spearman rank correlation was used to determine OCT correlations with clinical measures (adjusted partial correlation coefficients are presented in
the results section where stated).
OCT: optical coherence tomography, MS: multiple sclerosis, EDSS: Expanded Disability Status Scale, LA: letter acuity, RNFL: peri-papillary retinal nerve
fiber layer, AMT: average macular thickness, GCIP: ganglion cell layer þ inner plexiform layer, RRMS: relapsing–remitting MS, SPMS: secondary
progressive MS, PPMS: primary progressive MS.
*Visual acuity data was not available for all participants in the study – all MS: n ¼ 120, RRMS: n ¼ 86, SPMS: n ¼ 18, PPMS: n ¼ 16.

Downloaded from msj.sagepub.com at Nat. Taichung Univ. of Sci. & Tech. on April 22, 2014
Saidha et al. 1459

Figure 5. Panels A and C represent scatter plots of peri-papillary retinal nerve fiber layer (RNFL) thickness with 2.5% and 1.25%
low-contrast letter acuity respectively in the entire MS cohort. Panels B and D represent scatter plots of GCIP (ganglion cell layer plus
inner plexiform layer) thickness with 2.5% and 1.25% low-contrast letter acuity respectively in the entire MS cohort. The solid red
lines in each panel represent the lines of correlation. Note that in comparing panels A and B, the distribution of data points lies closer
to the line of correlation in panel B than panel A, consistent with the higher correlation coefficient between GCIP thickness and 2.5%
low-contrast letter acuity (r ¼ 0.49), than between RNFL thickness and 2.5% low-contrast letter acuity (r ¼ 0.38). Likewise, the
distribution of data points lies closer to the line of correlation in panel D than panel C, consistent with the higher correlation
coefficient between GCIP thickness and 1.25% low-contrast letter acuity (r ¼ 0.52), than between RNFL thickness and 2.5% low-
contrast letter acuity (r ¼ 0.39). These scatter plots illustrate that 2.5% and 1.25% low-contrast letter acuity correlate better with
GCIP thickness than RNFL thickness. Although not included in this figure, 100% high-contrast visual acuity also correlates better with
GCIP thickness (r ¼ 0.41) than RNFL thickness (r ¼ 0.25) in MS, with similar appearing scatter plots to those presented above. GCIP
thickness also maintains its significant correlations with visual function (both high and low-contrast) after adjusting for RNFL thickness,
while RNFL thickness loses its significant correlations with visual function (high and low-contrast) after adjusting for GCIP thickness,
further supporting greater association between visual function and GCIP thickness than RNFL thickness in MS.

Discussion
Results of this study demonstrate objective in vivo evi-
dence suggesting that GCIP thinning occurs in all MS
subtypes, is most prominent in SPMS and increases
with disease duration (as does RNFL thinning). Eyes
with AON history demonstrated greater GCIP thinning
than non-AON eyes. High correlation was found
between GCIP and RNFL thickness values, but not
other retinal layers. These data provide human in vivo
evidence supporting association between optic neurop-
athy and GCL pathology, as has been proposed on the
basis of animal and post-mortem studies, but until now
not demonstrated with in vivo investigations.11,14,15
Notwithstanding our observations, we acknowledge
that only a longitudinal study assessing retinal layer
changes following AON may ultimately confirm that
optic neuropathy results in GCL degeneration.
Although AMT correlated strongly with RNFL thick-
ness and retinal neuronal layer thicknesses, results of
this study demonstrate that significant GCIP thinning
may be present, despite normal AMT (as observed in
the PPMS cohort). Likewise, approximately 6% of MS

Downloaded from msj.sagepub.com at Nat. Taichung Univ. of Sci. & Tech. on April 22, 2014
1460
participants in this study demonstrated INL/ONL thin-
ning, despite normal AMT. These factors highlight the
potential benefit of systematic OCT segmentation over
conventional AMT in the assessment of retinal neuro-
nal architecture in MS.
