Introduction

Optic neuritis (ON) is an acute inflammatory demyelinating disorder of the optic nerve, primarily an isolated phenomenon or secondarily associated with other neurological diseases such as neuromyelitis optica or multiple sclerosis (MS) and represents the most common cause of visual loss in young adults [1].

The most common symptoms include unilateral, subacute visual loss with variable severity, retrobulbar pain worsened with eye movement, dyschromatopsia, without systemic or other neurological symptoms.

The onset of ON is mostly monophasic, but it can also sometimes be polyphasic with recurrent relapses.

ON is most frequent in females (F:M = 3:1) and is usually diagnosed in young adults (20–45 years), with a mean age at onset of 36 years.

ON may be the initial presentation in ~ 20% of patients affected by MS, and up to 50% of patients with MS may experience during their life [2].

ON is normally a self-limiting event, and recovery of visual acuity typically occurs within the first few weeks from the onset, even though persistent visual impairment has been observed in some patients [3, 4].

Visual recovery after an episode of ON is variable and difficult to predict [5]. It has been reported that patients who have symptoms at onset have a poorer prognosis, and this is mainly related to the fact that vision reestablishment starts earlier compared with patients with mild symptoms, but restoration of visual function is still possible [6, 7]. Corticosteroid treatment hastens recovery following ON, but this does not improve final visual function.

Magnetic resonance imaging (MRI) may help ON diagnosis as it allows detecting optic nerve inflammation, to rule out differential diagnoses, and to predict the risk for developing MS [1, 4].

The application of MRI in the assessment of optic nerve damage in a single episode of acute ON has been rarely performed [4], and its role in predicting the prognosis of visual impairment and visual outcome after ON is not clear [4, 6,7,8,9,10].

Optical coherence tomography (OCT) is a noninvasive imaging technique used in ophthalmology to visualize the layers of the retina. It allows the depiction of unmyelinated central nervous system axons within the retina, the so-called retinal nerve fiber layer (RNFL), and provides a measurement of its thickness (micron-scale). The RNFL thickness is used as a noninvasive biomarker of neurodegeneration and axonal loss in MS [11, 12].

Our aim was to assess brain and orbital MRI findings of patients with a first attack of acute ON, and their correlation with visual acuity at presentation, visual recovery, MS development and a possible correlation between orbits MRI pattern and RNFL on OCT.

Methods and materials

This retrospective study has been approved by our Institutional Review Board (Institutional Review Board Area 1 Milan; reference number: 2018/ST/116). All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and national research committee and with the 1964 Helsinki Declaration and its later amendments. Informed consent was obtained from all individual participants included in the study.

Patients

We retrospectively revised the ophthalmological, neurological and imaging data of patients who were admitted to our Emergency Department (ED) with the first episode of acute ON from January 2015 to January 2017 (N = 85).

The exclusion criteria were: patients aged < 18 years; previous episodes of ON; history of MS or optic neuromyelitis (NMO); patients affected by known ophthalmological diseases; patients affected by immunological/infective disorders; patients with previous neurological events; patients with family history of Leber hereditary optic neuropathy; lack of complete ophthalmological, neurological, imaging data; time between symptoms onset and presentation > 14 days; time between ED arrival and MRI execution > 14 days; MRI executed in other Institutions; lack of follow-up data; clinical follow-up < 1 year.

We therefore included in our study 37 patients (age range 21–57 years; mean age 33 years), 26 females and 11 males. Twenty had right ON, 13 had ON in the left eye, and 4 had bilateral ON, for a total of 41 affected eyes.

After the arrival, each patient underwent a complete ophthalmological examination, including fundoscopic examination, visual acuity assessment, visual field and OCT, brain and orbits unenhanced computed tomography, according to the protocol of our Emergency Department, to exclude intracranial and orbital abnormalities, blood tests to exclude infective/autoimmune causes of ON and to test AQP4 antibodies myelin oligodendrocyte glycoprotein (MOG) IgG and cerebrospinal fluid (CSF) collection for detection of the oligoclonal bands, as previously described [13].

Following the neuro-ophthalmological confirmation of the diagnosis of ON, steroid therapy was administered intravenously within 24 h from the patient access (methylprednisolone 1 g/day for 5 days, followed by a low-dose steroid oral regimen for 15 days).

MRI

MRI of brain, orbits, cervical spine was performed after the beginning of the steroid treatment, within 7 ± 6 days from ON onset on the same 1.5 T MRI scanner (Achieva, Siemens Forchheim, Germany) with the acquisition protocol shown in Table 1 with intravenous administration of contrast medium (Gadovist, 0.1 mmol/kg) with an 8-channel head coil [13].

