Региональная уязвимость сетчатки при множественном склерозе: интеграция OCT, MRI и клинических данных для улучшения диагностики и автоматизированного мониторинга.

Introduction

Multiple sclerosis (MS) is a chronic, autoimmune, demyelinating disorder of the central nervous system (CNS) that affects both white matter (WM) and gray matter (GM), leading to a wide range of neurological symptoms. Among the many manifestations of MS, visual disturbances are common, often presented as optic neuritis (ON) with decreased visual acuity (VA) and ocular motility disorders with diplopia. Optical coherence tomography (OCT) has emerged as a powerful, noninvasive imaging technique capable of assessing structural changes in the retina, making it a valuable tool in MS diagnosis and disease monitoring [ 1 ].

OCT was originally developed for ophthalmic applications, particularly in diagnosing and managing retinal diseases such as glaucoma and macular degeneration. However, its utility in MS became apparent when studies revealed a strong correlation between retinal nerve fiber layer (RNFL) thinning and MS-related neurodegeneration [ 2 ]. Since then, numerous investigations have confirmed that OCT-derived parameters, including peripapillary RNFL thickness, inner plexiform layer (IPL) thickness, and macular volume (MV), serve as biomarkers for neuroaxonal loss in MS patients [ 3 ].

Studies have demonstrated that MS patients exhibit progressive thinning of the RNFL and macular layers even in the absence of acute ON, suggesting that OCT can detect subclinical neurodegeneration [ 4 ].

Moreover, longitudinal OCT assessments have provided insights into MS disease progression, with faster rates of retinal thinning correlating with increased disability and brain atrophy seen on magnetic resonance imaging (MRI) scans.

More recent studies have explored the application of spectral-domain OCT and OCT angiography (OCT-A) in MS. These advanced imaging techniques provide enhanced resolution and additional layers of analysis, allowing for more precise measurement of retinal and vascular changes. Research suggests that alterations in the microvasculature detected by OCT-A could serve as early indicators of MS [ 5 ]. Additionally, machine learning (ML) algorithms have been applied to OCT data to improve the early diagnosis of MS and differentiate it from other demyelinating diseases [ 6 ].

A systematic review and meta-analysis further highlighted that OCT parameters correlate with cognitive performance in MS patients, underscoring the potential of retinal imaging as a surrogate marker for global neurodegeneration [ 7 ]. The posterior pole analysis of the retina using the Spectralis® OCT protocol has been found to provide precise data for monitoring MS severity [ 8 ].

One area of growing interest is the specific vulnerability of the temporal quadrant of the RNFL and macular layers in MS patients. Several studies indicate that the temporal RNFL exhibits the most pronounced thinning, particularly in patients with progressive MS [ 9 ]. This phenomenon has been hypothesized to result from selective neurodegeneration within the visual pathways and may correlate with cognitive impairment and motor dysfunction.

Furthermore, comparative analyses between OCT and MRI findings suggest that temporal retinal thinning may be an early marker of trans-synaptic degeneration due to MS lesions along the optic radiations [ 10 ]. These findings reinforce the importance of OCT in complementing MRI as part of a multi-modal approach to MS monitoring.

OCT has established itself as an essential tool in MS research and clinical practice, offering a window into the neurodegenerative processes affecting the retina. With ongoing advancements in high-resolution imaging and artificial intelligence (AI)-based analyses, the role of OCT in MS is expected to expand further. Understanding regional susceptibilities, such as the temporal quadrant’s vulnerability, may enhance our ability to predict disease progression and tailor personalized treatment strategies for MS patients.

Aim

The aim of this study was to perform a comprehensive statistical analysis of a cohort of MS patients in comparison with a matched control group, correlate OCT findings with clinical and MRI parameters, and investigate the underlying mechanisms contributing to the selective vulnerability of the temporal quadrant in MS-related retinal neurodegeneration.

Patients, Materials and Methods

This study was conducted on a cohort of 43 MS patients and 38 healthy controls. The study was approved by the Research Ethics Committee of the University of Medicine and Pharmacy of Craiova, Romania (Approval No. 146/
24.09.2021). Written informed consent was obtained from all subjects prior to their inclusion in the study.

Ophthalmological examination of the patients included the following procedures: assessment of VA using the Tomey TPC 2000 projector with a decimal scale, measurement of intraocular pressure (IOP) via Goldmann applanation tonometry, biomicroscopic examination of the anterior and posterior segments, fundus photography using the Tomey TFC 1000 fundus camera, and OCT examination.

As this was a prospective, observational, comparative study, a control group was selected, consisting of 38 healthy individuals, matched by age and sex to the patient group.

For both the MS group and the control group, the exclusion criteria included: systemic conditions such as diabetes mellitus, decompensated cardiovascular diseases, other neurological disorders, and ocular conditions such as any type of retinopathy, macular degeneration, glaucoma, refractive errors exceeding 4 diopters, and media opacities – including cataract or vitreous abnormalities.

Each subject underwent three separate OCT investigations for three consecutive years, focusing on ganglion cell layer (GCL), MV, optic nerve head (ONH) and RNFL.

Clinical evaluations, including Expanded Disability Status Scale (EDSS) scores and disease duration, were recorded for MS patients.