Retinal axons are unmyelinated, making the retina a
unique structure within the CNS. Given this and the
predilection of MS to affect the optic nerves,7,8 the eye
has been proposed as a model within which to study the
neurodegenerative processes associated with MS.1
Demyelinated, yet intact, retrobulbar axons may poten-
tially be remyelinated by local viable oligodendrocytes,
protecting against axonal degeneration and neuronal
loss, thereby resulting in RNFL and GCL preservation.
OCT has several compelling methodological character-
istics including highly significant structure–function
correlations, which make it equipped to detect and
monitor the course of disease-related neurodegenera-
tion in MS and to document the neuroprotective
and potentially neurorestorative effects of therapeutic
agents.1,13 The advent of OCT segmentation not only
allows objective characterization of optic nerve demye-
lination consequences upon proximal RNFL axons, but
now also associated retinal neuronal architecture. OCT
measures of RNFL thickness have been proposed as a
primary outcome measure in imminent trials of poten-
tial neuroprotectants in AON. However, superior struc-
ture–function correlation between GCIP thickness and
validated clinical measures found in this systematic ret-
inal layer segmentation analysis provide support that
retinal neuronal topography, particularly of the GCL,
may serve as a better instrument than RNFL to examine
MS associated neurodegenerative mechanisms within
the eye. Following complete degeneration, it might be
reasonable to anticipate equal numbers of axonal
(RNFL fibers) and neuronal (ganglion cells) death. If
this is the case, differences in structure–function correla-
tions between RNFL thickness and GCIP thickness may
not be expected. Bearing this in mind, it is worth consid-
ering the possibility that astrogliosis (which principally
occurs in the RNFL) may contribute to falsely increased
measures of RNFL thickness.15 In other words, the pres-
ence of astrogliosis may confound the ability of OCT to
measure RNFL thickness as accurately as GCIP thick-
ness. This may be a contributing factor underlying the
better structure–function correlations observed with
GCIP thickness than RNFL thickness.
No significant correlations were detected between
INL and ONL thicknesses and RNFL thickness.
Average INL and ONL thicknesses also did not differ
according to AON history, unlike GCIP thickness.
Taken together, these findings lend support to the
recently proposed hypothesis that INL/ONL neuronal
layer thinning in MS may represent primary retinal
neuronal pathology.30 Although the MTP phenotype
Downloaded from msj.sagepub.com at Nat. Taichung Univ. of Sci. & Tech. on April 22, 2014
Multiple Sclerosis Journal 17(12)
in MS is characterized by INL/ONL thinning on
OCT segmentation, criteria defining MTP are based
upon conventional OCT measures; the combination
of AMT <5th percentile with normal RNFL thick-
ness.30 According to this definition of MTP, our
group, and more recently another group have found
the prevalence of MTP to be approximately 10% in
MS.30,42 In the current study, which excluded patients
fulfilling these MTP criteria, 6% of MS participants
still had INL/ONL thinning in the presence of normal
RNFL thickness typical of the MTP phenotype, but did
not fulfill MTP criteria on account of having normal
AMT. This highlights that; AMT is less useful than
OCT segmentation in assessing discrete retinal neuro-
nal architecture in MS since it is a composite measure
of multiple retinal layers, that the existing definition of
MTP may benefit from refinement and inclusion of
OCT segmentation criteria, and that previous preva-
lence estimates of primary retinal pathology in MS
are thus likely to be underestimates. These new higher
prevalence estimates of INL/ONL pathology are in
accordance with post-mortem data, which found 40%
of MS eyeballs had INL atrophy.15 Furthermore, the
MTP definition is limited by virtue of excluding eyes
with RNFL thinning (which may be the result of two
diametrically polar pathological processes; dying back
versus dying forward), and mixed retinal pathology as a
result.30 Approximately 8% of MS participants fulfilled
our criteria for mixed retinal pathology, implicating
dual mechanisms of retinal changes in some MS
patients. In total, approximately 14% of MS partici-
pants in this study demonstrated INL/ONL thinning.
