Introduction

Multiple sclerosis (MS) represents a potentially severe cause of disability throughout adult life [1]. Alongside with chronic inflammation of the central nervous system (CNS), neurodegenerative processes may be present since the early stages of MS, constituting the primary substrate of irreversible disability [2,3,4,5,6,7,8].

As part of the CNS, the optic nerve (ON) is a major target of MS [3]. Optical coherence tomography (OCT), and particularly macular segmentation using spectral-domain-OCT (SD-OCT), represents a sensitive easy-accessible tool to investigate retinal damage in MS, and is currently considered a reliable prognostic biomarker of neurodegeneration [9,10,11,12,13]. Peripapillary and macular retinal nerve fiber layer (pRNFL and mRNFL), and macular ganglion cell/inner plexiform layer (mGCIPL) thinning results from axonal loss and neuronal damage of the inner retina. They correlate with visual function, global disability, and brain atrophy [14,15,16,17], and reflect MS-related ON neurodegeneration [11]. Conversely, inner nuclear layer (INL) thickening is linked to CNS inflammation, possibly due to the formation of microcystic macular oedema [18, 19].

The reduction of RNFL and mGCIPL results primarily from dying-back axonal loss in the ON after retrobulbar acute optic neuritis (ONe) [3], and is accelerated in MS patients exhibiting clinical-radiological disease activity [20,21,22]. Nevertheless, RNFL and mGCIPL thinning are described in MS even in the absence of clinical ONe [23, 24]. This is possibly due either to subclinical inflammation of the retina/ON, or to lesions in the visual cortex/optic radiations that lead to retinal structural changes (retrograde trans-synaptic degeneration) [9, 25]. Independently from optic neuropathy, several studies have shown that a primary process targeting retinal neurons may also act in MS, in a way analogous to the early damage of the gray matter (GM) [26, 27]. In a post-portem study, Green et al. observed retinal atrophy beyond the macular ganglion cell layer (mGCL) in MS patients and hypothesized that outer retina cell loss could be the consequence of a direct immune-mediated process [28]. Saidha et al. described a subset of MS patients with normal pRNFL but reduced macular thickness, not attributable to retrograde degeneration [29].

We recently observed in a small group of newly diagnosed MS without ONe (MSNON) patients a significant mRNFL and mGCIPL thinning compared to healthy controls (HC), without a concomitant reduction of the pRNFL [30]. Furthermore, we proposed cerebrospinal fluid (CSF) β-amyloid1–42 (Aβ) levels as a predictive biomarker of disease progression in MS. CSF Aβ levels were found to predict patients’ expanded disability status scale (EDSS) increase at 3 and 5 year of follow-up [31], and to correlate with GM atrophy and white matter (WM) damage [31, 32]. In this scenario, a possible cut-off value of CSF Aβ levels to identify patients with a worse prognosis has been proposed [32, 33].

Given these premises, the main aim of this study was to evaluate longitudinal changes in OCT measurements over a 12-month period in a cohort of newly diagnosed relapsing-remitting (RR)-MSNON patients in comparison with HC. Moreover, we evaluated whether CSF Aβ levels at diagnosis may be related to the structural retinal imaging findings, which are important biomarkers for early recognition of neurodegeneration in MS.

Materials and methods

Subjects

Table 1 summarized the main demographic and clinical characteristics of the population. Twenty-three patients with a new diagnosis of RR-MS according to the 2017 revisions of the McDonald criteria [34] were recruited and followed-up for 12 months.

Table 1 Demographic and clinical characteristics of the included population
Full size table

All patients underwent lumbar puncture (LP) at baseline. Neurological evaluation with assessment of the EDSS, ophthalmological exam, SD-OCT, and brain magnetic resonance imaging (MRI) were performed at baseline and repeated after 12 months. Exams were performed before starting any treatment, including corticosteroids. After diagnosis the patients started a MS specific treatment, as specified in Table 1. No evidence of disease activity (NEDA) status was assessed at follow-up for each recruited subject as the absence of clinical relapses, increase in disability (as measured by EDSS), and new or enhancing lesions on their MRI scans. Ten age- and sex-matched healthy volunteers (HC) were also included and underwent ophthalmological visit and SD-OCT at baseline and follow-up.