Table 1 MRI acquisition protocol
Full size table

Dorso-lumbar MRI was added in three patients, according to the Neurologist’s indication.

Steroid treatment was started after the diagnosis of ON, before MRI execution.

Imaging exams were transferred to the MRI workstation (Leonardo, Siemens, Forchheim, Germany) and revised by two experienced neuroradiologists in consensus, who were blinded to all clinical and ophthalmologic information such as the affected eye, laboratory results and final clinical diagnoses.

Brain MRIs patterns were classified as (0) normal, (1) non-specific, if millimetric hyperintensities in T2 and FLAIR with location and morphology not typical for demyelinating lesions were observed, (2) lesion with MS-like shape, localized in typical MS sites—periventricular, juxtacortical, infratentorial, or spinal cord—with dissemination in space [14], (3) or lesions with MS-like shape, localized in typical MS sites—periventricular, juxtacortical, infratentorial, or spinal cord—with dissemination in space and coexistence of gadolinium-enhancing and non-enhancing lesions [15, 16]. The number of brain and spinal cord lesions was recorded.

Optic nerves MRIs were reported as (1) normal, (2) STIR—alteration without contrast enhancement (CE), (3) STIR signal abnormalities + CE. Optic nerves pathologic findings were localized into three sites [intra-orbital (IO), canalicular (CA) and chiasmal (CH)], and the length of altered signal or CE was measured in cm.

Visual acuity

Visual acuity was measured at patients’ arrival, according to Snellen chart (decimal). Light perception and motus manus (mm) were assigned a decimal visual acuity of 0.

The degree of visual acuity was further divided into three groups: severe deficit (mm—4/10), intermediate deficit (5–7/10), slight deficit (8/10, 9/10). Each patient was followed-up for at least 12 months (mean follow time: 16 ± 4 months).

OCT

OCT test was acquired using the same machine (Spectralis version 6.3.2, Eye Explorer Software 1.6.1.0, Heidelberg Engineering™, Heidelberg, Germany), after mydriasis induced with 1% tropicamide.

OCT produces a cross-sectional image of the peripapillary retina which is color-coded and shows the temporal, superior, nasal, inferior and temporal sections on the same image.

The software automatically recognizes the retinal nerve fiber layer (RNFL) layer, measures the RNFL thickness along the scan and reports the results on a graphic display showing the RNFL thickness (in μm). We considered the average global value of RNFL and classified OCT results into four groups, according to its value: (1) normal; (2) increased; (3) reduced; (4) borderline.

Follow-up

Ophthalmological evaluation including visual acuity assessment was executed at 6 months to establish the visual outcome. The timing of 6 months for follow-up has been decided in accordance with previous study that described dominant visual recovery within this period [17, 18].

Visual outcome was classified into four groups: (1) complete visual recovery: visual acuity back to the value before ON, (2) partial visual recovery: recovery of visual acuity of at least 50% compared to the value before ON, (3) stable deficit: recovery of visual acuity < 50% compared to the value before ON, or persistence of the visual deficit, as during ON episode, (4) visual worsening.

Patients were followed-up from a neurological point of view for at least 1 year (range 12–29 months) and were diagnosed as MS or clinically isolated syndrome (CIS).

CIS or MS diagnosis was based on revised McDonald criteria [16].

Statistical analysis

Statistical analysis was performed using SPSS (IBM SPSS Statistics for Windows, Version 22.0. Armonk, NY: IBM Corp). Pearson chi-square test for categorical variables was used to assess the correlation between brain MRI patterns and visual acuity, visual outcome and diagnosis, and between orbits MRI patterns and visual acuity, visual outcome and diagnosis. Logistic regression model was applied to variables with p < 0.05.

We then performed Mann–Whitney U test to assess the correlation between the lesions number and the diagnosis, applied the logistic regression model and finally performed ROC curve analysis, and to evaluate the correlation between orbital MRI pattern and extension of the altered signal of the optic nerve.

Kruskal–Wallis test was performed to assess possible correlation between orbits MRI pattern and RNFL thickness at OCT.

Results

Brain MRIs were classified as follows: (1) normal in 9 patients out of 37 patients; (2) millimetric hyperintensities in T2 and FLAIR with location and morphology not typical for demyelinating lesions in 9 out of 37 patients; (3) lesions with MS-like shape, localized in typical MS sites with dissemination in space without contrast enhancement in 13 out of 37 patients and (4) lesions with MS-like shape, localized in typical MS sites with dissemination in space and coexistence of gadolinium-enhancing and non-enhancing lesions in 6 out of 37 patients.

MS-like shape lesions number range was: 0–25, with a mean number of 8 ± 2.

