As mentioned previously, dropout of ganglion cell axons located in the RNFL, as well as of ganglion cell neurons themselves (located in the GCL), is primarily felt to be the derivative of inflammatory axonal transection and chronic demyelination related to optic neuropathy. Moreover, neuronal dropout within the INL of MS eyes has also been demonstrated on postmortem examination, although whether the etiology of this finding relates more to retrograde transsynaptic degeneration as opposed to primary neuronal mechanisms of retinal injury remains unclear. Additional evidence supporting the occurrence of deeper retinal pathology in MS is derived from a number of electroretinography (ERG) studies. In a study of 27 advanced MS patients, light-adapted and dark-adapted flicker ERG demonstrated a reduction of 2 or more standard deviations in some or all components of the ERG responses. ERG findings of eyes with chronic optic atrophy in MS patients were not found to be the same as ERG abnormalities in eyes with surgical section of the optic nerves, raising the possibility that detected ERG abnormalities in MS patients may not be the derivative of optic neuropathy and therefore not related to retrograde transsynaptic degeneration [45]. Another study compared ERG under photopic conditions to pattern reversal visual evoked potentials (PR-VEP) in MS patients with demyelinating optic neuropathy, and patients felt to have monocular demyelinating optic neuropathy. Although optic atrophy resulting from section or optic nerve compression has not been reliably found to result in diminution in amplitudes of flash ERG [46–48], patients with demyelinating optic neuropathy in this study demonstrated early and selective attenuation of the b wave (mostly thought to receive contributions from bipolar cells in the INL), which did not seem to correlate with PR-VEP results. Given the disconnect between PR-VEP (primarily thought of as reflecting the integrity of the optic nerve) and detected ERG abnormalities (predominantly thought to reflect retinal integrity) in this study, it was postulated that the derivative of the ERG findings may be independent of optic nerve pathology [49]. ERG was also included as part of a study investigating MS patients for possible retinal autoantibodies [50]. Standard and bright flash ERG was performed in 34 MS patients. Bright flash a-wave, cone b-wave, and rod–cone b-wave implicit times were significantly delayed in MS patients, and the amplitude of the sums of photopic oscillatory potentials was also significantly reduced in MS patients. No correlation was detected between the ERG parameters and visual acuity, contrast sensitivity, color vision, or visual fields. Detected ERG abnormalities in this study implied dysfunction in several retinal layers but particularly so in the photoreceptor layer, and again, the study findings raised the possibility that the etiology of the ERG findings may not be related to optic neuropathy but instead primary retinal pathology.

The RNFL, GCL, and INL atrophy observed in ocular histological studies of MS patients has been based on qualitative assessment [40, 41]. Qualitative analyses of the plexiform layers (inner and outer plexiform layers), as well as deeper retinal structures such as the outer nuclear layer, photoreceptors, and retinal pigment epithelium in MS eyes, are lacking. Quantitative pathologic ultrastructural retinal analysis has not been previously performed in MS eyes and may relate to postmortem retinal detachment, rapid degradation of retinal tissue, difficulty with obtaining MS eyes for postmortem assessment, and challenges surrounding fixation of retinal tissue among other potential reasons. Similarly, clinicopathologic correlation of ultrastructural retinal changes from eyes of MS patients is lacking. The majority of retinal pathologic descriptions in MS are limited by being restricted to end-of-life analysis and as a result are not necessarily informative regarding the pathobiological mechanisms underlying observed findings or their temporal evolution.