This study has a number of limitations warranting
discussion. The retinal segmentation employed has not
been extended to separate the INL from the OPL, and
is likely incapable of separating the GCL from the IPL
since the boundaries between these two layers cannot
be visually discriminated on Cirrus HD-OCT. Indeed,
the GCL and IPL have also not been distinguished by
other OCT segmentation techniques.29,43 Since post-
mortem studies14,15 demonstrate atrophy of the GCL
and INL in MS, we strongly suspect that the reduction
in OCT segmentation parameters principally relate to
neuronal atrophy/loss. However, it is plausible that
INL or ONL pathology may be associated with plexi-
form layer sequelae. Future advances in OCT segmen-
tation algorithms with refinements in layer recognition
will facilitate more accurate assessment of MS related
neuroretinal pathology. Further, the majority of MS
patients in this study had RRMS. To more precisely
and accurately characterize the correlations of OCT
segmentation measures according to MS subtype, the
enrollment of greater numbers of not only SPMS and
PPMS patients, but also RRMS patients in future stud-
ies is warranted. Limited participant enrollment to this
Saidha et al.
study and under powering may have contributed to the
lack of significance for some of the analyses, particularly
in the progressive MS cohorts. Additionally, we have
previously demonstrated that Cirrus HD-OCT has
inter-rater ICCs of 0.97 for peripapillary RNFL thick-
ness41 and 0.99 for GCIP thickness.30 Improved signal-
to-noise ratios within the GCIP with Cirrus HD-OCT, as
well as the lower variance of GCIP as compared to
RNFL thickness measures may have contributed to the
better structure–function correlations observed with this
measure. These findings should thus be recapitulated
with other OCT devices to ensure they are not specific
to the OCT device used in this study.
The formulation and application of a mixed retinal
pathological designation may represent a fundamental
preliminary step in studying this novel pathobiological
phenotype, although future refinement of this definition
may be necessary. In addition, further clinical and elec-
trophysiological characterization of this subgroup is
warranted in future studies. An important hypothetical
principle worthy of consideration is that INL and ONL
thinning may still represent regional neuronal suscepti-
bility to retrograde trans-synaptic degeneration in sub-
sets of MS patients. Longitudinal studies utilizing OCT
segmentation may help characterize and quantify reti-
nal neuronal loss following AON, and determine
whether deeper INL/ONL thinning is seen as a sequela
of AON, or whether it is unrelated to optic nerve
pathology, as suggested.30
In summary, retinal segmentation techniques extend
the utility of OCT in MS, enabling the in vivo objective
assessment of retinal neuronal in addition to axonal
integrity. GCIP thinning appears to occur in all MS
subtypes and GCIP thickness correlates better with
visual dysfunction and disability in MS than RNFL
thickness. The demonstration that a subset of MS
patients may have primary retinal neuronal pathology,
in addition to optic nerve retrograde mediated mecha-
nisms of retinal changes, supports recent findings and
may help further our understanding of the pathobiolo-
gical underpinnings of this highly complex and hetero-
geneous disorder.
