Optic radiations are thinner and show signs of iron deposition in patients with long-standing remitting-relapsing multiple sclerosis: an enhanced T₂^*-weighted angiography imaging study

We investigated 51 consecutive RRMS patients (nine patients with a disease duration [DD] < 2 years [group Les], 42 patients with a DD ≥ 2 years [group Mor]) with visual impairment and without a history of clinical optical neuritis (ON) in both eyes. RRMS was diagnosed according to the 2016 revised criteria [16]. None of these patients had been previously diagnosed with a cerebrovascular disease, other neurological diseases or cardiovascular morbidity. Additionally, none of the patients had a contraindication for having a contrast agent injection as part of an MRI examination. The patients were recruited from the Neurological Department of the First Affiliated Hospital of Chongqing Medical University (Chongqing, China) from September 2013 to August 2016. All the patients were clinically assessed on the day of the MRI using the Expanded Disability Status Scale (EDSS) [17]. Chun Zeng and Silin Du examined all the patients. For comparison, 51 consecutive healthy subjects (group Con) were enrolled in this study. According to the institutional guidelines, all subjects signed an informed written consent form approved by our institutional review board before undergoing MRI.

All the patients underwent brain scan using a 3.0-T magnetic resonance (MR) system (GE Medical system, Milwaukee, WI, USA) and an eight-channel phased-array head coil. A standard protocol for MS studies was performed that included axial dual-echo proton density (PD)-T₂-weighted imaging (T₂WI), and axial fluid-attenuated inversion recovery (FLAIR). In addition, post-contrast T₁WI was reperformed again for all patients after injecting 0.1 ml/kg body weight of gadolinium contrast agent (Bayer-Schering Pharma, Milan, Italy).

Before injecting gadolinium, 3-D ESWAN data were was acquired with eight echoes using the following parameters: repetition time (TR) = 60 ms, effective echo time (TE) = 6 ms, number of excitation (NEX) = 0.75, field of view (FOV) = 22 cm × 22 cm, matrix = 448 × 320, receiver bandwidth = +62.5 kHz, and flip angle = 20°. The sequence was acquired with 2-mm-thick contiguous sections and no space. All scans were oriented parallel to the anterior-posterior commissural (AC-PC) line with 56–64 locations on the middle sagittal plane and covered the entire brain area.

The lesions affecting the OR without enhancement were hypointense on T₁WI and hyperintense on correponding T₂WI and FLAIR, and were confirmed by one clinical neuroradiologist with more than 10 years of experience in clinical and scientific MS imaging (*BLINDED*), who was not aware of the paraclinical and clinical data. Virchow-Robin spaces were excluded by their tubular appearance and FLAIR hypointense.

All ESWAN images were post-processed using research software (FuncTool 6.3.1 software; GE Healthcare) on an ADW4.4 workstation (Sun Microsystems, Santa Clara, CA, USA). A high-pass filter was applied in the subsequent processing steps to reduce aliasing artefacts, which arise predominantly in images involving air-tissue interfaces and background field inhomogeneities. The next step was to create a new type of phase image, which was referred to as a phase mask. Separate acquisition of phase and magnitude data allowed postprocessing to be performed offline instead of using the MRI system. Offline processing used the SPIN software (Signal Processing In NMR; http://​www.​mrc.​wayne.​edu/​download). For paramagnetic substances, an increase in magnetic field led to a negative phase relative to that of the surrounding parenchyma and CSF. First, an observer adjusted the brightness and contrast of each of the phase and magnitude images to obtain improved definition of the anatomical reference structures, consequently allowing greater accuracy in the measurement. Second, the ORs were magnified to the same degree for a clearer definition of the OR margins. Third, the mean phase values (MPVs) of the OR were measured on the phase images (Fig. 1), and thicknesses were measured and signals were observed on the magnitude images (Fig. 2). For each subject, three continuous axial slices characterised by a high inter-individual comparability were selected: (a) one slice tangential to the inferior splenium of the corpus callosum and superior to the vein of Galen, (b) one slice parallel and inferior to the first slice, and (c) one slice parallel and superior to the first slice. To ensure data accuracy, all MPVs and thicknesses were the mean value for three slices [14]. All post-processing of structural MRI was performed three times with repeated measurements by an observer (Yongmei Li), and the final result was the mean value of the three measurements.