The retina, although unmyelinated, appears to be a frequent site of inflammation, blood–retinal barrier disruption, and neuronal loss in MS. Retinal perivascular inflammation (periphlebitis), indicating blood–retinal barrier disruption, is estimated to occur in up to 20 % of MS patients [40, 51]. This ophthalmoscopic finding refers to visible cuffs of inflammatory cells surrounding retinal veins analogous to cellular infiltrates seen around cerebral veins in prototypical MS lesions [52]. There may be a tendency for active retinal periphlebitis to occur simultaneously with blood–brain barrier disruption in MS [53], retinal periphlebitis may be related to overall MS severity, and may be a risk factor for the development of relapses and gadolinium-enhancing lesions, as well as an increase in T1 lesion volume during follow-up [54, 55]. Retinal periphlebitis may occur at the same time as acute MSON, and its presence in the setting of acute MSON may confer a greater risk for conversion to clinically definite MS in the future [56]. Intermediate uveitis, particularly of the pars planitis subtype, occurs in up to 16 % of MS patients [57]. Consistent with clinical observations, postmortem analyses reveal retinal inflammation with activated microglia in MS eyes [41]. Intriguingly, these collective findings raise the contentious possibility that myelin may not be necessary for instigating or propagating inflammation in MS and highlight the unmyelinated retina as an opportune site within which to study inflammation and neurodegeneration in MS.

First reported in 1991, optical coherence tomography (OCT) is a noninvasive imaging technology that uses near-infrared light to generate cross-sectional or 3-dimensional (3D) images of tissues such as the retina [58, 59]. This enables quantification of the thicknesses of retinal structures with very high resolution (<10 μm) [60]. Ophthalmologists have been using this technology for some time in the evaluation and management of a variety of disorders, including glaucoma and age-related macular degeneration among many more. In neurologic disorders, the most relevant structure conventionally imaged by OCT has been the peripapillary RNFL (RNFL located around the optic disk where the thickness of the RNFL is in general regarded as being at its greatest). In addition to providing a measurement of axonal integrity through the measurement of RNFL thickness, the other conventional measure OCT has traditionally provided is total macular volume (TMV) or average macular thickness (AMT). Since the macula is thought to be relatively enriched by ganglion cell bodies, OCT-derived measures of TMV/AMT have been used in the past as estimates of neuronal integrity in the retina [61, 62]. However, it is worth noting that AMT/TMV is entirely nonspecific and simply represents composite measures of all of the intervening layers of the retina [10]. Except in situations where total retinal thickness is of interest, AMT and TMV have been largely superseded by specific measurements of discrete retinal layers within the macula, as generated by modern intraretinal layer segmentation techniques.

The cross-sectional or 3D images produced by OCT are generated through the measurement of magnitude and echo time delay of backscattered light from an optical beam scanned across tissue such as the retina [9]. OCT may be regarded as the optical analog of ultrasound B-mode imaging. Contemporary OCT devices were based on Michelson’s principle of low-coherence interferometry [63, 64]. With this technique, measurements are performed by the use of a Michelson-type interferometer with a low-coherence-length, superluminescent diode light source. One arm of the interferometer directs light and collects the backscattered signal from the object of interest. A second reference arm with a reflecting mirror is mechanically controlled to vary the time delay and measure interference. The use of low-coherence-length light means that interference occurs when the distance traveled by the light in the sample and reference arms of the interferometer matches to within the coherence length. This allows echo delays of light from a tissue to be measured with a high degree of accuracy. The resulting data is a representation of optical backscattering in a cross section or volume of tissue.

Within the past decade, OCT technologies have dramatically evolved, leading to the generation of OCT technology known as “Fourier domain” or “spectral domain” OCT. In spectral domain OCT, light echoes are detected by measuring the interference spectrum of infrared light by the use of an interferometer with a stationary reference arm [65]. This advanced OCT technology detects light echoes simultaneously, allowing high-speed scanning to be performed. Most current commercial spectral domain OCT devices achieve axial image resolution of approximately 5 μm, with dramatically faster imaging speeds than earlier generations of OCT technology [66–69]. Currently, the fastest commercially available spectral domain OCT (Spectralis OCT, Heidelberg Engineering, Heidelberg, Germany) has imaging speeds of approximately 40,000 axial scans per second—roughly 100 times faster than older time domain OCT.