Eyes with a history of ONe were excluded from the study, at baseline as well as the follow-up. Visuospatial abilities were carefully evaluated, with no evidence of impairment requiring further specific investigation. Regarding a potential subclinical ophthalmological involvement, subjects’ visual acuity and color vision were evaluated. All participants with a refractive error greater than 5.0 diopters (D), media opacity, systemic conditions that could affect the visual system, history of ocular trauma, or concomitant ocular diseases (including glaucoma or other known optic neuropathy) were excluded.

The current study was approved by the Institutional Review Board of the Fondazione IRCCS Ca′ Granda, Ospedale Maggiore Policlinico (Milan, Italy). All MS patients and HC gave their written informed consent for this research before entering the study.

Optical coherence tomography

OCT imaging in all subjects were performed with a SD-OCT device, using the built-in segmentation software of the Heidelberg Eye Explorer, version 1.10.2.0 (Spectralis, Heidelberg Engineering, Germany). All the sequences were acquired using the machine follow-up acquisition mode. The standard scanning protocol used at the baseline of the study included 61 high-speed B-scans and each scan was approximately 8.5 mm in length and spaced 118 μm apart. All 61 B-scans were acquired in a continuous, automated sequence, and covered a 30°× 25° area. A minimum of 20 frames were averaged automatically and used to obtain a good quality image. The central fixation target was used to center the raster scan to the fovea. Color-coded retinal thickness maps were generated automatically by the built-in software of the device by applying the ETDRS grid on the fovea and measurements of the retinal thickness were recorded. The ETDRS grid divides the macula into 3 concentric rings (center, inner, and outer), with the inner ring measuring 1 to 3 mm and the outer ring measuring 3 to 6 mm of diameter (referring to a ring with a diameter of 1-mm centered on the fovea). The grid further divides inner and outer rings into 4 quadrants (superior, inferior, temporal, and nasal) (Fig. 1). All individual retinal layers were measured with the new SD-OCT automatic segmentation explorer mapping software of the device (Heidelberg Engineering, Heidelberg, Germany). Good scan quality and automatic segmentation were assessed prior to the analysis by a trained ophthalmologist, and poor-quality images were rejected. The automated retinal segmentation software was applied to determine thicknesses of the following parameters: mRNFL, mGCL, macular inner plexiform layer (mIPL), macular inner nuclear layer (mINL), macular outer plexiform layer (mOPL), macular outer nuclear layer (mONL), and macular retinal pigment epithelium layer (mRPE). The mGCIPL was determined by combining the mGCL and mIPL parameters. For each layer, average thicknesses were calculated within ETDRS grid quadrants and were compared between the two groups. Using the follow-up acquisition mode, pRNFL thickness was measured with a 12° circular scan around the optic nerve with the activated eye tracker. The global average pRNFL thickness and the temporal sector RNFL thickness (pRNFL-T) were evaluated.

Fig. 1
figure 1

Spectral-domain optical coherence tomography (SD-OCT) follow-up analysis. Left column a: example, pRNFL thickness measurement follow-up analysis. First row: baseline exam; second row: follow-up exam.

Right column b: example, follow-up analysis of the automatic retinal segmentation follow-up analysis, specifically GCL thickness and color-coded thickness map. First row: baseline exam; second row: follow-up exam

Full size image

CSF collection and Aβ determination

CSF samples were collected by LP in the L3/L4 or L4/L5 interspace. CSF samples were centrifuged at 8000 rpm for 10 min. The supernatant was aliquoted in polypropylene tubes and stored at − 80 °C until use. CSF cell counts, glucose, and proteins were determined. Albumin was measured by rate nephelometry. CSF Aβ was measured using a commercially available sandwich enzyme-linked immunosorbent assay (ELISA) kit (Fujirebio, Ghent, Belgium). According to our previous findings [40], we considered the threshold value of 813 pg/mL to divide patients into two groups (Aβhigh versus Aβlow).

