Differentiating Multiple Sclerosis and Glaucoma With Sectoral Pattern Analysis of Peripapillary Nerve Fiber Layer

Abstract

Purpose: To distinguish between multiple sclerosis (MS) and glaucoma by nerve fiber layer (NFL) thinning patterns.

Methods: MS patients were diagnosed by the 2017 McDonald Criteria; glaucoma patients had disc rim thinning or an NFL defect, with or without perimetric defect. The peripapillary NFL thickness was divided into eight sectors, and percentage reduction (% reduction) was calculated relative to normative reference values. The MS and glaucoma eyes were grouped based on the severity of NFL thinning in the worst sector: significant reduction (<1 percentile of normal reference), borderline reduction (1%∼5%), and no reduction (>5%). We devised four diagnostic indexes, and the area under the curve of receiver operating characteristics (AROC) and accuracy were used to evaluate the indexes.

Results: We enrolled 58 control subjects (58 eyes), 56 MS subjects (112 eyes), and 92 glaucoma subjects (92 eyes) at two centers. The most pronounced percent reduction in MS eyes occurred in the temporal-upper and temporal-lower sectors. In glaucoma eyes, this occurred in the inferior-temporal, inferior-nasal, and superior-temporal sectors. The temporal pattern index had the best AROC (0.96, 0.91–1.00) and accuracy (92.6%) in the significant reduction group. It had good AROC (0.88, 0.78–0.99) and accuracy (76.7%) in the borderline reduction group.

Conclusions: Normalizing NFL reduction as a percentage of normal reference accentuated patterns characteristic of MS and glaucoma. Quantitative pattern indexes were effective in differentiating the two diseases.

Translational Relevance: The utility of optical coherence tomography in the differential diagnosis of optic neuropathies is enhanced by analyzing the retinal nerve fiber layer percentage reduction pattern.

Introduction

Multiple sclerosis (MS) is an inflammatory neurodegenerative disease of the central nervous system. Although optic neuritis (ON) is a common ocular component of MS,^1,2 MS patients can develop optic neuropathy without ON episodes.³ Glaucoma is a progressive optic neuropathy and one of the leading causes of blindness worldwide.⁴ Optic neuropathy caused by glaucoma and MS is often distinguishable by the presenting clinical signs and symptoms, such as elevated intraocular pressure (glaucoma) or acute unilateral loss of vision with pain in eye movement (acute optic neuritis in MS). But these classic findings are sometimes absent. Individuals with MS may experience optic nerve involvement even if they have no history of ON.^5,6 Besides, half of glaucoma patients initially present with intraocular pressure within the normal range.^7–9 

Optical coherence tomography (OCT) can provide an accurate and objective measurement of the retinal nerve fiber layer (NFL). It is useful in diagnosing and monitoring glaucomatous optic neuropathy and nonglaucomatous optic neuropathy.^10–12 Both MS and glaucoma result in NFL thinning, making it challenging to differentiate between them based on OCT scans alone.^13–16 Nevertheless, each disease has its own distinctive pattern of NFL thinning. MS tends to produce papillomacular bundle defects, and glaucoma most commonly produces defects in the superior and inferior quadrants.^17,18 Previous studies have also found that nonglaucomatous optic neuropathies tend to result in thinner nasal and temporal NFL compared to glaucomatous optic neuropathy.^19,20 However, they did not find a significant difference in NFL thickness in any particular sector or quadrant.²⁰ Although optic disc morphology is very helpful, some patients have large physiological cups, and the determination of disc rim loss and pallor is highly subjective.²¹ Moreover, an increasing number of neurologists are using OCT,²² and intereye differences in NFL thickness were tried to serve as a diagnostic criterion of MS.^23–25 However, using intereye differences as a diagnostic factor can be nonspecific because glaucoma, which is more prevalent than MS, can also present with intereye differences in NFL thickness.²⁶ Therefore it is helpful to have a readily available objective test that can quickly distinguish between the optic neuropathies caused by MS and glaucoma. 