Phase images served as a direct measure of the magnetic field variation based on the following formulas: ф (phase) = -r ΔBTE (where r signifies gyromagnetic, ΔB is the change in the magnetic field between tissues and TE is the echo time); ΔB = cVΔxBo (where c is the iron content, V is the voxel volume and Δx is the variation in molar susceptibility between tissues. When iron is present, at a given TE, the more iron that exists in the tissue, the more the phase will decrease. Therefore, the phase contrast will depend on how much iron is present [18]. The characteristics of the OR thickness and signal were evaluated and categorised into the following three types: (1) bilateral thinning of the OR with a heterogeneous signal; (2) unilateral thinning of the OR with a heterogeneous signal; and (3) a bilateral OR with a normal thickness and homogeneous signal. MPVs and thicknesses of the OR were averaged between hemispheres.

One hundred and eight non-enhancing lesions, which were hypointense upon T₁WI and hyperintense upon T₂WI, were observed in the OR regions of 33 patients (26 patients in group Mor and seven patients in group Les). Of the 33 patients, 18 in group Mor and two in group Les had bilateral OR lesions (Fig. 3) and eight patients in group Mor and five patients in group Les had unilateral OR lesions (Fig. 4). No lesions were detected in the OR regions of the remaining 18 patients (16 patients in group Mor and two patients in group Les).

In healthy controls, the OR showed bilateral symmetrical high-signal bands (HSBs) on the phase images (Fig. 1) and LSBs on the corresponding magnitude images (Fig. 2) that were lateral to the lateral ventricles.

The average MPVs for the OR were 1,981.55 ± 7.75 in group Mor, 1,998.45 ± 2.01 in group Les, and 2,000.48 ± 5.53 in group Con (Table 2). Group analysis revealed that this difference was driven by the DD of the patients. The MPV of the OR was significantly lower in group Mor than in group Les (p < 0.05) and group Con (p < 0.01), and the MPV of the OR was significantly lower in group Les than in group Con (p < 0.01).

Pearson’s correlation analysis showed a remarkable negative correlation between the DD and MPVs of the OR in the RRMS patients (R = 0.91, p < 0.001); However, no correlation was found between the MPV and EDSS (p > 0.05).

In group Mor, 28 patients with bilateral OR lesions showed thinning of the bilateral OR with a heterogeneous signal (Fig. 5) and 14 patients with unilateral OR lesions showed thinning of the ipsilateral OR with a heterogeneous signal (Fig. 6). In the group Les, for the remaining nine patients without OR lesions, the bilateral OR appeared normal with a homogeneous signal.

The average thicknesses of the OR were 3.2 mm, 3.8 mm, and 3.8 mm in group Mor, group Les, and group Con, respectively. There was no significant difference in the thickness of the OR between group Con and group Les (p > 0.05). Additionally, the thickness of the OR was lower in group Mor than in group Con (p < 0.05).

Pearson’s correlation analysis showed a remarkable negative correlation between the DD and thickness of the OR in the patients (R = 0.9, p < 0.001); however, no correlation was found between the thickness and EDSS in RRMS patients (p > 0.05).

Our results showed that the OR was slightly hyperintense on the phase image of ESWAN in both the patients and the controls, whereas the average MPV of the OR in the patients was lower than that in the controls. Chen et al. [19] and Du et al. [20] previously had demonstrated a negative correlation between MPV and brain iron concentration in MS patients. Thus, we speculated that a small number of iron depositions that cannot be distinguished with the naked eye may exist in the OR in patients. Recently, a study found that the phase value, which is indicative of greater iron content, was significantly higher in MS patients than in healthy controls and was associated with more severe lesion burden and brain atrophy [21]. Many studies have demonstrated increased iron deposition in the brain in MS [22, 23]. One possible source of iron deposition in the OR is macrophages, and microglia may acquire high levels of iron by phagocytosing myelin/oligodendrocyte debris.

