Multiple sclerosis (MS) is a chronic immune-mediated disorder of the central nervous system characterized by multifocal demyelination, axonal injury, and highly heterogeneous clinical manifestations.
The diagnosis of MS has evolved over time. The McDonald criteria, introduced in 2001, were developed to incorporate MRI into the diagnostic framework by providing objective evidence of dissemination in space (DIS) and dissemination in time.1 Over successive revisions, the criteria have increasingly integrated paraclinical biomarkers to enable earlier and more accurate diagnosis, with an emphasis on maintaining specificity and minimizing the risk of misdiagnosis or overdiagnosis.2
The 2024 revised criteria include the optic nerve as a fifth anatomic location for demonstrating DIS.3 This review discusses the utility and limitations of the paraclinical tests used to assess optic nerve structure and function, particularly in the context of MS and the updated diagnostic framework.
Anatomic Overview
The optic nerve originates from the axons of retinal ganglion cells, which converge at the optic disc to form ~1.2 million fibers. These axons constitute the retinal nerve fiber layer. They remain unmyelinated within the retina but become myelinated immediately posterior to the lamina cribrosa, where they are ensheathed exclusively by oligodendrocytes rather than Schwann cells.4 As a result, the optic nerve is unique among the cranial nerves in being myelinated along its entire course by oligodendrocytes, a feature that contributes to its frequent involvement in MS.
The optic nerve has long been recognized as a common site of pathology in MS, with up to 20% of individuals presenting with optic neuritis (ON) as their initial manifestation and an even higher proportion demonstrating subclinical involvement.3,4 Early autopsy studies and clinical observations of optic disc pallor in individuals without a history of ON indicate that optic nerve injury often occurs silently and that MS pathology encompasses both demyelination and axonal or neuronal loss.
Modern paraclinical tools—particularly pattern-reversal visual evoked potentials (VEPs) and optical coherence tomography (OCT)—have substantially improved the detection of subclinical optic neuropathy.5 These modalities are noninvasive, relatively inexpensive, and highly sensitive, although each has limitations and is susceptible to artifacts or misinterpretation. With the 2024 revision of the McDonald criteria recognizing the optic nerve as a distinct anatomic region for demonstrating DIS, these tests assume increasing importance in MS diagnosis.6
This review summarizes the advantages and limitations of OCT, VEPs, and orbital MRI in detecting predominantly subclinical optic nerve involvement in MS.
Demyelinating ON
Acute demyelinating ON is the presenting manifestation in ~20% to 25% of individuals with relapsing–remitting MS and occurs in up to half of individuals at some point during their disease course.7 The diagnosis typically relies on a characteristic patient history and examination, which are well established in the literature.8,9
Despite the generally recognizable presentation of demyelinating ON, misdiagnosis and overdiagnosis are not uncommon. In a recent study, 60% of individuals referred to a tertiary neuro-ophthalmology clinic for presumed demyelinating ON were ultimately given an alternative diagnosis.10 A major contributor to misdiagnosis was misinterpretation of diagnostic testing results, including OCT in some cases.10 Misdiagnosis can lead to unnecessary testing, inappropriate treatment, and emotional and financial burden. In the context of MS diagnosis, inaccurate interpretation of OCT scans or VEP test results may result in erroneously designating the optic nerve as an involved site. Thus, confirming ON—and, by extension, optic nerve involvement as a distinct anatomic region in the McDonald criteria—requires careful integration of the patient history, thorough neuro-ophthalmic examination, and judicious interpretation of diagnostic tests.
Assessment of the Visual System: Overview
Assessment of the visual system necessarily includes evaluation of both structure and function. As in many physiologic systems, structure–function mismatch is common, making it essential to measure both to fully characterize visual pathway injury.
Most functional assessments in ophthalmology—particularly those involving the afferent visual pathways—are psychophysical and rely on individual patient responses. These include visual acuity, color vision, and perimetry tests. The strengths of these assessments are their convenience, widespread availability, and strong correlations with other functional outcomes and quality of life measures.5,11
Two paraclinical tests allow more objective assessment of optic nerve structure and function and are incorporated into the revised MS criteria to determine whether the optic nerve is involved: OCT and pattern-reversal VEPs. These tests are widely available, noninvasive, and relatively inexpensive, but—like all diagnostic tools—have limitations, and should be interpreted within the clinical context. Table 1 summarizes key features of OCT and VEP, their potential roles in early MS diagnosis, and important limitations. The results of these paraclinical tests should be integrated with the history and a careful neuro-ophthalmic examination to ensure accurate diagnosis and avoid misinterpretation.