We hypothesized that the approach of comparing individual sectors is unlikely to be successful in differentiating between disease types because the sector parameters are affected by both the global disease severity and localizing patterns. To remove the confounding effect of disease severity, it is necessary to develop pattern analysis methods that compare the relative severity of NFL defects between sectors within an individual eye. Therefore in this study we developed new diagnostic indexes based on the pattern of percent reduction in NFL thickness to distinguish between MS and glaucoma eyes. 

Methods

Participants

This case-control study was performed at the Casey Eye Institute, Oregon Health & Science University, and Portland Veteran Affairs Medical Center from January 2, 2020, to May 12, 2023. The research protocol was approved by the Institutional Review Board at Oregon Health & Science University and adhered to the tenets of the Declaration of Helsinki. Written informed consent was obtained from each participant. 

The participants with MS were part of the “Race to Erase” (R2E) study and “Retinal Microvasculature as a Predictor of Neurodegeneration in Multiple Sclerosis” study. The inclusion criteria for MS were diagnosis of MS by 2017 McDonald Criteria and age between 18 to 70 years. The exclusion criteria for MS were (1) Other ocular diseases or pathology that would interfere or confound the assessment of MS and ON for both eyes, including glaucoma, diabetic or hypertensive retinal disease, age-related macular degeneration, history of uveitis, and amblyopia; (2) Previous intraocular surgery except for uncomplicated cataract extraction with posterior chamber intraocular lens implantation; (3) Spherical equivalent refractive error greater than 3 or −7 diopters. Both eyes of the MS participants received OCT scanning. 

Glaucoma and normal control participants were part of the “Functional and Structural Optical Coherence Tomography for Glaucoma” study.²⁷ The inclusion criteria for glaucoma were optic disc appearance consistent with glaucoma: (1) diffuse or localized thinning of the optic disc rim visible on funduscopy; (2) well-defined NFL bundle defect visible on funduscopy. The participants with perimetric glaucoma (PG) should have visual field (VF) tests that showed glaucomatous defect with pattern standard deviation outside normal limits (P < 0.05) or glaucoma hemifield test outside normal limits (P < 0.05). The participants with preperimetric glaucoma (PPG) should have no glaucomatous defects on the VF test. The inclusion criteria for normal control were (1) no history of glaucoma, retinal pathology, or current corticosteroid use; (2) no history of ocular hypertension as defined by IOP ≥22 mm Hg; (3) normal Humphrey 24-2 VF test; (4) normal optic nerve head and NFL appearance on funduscopy; (5) symmetric optic nerve head appearance between both eyes; (6) central pachymetry >470 µm. The exclusion criteria for both normal control and glaucoma participants were (1) best-corrected visual acuity less than 20/40; (2) spherical equivalent refractive error greater than 3 or −7 diopters; (3) previous intraocular surgery except for uncomplicated cataract extraction with posterior chamber intraocular lens implantation; (4) any diseases that may cause VF loss or optic disc abnormalities; (5) narrow anterior chamber angle by gonioscopy. The glaucoma patients with untreated baseline IOP more than 21 mm Hg were classified as high-tension glaucoma (HTG), and those with untreated baseline IOP equal to or less than 21 mm Hg were considered as normal-tension glaucoma (NTG).^28,29 Only one eye of each glaucoma or normal participant received OCT scanning and analysis. For normal participants, the eye was randomly selected. For glaucoma participants, the eye with the worse VF was selected. 

Ancillary Tests

The axial length was measured by IOLMaster500 (Carl Zeiss Meditec, Inc., Dublin, CA, USA). The VF tests were performed with the Humphrey Field Analyzer II (Carl Zeiss Meditec, Inc.) with the 24-2 threshold test, size II white stimulus, and SITA standard algorithm. A test was considered unreliable if the false-positive rate, false-negative rate, or fixation loss was higher than 33%. 