The OCT scans were performed using 3D Cirrus HD OCT 5000 (Carl Zeiss Meditec, Inc., Dublin, CA, USA) with standardized protocols. The temporal quadrant was specifically analyzed to investigate its vulnerability to MS-related neurodegeneration.

In all patients from both the MS group and the control group, OCT examination was performed using two protocols: macular scanning to obtain macular parameters and data on the GCL+IPL complex, and the peripapillary RNFL protocol, which measures RNFL thickness around the ONH.

Macular area measurements were centered on a 6×6 mm region and obtained via macular cube scanning – macular cube 512×128 or 200×200 – allowing for both quantitative and qualitative analysis of the retina. The early treatment of diabetic retinopathy study grid was automatically centered on the fovea, calculating retinal thickness values (expressed in μm) from the internal limiting membrane to the retinal pigment’ epithelium.

Quantitative and qualitative assessment of the GCL–IPL layer was based on the same macular cube scan, with data collected over a 14.13 mm2 area centered on the fovea. GCL–IPL thickness maps were generated using elliptical measurements centered on the fovea. The Cirrus OCT algorithm enabled segmentation and identification of the outer boundaries of the IPL and RNFL.

RNFL and ONH analysis was based on the optic disc cube 200×200 protocol, which positions the optic disc within a 6×6 mm scanning area. The average RNFL thickness was analyzed for each eye, including distribution across the four quadrants (superior, nasal, inferior, temporal) as well as the clock-hour sectors.

All OCT data were compared against the normative database. Scans with a quality score <7 were excluded from analysis and repeated.

Additionally, MRI data were collected using MRI Ingenia 3T Philips to assess correlations between retinal changes and brain changes. Parameters such as whole brain volume, % WM and lesion volume were logged. MRI was available for only 15 patients.

Statistical analysis was performed using descriptive statistics, including mean and standard deviation. Group comparisons were conducted using Student’s t-test (for two groups) and analysis of variance (ANOVA) (for more than two groups). Pearson’s correlation coefficients were calculated to assess the relationships between OCT parameters, MRI findings, and clinical data. A p-value less than 0.05 was considered statistically significant.

First, a case-control assessment was employed, followed by a longitudinal examination. Samples from the three OCT modalities are presented in Figure 1A , 1B , 1C .

Normal appearance of OCT images across three scan types: GCL+IPL (A), the macula (B), and the RNFL (C), with all sections displaying the full retinal thickness from the ILM to the RPE. The fovea is located at the center of images (A) and (B), while the OD is at the center of image (C). GCL: Ganglion cell layer; I: Inferior; ILM: Inner limiting membrane; IPL: Inner plexiform layer; N: Nasal; OCT: Optical coherence tomography; OD: Optic disc; RNFL: Retinal nerve fiber layer; RPE: Retinal pigment epithelium; S: Superior; T: Temporal

Results

The MS cohort included 43 patients (mean disease duration: 9±7.5 years; mean EDSS: 2.3±1.7; range: 0–6.5). The control group consisted of 38 age- and sex-matched healthy individuals. VA was slightly reduced in the MS group (average: 0.9) compared to controls (1.0). A total of 12 patients experienced ON (five at onset, seven post-diagnosis), while 26 reported diplopia either at onset (n=14) or during disease evolution (n=12). All subjects completed three OCT exams over three years.

All participants underwent the three separate OCT modality scans. For each OCT modality, sample images showing mild, moderate, and severe cases are presented in Figure 2 . Figure 3A , 3B , 3C , 3D , 3E , 3F , 3G , 3H , 3I presents sample images from all OCT modalities.

Expanded Disability Status Scale (EDSS) Class distribution of the patients included in the study

Pathological appearance of OCT images across three scan types: GCL+IPL (A, D and G), macula (B, E and H), and RNFL (C, F and I). All images show the full thickness of the retina. The first row (A–C) illustrates mild changes, with slight thinning of the retinal layers and no clinical symptoms. The second row (D–F) shows moderate pathology, associated with limited clinical effects and a slight reduction in visual acuity. The third row (G–I) reveals severe alterations in the retinal layers, with significant clinical impact, potentially progressing to blindness. GCL: Ganglion cell layer; IPL: Inner plexiform layer; N: Nasal; OCT: Optical coherence tomography; RNFL: Retinal nerve fiber layer; T: Temporal

The results of the first assessment for GCL thickness are presented in Table 1 , macular thickness measurements are detailed in Table 2 , and RNFL thickness findings are summarized in Table 3 .

Table 1 GCL thickness assessment

Eye/Parameter



MS



Control



p
-value (Student’s
t
-test)



RE GCL+IPL thickness [μm]



74.3±8.0



80.6±6.1



0.007



RE GCL+IPL thickness [μm]



70.6±9.4



78.2±7.6



0.006



LE GCL+IPL thickness [μm]



74.2±6.5



80.2±7.8



0.005



LE GCL+IPL thickness [μm]



71.1±6.8



75.5±13.8



0.137

Values are expressed as mean ± SD. GCL: Ganglion cell layer; IPL: Inner plexiform layer; LE: Left eye; MS: Multiple sclerosis; RE: Right eye; SD: Standard deviation

GCL thickness assessment

Eye/Parameter

MS

Control

p
-value (Student’s
t
-test)

RE GCL+IPL thickness [μm]

74.3±8.0

80.6±6.1

0.007

RE GCL+IPL thickness [μm]