Optical Coherence Tomography
OCT, first described in 1991, applies low-coherence interferometry to generate high-resolution, in vivo cross-sectional images of the retina. Early time-domain systems introduced in the mid-1990s have been largely replaced by modern spectral-domain and swept-source devices, which provide rapid, reproducible imaging with axial resolution on the order of 3 to 4 μm12 (Figure 1).
Figure 1. Cross-sectional macular optical coherence tomography image illustrates normal layered retinal architecture. The laminar distinctions reflect the highly organized structure of the neurosensory retina and form the basis for quantitative analysis in neuro-ophthalmic disease. Image provided courtesy of Gregory Van Stavern, MD.
Most OCT platforms segment the peripapillary retinal nerve fiber layer (RNFL) and the macular ganglion cell–inner plexiform layer (GCIPL)—the former reflecting axons and the latter neuronal cell bodies and dendrites. However, OCT tests cannot perfectly distinguish axons or neurons from other tissue components, such as gliosis, edema, or blood vessels. Some devices can segment additional retinal layers, although these capabilities vary by manufacturer.13
OCT enables quantitative assessment of axonal and neuronal integrity in the anterior visual pathway and has multiple clinical applications. It is useful in individuals with optic atrophy to document progression, and in those with papilledema to monitor for resolution or worsening. In all settings, OCT serves as a complement to the clinical history and examination.14
A substantial body of literature supports the value of OCT tests in MS diagnosis, particularly in detecting both symptomatic and asymptomatic optic nerve involvement. Key findings include the following:
- Progressive axonal and neuronal loss can be reliably tracked with OCT over time.15,16
- RNFL and GCIPL thicknesses are reduced in people with MS compared with controls, including in eyes without previous ON.17
- Inter-eye differences of ~5 µm (RNFL) and ~4 µm (GCIPL) are characteristic after unilateral ON.18,19
- Longitudinal OCT changes correlate loosely with global MS progression; for example, macular volume loss correlates with brain volume loss.16,20
- RNFL and GCIPL thickness correlate with functional outcomes, particularly low-contrast visual acuity.17,21
Limitations of OCT. As with any diagnostic test, OCT has limitations related to operator technique, patient-specific factors, and inherent device constraints. Poor-quality scans may occur with inadequate cooperation, unstable fixation, small pupils, or media opacity, such as corneal scars or cataracts. Operators must recognize these issues to avoid misinterpretation. Chen and Kardon22 provide an overview and summary of these issues.
There is a well-established relationship between refractive status and OCT measurements.22-24 Individuals with significant axial hyperopia (very short eyes) or axial myopia (very long eyes) may exhibit RNFL and GCIPL thicknesses that fall outside the normative range. Hyperopia greater than +3.00 diopters is generally associated with increased thickness; myopia beyond –5.00 diopters is associated with reduced thickness. These differences typically represent true anatomic variation rather than pathology, but they can complicate interpretation if refractive status is not documented. Formal refraction or lensometer measurement of corrective lenses should therefore be included when interpreting OCT test results.
Segmentation failure is another important limitation associated with OCT tests.22 It may result from poor fixation, media opacity, vitreomacular traction, or substantial retinal or optic disc edema. Swelling can distort retinal architecture enough to prevent accurate boundary detection, producing spuriously low or high values. Figure 2 demonstrates normal OCT scan results; Figures 3 and 4 illustrate examples of refractive and segmentation-related artifacts.
Figure 2. Representative normal optical coherence tomography images obtained using the Zeiss Cirrus high-density optical coherence tomography platform. The peripapillary retinal nerve fiber layer (RNFL) thickness map demonstrates normal global and sectoral values with high signal strength (left). The ganglion cell–inner plexiform layer thickness map shows symmetric macular inner retinal structure (right). Color-coded thickness and deviation maps reflect comparison with the device’s normative database, with green indicating values within normal limits. Images provided courtesy of Gregory Van Stavern, MD.
Figure 3. Ganglion cell–inner plexiform layer thickness map (left) and peripapillary retinal nerve fiber layer (RNFL) thickness map (right) from a patient with high myopia. Both scans show characteristic temporal displacement of the TSNIT peaks, a useful clue to underlying axial elongation rather than optic nerve pathology. Although not shown, signal strengths were within normal limits, and segmentation was anatomically accurate. The diffuse thinning patterns and temporal shifts reflect baseline structure in high myopia rather than glaucomatous or demyelinating optic neuropathy. Images provided courtesy of John J. Chen, MD.