OCT and NFL Thickness Measurements

Participants were scanned with the 6 × 6-mm disc scan using a 120-kHz spectral-domain OCT system (Optovue Solix; Visionix, Fremont, CA, USA). A scan with a signal strength index lower than 50 or significant eye motion would be excluded to ensure image quality. The retinal layer segmentation and disc center detection were performed using Solix software (V1.0.0.342; Visionix). Manual correction of the segmentation or disc centration was conducted if needed. The NFL thickness was measured in eight modified Garway-Heath sectors (Fig. 1A).³⁰ The eight sectors were shown in NSTIN order: nasal-upper, superior-nasal (SN), superior-temporal (ST), temporal-upper (TU), temporal-lower (TL), inferior-temporal (IT), inferior-nasal (IN), and nasal-lower.³¹ 

Figure 1.

 

An example of how we get the sector NFL thickness and convert it into percent reduction. (A) The 6.0 mm × 6.0 mm NFL thickness map of a glaucoma eye was divided into eight sectors using a modified Garway-Heath scheme. (B) The NFL thickness of each sector was shown on the bar plot. The black bars are for the NFL thickness of each sector from the glaucoma eye, and the light gray bars are for the NFL thickness of each sector from normal reference. The difference between the sectoral NFL thickness and the normal reference is the light gray area. (C) The bar plot showed the percent reduction of each sector. The largest percentage reduction was in the ST sector, and there was no percentage reduction in the TL sector. ILM, internal limiting membrane; NU, nasal upper; NL, nasal lower.

Normal Reference and Grouping

The sectoral NFL thickness was first adjusted for age and axial length based on multiple linear regression with a mixed-effect model of normal eyes. This adjustment was then applied to all eyes in the study. After the adjustment, the sectoral normal references were determined based on the average NFL thickness in normal eyes. The one and five-percentile cutoffs were established by identifying points 1.64 and 2.33 standard deviation (SD) below the average, following the principles of a normal distribution. 

We grouped MS eyes and glaucoma eyes by the severity of NFL reduction: significant reduction group (worst sector NFL thickness < 1 percentile cutoff), borderline reduction group (worst sector between 1 and 5 percentile cut points), and no reduction group (all sectors > 5 percentile cutoff). The no reduction group was excluded because these patients may not have any optic neuropathy or such an early stage that no distinct pattern can yet emerge on OCT. 

NFL Percentage Reduction

To compare the severity of NFL reduction among different sectors within individual eyes, it is necessary to normalize the reduction against healthy control values in each sector. We used the percentage NFL reduction (percent reduction), defined as the difference between the sectoral NFL thickness and the normal reference, as a percentage of the normal reference (Figs. 1B, 1C):  

\begin{eqnarray*} && Sectoral\ \% \ reduction \\ && =\frac{{Sectoral\ \textit{NFL}\ thickness - \textit{sectoral}\ normal\ \textit{reference}}}{{Sectoral\ normal\ \textit{reference}{\rm{\ }}}}\end{eqnarray*}

 

If the sectoral NFL is thicker than the respective normal reference, we cap the percent reduction value at zero. 

Indexes of Sectoral Pattern Analysis

We devised pattern indexes for the differential diagnosis of MS and glaucoma based on previous observations that MS and associated ON predominantly reduce NFL thickness in the temporal quadrant, and glaucoma predominantly leads to NFL thinning in superior and inferior quadrants.^17–19 

The pattern index takes the difference between the average percent reduction in sectors primarily affected by MS (TU, TL) and the average percent reduction in sectors primarily affected by glaucoma (ST, IT, SN, IN):  

\begin{eqnarray*} && Pattern\ index = Average\left( {\textit{TU}\% loss,\textit{TL}\% loss\ } \right) \\ && - Average\left( {ST\% loss,\ IT\% loss,\textit{SN}\% loss,\textit{IN}\% loss} \right)\end{eqnarray*}

 

The formula for calculating the temporal pattern index is similar to that of the pattern index, but the SN and IN sectors were excluded because they are less affected by glaucoma compared to the ST and IT sectors:  

\begin{eqnarray*} && \textit{Temporal}\ pattern\ index = Average\left( {\textit{TU}\% loss,\textit{TL}\% loss} \right) \\ && - Average\left( {ST\% loss,\ \textit{IT}\% loss} \right)\end{eqnarray*}