70.6±9.4

78.2±7.6

0.006

LE GCL+IPL thickness [μm]

74.2±6.5

80.2±7.8

0.005

LE GCL+IPL thickness [μm]

71.1±6.8

75.5±13.8

0.137

Values are expressed as mean ± SD. GCL: Ganglion cell layer; IPL: Inner plexiform layer; LE: Left eye; MS: Multiple sclerosis; RE: Right eye; SD: Standard deviation

Table 2 Macula thickness assessment

Eye/Parameter



MS



Control



p
-value (Student’s
t
-test)



RE thickness central subfield [μm]



252.9±21.2



252.8±17.7



0.98



RE volume cube [mm

3

]



9.9±0.5



10.1±0.4



0.23



RE thickness average cube [μm]



276.1±14.1



280.0±10.2



0.19



LE thickness central subfield [μm]



253.4±21.5



255.8±18.0



0.63



LE volume cube [mm

3

]



10.0±0.5



10.1±0.5



0.86



LE thickness average cube [μm]



278.0±14.7



278.3±12.8



0.94

Values are expressed as mean ± SD. LE: Left eye; MS: Multiple sclerosis; RE: Right eye; SD: Standard deviation

Macula thickness assessment

Eye/Parameter

MS

Control

p
-value (Student’s
t
-test)

RE thickness central subfield [μm]

252.9±21.2

252.8±17.7

0.98

RE volume cube [mm

3

]

9.9±0.5

10.1±0.4

0.23

RE thickness average cube [μm]

276.1±14.1

280.0±10.2

0.19

LE thickness central subfield [μm]

253.4±21.5

255.8±18.0

0.63

LE volume cube [mm

3

]

10.0±0.5

10.1±0.5

0.86

LE thickness average cube [μm]

278.0±14.7

278.3±12.8

0.94

Values are expressed as mean ± SD. LE: Left eye; MS: Multiple sclerosis; RE: Right eye; SD: Standard deviation

Table 3 RNFL assessment

Eye/Parameter



MS



Control



p
-value (Student’s
t
-test)



RE average RNFL thickness [μm]



89.1±11.8



92.0±8.2



0.21



RE rim area [mm

2

]



1.4±0.3



1.4±0.3



0.88



RE disc area [mm

2

]



2.1±0.4



2.0±0.3



0.21



RE average C/D ratio



0.5±0.1



0.5±0.2



0.46



RE vertical C/D ratio



0.5±0.1



0.5±0.1



0.77



RE cup volume [mm

3

]



0.2±0.2



0.2±0.2



0.71



RE S [μm]



118.8±14.6



114.8±11.8



0.19



RE N [μm]



68.5±13.8



69.7±10.9



0.67



RE I [μm]



109.9±20.4



114.5±15.1



0.27



RE T [μm]



59.3±9.4



68.9±13.1



0.00



LE average RNFL thickness [μm]



90.0±13.1



87.9±13.2



0.51



LE rim area [mm

2

]



1.4±0.3



1.4±0.3



0.71



LE disc area [mm

2

]



2.0±0.3



2.0±0.4



0.72



LE average C/D ratio



0.5±0.2



0.5±0.2



0.55



LE vertical C/D ratio



0.5±0.2



0.5±0.2



0.61



LE cup volume [mm

3

]



0.1±0.2



0.2±0.2



0.20



LE S [μm]



119.6±21.2



116.1±24.1



0.51



LE N [μm]



67.2±10.0



63.7±14.2



0.23



LE I [μm]



116.1±26.1



115.3±19.2



0.88



LE T [μm]



56.5±12.9



64.2±11.4



0.01

Values are expressed as mean ± SD. C/D: Cup-to-disc; I: Inferior; LE: Left eye; MS: Multiple sclerosis; N: Nasal; RE: Right eye; RNFL: Retinal nerve fiber layer; S: Superior; SD: Standard deviation; T: Temporal

RNFL assessment

Eye/Parameter

MS

Control

p
-value (Student’s
t
-test)

RE average RNFL thickness [μm]

89.1±11.8

92.0±8.2

0.21

RE rim area [mm

2

]

1.4±0.3

1.4±0.3

0.88

RE disc area [mm

2

]

2.1±0.4

2.0±0.3

0.21

RE average C/D ratio

0.5±0.1

0.5±0.2

0.46

RE vertical C/D ratio

0.5±0.1

0.5±0.1

0.77

RE cup volume [mm

3

]

0.2±0.2

0.2±0.2

0.71

RE S [μm]

118.8±14.6

114.8±11.8

0.19

RE N [μm]

68.5±13.8

69.7±10.9

0.67

RE I [μm]

109.9±20.4

114.5±15.1

0.27

RE T [μm]

59.3±9.4

68.9±13.1

0.00

LE average RNFL thickness [μm]

90.0±13.1

87.9±13.2

0.51

LE rim area [mm

2

]

1.4±0.3

1.4±0.3

0.71

LE disc area [mm

2

]

2.0±0.3

2.0±0.4

0.72

LE average C/D ratio

0.5±0.2

0.5±0.2

0.55

LE vertical C/D ratio

0.5±0.2

0.5±0.2

0.61

LE cup volume [mm

3

]

0.1±0.2

0.2±0.2

0.20

LE S [μm]