Statistical analysis

Data were reported as mean (± standard deviation) and number (percentage) for continuous and categorical variables, respectively. Group differences between MS patients and HC were tested by Chi-square test for sex and Wilcoxon rank-sum test for age.

A generalized estimating equation (GEE) model was performed to assess differences between groups for RNFL variables (Table 3) investigated, in order to take into account the correlation between observations originated from the same patient (right and left eye). In the GEE model, the RNFL variables were used as dependent variables and an unstructured correlation matrix was used as correlation structure.

We also compared between-group changes at baseline and 12 months with a mixed model, using time and group (MS patients and HC) as fixed factors, interaction between time and group, patient as variable with a random effect, and an unstructured covariance matrix.

A p value lower than 0.05 was considered statistically significant. SAS 9.3 (Institute Inc., Cary, NC, USA) and R software (version 3.5.0) were used for all the analysis.

Results

Forty-two eyes of 23 RR-MS patients, without impairment of vision, color vision, and visuospatial abilities, were evaluated at baseline and follow-up, together with 20 eyes of 10 age- and sex-matched HC. Four eyes of 4 patients were excluded from the analysis due to previous history of ONe. At follow-up, 14 patients (27 eyes) were classified as NEDA, whereas 9 patients (15 eyes) do not reach all NEDA criteria. No lesions in the anterior visual pathway were observed at brain MRI at both time points of the study. A detailed list of all OCT measurements at the two time points of the study is listed in Table 3.

At baseline, mRNFL, mGCL, mIPL, and mGCIPL thickness values were significantly reduced in MS patients compared to HC (p = 0.03, p = 0.004, p = 0.02, and p = 0.008, respectively, Fig. 2), whereas no differences were found in pRNFL and pRNFL-T thickness. The longitudinal evaluation after 12 months confirmed that mRNFL, mGCL, mIPL, and mGCIPL thickness are reduced in MS patients compared to HC (p = 0.05, p = 0.008, p = 0.01, and p = 0.01, respectively, Fig. 3). At follow-up, mIPL was found significantly thinned compared to baseline in MS patients only (p = 0.02), and there was a trend towards mGCIPL and mINL thinning (p = 0.067 and p = 0.061, respectively). Follow-up revealed pRNFL and pRNFL-T loss in MS patients (both p = 0.005) and not in HC, but their thickness values still did not result significantly different in the two groups, except for a trend towards pRNFL-T thinning (p = 0.09). At both baseline and follow-up, mINL, mOPL, mONL, and mRPE did not differ between MS patients and HC, and mRNFL, mGCL, mOPL, mONL, and mRPE revealed no longitudinal changes in the two groups.

Fig. 2
figure 2

Baseline differences in mRNFL (a), mGCL (b), mIPL (c), and mGCIPL (d) thickness (in μm) between patients and controls. mRNFL, macular retinal nerve fiber layer; mGCL, macular ganglion cell layer; mIPL, macular inner plexiform layer; mGCIPL, ganglion cell layer-inner plexiform layer, HC, healthy control, MS, multiple sclerosis

Full size image
Fig. 3
figure 3

Differences in mRNFL (a), mGCL (b), mIPL (c), and mGCIPL (d) thickness (in μm) betwenn patients and controls at follow-up (12 months after baseline). mRNFL, macular retinal nerve fiber layer; mGCL, macular ganglion cell layer; mIPL, macular inner plexiform layer; mGCIPL, ganglion cell layer-inner plexiform layer; HC, healthy control; MS, multiple sclerosis

Full size image

No difference in all OCT measurements was found at baseline between active and non-active MS patients. Nevertheless, patients who do not reach all NEDA criteria at follow-up had a significant pRNFL, pRNFL-T, mGCL, mIPL, and mGCIPL loss (p = 0.026, p = 0.006, p = 0.017, p = 0.006, 0 = 0.014, respectively) that was not observed in patients classified as NEDA. EDSS score at follow-up was found to be associated with the thinning of mGCL (β = 0.029, p = 0.017) and mGCIPL (β = 0.47, p = 0.03) in MS patients.