 

The worst sector index identifies whether the worst sector is a sector primarily affected by MS (TU, TL) or primarily affected by glaucoma (ST, IT, SN, IN). The worst sector is defined as the one with the largest percentage reduction, which means it has the minimum value (most negative value):  

\begin{eqnarray*} && \textit{Worst}\ sector\ index = \textit{Min}\left( {\textit{TU}\% loss,\textit{TL}\% loss} \right) \\ && - \textit{Min}\left( {\textit{ST}\% loss,\ \textit{IT}\% loss,SN\% loss,\textit{IN}\% loss} \right)\end{eqnarray*}

 

The process for calculating the temporal worst sector index is similar to that of the worst sector index, except that the SN and IN sectors are excluded because they are less affected by glaucoma compared to the ST and IT sectors:  

\begin{eqnarray*} && \textit{Temporal}\ worst\ sector\ index = \textit{Min}\left( {\textit{TU}\% loss,\textit{TL}\% loss} \right) \\ && - Min\left( {ST\% loss,\ \textit{IT}\% loss} \right)\end{eqnarray*}

 

Analyze One Eye Per Participant To Prevent Bias Due to Intereye Correlation

In this study, we measured both eyes of each MS subject. To avoid the correlation between the two eyes of individual subjects from biasing the statistical analysis, the study analyses only used measurements from the eye with worse visual field mean deviation (VF-MD) if both eyes belonged to the same severity group. 

Statistics

All statistical analyses were performed with R Statistical Software (v4.2.2; R Core Team 2022, Vienna, Austria). The characteristic data of participants were summarized with descriptive statistics. Continuous variables were compared using an unpaired t-test. Sex was compared using a χ² test. The Shapiro-Wilk test was used to calculate the normality of the sectoral NFL thickness in the normal group. The indexes were evaluated with the area under the receiver operating characteristics (AROC) and the classification accuracy. The AROC comparison was made with the DeLong test.³² We used a cutoff value of 0 to distinguish between glaucoma and MS for easy clinical interpretation. This is very close to the optimal cutoff needed for maximum accuracy based on ROC analysis with Youden's J index.³³ Accuracy was estimated based on the number of eyes correctly classified as MS or glaucoma divided by the number of all eyes. The accuracy comparison was made with McNemar's test. 

Results

A total of 268 eyes were initially included in the study; however, six eyes were subsequently excluded because of inadequate quality OCT scans. Our final enrollment consisted of 58 healthy eyes from 58 participants, 112 MS eyes from 56 participants, and 92 glaucoma eyes from 92 participants. Among the MS eyes, 32 had a history of ON, and 80 had no history of ON. In the eyes with glaucoma, 43 had PPG, and 49 had PG. There were 56 eyes (61%) with HTG and 36 (39%) with NTG. Out of all the eyes with glaucoma, the majority had primary open-angle glaucoma. However, there were three cases (3%) of pigmentary glaucoma and one case (1%) of pseudoexfoliative glaucoma. Compared to MS eyes, the glaucoma eyes had older age, shorter axial length, thinner overall NFL thickness, more negative VF-MD, and higher pattern standard deviation in the visual field test (Table 1). 

Table 1.

 

Participant Characteristics

 

The NFL thickness measurements in the normal group followed a normal distribution, as confirmed by the Shapiro-Wilk test (P > 0.3). We classified the eyes according to the severity of NFL reduction in the worst sector. The normal reference, one-percentile cutoff, and the five-percentile cutoff of eight sectoral NFL thicknesses were obtained from the healthy control group (Table 2). The significant reduction group included 20 MS eyes (eight with ON and 12 without ON) and 61 glaucoma eyes (43 PG and 18 PPG). In the borderline reduction group, there were 24 MS eyes (six with ON and 18 without ON) and 19 glaucoma eyes (six PG and 13 PPG). There was no significant difference in VF-MD between MS and glaucoma eyes within each severity group. 