119.6±21.2

116.1±24.1

0.51

LE N [μm]

67.2±10.0

63.7±14.2

0.23

LE I [μm]

116.1±26.1

115.3±19.2

0.88

LE T [μm]

56.5±12.9

64.2±11.4

0.01

Values are expressed as mean ± SD. C/D: Cup-to-disc; I: Inferior; LE: Left eye; MS: Multiple sclerosis; N: Nasal; RE: Right eye; RNFL: Retinal nerve fiber layer; S: Superior; SD: Standard deviation; T: Temporal

A statistically significant reduction in GCL+IPL thickness was observed bilaterally in MS patients, most notably in the right eye (p<0.01), indicating early neurodegenerative involvement. Minimum GCL+IPL thickness in the left eye did not reach significance, possibly due to greater inter-subject variability. Quadrant-based analyses (not included in Table 1 ) confirmed that sectors 2–5 were most affected, pointing to regional GCL thinning rather than uniform atrophy.

No significant global macular thickness differences were noted between MS patients and controls (p>0.05). However, secondary analysis (not tabulated) revealed significant thinning in outer quadrants (see Figure 4 for quadrants reference), suggesting early, localized macular changes that may escape detection with global measures. These findings support the idea that peripheral macular regions are more vulnerable than the central foveal zone.

Macula thickness report example. Macular thickness map. Fundus image. Macular thickness measured between the ILM and the RPE. Table with macular parameters. ILM: Inner limiting membrane; OCT: Optical coherence tomography; OD: Oculus dexter (right eye); OS: Oculus sinister (left eye); RPE: Retinal pigment epithelium

Temporal RNFL thickness showed highly significant reductions in both eyes among MS patients (p<0.01), underscoring the selective vulnerability of the papillomacular bundle. Other quadrants (superior, nasal, inferior) did not differ significantly. Detailed 12-clock-hour analysis (not shown) further reinforced this pattern, highlighting the temporal RNFL as a specific and early biomarker of MS-related degeneration.

To further assess the longitudinal dynamics of OCT changes, the same parameters were compared across three investigations conducted at one-year intervals. The results indicated no significant changes in any of the three OCT modalities for both the MS group and the control group, as evidenced by the ANOVA test, which yielded a p-value greater than 0.05 for all comparisons.

Taking into consideration the results presented above it is clear that there were no evident changes in the MS group within the three years of our study, that the GCL seems to be the most sensitive examination, followed by the temporal zone of the RNFL, while the macula exhibited no significant changes.

Sample OCT results are illustrated in Figures 4 , 5 , 6 , providing representative examples of the structural retinal changes observed in MS patients. Figure 4 showcases the macular thickness maps, which, while not showing major differences globally, reveal subtle peripheral thinning in some patients – particularly in the outer quadrants – supporting the notion of region-specific macular vulnerability. Figure 5 presents the GCL+IPL thickness report, highlighting both absolute and deviation maps that reveal localized thinning when compared to age-matched normative data. These images allow for quick identification of early neurodegenerative changes, even in the absence of overt clinical symptoms. Figure 6 displays the RNFL thickness profile, including quadrant-based and clock-hour segmentation. Notably, temporal quadrant thinning is clearly visible in affected individuals, aligning with the statistical findings and reinforcing its value as an OCT biomarker for MS. Together, these images visually support the quantitative data presented in Tables 1 , 2 , 3 and underscore the diagnostic potential of OCT in detecting subclinical retinal neurodegeneration in MS.

GCL thickness report example. The GCIPL thickness map shows the absolute thickness of the GCL and IPL. The GCIPL deviation map illustrates the deviation from the age-matched normative control group. The GCIPL thickness by sectors exhibits color-coded map for assessing different regions of the GCIPL. GCIPL quantification table includes the average GCIPL thickness compared to normative data for the same age group. GCL: Ganglion cell layer; IPL: Inner plexiform layer; OD: Oculus dexter (right eye); OS: Oculus sinister (left eye)

RNFL report example. The RNFL thickness map shows the absolute thickness of the RNFL. The RNFL thickness deviation map shows comparison with age-matched normative values. Horizontal, vertical and circular RNFL thickness tomograms. Data table with average RNFL thickness and optic disc parameters. Graph of neuro-retinal and RNFL thickness with normative comparison. Thickness graph by quadrants and clock hours. C/D: Cup-to-disc; INF: Inferior; N (NAS): Nasal; OD: Oculus dexter (right eye); OS: Oculus sinister (left eye); RNFL: Retinal nerve fiber layer; SUP: Superior; T (TEMP): Temporal

The MRI findings are presented in Table 4 , with a maximum lesion volume of 31.83 mL.

Table 4 MRI brain assessment

Whole brain volume [mL]



% WM



Lesion volume [mL]



1001±93



24.7±23.6



6.3±7.59

Values are expressed as mean ± SD. MRI: Magnetic resonance imaging; SD: Standard deviation; % WM: White matter percentage

MRI brain assessment

Whole brain volume [mL]

% WM

Lesion volume [mL]

1001±93

24.7±23.6

6.3±7.59

Values are expressed as mean ± SD. MRI: Magnetic resonance imaging; SD: Standard deviation; % WM: White matter percentage

Discussions

This study aimed to evaluate retinal neurodegeneration in MS using OCT and its correlation with clinical and MRI parameters. Our findings reinforce the role of OCT derived biomarkers in assessing neuroaxonal damage and highlight the selective vulnerability of the temporal quadrant in MS-related retinal changes.