Figure 4. Ganglion cell–inner plexiform layer (GCIPL) thickness maps from both eyes demonstrate severe segmentation failure, characterized by warping and “propeller-like” distortion of the color map and a minimum GCIPL value <40 µm, which is nonphysiologic and strongly indicative of segmentation error. In this case, marked optic disc swelling appears to have distorted the retinal architecture and disrupted layer boundaries, leading to unreliable segmentation. Similar artifacts may also occur with macular pathology, media opacity, or poor fixation. Accurate interpretation requires correlation with B-scans, signal strength, and clinical findings to avoid misclassification as true GCIPL thinning. Image provided courtesy of Gregory Van Stavern, MD.
As with automated visual field testing, OCT interpretation benefits from a brief, systematic review to avoid common errors. At a minimum, patient age and scan quality should be confirmed, as measurements are compared with age-matched normative databases; for Cirrus OCT, a signal strength of ≥7/10 is generally acceptable. Refractive status and axial eye length should be considered, particularly in people with high myopia or hyperopia, pseudophakia, or previous refractive surgery, as these factors can shift RNFL and GCIPL measurements outside normative ranges. Very low focal RNFL or GCIPL values (eg, <40 µm) are more likely due to segmentation failure unless there is longstanding severe optic neuropathy. When assessing progression, comparisons should be made using the same OCT device and scan location, as measurements are not interchangeable across platforms.
Visual Evoked Potentials
VEPs provide an objective measure of retinocortical conduction and were among the earliest tools used to detect demyelination in MS.25 The most robust feature of the pattern-reversal VEP—the preferred and most reliable modality—is the P100 latency, a highly reproducible positive peak occurring ~100 ms after stimulus reversal (Figure 5). Before MRI became widely available, delayed P100 responses provided important evidence of DIS, and multiple studies demonstrated that latency prolongation predicted conversion from clinically probable to definite MS.25 With the advent and rapid evolution of MRI starting in the 1980s, the diagnostic role of VEP declined as neuroimaging achieved greater sensitivity for detecting white matter lesions throughout the CNS.
Figure 5. Pattern reversal visual evoked potential. A black-and-white checkerboard stimulus alternates contrast while an occipital scalp electrode records the cortical response. The resulting waveform includes the characteristic P100 peak, a highly reproducible component occurring ~100 ms after stimulus onset. Prolonged P100 latency is a sensitive indicator of retinocortical conduction delay and visual pathway dysfunction.
Figure created with assistance of AI (ChatGPT; San Francisco, CA).
Limitations of VEPs. As with OCT, several limitations must be considered when interpreting VEP results in the diagnostic evaluation of MS.
The most important limitation is that VEP is nonlocalizing: prolonged P100 latency indicates dysfunction anywhere along the pathway from the retina to the primary visual cortex. Although certain features can suggest anterior visual pathway involvement—for example, a significant interocular difference (>2.5 SD for a given laboratory’s normative database) may indicate unilateral or asymmetric optic nerve injury—the test cannot reliably distinguish optic nerve pathology from retinal (particularly macular) dysfunction, both of which may produce latency delay.25,26
Accurate interpretation therefore requires adherence to standardized testing guidelines and the use of laboratory-specific normative values. Latency prolongation >2.5 SD from the normative mean is generally considered abnormal.26
VEPs depend on patient cooperation, attention, and stable fixation. Individuals may inadvertently “look past” the stimulus, artificially prolonging latency. Refractive error is also relevant; performing the test without appropriate optical correction, particularly in individuals with high hyperopia or myopia, can similarly result in delayed P100 responses.26,27
Orbital MRI
Orbital MRI is the preferred imaging modality for detecting optic nerve pathology because it provides high-resolution visualization of the intraorbital and intracanalicular segments—areas difficult to assess on standard brain MRI. Dedicated orbital protocols rely on fat suppression to remove the bright signal from intraorbital fat that would otherwise obscure the small, low-contrast optic nerve. Effective fat suppression also reduces chemical-shift artifact, improving differentiation between intraneural and perineural abnormalities.28
High-quality orbital MRI typically includes thin-slice (2 to 3 mm), short tau inversion recovery, or fat-saturated T2-weighted sequences, with many centers adding postcontrast fat-suppressed T1 imaging and high-resolution 3-dimensional acquisitions (eg, VISTA, CUBE, SPACE) to enhance depiction of the nerve along its entire course.28,29
Orbital MRI in Asymptomatic Optic Nerve Involvement
Subclinical optic nerve involvement is common in MS, and orbital MRI may demonstrate isolated T2/fluid-attenuated inversion recovery (FLAIR) hyperintensity without gadolinium enhancement (Figure 6). In the appropriate clinical context, such findings may support optic nerve involvement; however, T2/FLAIR hyperintensity alone is nonspecific. In a large retrospective series,30 368 of 698 eyes exhibited increased T2/FLAIR signal without enhancement or atrophy, reflecting diverse chronic optic neuropathies. Some eyes with increased signal had normal clinical examination results, whereas others had severe retinal disease. These findings highlight that isolated T2/FLAIR abnormalities must be interpreted cautiously and are insufficient on their own to establish MS-related optic nerve involvement.