Table 2.

 

The Sectoral Normal Reference and Cutoffs From The Control Group

 

Converting the thickness to percent reduction made it easier to observe the differing patterns between MS and glaucoma in the significant reduction group (Fig. 2). The worst percent reduction for MS eyes occurred in the TU (mean ± SD, −39% ± 13%) and TL (−39% ± 12%) sectors, whereas for glaucomatous eyes, it occurred in the IT (−46% ± 20%), IN (−37% ± 17%) and ST (−36% ± 23%) sectors. It was typical to see both TU and TL reduction in MS cases and mainly inferior (IT or IN) or superior (ST) sector reduction in glaucoma cases (Fig. 3). 

Figure 2.

 

The boxplots of sectoral nerve fiber layer thickness and percent reduction between MS and glaucoma in the significant reduction group. (A) The boxplot illustrates the distribution of peripapillary NFL thickness across different sectors in the significant reduction group. The plot shows that the sectors with the lowest thickness values are TU and TL in both MS and glaucoma. The patterns of MS and glaucoma were qualitatively similar. (B) The boxplot illustrates the distribution of NFL thickness reduction (µm) across different sectors. The greatest reduction occurs in the IT and ST sectors for both MS and glaucoma. (C) The boxplots illustrate the distribution of NFL percentage reduction across different sectors. There are distinct patterns between MS and glaucoma, especially in the temporal sectors (ST, TU, TL, IT). NU, nasal upper; SN, superior nasal; NL, nasal lower.

Figure 3.

 

Typical NFL % reduction of multiple sclerosis and glaucoma. Two images are present in each set. The left bar plot shows the percentage reduction per sector in a specific case, while the right picture is the en-face OCT image of the NFL slab. The first two sets (A and B) show typical cases of MS with varying degrees of involvement, where the TU and TL sectors experience the most significant percentage reduction. In contrast, the next two sets (C and D) depict typical scenarios of glaucoma. In C, the IT sector has the worst percent reduction in an eye with an inferior NFL bundle defect. (D) The ST sector has the worst percent reduction in an eye with superior NFL bundle defects. NU, nasal upper; SN, superior nasal; NL, nasal lower; PI, pattern index; tPI, temporal pattern index; WSI, worst sector index; tWSI, temporal worst sector index.

For the differentiation between MS and glaucoma, all the indexes performed well, and there is no significant difference between them (P > 0.06). The indexes had the best AROC (0.90–0.96) and accuracy (87.7%–92.6%) in the significant reduction group (Table 3). The performance decreased if the severity decreased, but the index still performed well in the borderline reduction group (AROC = 0.84–0.88, accuracy = 72.1%–81.4%). 

Table 3.

 

Diagnostic Accuracy of Percentage Reduction Indexes in Each Group

 

We analyzed the cases misclassified by three or more indexes to gain insight. Notably, two glaucoma eyes, both of them of the normal tension variety, had NFL thinning mainly in the temporal quadrant, similar to MS eyes (Fig. 4). Of the three misclassified MS eyes in the significant reduction group, two had high myopia (Figs. 5A, 5B), and another eye had a history of a cotton wool spot at the location of subsequent thinning, which is not typical of MS (Fig. 5C). 

Figure 4.

 

Two normal-tension glaucoma eyes were misclassified by all four indexes. They had the worst NFL thinning in the TL sector. The cases were illustrated by sector NFL % reduction bar plots and the en-face OCT image of the NFL slab. NU, nasal upper; SN, superior nasal; NL, nasal lower; PI, pattern index; tPI, temporal pattern index; WSI, worst sector index; tWSI, temporal worst sector index.

Figure 5.