GCL and IPL demonstrated the most significant reductions in thickness in MS patients compared to the control group, particularly in the right eye’s minimum thickness. This suggests that GCL thinning is a sensitive indicator of neurodegeneration, aligning with prior studies that emphasize GCL loss as an early marker of MS-related axonal damage. Furthermore, quadrant analysis revealed that quadrants 2–5 exhibited significant differences between MS and control groups, whereas quadrants 1 and 6 did not show notable variation. These results reinforce the importance of localized assessments in MS-related retinal damage.

Macular thickness did not exhibit statistically significant differences between the MS and control groups, though the control group showed slightly higher values. This suggests that MV loss is not as prominent or as early a marker as GCL thinning. However, quadrant-based analysis highlighted significant changes in the outer quadrants, indicating that while central macular thickness remains stable, peripheral macular regions may undergo subtle degeneration in MS.

Analysis of RNFL thickness revealed that the temporal quadrant was the only region with a statistically significant reduction in MS patients compared to controls. This finding supports the hypothesis that the temporal quadrant is particularly susceptible to MS-related neurodegeneration, potentially due to trans-synaptic degeneration linked to damage in the visual pathways.

Despite the differences observed between MS patients and controls, longitudinal analysis showed no significant changes over three years in any of the OCT modalities. The ANOVA test yielded p-values >0.05, indicating that retinal layer thickness remained relatively stable over time. This suggests that while OCT detects neurodegeneration in MS, its progression may be slow or variable, particularly over short observational periods.

MRI findings revealed a maximum lesion volume of 31.83 mL, which showed a strong correlation with EDSS scores (r=0.61), supporting the association between lesion burden and disability severity. A moderate correlation (r=0.42) was observed between lesion volume and GCL+IPL thickness, reinforcing the link between brain atrophy and retinal neurodegeneration. Additionally, a weak correlation was noted between lesion volume and temporal RNFL thickness, further supporting the notion that MS-related damage extends beyond the brain to affect retinal structures.

Summarizing, our study shows that:

(1) GCL+IPL thickness is the most sensitive OCT parameter for detecting MS-related neurodegeneration;

(2) Macular thickness remains largely unchanged, with minor differences only in outer quadrants;

(3) RNFL temporal quadrant thinning is the most distinctive feature, reinforcing its role as an MS biomarker;

(4) No significant changes were observed over three years, suggesting slow progression detectable via OCT;

(5) MRI findings correlate with EDSS and retinal thinning, confirming a link between brain atrophy and retinal damage.

Our findings align with and contribute to the growing body of evidence on OCT-derived biomarkers in MS. The observed thinning of the RNFL, particularly in the temporal quadrant, is a well-documented phenomenon in MS research. Previous studies have consistently highlighted the susceptibility of the temporal quadrant to MS-related neurodegeneration, suggesting a link to trans-synaptic degeneration along the optic pathways. The temporal quadrant appears to be more susceptible to damage and may represent the earliest and most prominent site of involvement in patients with MS, in contrast to the nasal quadrant, which tends to be more resistant [ 11 ]. The underlying mechanism responsible for this phenomenon remains unclear or poorly elucidated [ 12 , 13 ], but a plausible explanation may lie in the axonal composition of the temporal quadrant, which primarily consists of small parvocellular axons within the papillomacular bundle. Additionally, according to several studies, the temporal RNFL exhibits a distinct developmental pattern compared to the other RNFL quadrants. During development, scleral expansion leads to a broader distribution of nerve fibers across a larger surface area, which may contribute to the earlier thinning of the temporal RNFL [ 14 ].

Thinning of the temporal RNFL in MS has been consistently reported in various publications over time [ 7 ]. Parisi et al. (1999) [ 15 ] published the first OCT study in ON, comparing average RNFL thickness in 14 patients with ON and MS to 14 healthy controls. The study was conducted at least one year after the onset of initial ON symptoms and revealed a significant thinning of both the average RNFL and the temporal RNFL quadrant in the ON group compared to the control group. Histopathological findings by Kerrison et al. [ 16 ] also demonstrated RNFL thinning in the temporal sector in MS patients, reinforcing the vulnerability of this region. More recently, Pillay et al. reported that temporal quadrants are particularly sensitive to damage in early MS and ON, with more pronounced changes over time, while the nasal quadrant appears more resistant to RNFL loss [ 17 ]. In a large 3-year prospective multicenter study, Paul et al. also detected a significant reduction in temporal RNFL thickness in MS patients compared to healthy controls, with statistically significant differences in average RNFL thickness [ 18 ].

These findings are consistent with other studies that have documented distinct patterns of RNFL loss in MS patients – with or without a history of ON – and highlighted the relative resilience of the nasal quadrant [ 19 , 20 ].

This is further supported by studies that have explored different MS subtypes, demonstrating that RNFL loss is more pronounced in progressive forms of the disease [ 21 ].