Funding
National Multiple Sclerosis Society (TR 3760-A-3 to P.A.C
and RG 4212-A-4 to Laura J. Balcer subcontracted to
P.A.C), National Eye Institute (R01 EY 014993 and R01
EY 019473 to Laura J. Balcer subcontracted to P.A.C), and
Braxton Debbie Angela Dillon and Skip (DADS) Donor
Advisor Fund (to P.A.C, E.M.F)
Conflict of interest statement
Dr Shiv Saidha has received consulting fees from
MedicalLogix for the development of continuing
Downloaded from msj.sagepub.com at Nat. Taichung Univ. of Sci. & Tech. on April 22, 2014
1461
medical education programs in neurology. Dr
Christopher Eckstein receives funding through a
Sylvia Lawry Physician Fellowship from the National
Multiple Sclerosis Society. Jonathan Oakley, Mary
Durbin and Scott Meyer are employed by Carl Zeiss
Meditec Inc. Dr Elliot Frohman has received speaker
honoraria and consulting fees from Biogen Idec,
TEVA, and Athena. He has also received consulting
fees from Abbott Laboratories. Dr Scott Newsome
has received speaker honoraria and consulting fees
from Biogen Idec. Dr John Ratchford has consulted
for Sun Pharmaceuticals. Dr Peter Calabresi has pro-
vided consultation services to Novartis, EMD-Serono,
Teva, Biogen-IDEC, Vertex, Vaccinex, Genzyme,
Genentech; and has received grant support from EMD-
Serono, Teva, Biogen-IDEC, Genentech, Bayer, Abbott,
and Vertex. Stephanie Syc and Teresa Frohman have no
disclosures.
References
1. Frohman EM, Fujimoto JG, Frohman TC, Calabresi
PA, Cutter G and Balcer LJ. Optical coherence tomog-
raphy: a window into the mechanisms of multiple sclero-
sis. Nat Clin Pract Neurol 2008; 4: 664–675.
2. Hsu IJ, Sun CW, Lu CW, Yang CC, Chiang CP and Lin
CW. Resolution improvement with dispersion manipula-
tion and a retrieval algorithm in optical coherence tomog-
raphy. Appl Opt 2003; 42: 227–234.
3. Huang D, Swanson EA, Lin CP, Schuman JS, Stinson
WG, Chang W, et al. Optical coherence tomography.
Science 1991; 254: 1178–1181.
4. Fisher JB, Jacobs DA, Markowitz CE, Galetta SL,
Volpe NJ, Nano-Schiavi ML, et al. Relation of visual
function to retinal nerve fiber layer thickness in multiple
sclerosis. Ophthalmology 2006; 113: 324–332.
5. Costello F, Coupland S, Hodge W, Lorello GR, Koroluk
J, Pan YI, et al. Quantifying axonal loss after optic neu-
ritis with optical coherence tomography. Ann Neurol
2006; 59: 963–969.
6. Pulicken M, Gordon-Lipkin E, Balcer LJ, Frohman E,
Cutter G and Calabresi PA. Optical coherence tomogra-
phy and disease subtype in multiple sclerosis. Neurology
2007; 69: 2085–2092.
7. Ikuta F and Zimmerman HM. Distribution of plaques in
seventy autopsy cases of multiple sclerosis in the United
States. Neurology 1976; 26: 26–28.
8. Toussaint D, Perier O, Verstappen A and Bervoets S.
Clinicopathological study of the visual pathways, eyes,
and cerebral hemispheres in 32 cases of disseminated scle-
rosis. J Clin Neuroophthalmol 1983; 3: 211–220.
9. Trapp BD, Peterson J, Ransohoff RM, Rudick R, Mork
S and Bo L. Axonal transection in the lesions of multiple
sclerosis. N Engl J Med 1998; 338: 278–285.
10. Naismith RT, Xu J, Tutlam NT, Trinkaus K, Cross AH
and Song SK. Radial diffusivity in remote optic neuritis
discriminates visual outcomes. Neurology 2010; 74:
1702–1710.
1462
11. Shindler KS, Ventura E, Dutt M and Rostami A.
Inflammatory demyelination induces axonal injury and
retinal ganglion cell apoptosis in experimental optic neu-
ritis. Exp Eye Res 2008; 87: 208–213.
12. Burkholder BM, Osborne B, Loguidice MJ, Bisker E,
Frohman TC, Conger A, et al. Macular volume deter-
mined by optical coherence tomography as a measure
of neuronal loss in multiple sclerosis. Arch Neurol 2009;
66: 1366–1372.
13. Saidha S, Eckstein C and Ratchford JN. Optical coher-
ence tomography as a marker of axonal damage in mul-
tiple sclerosis. CML – Multiple Sclerosis 2010; 2: 33–43.