Optic nerves MRIs were reported as: (1) normal in 14 out of 41 eyes (34%); (2) STIR- alteration without CE in 13 out of 41 eyes (31.7%); STIR signal abnormalities + CE in 14 out of 41 eyes (34.1%). Optic nerves pathologic findings were localized into the IO segment in 9 out of 27 eyes with signal abnormalities (33.3%); into the CA segment in 16 out of 27 eyes (59.2%); IO + CA in 2 out of 27 eyes (7.4%), whereas no localization was observed in the CH segment. The length of altered STIR signal or CE measured as follows: extension range: 0.5–2 cm; mean extension: 1.5 ± 0.2 cm. The extent of signal alterations was greater in the optic nerves showing contrast enhancement (p = 0.03; Fig. 1). Three patients showed spinal cord lesions at both cervical and dorsal level; one patient showed one cervical spinal cord lesion.

Fig. 1
figure 1

The box plot shows the extent of signal alterations and was greater in the optic nerves that demonstrated contrast enhancement

Full size image

Visual outcome was classified as: complete recovery in 12 out of 37 patients (32.4%), partial recovery in 16 out of 37 patients (43.2%) and deficit persistence in 9 out of 37 patients (24.3%). No patients in our case series presented a worsening in visual acuity at 6-months follow-up. Mean visual acuity at follow-up was 7 ± 1/10.

Twenty-five patients (67.5%) received a diagnosis of MS; 12 patients (32.4%) received a diagnosis of CIS; in our group no patients received a diagnosis of NMO. All patients with spinal cord lesions received a diagnosis of MS.

We observed statistically significant correlation (p = .001) between brain MRI pattern and diagnosis [p = .004; OR 45.5 (95% CI 3.48–594.6)].

Lesions number was statistically correlated with final diagnosis (p = .001; OR 1.342; 95% CI 1.102–1.631) at Mann–Whitney test (Fig. 2). The presence of four lesions as a threshold has a sensitivity of 87.5% and a specificity of 66.6% for the diagnosis of MS with area under the curve (AUC) of 0.8 (Fig. 3).

Fig. 2
figure 2

1: SM; 2: CIS. The box plot shows the correlations between the lesions number and final diagnosis

Full size image
Fig. 3
figure 3

ROC curve shows the correlation between the lesions number and the diagnosis of MS

Full size image

No statistically significant correlation was found between brain MRI pattern and visual acuity (p = 0.6), neither between orbital MRI pattern and visual acuity (p = 0.4). No statistically significant correlation was found between brain MRI pattern and visual outcome (p = 0.2), neither between orbital MRI pattern and visual outcome (p = 0.9).

At Kruskal–Wallis test, no statistically significant correlation was observed between optic nerve abnormalities extension and visual acuity (p = 0.8), nor between optic nerve abnormalities extension and visual outcome (p = 0.2).

We observed a significant correlation between orbits MRI pattern and RNFL at Kruskal–Wallis test (p = 0.037).

No statistically significant correlation was observed between orbits MRI pattern and final diagnosis (p = 0.6).

Discussion

Different disorders determine the inflammation of the optic nerve, but the form caused by MS related demyelination is the most common [2]. Demyelinating ON represents the first manifestation of MS in about 20% of patients [19] and a diagnosis that brings great concerns to patients who are often young and previously healthy.

The execution of MRI, soon after the onset of symptoms, has important implications: first, MRI helps clinicians to confirm the diagnosis of ON, but it can also provide with useful information for patient management and treatment planning.

In our case series, most optic nerve lesions were in the intra-orbital segment, and this is in line with the evidence of another study by Soelberg et al. [4], but our frequency of optic nerve abnormalities was lower: 65.8% versus 80.6%. Berg et al. [10], in their study on 104 patients observed T2 lesions of the optic nerve in 79.8% of patients and gadolinium enhancement in 74% of patients. In a study on 37 adult patients with a recent or past attack of ON who executed obits MRI with STIR sequence [20], high-signal abnormalities of the optic nerves were found in 84% of symptomatic subjects and poor visual outcome was observed in patients with more extensive lesions, or with lesions located in the optic canal. Our study shows that visual evoked potentials were more sensitive than MRI in detecting lesions and in diagnosing demyelinating disease of the optic nerve; however, we must consider that this study was conducted in 1998 and the MR equipment was certainly less efficient than the currently available scanners.

STIR sequence showed abnormalities in 88% of acute ON in another retrospective study [21].

We observed a lower rate of optic nerves altered signal; however, differences in rates of MRI abnormalities detected at optic nerves can be explained by different timings of MRI execution after symptoms onset and by the beginning of the steroid treatment in our patients: Berg et al., for example, performed the MRI within 24 h form the patient’s arrival, before the corticosteroid therapy.