OCT is inexpensive, reproducible, well tolerated, fast, painless, and easily repeatable. The high resolution of the images generated by OCT enables both quantitative and qualitative assessment of retinal structures. Typically OCT scan acquisition takes less than 5 min and can be performed by nonmedical personnel. Intervisit intraclass correlation (ICC) coefficients for average RNFL thickness using spectral domain OCT in healthy controls have been shown to be approximately 0.97 [70]. Modern high-speed, high-definition, spectral domain OCT is capable of rendering extremely high-resolution images, from which the individual retinal layers can be discriminated. This has led to the development of precise and fully automated segmentation algorithms allowing objective, reliable quantification of distinctive retinal layers in the macular region, in addition to conventional peripapillary RNFL thickness [9, 10, 71, 72]. Such techniques are now transitioning into clinical practice, enabling real-time quantification of the macular RNFL, combined GCL and inner plexiform layer (GCIP), combined INL and outer plexiform layer (INL), and ONL including the photoreceptor segments (ONL), in addition to peripapillary RNFL thickness [71]. Intervisit ICCs for average GCIP, INL, and ONL thickness using spectral domain OCT in healthy controls have been shown to be approximately 0.99 (0.99 in MS patients), 0.91 (0.94 in MS patients), and 0.93 (0.92 in MS patients), respectively [71].

The first study to report OCT findings in MS eyes was reported in 1999 [73]. In this pilot study using first-generation OCT technology, 14 MS patients with a history of acute MSON with a complete recovery of visual acuity and 14 age-matched controls underwent OCT. In eyes previously affected by acute MSON, the RNFL thickness was reduced by an average of 46 % as compared to the eyes of healthy controls and by an average of 28 % when compared to the opposite “unaffected” eye of the same patient. In eyes without acute MSON history (“unaffected” eyes), the RNFL thickness was reduced by an average of 26 % relative to the eyes of healthy controls. While RNFL thickness was found to be associated with pattern ERG, no association with visual evoked potentials (VEPs) was detected. However, subsequent studies have found associations between OCT-derived measures and VEPs. In 2005, one of the earliest studies of OCT in RRMS patients with a history of acute MSON (as well as patients with a history of isolated acute MSON) to also assess macular volume was performed [74]. In eyes previously affected by acute MSON, the RNFL thickness was reduced by an average of 33 % when compared to the eyes of controls and by an average of 27 % when compared to the opposite “unaffected” eye of the same patient. Furthermore, TMV was reduced by an average of 11 % when compared to the eyes of controls and by 9 % when compared to the opposite “unaffected” eye of the same patient. In this same study, RNFL thinning correlated better with VEP P100 amplitude (thought to mainly reflect axonal integrity) rather than P100 latency (thought to mainly reflect myelin integrity), supporting the concept that reductions in RNFL thickness may be attributable to axonal degeneration. The literature is now replete with similar findings reproduced across a multitude of different studies, with many studies consistently showing that both RRMS eyes with and without a history of acute MSON have a reduced RNFL thickness as compared to healthy controls.

OCT-derived findings of reductions in RNFL thickness in RRMS confirm what is already known from postmortem studies: that subclinical optic neuropathy affects MS eyes and that similar but accelerated RNFL thinning occurs following acute MSON. Along these lines and since the advent of OCT retinal layer segmentation, there are now several studies showing reductions in GCIP thickness in MS eyes, both with and without a history of acute MSON, again with greater reductions in GCIP thickness similarly observed in eyes with acute MSON history [71, 75, 76]. RNFL thickness has been found to correlate cross-sectionally with visual function, visual quality of life, and expanded disability status scale (EDSS) scores [72]. Since the RNFL represents the axonal tracts of ganglion cell neurons, GCIP reduction is believed to be the result of retrograde axonal degeneration of the retinal nerve fibers. In other words, reduction in both RNFL and GCIP thicknesses is thought to be the derivative of the same pathologic processes—namely, optic neuropathy. Despite their similarity (the RNFL and GCIP may be regarded as biological extensions of one another), OCT-derived measures of GCIP thickness seem to have superior structure–function relationships to those of RNFL thickness [76]. This appears to be the case with high-contrast (100 %) and low-contrast (2.5 and 1.25 %) visual acuity, as well as EDSS scores. This may relate to (1) better reproducibility, (2) astrogliosis within the RNFL confounding RNFL measurements, and (3) reduced susceptibility of the GCIP to edema during optic nerve inflammation [70, 71, 75]. In a study of 132 MS patients, of which the majority had RRMS (73 %), the correlation coefficient between OCT-derived RNFL thickness and 2.5 % low-contrast letter acuity scores was 0.38, and with 1.25 % low-contrast letter acuity scores, it was 0.39 [76]. In this same study, the correlation coefficient between OCT-derived GCIP thickness and 2.5 % low-contrast letter acuity scores was 0.49, and with 1.25 % low-contrast letter acuity scores, it was 0.52. GCIP thickness also correlated better with EDSS than RNFL thickness in this same study.