14. Kerrison JB, Flynn T and Green WR. Retinal pathologic
changes in multiple sclerosis. Retina 1994; 14: 445–451.
15. Green AJ, McQuaid S, Hauser SL, Allen IV and Lyness
R. Ocular pathology in multiple sclerosis: retinal atrophy
and inflammation irrespective of disease duration. Brain
2010; 133: 1591–1601.
16. Gills Jr JP. Electroretinographic abnormalities and
advanced multiple sclerosis. Invest Ophthalmol 1966; 5:
555–559.
17. Papakostopoulos D, Fotiou F, Hart JC and Banerji NK.
The electroretinogram in multiple sclerosis and demyelin-
ating optic neuritis. Electroencephalogr Clin Neurophysiol
1989; 74: 1–10.
18. Forooghian F, Sproule M, Westall C, Gordon L,
Jirawuthiworavong G, Shimazaki K, et al. Electroretino-
graphic abnormalities in multiple sclerosis: possible role
for retinal autoantibodies. Doc Ophthalmol 2006; 113:
123–132.
19. Gundogan FC, Demirkaya S and Sobaci G. Is optical
coherence tomography really a new biomarker candidate
in multiple sclerosis? – A structural and functional eval-
uation. Invest Ophthalmol Vis Sci 2007; 48: 5773–5781.
20. Schuman JS, Pedut-Kloizman T, Pakter H, Wang N,
Guedes V, Huang L, et al. Optical coherence tomography
and histologic measurements of nerve fiber layer thick-
ness in normal and glaucomatous monkey eyes. Invest
Ophthalmol Vis Sci 2007; 48: 3645–3654.
21. Knott EJ, Sheets KG, Zhou Y, Gordon WC and Bazan
NG. Spatial correlation of mouse photoreceptor–RPE
thickness between SD-OCT and histology. Exp Eye Res
2011; 92: 155–160.
22. Cohen MJ, Kaliner E, Frenkel S, Kogan M, Miron H
and Blumenthal EZ. Morphometric analysis of human
peripapillary retinal nerve fiber layer thickness. Invest
Ophthalmol Vis Sci 2008; 49: 941–944.
23. Fujimoto JG. Optical coherence tomography for ultra-
high resolution in vivo imaging. Nat Biotechnol 2003;
21: 1361–1367.
24. Chen TC, Cense B, Miller JW, Rubin PA, Deschler DG,
Gragoudas ES, et al. Histologic correlation of in vivo
optical coherence tomography images of the human
retina. Am J Ophthalmol 2006; 141: 1165–1168.
25. Srinivasan VJ, Monson BK, Wojtkowski M, Bilonick RA,
Gorczynska I, Chen R, et al. Characterization of outer
retinal morphology with high-speed, ultrahigh-resolution
optical coherence tomography. Invest Ophthalmol Vis Sci
2008; 49: 1571–1579.
Downloaded from msj.sagepub.com at Nat. Taichung Univ. of Sci. & Tech. on April 22, 2014
Multiple Sclerosis Journal 17(12)
26. Kim KH, Puoris’haag M, Maguluri GN, Umino Y,
Cusato K, Barlow RB, et al. Monitoring mouse retinal
degeneration with high-resolution spectral-domain opti-
cal coherence tomography. J Vis 2008; 8: 171–1711.
27. Sarunic MV, Yazdanpanah A, Gibson E, Xu J, Bai Y,
Lee S, et al. Longitudinal study of retinal degeneration in
a rat using spectral domain optical coherence tomogra-
phy. Opt Express 2010; 18: 23435–23441.
28. Wojtkowski M, Srinivasan V, Fujimoto JG, Ko T,
Schuman JS, Kowalczyk A, et al. Three-dimensional
retinal imaging with high-speed ultrahigh-resolution
optical coherence tomography. Ophthalmology 2005;
112: 1734–1746.