 

Three MS eyes in the significant reduction group were misclassified by most indexes. The cases were illustrated by sector NFL percent reduction bar plots and the en-face OCT image of the NFL slab. (A) An MS eye with high myopia. The spherical equivalent was −6.75 diopters, and the axial length measured 24.8 mm. (B) An MS eye with high myopia. The spherical equivalent is −6.50 diopters, and the axial length measured 25.3 mm. (C) An MS eye with a superior NFL bundle defect, associated with a history of cotton wool spot at the same location. NU, nasal upper; SN, superior nasal; NL, nasal lower; PI, pattern index; tPI, temporal pattern index; WSI, worst sector index; tWSI, temporal worst sector index.

Discussion

Optical coherence tomography is a commonly available imaging technology in eye clinics.³⁴ It is a standard tool for the assessment of glaucoma and other optic neuropathies.^35–37 Thinning of the peripapillary NFL on OCT is a sensitive indicator of glaucoma damage³⁸ and a common finding in all optic neuropathies.³⁶ In glaucoma, NFL defect is most prominent in the arcuate bundles in the IT sector, ST sector, or both. This can be seen on OCT and sometimes directly visualized through biomicroscopy.^18,39 In contrast, NFL defect in MS is often noted temporally in the papillomacular bundle, consistent with temporal pallor in the optic disc rim.^17,40 Temporal loss is common in other nonglaucomatous optic neuropathy as well.¹⁹ Therefore analysis of the NFL pattern is a potential way to distinguish between MS-associated optic neuropathy and glaucoma. 

However, a previous study found no significant differences in the NFL thickness patterns between MS and glaucoma.²⁰ Therefore we seek to develop novel methods of NFL pattern analysis to improve the contrast between the two diseases and optimize the accuracy of differential diagnosis. 

We found that the key to uncovering the distinct difference in the NFL patterns between MS and glaucoma was to convert the sectoral thickness values into percent reduction relative to a normative reference. In the healthy human eye, the NFL is thickest by far in the IT and ST sectors where the arcuate bundles reside. This pattern remains in both MS and glaucoma eyes. Even when the NFL thickness values are converted to NFL reduction (in micrometer units) by subtracting the normative value, we still found that, on average, the IT and ST sectors incurred the greatest µm thinning in both MS and glaucoma eyes. It was only when we converted the thinning to units of percent reduction that we found a distinct difference between MS and glaucoma in the pattern of reduction. On the percent reduction scale, MS eyes tended to have the most severe reduction in one or both of the temporal sectors (TU or TL), whereas glaucomatous eyes tended to have the most severe reduction in either the IT or ST sectors. 

By looking for the sector with the worst percent reduction on the OCT peripapillary scan, it is already possible to distinguish between MS and glaucoma with high accuracy. Our results showed 87.7% accuracy for eyes with significant NFL sector thinning and 72% accuracy for eyes with borderline NFL sector thinning for our temporal worst-sector index. Going beyond simple identification of the worst sectors, combining quantitative information from multiple sectors further adds to diagnostic power, as demonstrated by the temporal pattern index, which achieved a very high performance of 0.96 AROC and 92.6% accuracy in the significant reduction group. 

Because NFL in the nasal hemisphere is relatively spared in both glaucoma and MS,^41–44 including the nasal sectors in the diagnostic indexes did not improve diagnostic power. However, if we are to generalize our approach to include other optic neuropathies, such as compressive optic neuropathy, the nasal sectors would then become important.^45,46 So the selective importance of the temporal hemisphere is a consequence of the selection of the diseases we are targeting in this study. 

An important potential clinical use of our approach is the detection of optic nerve damage in MS patients, many of whom can develop NFL reduction without presenting with ON symptoms and findings.^3,47 In the significant reduction and borderline reduction groups, there were 26 MS eyes without a history of ON, yet our indexes demonstrated remarkable accuracy in classifying these cases. The pattern of NFL reduction in MS patients without ON is similar to those with ON, in agreement with previously published results.⁴⁸ 

There are some limitations to our study. First, only a limited number of patients participated in this study, and the number of eyes with significant NFL thinning is relatively small. Second, there was a significant age difference between the MS and glaucoma cohorts. To account for this, the percent NFL reduction was calculated after adjusting for the effects of age. Third, some participants in the study may have had no or very little optic nerve atrophy. Some MS patients may have had no optic nerve involvement. The PPG group in the study was diagnosed based on disc assessment alone and the diagnosis was not confirmed by structure-function match or longitudinal progression. Clinical disc assessment has limited reliability in establishing optic neuropathy. Therefore, for the assessment of diagnostic accuracy, we only included eyes with at least borderline NFL reduction that was measurable by OCT.^49–51 Although the diagnostic accuracy is excellent in patients with definitive focal NFL thinning, the accuracy decreases in patients with only borderline reduction. 