Thus far, a few theories have been proposed to explain the relationship between iron and MS. Some researchers have considered that the breakdown of the blood-brain barrier (BBB) in MS can result in the extravasation of red blood cells into the CNS [24]. Other researchers have considered iron deposition to be only a secondary phenomenon of MS [25]. Khalil et al. showed that iron accumulation did not precede the development of MS but rather occurred as a by-product of the pathophysiological processes of the disease and was related to morphological brain damage [26]. However, the precise mechanisms underlying this excessive iron accumulation in MS are still uncertain.

In addition, we found a remarkable negative correlation between DD and MPV in the OR in patients, which was consistent with several prior studies. Zhang et al., using a 3-D T₂^*-weighted spoiled multi-echo GRE sequence in a 3T MRI system, demonstrated that non-enhancing lesions with a longer DD showed a higher relative susceptibility/R₂^* value, reflecting iron deposition [2]. Hammond et al. demonstrated that the local field shift (LFS), which is caused by magnetic susceptibility-shifted compounds such as iron, was significantly correlated with DD in the caudate and putamen in MS [27]. Similarly, Reshiana et al. demonstrated that precentral grey matter (GM) MPVs were negatively correlated with DD [21].

Our study found no correlation between MPVs and EDSS scores. This result can be partially explained by the small amount of iron deposition that may not cause immediate effects and are not reflected in the EDSS score. Another possible reason is that the OR is part of the visual system but EDSS is a measure of physical disability in MS and is heavily weighted toward motor disability. Similarly, Khalil et al. found that short-term changes in iron concentration were not associated with changes in disability in 76 patients with clinically isolated syndrome (CIS) and 68 patients with MS [23], and Du et al. found no correlation between the MPVs of all regions of interest (ROIs) and EDSS scores in the study. Conversely, Rudko et al. showed that quantitative susceptibility (QS) maps more greatly aided the identification of significant, voxel-level increases in iron deposition in the subcortical GM of MS patients compared with control subjects. ROI analysis of the mean R₂^* and QS in the subcortical GM demonstrated that R₂^* and QS were strongly correlated with EDSS [28]. We considered that discrepancy may be due to differences in the population characteristic and region in anatomy.

Our results showed that the bilateral OR around the trigone of the lateral ventricle was hypointense, which is consistent with other reports [12, 14]. A histological analysis of a post-mortem specimen confirmed that the identified T₂^*W-hypointense periventricular structures reflect the OR [13]. Moreover, we demonstrated that patients with a longer DD had a thinner OR than controls. A plausible explanation is focal axonal trans-section within MS plaques after a long DD [13, 29]. Daniel et al. demonstrated that the median MRI indices along the OR were also significantly abnormal in all MS subgroups [30]. Interestingly, 28 patients with bilateral OR lesions showed bilateral thinning of the OR and a heterogeneous signal, and 14 patients with unilateral OR lesions showed homolateral thinning of the OR and a heterogeneous signal. Consistent with the results of a previous study [4], the analysis of DTI demonstrated that T₂ OR+ patients had a more severe and distributed pattern of DTI abnormalities along the ORs than did healthy controls and T₂ OR- patients. The notion is that trans-synaptic retrograde degeneration of the OR due to MS lesions might occur in these patients.

Surprisingly, in the nine patients with a shorter DD, the OR had a normal thickness and homogeneous signal. However, these nine patients had a lower MPV than the controls, possibly because the iron accumulation in the OR is more pronounced in early stages and occurs independent of its morphology. Chen et al. demonstrated that the fractional anisotropy (FA) values were significantly lower, and the mean diffusivity (MD), λ_∥, λ_⊥values were significantly higher in the unaffected OR in MS patients than in controls. Similarly, using 3D-ESWAN, Zeng et al., observed that the ICVs and their main tributaries had a score of 3, with the presence of continuous vessel walls and homogeneous signals in MS patients with a relatively short DD.

Although we believe that our findings in this study are substantial and have a good pathophysiological correlation, they are not without limitations. First, the cross-sectional design and relatively small number of subjects studied may not provide an accurate reflection of the iron deposition and thinning of the OR in MS patients. Second, owing to the nature of ESWAN phase images and the associated ‘blooming’ effects, our measurements of iron are inherently relative rather than absolute. Third, OR quantification may be influenced by MS lesions; however, the assessment of the correlation between OR damage and the severity of clinical involvement has been a topic of previous investigation and is beyond the scope of our study.