While RNFL thinning remains a widely accepted biomarker, the GCL+IPL thickness has emerged as a more sensitive indicator of early neurodegeneration. Our study supports the notion that GCL+IPL thinning occurs even in the absence of acute ON, a finding consistent with prior research demonstrating that GCL atrophy precedes RNFL loss [ 6 ]. RNFL thinning in the absence of a history of acute ON is believed to result from chronic inflammatory processes within the CNS or subclinical demyelination along the visual pathways. Oxidative stress and mitochondrial dysfunction are also thought to contribute to the ongoing neurodegenerative mechanisms. In systematic reviews and meta-analyses, a limitation has been noted regarding the inclusion of RNFL thickness data, as these values may be confounded by optic disc edema during the acute phase of ON [ 22 ]. Viladés et al. [ 8 ] assessed MS severity using a novel OCT-based tool. Their findings suggest that the GCL is affected prior to RNFL thinning, although detectable changes in GCL appear only after approximately five years. However, the correlation between average GCL thickness and disease severity (measured by EDSS score) did not reach statistical significance. Still, GCL appears to correlate more consistently with disease duration, even from the onset of initial symptoms. There is often a substantial delay between symptom onset and definitive MS diagnosis, during which retinal changes may already be well established. With regard to RNFL findings, the same authors propose that RNFL thickness may serve as a reliable biomarker for disease prognosis and severity.

Several publications have reported significant correlations between EDSS scores and RNFL thickness, but not with GCL thickness, in patients with MS. Similar findings were reported by Shi et al. [ 23 ] who found that visual function and disability were more strongly associated with focal thinning of the GCL+IPL. Eslami et al. [ 24 ] also demonstrated a relationship between RNFL thickness, MV, and MS subtype, noting correlations with disease severity. These findings suggest that the EDSS is not a strictly linear marker of disease progression, as it may fluctuate depending on disease relapses and MS subtype. Nonetheless, areas of correlation between RNFL thickness and EDSS have been observed as early as one year after OCT assessment. Montolío et al. [ 25 ] developed a mathematical model to predict RNFL thinning in MS and found consistent correlations between RNFL loss and EDSS progression over time. Rothman et al. [ 26 ] further demonstrated that retinal measurements predict long-term disability, reporting a significant inverse correlation between MV and EDSS over a 10-year follow-up period. Importantly, several studies have shown that GCIPL thickness offers greater reliability and reproducibility than RNFL thickness when assessing structure–function relationships in MS. This may be explained by the fact that GCIPL measurements are less affected by astrocytic gliosis or optic disc edema, which primarily influence the RNFL.

In support of this, Saidha et al. [ 27 ] documented GCL–IPL thinning across all MS subtypes and concluded that GCL–IPL thickness correlates more closely with visual dysfunction and overall disability than RNFL thickness.

In addition, other studies have found that GCL thinning is correlated with cognitive decline in MS, reinforcing the hypothesis that retinal neurodegeneration mirrors broader neuroaxonal loss in the CNS [ 7 ].

Retinal layer thickness measurements may hold predictive value for disease progression in MS as Christensen et al. [ 28 ] investigated cognitive decline in relation to OCT parameters and MRI changes in patients with relapsing-remitting MS (RRMS). The study found a correlation between GCL+IPL thickness and performance on the Symbol Digit Modalities Test (SDMT), supporting the hypothesis that OCT may serve as a biomarker for cognitive dysfunction. However, GCIPL thickness did not independently predict SDMT performance when MRI measures were also considered, suggesting that OCT has limited utility as a standalone predictor of cognitive performance compared to MRI in RRMS.

Another study [ 29 ] further demonstrated an association between cognitive decline and retinal thinning in newly diagnosed MS patients, including both RRMS and primary progressive MS subtypes. Over a 3-year follow-up, both groups exhibited reduced RNFL and GCL thickness, along with worse cognitive performance. GCIPL was identified as a more accurate marker of neurodegeneration, correlating significantly with cerebral atrophy.

In another study involving 60 MS patients without ON, followed for three years, those with baseline GCIPL thickness <77 μm and RNFL thickness <88 μm showed a 3- to 4-fold increased risk of disability progression. Moreover, patients with a GCIPL thinning rate of >1 μm/year or >1.5 μm/year had a 6-fold higher risk of worsening disability. Interestingly, no significant differences in retinal layer thickness were observed between patients with and without relapses, suggesting that thinning rates were not influenced by relapse activity [ 30 ].

Our findings also revealed that GCL+IPL reductions were most pronounced in specific quadrants, aligning with studies that suggest region-specific vulnerability rather than uniform atrophy across the entire retinal layer. With regard to the quadrant-specific distribution of GCL+IPL thinning, the data obtained in our study were consistent with several previously published findings. A study involving 44 patients with MS (24 with ON and 20 without), 29 patients with idiopathic ON, and 20 healthy controls concluded that the average GCIPL thickness, as well as values in the superior, superonasal, superotemporal, infero-nasal, and inferotemporal quadrants, were significantly reduced across all groups. These findings support the view that GCL–IPL thickness is a more clinically sensitive structural marker than RNFL in patients with early MS, regardless of ON history [ 17 ]. Similarly, Estiasari et al. [ 31 ] observed a significant reduction in average GCIPL thickness particularly in the inferior, inferotemporal, infero-nasal, and superonasal quadrants among MS patients, further highlighting regional vulnerability. In another study, Shi et al. [ 23 ] identified that GCIPL thinning in MS predominantly occurs 1.98 mm nasal and 0.42 mm inferior to the fovea, within a circular region approximately 1 mm in diameter, referred to as the “M zone”. Histopathological studies have demonstrated a preferential loss of small parvocellular axons in MS, suggesting that the “M zone” reflects involvement of the papillomacular bundle. Moreover, mitochondrial dysfunction in this region parallels that seen in mitochondrial optic neuropathies, further implicating the parvocellular pathway in MS-related neurodegeneration.