In addition, myopia distorts the NFL thickness pattern by causing thinning in the superior, inferior, and nasal sectors,⁵² which could reduce the accuracy of diagnostic indexes. Compensation for myopic refractive error or axial length may help resolve this limitation.⁵³ Finally, atypical patterns can present in almost any disease, including MS and glaucoma. Some glaucomatous eyes can have predominant defects at the papillomacular bundle—a pattern similar to that of MS. The frequency of such anomalous patterns has been estimated to occur in around 5% of eyes with early glaucoma.^54,55 This pattern may be more common in those with NTG.⁵⁶ Indeed, we found that our indexes misclassified two NTG patients with predominantly temporal NFL thinning. Thus clinicians should take particular care in interpreting the pattern of NFL reduction in patients with normal intraocular pressure or minimal NFL reduction. In such patients, more attention should be paid to other distinguishing clinical findings, such as the optic disc morphology. Rim thickness measurements with OCT can also provide an additional way to differentiate the glaucomatous and non-glaucomatous optic neuropathy.⁵⁷ Among the MS eyes in our series, there was one eye with a superotemporal NFL defect associated with a cotton wool spot, which may be a very atypical manifestation of MS or other unrelated diseases (e.g., hypertensive retinopathy). The presence of these atypical patterns implies that the NFL pattern cannot be used alone for reliable diagnosis but must be interpreted in the context of other diagnostic information. 

The diagnostic indexes investigated in this study are simple and based on known patterns of NFL reduction in MS and glaucoma that have been described in previous literature. We have already achieved very good diagnostic accuracy. However, it is possible that machine learning and artificial intelligence approaches may achieve even better results. The current study only examined NFL thickness, and additional diagnostic information may be obtained from OCT and OCT angiography measurements of both structure and perfusion of the optic disc rim and the macular ganglion cells. The pattern analysis approach can also be used to differentiate between other types of optic neuropathies and neurodegenerative diseases. We plan to expand our investigation to these additional possibilities in the future. 

In conclusion, it was possible to identify distinct patterns of NFL thinning in MS and glaucoma eyes by calculating the percent reduction in eight peripapillary sectors. Quantitative indexes based on these patterns were found to be effective in distinguishing between MS and glaucoma eyes, provided that significant or at least borderline NFL thinning exists. This is the first article to use percent reduction to find disease-characteristic patterns of NFL defect, and we welcome other investigators to try this approach. Our clinical study has a modest sample size, and larger studies are needed. 

Acknowledgments

Supported by the VA grant CDA IK2RX003407-01A1, Race to Erase MS Innovation award, National Institutes of Health (NIH) NIH grants R01 EY023285, R21 EY032146, P30 EY010572, P30 EY010572 core grant, the Malcolm M. Marquis, MD Endowed Fund for Innovation, and an unrestricted grant from Research to Prevent Blindness to Casey Eye Institute. The sponsor or funding organization had no role in the design or conduct of this research. 

Disclosure: P.-H. Yeh, None; O. Tan, Visionix/Optovue (P, R); E. Silbermann, None; E. White, None; D. Choi, None; A. Chen, None; E. Ing, None; D. Bourdette, None; J. Wang, Visionix/Optovue (P, R), Genentech (P, R); Y. Jia, Visionix/Optovue (P, R), Optos (P), Genentech (P, R, F); D. Huang, Visionix/Optovue (F, P, R), Boehringer Ingelheim (C), Canon (F), Cylite (F), Intalight (F), Genentech (P, R), Kugler (R)