Figure 6. Coronal T2-weighted, fat-suppressed, thin-section MRI scan demonstrates hyperintensity of the right optic nerve, consistent with intrinsic optic nerve signal abnormality. Image provided courtesy of Gregory Van Stavern, MD.
Technical Standards and Guideline Recommendations
The revised McDonald criteria do not recommend routine orbital MRI unless ON is part of the clinical presentation.3 When orbital imaging is performed, recommended protocols include ≥1.5T magnet strength, 2–3 mm slice thickness with no gap, coronal fat-suppressed T2 or short tau inversion recovery sequences, and fat-suppressed postcontrast T1 imaging. Some guidelines suggest including 3-dimensional double inversion recovery sequences, which may increase sensitivity for detecting subclinical lesions.
Proper interpretation requires close attention to protocol quality and correlation with the clinical picture to avoid overcalling nonspecific findings.
Integrating the Optic Nerve Into the Diagnosis of MS
Several studies have examined whether including the optic nerve as a distinct anatomic site improves the diagnostic performance of the McDonald criteria. These analyses apply the revised DIS definition—now incorporating the optic nerve—to earlier cohorts with clinically isolated syndrome (CIS) or clinically definite MS. Table 2 provides a summary of studies incorporating optic nerve assessment into DIS criteria.
In one of the earliest multimodal assessments, Sisto et al31 evaluated visually asymptomatic participants with MS using MRI, VEPs, and automated perimetry. Although the study sample size was small (22 eyes of 11 individuals), a substantial proportion of participants demonstrated subclinical visual pathway involvement: 72% had abnormal MRI scan findings (intracranial rather than optic nerve), 63.6% had abnormalities on automated perimetry tests, and 54.4% had abnormal VEP test results. Some of these measures were nonspecific, but the study highlighted consistent structural and functional abnormalities in participants with MS without a history of ON—findings in line with later work demonstrating asymptomatic RNFL and GCIPL thinning.
Brownlee et al32 evaluated 160 participants with CIS (129 with ON, 31 without ON) and compared the 2017 McDonald DIS criteria with a modified version that incorporated optic nerve involvement. When symptomatic ON was included (clinically or by VEP test), sensitivity increased from 74% to 83% without a change in specificity (77%). However, adding asymptomatic optic nerve involvement did not improve diagnostic performance, likely due to the absence of OCT as an assessment modality in that study.
In a multicenter prospective cohort, Bsteh et al33 assessed 267 participants with MS using modified DIS criteria that incorporated OCT-defined optic nerve involvement based on validated intereye thresholds. Adding the optic nerve as a fifth region improved sensitivity (84.2% vs 77.9%) without reducing specificity (52.2% in both), and shortened the time to a second clinical attack—the study’s primary end point.
Vidal-Jordana et al34 evaluated a prospective CIS cohort in which participants underwent 2 of 3 assessments (optic nerve MRI, OCT, or VEP) within 6 months of symptom onset. Asymptomatic optic nerve involvement was identified in 27.5% of participants with MS without ON. Incorporating optic nerve involvement into the DIS definition increased sensitivity (92.5% vs 88.2%) with a modest decrease in specificity (80.0% vs 82.2%) and essentially unchanged accuracy (86.6% vs 86.5%). Although accuracy may appear to improve when optic nerve involvement is included, accuracy weights true positives and true negatives equally and does not reflect the asymmetric clinical consequences of false-positive MS diagnoses. Given the cost, potential toxicity, and long-term implications of DMTs, even small reductions in specificity may have considerable clinical impact. Thus, incorporating optic nerve findings into DIS criteria requires cautious interpretation, with careful consideration of alternative explanations for OCT asymmetry or optic nerve signal abnormalities.
Summary
In the 2024 McDonald criteria updates, 5 anatomic locations are now recognized for establishing DIS: periventricular, juxtacortical or cortical, infratentorial, spinal cord, and optic nerve regions.3 OCT scans and pattern-reversal VEP tests can be used to detect subclinical optic nerve involvement, providing objective quantitative measures of structural and functional integrity within the visual system. These tests are powerful diagnostic tools, but must be interpreted with an understanding of their limitations and within the broader clinical context.