We split our cohort of patients into two groups based on CSF Aβ levels (Aβlow under 813 pg/ml and Aβhigh over 813 pg/ml) [39]. According to their CSF Aβ levels, 9 MS patients were Aβlow (18 eyes) and 14 Aβhigh (24 eyes). No significant differences were found between Aβlow and Aβhigh subgroup of patients at baseline, i.e., age, gender, disease duration, and EDSS score (Table 2). Aβlow MS patients displayed reduced pRNFL-T and mRNFL thickness values as compared with Aβhigh already at baseline (p = 0.003 and p = 0.03, respectively), and the same result was observed longitudinally at follow-up for pRNFL-T (p = 0.003; Fig. 4), but not for mRNFL. However, the longitudinal evaluation revealed a significant mRNFL loss in the Aβlow group (p = 0.05) that was not observed in the Aβhigh group.

Table 2 Baseline demographic and clinical characteristics of MS patients divided according to their CSF Aβ levels
Full size table
Fig. 4
figure 4

Box plot showing pRNFL-T thickness (in μm) in Aβlow and Aβhigh multiple sclerosis patients at baseline and after 12-month of follow-up. pRNFL-T, temporal sector peripapillary retinal nerve fiber layer; Aβ, β-amyloid1–42

Full size image
Table 3 OCT data at baseline and follow-up of the included
Full size table

Discussion

In this study, we first confirmed that, at very early clinical stages, mRNFL, mGCL, mIPL, and mGCIPL thickness values were significantly reduced in MSNON patients compared to controls, whereas there were no differences in pRNFL thickness. Secondly, the longitudinal assessment after 12 months demonstrated pRNFL and pRNFL-T loss in MS patients and not in HC, without a significant and concomitant progressive thinning of mGCIPL.

Different studies have described longitudinal data regarding OCT finding in MS patients, showing both pRNFL and inner retinal thinning. According to a recently published systematic review and meta-analysis [12], the reported annual pRNFL atrophy rate was higher in MSON eyes (− 0.91 μm/year) compared with MSNON eyes (− 0.53 μm/year) [34], while the highest annual atrophy rate in MSNON patients was found in those who had a shorter disease duration [8]. A plateau effect was observed in patients with a longer disease duration (> 20 years).

Consistent with the data regarding the RNFL [12], atrophy of mGCIPL was more severe in MSON eyes than in MSNON eyes. Inner retinal layers thinning preceded pRNFL loss: most of the studies showed significant thinning of the GCL within the first month from symptoms onset [35], while RNFL atrophy became detectable only after at least 3 months.

Likewise, in our longitudinal study, we found a significant thinning over the time of both pRNFL and pRNFL-T in MS patients in comparison with HC, with an atrophy rate of − 2.5 μm/year, higher than the previous studies. Regarding the inner retinal layers longitudinal data, we found a decrease of their thickness (mRNFL, GCL, IPL, and GCIPL), although it was statistically significant only for IPL (atrophy rate 0.5 μm/year). According to our results, in MSNON patients pRNFL loss was more evident than inner retinal layers throughout the course of 12 months of the follow-up time.

We speculate that these data can reflect (1) the progressive, incipient axonal atrophy, that is known to happen later in time; (2) the plateau of the GCIPL atrophy, which was already detectable at the very beginning of the disease (the baseline of the study). The difference in annual atrophy rates of this study compared with previous ones (− 2.5 μm/year vs. − 0.53 μm/year) could be explained by differences in the demographic data, since the disease duration of our population was only 18 months, significantly shorter than the disease duration of the other cohorts.

Since a significant difference of pRNFL thinning between patients and HC was not found, we hypothesize that the amount of atrophy that occurred over a period of 12 months was not enough to reach the statistical significance. A longer follow-up is needed to support these hypotheses. Furthermore, considering that some studies conducted over a longer period of time described the progressive thinning of both pRNFL and inner retinal layer, we speculate that the plateau of the inner retinal atrophy might be transient in time, possibly mirroring the relapsing-remitting nature of the disease. An alternative hypothesis, or at least additional, that could explain our findings is based on the assumption that the ganglion cells’ damage could potentially precede peripapillary axons abnormalities in MS patients. In other words, the earliness of ganglion cell thinning may reflect a primary process targeting retinal neurons’ bodies before affecting the peripapillary axons. As already proposed [30], it is likely that previous studies reported a concomitant thinning of GCIPL and pRNFL because their results were not obtained during the very early stages of the disease. In line with this framework, we speculated that the retina itself may be a primary target of degenerative processes in MS, possibly in combination with inflammatory mechanisms, and as a finding of this study, we supported the previous hypothesis that retinal damage could begin from the macular ganglion cells and might not be only a consequence of inflammatory attacks to the anterior optic pathway. Supporting this theory, the thinning of the IPL observed in this study could reflect a retrograde atrophy secondary the ganglion cell loss.