Although cross-sectional findings of OCT-derived RNFL thickness and indeed GCIP thickness in RRMS have been relatively consistent across studies, longitudinal findings have been less consistent. Although baseline RNFL thicknesses in a recent longitudinal study (consisting of 164 MS patients, of which 116 had RRMS) were consistent with thicknesses observed in previous studies [72–74, 77, 78], the rate of RNFL thinning (−0.21 μm/year) was lower than that observed in other longitudinal studies and was not significantly different from healthy controls. In one study of more than 1000 MS patients (83 % had RRMS), the rate of RNFL thinning in MS eyes was −2.0 μm/year [79], and in another study of 166 MS patients (94 % RRMS), it was −2.7 μm/year [80]. In the latter study, RRMS patients with disease durations <10 years, as well as those experiencing non-ocular relapses, were interestingly found to have faster rates of RNFL thinning. However, in addition to the more recent study, two other studies similarly failed to demonstrate a significant decrease in RNFL thickness during follow-up of MS patients. This being said, in one of these studies, the cohort was primarily composed of PPMS patients, and in the other study, only 37 MS patients (of which 27 had RRMS) were investigated [81, 82]. Discrepancies in RNFL change across studies may relate to differences in cohort characteristics, statistical analyses employed, and differences in the use of disease-modifying therapies between patients in studies. Although these longitudinal studies excluded patients who developed acute MSON during follow-up, failure to capture and exclude all patients that developed acute MSON, as well the possible effects of subclinical and/or subradiological inflammation of the optic nerves, may also be a contributor to these discrepancies.

There is a paucity of studies examining longitudinal change in GCIP thickness in RRMS and MS in general. In the recent longitudinal study mentioned previously in which significant RNFL thinning was not observed during follow-up, it is interesting that significant GCIP thinning was observed during the course of follow-up [83]. Moreover, GCIP thinning was found to be accelerated in patients exhibiting non-ocular relapses, new T2 lesions, and new gadolinium-enhancing lesions during the course of study follow-up. Patients exhibiting disability progression were also found to have faster rates of GCIP thinning. Furthermore, rates of GCIP thinning were faster in patients with disease duration <5 years, which may reflect a greater availability of retinal ganglion cells for neurodegeneration earlier in the course of the disease, or alternatively a greater tendency for inflammatory disease activity to occur earlier in the disease process. Rates of GCIP thinning were augmented when these independent factors were present in combination. For example, patients with new T2 lesions, new enhancing lesions, and disease durations <5 years exhibited 70 % faster rates of GCIP thinning. The association between faster rates of GCIP thinning and disease activity outside of the visual pathway may raise the possibility that microscopic (subclinical and subradiological) inflammation occurs in the optic nerves. This could be a reflection of more diffuse opening of the blood–brain barrier (than can be appreciated clinically or radiologically), including in the optic nerves, during disease activity in RRMS. In this context, it is interesting to consider that RNFL thickness increases during acute MSON, the basis of which is thought to be related to a combination of interstitial edema and axonal swelling related to impaired transport mechanisms [9]. On the other hand, at least compared to fellow eyes, GCIP thickness is not thought to be increased during acute MSON (Fig. 8.2) [75]. Since optic nerve inflammation is associated with RNFL swelling, but not GCIP swelling [75], subclinical inflammation within the optic nerves (which may not be detectable by MRI) could theoretically result in the pseudo-normalization or swelling of RNFL thicknesses, thereby underestimating true rates of RNFL thinning. The absence of GCIP swelling during optic nerve inflammation, as well as minimal astroglial influence on GCIP thickness measures (the retinal astrocytes are predominantly located in the RNFL), may contribute toward the better reproducibility and lower variance of GCIP over RNFL thickness measures and help explain the discrepancy in RNFL and GCIP findings observed in this longitudinal study. In keeping with evidence-based characterizations of postmortem visual system analysis in MS [40, 41], OCT has provided replete in vivo evidence of RNFL and GCIP thinning in RRMS eyes, irrespective of a history of acute MSON.