In accordance with the literature, in our group of patients, all optic nerve abnormalities were in the IO or CA segment without any involvement of the CH portion, which is instead considered typical for neuromyelitis optica spectrum disorder [22, 23].

In our study, we did not demonstrate a significant correlation between orbits MRI pattern and visual outcome at follow-up.

In a prospective study by Hickman et al. [6] in 33 patients, 15 of whom performed serial gadolinium-enhanced orbits MRI until enhancement ceased, short acute lesion on triple-dose gadolinium-enhanced imaging are associated with initial improvement in vision (p < 0.01), but recovery was not related to the duration of enhancement. Due to the growing evidence of gadolinium deposition in the brain, especially in patients with ON (who are usually young and will have to undergo multiple MRIs during their lifetime in case of diagnosis of MS), these results are difficult to compare.

Kupersmith et al. [24] analyzed a group of 107 patients and observed that optic nerves with gadolinium enhancement in the optic canal had poorer color vision, compared with optic nerves with abnormalities in other segments. They also report that that when there was complete involvement of all segments of the nerves the threshold perimetry and color vision were mostly impaired, and that optic nerves with enhancing segments longer than 17 mm had poorer baseline visual acuity, threshold perimetry and color vision. However, visual recovery was similar regardless of location or length of abnormal enhancement at baseline, and this fact is in line with our results.

The extension of optic nerve abnormalities has been proposed as predictor of visual outcome in a study on 22 patients [25], where complete visual recovery was observed in patients with lesions extension less than 17.5 mm, whereas lesions greater than 17.5 mm and/or lesions with intracanalicular location were associated with incomplete or partial recovery.

The intracanalicular location was identified as a risk factor for poor outcome also in a retrospective study of 50 patients acute ON, who executed MRI including STIR sequence of the optic nerve [21], together with initial low visual acuity (less than 2/10) and the absence of orbital pain.

In the attempt to predict visual recovery after ON, Zhang et al. [26] proposed the use of texture analysis in a study on 25 patients with acute optic neuritis: open source software ImageJ was applied for the assessment of texture heterogeneity calculated on the coronal STIR image showing the largest cross-sectional area of the lesion in the optic nerve. They observed that only baseline lesion texture independently correlated with visual recovery at 6 and 12 months, whereas no other MRI variables at baseline, as altered signal length and gadolinium enhancement length, nor ophthalmic variables, as RNFL, correlated with visual outcome. The absence of a correlation between signal abnormalities in STIR and gadolinium enhancement of the optic nerves and visual recovery is in line with our results.

We confirmed that MRI brain lesions at presentation with optic neuritis (ON) increase the risk of developing clinically definite multiple sclerosis. The evidence that the presence of brain lesions at the onset of ON is related to the diagnosis of MS is in accordance with the literature: previous studies have reported that patients with no brain lesions at the onset of ON have a 15–22% risk of MS compared to a 56–88% risk in patients with one or more brain lesions [14, 27, 28]. This evidence has important clinical implications, since these patients may be later at a greater risk of disability [4].

All patients with spinal lesions received a diagnosis of MS, and this agrees with other studies [4, 29].

Another option for the assessment of patients affected by ON is represented by OCT. Data in the literature demonstrated the relationships between RNFL thickness and visual acuity, visual field and color vision, and previous results affirmed that OCT can be used as a noninvasive biomarker of neurodegeneration and axonal loss in MS [30] and of optic nerve atrophy [31, 32]. Moreover, a strong correlation between RNFL thickness and visual outcome has been observed in MS patients [33,34,35], suggesting that a lower RNFL thinning in early MS is related to a better visual outcome [33].We use this ophthalmological technique in combination with MRI examination.

Limitations

Our study has some limitations, first of all the 1-year clinical follow-up: it is indeed possible for the patient to develop MS at a greater distance of time; however, the same follow-up period was also considered in other studies in patients with ON [4, 8, 20, 24]. A longer follow-up is needed to strengthen our results. This is a retrospective study; however, we considered only the patients with complete clinical, ophthalmological and imaging documentation, eliminating from our case series all patients with partial or incomplete data; the patient population is not huge, but comparable to that proposed in other similar studies. Another limitation is the execution of MRI after the beginning of the steroid treatment, since the administration of steroids reduces active lesions contrast enhancement of the optic nerves, brain and spinal cord.

Conclusions

In conclusion, the correlation between orbits MRI pattern at baseline and visual recovery remains unclear, whereas the observation of brain lesions and their number at the first episode of ON can help predict the development of MS. The identification of those patients who may convert to MS is important, as they may later be at greater risk of disability [30].