In contrast to the changes observed in RNFL and GCL+IPL thickness, our study found no significant differences in overall macular thickness between MS patients and controls. This result agrees with studies reporting that MV loss is a later-stage phenomenon and may not be as reliable as RNFL or GCL thinning for detecting early disease-related changes [ 32 ]. However, some studies have suggested that outer macular quadrants may exhibit subtle thinning, which was also observed in our dataset, reinforcing the idea that specific regions within the macula may be more susceptible to MS-related damage than the entire structure [ 33 ].

Retinal morphology, as assessed by OCT, has gained increasing interest as a non-invasive biomarker in MS. In a study conducted by Ziccardi et al. [ 34 ], which included 93 patients with RRMS, no significant reduction in MV was found among patients without a history of ON when compared to healthy controls. The macular area was analyzed in three concentric regions: area 1 (0–1 mm), area 2 (1–3 mm), and area 3 (3–6 mm) from the foveal center. Among patients with a history of ON, significant MV reduction was observed in areas 2 and 3, regardless of the degree of VA recovery. However, in area 1, volume reduction was significant only in those with ON and poor visual recovery, indicating a possible relationship between structural central macular damage and long-term visual impairment.

In a separate cross-sectional study involving 120 MS patients, of which 116 had experienced at least one ON episode, Eslami et al. [ 24 ] reported that total MV was significantly lower in patients with secondary progressive MS compared to those with RRMS. Within the RRMS subgroup, both RNFL thickness and MV were significantly reduced in patients with a history of ON in comparison to those without. Moreover, the authors found that longer disease duration was associated with greater OCT abnormalities, including increased RNFL thinning, MV reduction, and worsening visual deficits. Based on these findings, they hypothesized that RNFL thickness and MV may be associated with EDSS scores and reflect overall disease severity.

The question of longitudinal retinal changes remains a challenge in MS research. Our study found no significant changes in OCT parameters over a three-year period, a finding that has been echoed in some previous research. While RNFL and GCL thinning are well-established markers, their progression may not be linear over short time frames, as demonstrated by studies indicating that significant atrophy may require longer follow-ups to be detected [ 35 ].

The prognostic value of OCT measurements in predicting long-term disability progression in MS has been further supported by a study conducted by Lambe et al. [ 36 ], which followed 132 MS patients (including 106 with RRMS) over a period of 10.4 years. After excluding eyes with a history of ON, the study compared patients with a baseline average GCIPL thickness ≥70 μm to those with GCIPL thickness <70 μm. The results revealed that individuals in the lower GCIPL thickness group had a fourfold increased risk of disability worsening, as measured by a significant increase in EDSS scores over time. Importantly, the study demonstrated that lower baseline GCIPL thickness was independently associated with long-term disability progression, suggesting that early retinal thinning is not merely a consequence of acute ON or disease duration but may reflect a broader neurodegenerative process inherent to MS. These findings highlight GCIPL thickness as a valuable OCT-derived biomarker with potential to inform long-term clinical prognosis and provide insight into global disease burden, even in the absence of overt visual symptoms.

This raises important considerations for clinical practice: while OCT is valuable for cross-sectional assessment, its ability to track disease progression over short intervals may be limited.

Finally, our findings provide further evidence that OCT and MRI findings are complementary rather than redundant. The strong correlation between lesion volume and EDSS scores observed in our study aligns with previous research confirming that MRI remains the “gold standard” for assessing brain atrophy and lesion load [ 37 ].

The relationship between retinal structural changes and brain atrophy in MS has been further explored in a study involving 95 MS patients, which investigated the associations between OCT parameters, regional and global brain volume loss measured by MRI, and both physical and cognitive disability. The study identified significant associations between the average thickness of RNFL and GCIPL and the total brain volume, thalamic volume, and cortical GM volume. However, no such correlations were observed with cerebral WM volume. Importantly, there was a negative correlation between RNFL and GCIPL thickness and T2-lesion volume, particularly in the optic radiations, suggesting that retinal thinning may reflect the burden of posterior visual pathway demyelination. Furthermore, the brain regions whose volumes correlated with EDSS scores and SDMT performance (a cognitive function measure) significantly overlapped with those associated with RNFL and GCIPL thickness, emphasizing the potential of OCT to serve as a non-invasive marker of clinically relevant GM damage in MS [ 38 ].

However, the moderate correlation between MRI lesion volume and GCL thinning supports the growing view that retinal measurements can serve as surrogate markers for neurodegeneration, particularly in patients where MRI monitoring is less feasible or frequent [ 39 ]. Our findings reinforce the importance of multimodal imaging approaches, where OCT can provide additional insights into subclinical neurodegeneration that might not yet be reflected in MRI changes.