29. Hood DC, Lin CE, Lazow MA, Locke KG, Zhang X and
Birch DG. Thickness of receptor and post-receptor reti-
nal layers in patients with retinitis pigmentosa measured
with frequency-domain optical coherence tomography.
Invest Ophthalmol Vis Sci 2009; 50: 2328–2336.
30. Saidha S, Syc SB, Ibrahim MA, Eckstein C, Warner CV,
Farrell SK, et al. Primary retinal pathology in multiple
sclerosis as detected by optical coherence tomography.
Brain 2011; 134: 518–533.
31. Hollander H, Bisti S, Maffei L and Hebel R
Electroretinographic responses and retrograde changes
of retinal morphology after intracranial optic nerve sec-
tion. A quantitative analysis in the cat. Exp Brain Res
1984; 55: 483–493.
32. Levkovitch-Verbin H, Quigley HA, Kerrigan-Baumrind
LA, D’Anna SA, Kerrigan D and Pease ME. Optic nerve
transection in monkeys may result in secondary degener-
ation of retinal ganglion cells. Invest Ophthalmol Vis Sci
2001; 42: 975–982.
33. Williams RR, Cusato K, Raven MA and Reese BE.
Organization of the inner retina following early elimina-
tion of the retinal ganglion cell population: effects on cell
numbers and stratification patterns. Vis Neurosci 2001;
18: 233–244.
34. Dawson WW, Maida TM and Rubin ML. Human pat-
tern-evoked retinal responses are altered by optic atro-
phy. Invest Ophthalmol Vis Sci 1982; 22: 796–803.
35. Kaufman D and Celesia GG. Simultaneous recording of
pattern electroretinogram and visual evoked responses
in neuro-ophthalmologic disorders. Neurology 1985; 35:
644–651.
36. Seiple W, Price MJ, Kupersmith M, Siegel IM and Carr
RE. The pattern electroretinogram in optic nerve disease.
Ophthalmology 1983; 90: 1127–1132.
37. Polman CH, Reingold SC, Edan G, Filippi M, Hartung
HP, Kappos L, et al. Diagnostic criteria for multiple scle-
rosis: 2005 revisions to the ‘‘McDonald Criteria’’. Ann
Neurol 2005; 58: 840–846.
38. Lublin FD and Reingold SC. Defining the clinical course
of multiple sclerosis: results of an international survey.
National Multiple Sclerosis Society (USA) Advisory
Committee on Clinical Trials of New Agents in
Multiple Sclerosis. Neurology 1996; 46: 907–911.
39. Kurtzke JF. Rating neurologic impairment in multiple
sclerosis: an expanded disability status scale (EDSS).
Neurology 1983; 33: 1444–1452.
Saidha et al.
40. ETDRS Research Group Early Treatment Diabetic
Retinopathy Study Research Group. Photocoagulation
for diabetic macular edema. Early Treatment Diabetic
Retinopathy Study report number 1. Arch Ophthalmol
1985; 103: 1796–1806.
41. Syc SB, Warner CV, Hiremath GS, Farrell SK,
Ratchford JN, Conger A, et al. Reproducibility of high-
resolution optical coherence tomography in multiple scle-
rosis. Mult Scler 2010; 16: 829–839.
42. Winges KM, Werner JS, Harvey DJ, Calabresi PA,
Balcer LJ, Cello KE, et al. Comparison of retinal nerve
1463
fiber layer thickness and macular thickness/volume
among patients with relapsing–remitting multiple sclero-
sis: baseline data from a phase 3 trial of fingolimod.
Association of Research in Vision and Ophthalmology,
Fort Lauderdale, FL, 5 May 2010. (Abstract published in
Investigative Ophthalmology & Visual Science, 52, ARVO
E-Abstract 5308.).
43. Tan O, Chopra V, Lu AT, Schuman JS, Ishikawa H,
Wollstein G, et al. Detection of macular ganglion cell
loss in glaucoma by Fourier-domain optical coherence
tomography. Ophthalmology 2009; 116: 2305–14. (e1–2.