In our view, a pRNFL-T significant loss with a concomitant plateau of mGCIPL thinning, already in the very early stages of the disease, is the most significant finding that seems to emerge from our study, as if to suggest a subsequent involvement of the peripapillary axons shortly after affecting the retinal neurons’ bodies.

In agreement with the previous findings in the literature [8, 9, 29], we did not find a significant progressive INL thinning, although a trend was observed. However, controversial results have been reported. Indeed, a thickening of this layer has been previously described and some authors tried to explain these differences by the occasional presence of macular oedema. Furthermore, although the INL has been described as a physiological barrier to retrograde trans-synaptic axonal degeneration [7], in human retinal post-mortem samples, extensive loss of neurons in the INL has been demonstrated after about 20 years of disease duration. Most studies suggested that the outer retinal layers do not differ in either MSON eyes or MSNON eyes compared with control eyes and that they do not change over a period of time. Likewise, we could not find any statistical difference or any progressive thinning of these layers over 12 months.

Interestingly, EDSS score was found to be associated with the thinning of mGCL and mGCIPL at follow-up. Several studies have examined the relationship between OCT measures and EDSS scores and inconsistent results have been reported [8, 22, 26]. Nevertheless, in a study performed in 132 patients, Saidha et al. [29] demonstrated that the GCIP thickness in MS patients correlated better than pRNFL thickness with EDSS score. Similarly, in our study, patients without all NEDA criteria at follow-up had a significant inner retinal layers loss that was not observed in patients classified as NEDA. Further follow-up time is needed to better elucidate these correlations over the time.

Moreover, we recently described a relationship between low CSF Aβ levels and worse prognosis in MS [31,32,33, 36, 37]. In line with these findings, Aβlow MS patients interestingly displayed a reduced internal retinal layers thickness in its entirety, compared to Aβhigh, although without reaching statistical significance for each layer. More in details, the most significant data concerned the reduced pRNFL-T thickness values in the Aβlow subgroup compared to Aβhigh, both at baseline and at follow-up. mGCIPL thickness values appeared reduced already at the baseline in MS patients compared to HC, although without significant differences between the two patient subgroups. We speculate that since mGCIPL are reduced at very early clinical stages, no difference could be detected among affected subjects. Taken together, this data seems to confirm a worse prognosis in Aβlow patients compared to Aβhigh, due to an early retinal involvement. In line with previous findings and speculations [31,32,33, 36, 37], the link we describe between CSF Aβ concentration and early retinal volume loss suggests the retinal layers may represent a site particularly susceptible both to inflammation and neurodegeneration.

There are some limitations when considering our study. First, we acknowledge that this represents an exploratory study and that a larger cohort of patients will be needed to confirm our findings. Second, it would be necessary to prolong the follow-up time to clarify the longitudinal changes of retinal layers in MS. Third, data from visual evoked potentials were not available. Further studies combining neurophysiological, neuroradiological, and OCT data are needed.

In conclusion, in this study we documented that, at very early clinical stages, mRNFL, mGCL, mIPL, and mGCIPL thickness values were significantly reduced in MSNON patients without a concomitant pRNFL thinning. The longitudinal assessment after 12 months demonstrated a pRNFL-T significant loss in MSNON patients compared to HC, together instead with a sort of plateau of mGCIPL thinning. Lastly, Aβlow subgroup of MSNON patients showed a significant reduction of retinal nerve fiber layer thickness compared to Aβhigh subgroup, already at the baseline.