The application of AI and ML in the analysis of OCT scans has brought a new level of precision to diagnosing and monitoring MS. Traditional OCT analysis, while effective, often relies on manual measurements or semi-automated software that require clinical interpretation. The introduction of AI-driven segmentation and classification models has enabled a more objective, reproducible, and scalable approach to detecting retinal neurodegeneration associated with MS.

One of the most significant advancements in this field is the use of deep learning (DL) architectures, particularly convolutional neural networks (CNNs), to classify MS patients based on OCT-derived biomarkers. Studies have shown that ML classifiers, including support vector machines (SVMs) and random forests, can accurately distinguish between MS and healthy controls by identifying patterns of RNFL and GCL thinning. These AI models outperform traditional statistical methods by leveraging vast amounts of imaging data to detect subtle structural changes that might be overlooked in routine clinical evaluations [ 40 ].

Automated retinal layer segmentation has also evolved, improving the precision of RNFL, GCL+IPL, and macular thickness measurements. AI-driven OCT platforms, such as Heidelberg Eye Explorer (HEYEX), use advanced segmentation algorithms to detect greyscale differences in retinal structures, offering a high degree of accuracy and consistency in evaluating MS-related neurodegeneration. A recent study found that AI-assisted segmentation methods significantly enhanced early MS detection, even in cases without clinically apparent ON, reinforcing the potential of OCT as a subclinical biomarker of disease progression [ 41 ].

Beyond initial diagnosis, AI models are increasingly being applied to monitor disease progression. Longitudinal retinal assessments have historically been challenging due to slow rates of retinal atrophy over short time frames. However, AI-driven algorithms trained on serial OCT scans have demonstrated an ability to detect microstructural changes in the retina that correlate with disease worsening before they manifest clinically. A study evaluating DL-based progression monitoring reported that AI could predict MS worsening with an accuracy exceeding 90%, particularly when incorporating multicenter datasets [ 42 ].

Despite these promising developments, several challenges remain in the widespread adoption of AI-driven OCT analysis for MS. One key issue is inter-device variability, as different OCT systems employ distinct imaging protocols and segmentation algorithms. This heterogeneity complicates the development of generalizable AI models, necessitating the inclusion of multicenter, multimodal datasets to ensure robustness across various clinical settings [ 43 ]. Another challenge is the “black box” nature of DL models, which can produce highly accurate results without clear explanations of the decision-making process. Efforts are being made to develop explainable AI (XAI) models that offer greater transparency in how predictions are made, ensuring that clinicians can interpret and trust AI-assisted OCT assessments [ 10 ].

As AI-driven tools continue to advance, the integration of automated OCT analysis into routine MS monitoring seems increasingly viable. The ability to detect early neurodegenerative changes, track disease progression, and predict disability worsening makes AI a powerful adjunct to traditional clinical and imaging assessments. Future research should focus on validating these models in large-scale, prospective clinical trials, ensuring that AI-assisted OCT analysis meets the reliability and clinical utility standards necessary for widespread adoption. If successfully implemented, these technologies could revolutionize patient care, offering a non-invasive, cost-effective, and highly accurate tool for personalized MS management.

This study has several limitations. The sample size is modest, limiting generalizability, and the three-year follow-up period may be too short to capture significant retinal changes in MS progression. Larger, long-term studies are needed to confirm the findings. While OCT is a valuable tool, it provides only structural insights and lacks direct assessment of neuroinflammatory processes. The absence of OCT-A also prevents evaluation of microvascular changes that may be relevant in MS. Additionally, variability in OCT measurements due to differences in segmentation algorithms and operator techniques could influence results, despite standardized protocols. The study’s cross-sectional design limits the ability to establish causal relationships between OCT changes, disability progression (EDSS), and MRI findings. Future research should incorporate multi-modal imaging and AI-driven analysis to enhance accuracy and predictive capabilities.

Despite these limitations, the study underscores the potential of OCT as a non-invasive biomarker in MS, particularly in assessing GCL+IPL loss and temporal RNFL thinning. Further longitudinal studies and automated diagnostic tools are needed to refine its clinical applications.

Overall, our study contributes to the existing body of literature by confirming established trends while providing additional insights into regional retinal vulnerability in MS. The temporal quadrant’s selective thinning, the prominence of GCL loss, and the limitations of macular thickness as a biomarker align with previous studies and further emphasize the need for targeted OCT assessments. Future research should focus on longitudinal studies with larger cohorts to determine the optimal role of OCT in tracking disease progression and treatment response in MS patients.

Conclusions

This study reinforces OCT as a crucial tool for assessing retinal neurodegeneration in MS, particularly through GCL+IPL thinning and temporal RNFL loss. While macular thickness showed no significant differences, the selective vulnerability of the temporal quadrant provides further insight into MS-related retinal damage. The lack of significant longitudinal OCT changes over three years raises questions about the optimal timeframe for detecting progression, suggesting that longer follow-ups or AI-enhanced analysis may be necessary. The moderate correlation between MRI lesion volume and retinal changes supports a multimodal approach that integrates OCT, MRI, and clinical assessments for a more comprehensive evaluation of MS neurodegeneration. Given the promise of AI-driven OCT analysis, future studies should focus on developing standardized, automated diagnostic tools to enhance early detection and disease monitoring. Further longitudinal and multimodal research is needed to refine OCT’s role in personalized MS management.

Conflict of interests

The authors declare that they have no conflict of interests.

References