Consistent with ERG and postmortem findings [41, 45, 49, 50, 84], OCT segmentation demonstrates quantitative INL and ONL abnormalities in MS. Although the basis for INL and potentially ONL involvement (the latter of which has not been confirmed to occur in MS eyes pathologically, although is suggested on the basis of ERG and OCT studies in MS in vivo) remains to be determined, highly conspicuous observations raise the possibility that a primary retinal neuronal mechanism of pathology may be operative in the anterior visual system in a limited, but not necessarily insignificant, proportion of MS patients. OCT identification of INL and ONL thinning in MS eyes without a prior history of optic neuritis, in which there is relative preservation of the RNFL and GCIP, has been referred to as the macular thinning predominant (MTP) phenotype of MS and may occur in up to 10 % of MS patients, of which the vast majority appear to have RRMS (Fig. 8.3) [71]. The MTP phenotype may be associated with more rapid accumulation of sustained disability in MS, as evidenced by higher multiple sclerosis severity scale (MSSS) scores, as well as unique visual symptoms not typical of optic neuropathy including excessive glare and photopsia [71]. The thinning of INL and ONL that is a characteristic of the MTP phenotype may be pathobiologically distinct from those that characterize tissue injury within the RNFL and GCIP related to optic neuropathy—although this and indeed even the existence of an MTP MS phenotype remain the subject of debate.

Although ocular histopathological examination reveals dropout of INL neurons in up to 40 % of MS eyes [41], qualitative assessment of OCT scans derived from MS patients in vivo reveals macular microcystoid changes in a small proportion (approximately 5 % of patients), which are predominantly localized to the INL (Fig. 8.4) [85, 86]. Macular microcystoid changes appear to mainly occur in the eyes of patients with RRMS and seem to occur more commonly in eyes with a prior history of acute MSON. The etiology and significance of macular microcystoid changes in RRMS eyes have been the source of lively scientific debate. Some proponents have raised the possibility that macular microcystoid changes may represent sites of primary retinal inflammation, on the basis that 25 % of patients with macular edema demonstrating fluorescein leakage examined in an ophthalmological study exhibited similar appearing macular microcystoid changes [87], that macular microcystoid changes in MS eyes may be dynamic in nature over time, and that they may be a harbinger of more aggressive MS. Given that macular microcystoid changes may be dynamic in nature, some authors have suggested that increased INL thickness in the absence of visible macular microcystoid changes may enable capturing of the same process. As such, a longitudinal study of 164 MS patients (123 RRMS) found that increased INL thickness at baseline predicted disease activity (non-ocular relapses, formation of new T2 and/or contrast-enhancing lesions) and disability progression during follow-up, suggesting that processes within the INL might somehow be relevant or related to global disease activity in RRMS and more specifically inflammation [86]. On the other hand, some investigators have proposed that macular microcystoid changes in MS eyes are of little significance, since they may be seen in other neuroinflammatory disorders (including neuromyelitis optica and chronic relapsing inflammatory optic neuropathy) as well as noninflammatory disorders such as neurofibromatosis, Leber’s hereditary optic neuropathy, and glaucoma [88–91]. Given the array of conditions in which macular microcystoid changes have been described, some consider them a final common pathway of retrograde degeneration, especially since there are histopathological descriptions of cavitation occurring within the INL following optic nerve transection dating back to the 1960s [44]. Moreover, some investigators feel these macular microcystoid changes may simply represent mechanical changes resulting from vitreomacular traction [90]. This being said, a recent study utilizing a mechanical model to examine for the presence of vitreomacular traction in OCT scans demonstrating macular microcystoid changes did not find support in favor of this hypothesis [92].