Correlation between cerebral cortex changes and clinical features in patients with neuromyelitis optica spectrum disorder with normal-appearing brain tissue: a case-control study

Abstract

Neuromyelitis optica spectrum disorder (NMOSD) is an inflammatory demyelinating disease of the central nervous system. However, whether and how cortical changes occur in NMOSD with normal-appearing brain tissue, or whether any cortical changes correlate with clinical characteristics, is not completely clear. The current study recruited 43 patients with NMOSD who had normal-appearing brain tissue and 45 healthy controls matched for age, sex, and educational background from December 2020 to February 2022. A surface-based morphological analysis of high-resolution T1-weighted structural magnetic resonance images was used to calculate the cortical thickness, sulcal depth, and gyrification index. Analysis showed that cortical thickness in the bilateral rostral middle frontal gyrus and left superior frontal gyrus was lower in the patients with NMOSD than in the control participants. Subgroup analysis of the patients with NMOSD indicated that compared with those who did not have any optic neuritis episodes, those who did have such episodes exhibited noticeably thinner cortex in the bilateral cuneus, superior parietal cortex, and pericalcarine cortex. Correlation analysis indicated that cortical thickness in the bilateral rostral middle frontal gyrus was positively correlated with scores on the Digit Symbol Substitution Test and negatively correlated with scores on the Trail Making Test and the Expanded Disability Status Scale. These results are evidence that cortical thinning of the bilateral regional frontal cortex occurs in patients with NMOSD who have normal-appearing brain tissue, and that the degree of thinning is correlated with clinical disability and cognitive function. These findings will help improve our understanding of the imaging characteristics in NMOSD and their potential clinical significance.

Introduction

Neuromyelitis optica spectrum disorder (NMOSD) is a group of severe autoimmune inflammatory demyelinating disorders of the central nervous system with a prevalence of approximately 0.5–4/100,000 people worldwide (Hor et al., 2020). NMOSD can affect the optic nerve, spinal cord, and brain parenchyma, and cause a range of clinical symptoms, including paralysis, visual impairment, and encephalopathy (Wingerchuk et al., 2015). Research on NMOSD has evolved substantially since its initial description as neuromyelitis optica consisting of simultaneous bilateral optic neuritis and transverse myelitis (Cree et al., 2002). The discovery of the specific antibody to aquaporin-4 (AQP4) water channels expanded the spectrum of neuromyelitis optica (Dutra et al., 2018). Subsequently, the term “NMOSD” was proposed to represent the entire clinical syndrome in the 2015 diagnostic criteria (Wingerchuk et al., 2015), and the disorder was divided into seropositive and seronegative types, depending on the anti-AQP4 status.

Patients with NMOSD typically present with spinal cord and optic nerve involvement, although the brain can also be affected. Brain lesions typically appear in regions with high AQP4 expression, such as the periependymal surface of the corpus callosum, third and fourth ventricle, hypothalamus, and area postrema. In addition, a distinctive feature of corticospinal tract involvement has been observed in some patients with NMOSD, despite low expression of AQP4 in this region (Dutra et al., 2018). Although some patients present with brain lesions, many people with NMOSD appear to exhibit either normal brain structure or non-specific white matter (WM) dots and patches, as assessed via conventional magnetic resonance imaging (MRI) (Kim et al., 2015; Wingerchuk et al., 2015; Dutra et al., 2018). Because conventional MRI is insufficient for detecting and quantifying certain occult abnormalities, researchers have used multiple neuroimaging methods, including high-resolution three-dimensional T1-weighted imaging (Zhang et al., 2022), resting-state functional imaging (Han et al., 2020), and diffusion tensor imaging (Yan et al., 2022) to search for potential changes in the brains of those with NMOSD. These studies have detected changes in gray matter (GM) volume and cortical thickness (Kim et al., 2016), functional connectivity and networks (Han et al., 2020), and structural WM integrity (Yan et al., 2022).

Cortical thinning in multiple sclerosis has been demonstrated to correlate with the extent of WM lesions, suggesting that cortical atrophy might result from axonal transection by WM lesions with an influence on highly connected regions (Charil et al., 2007; Treaba et al., 2021). Patients with NMOSD who had brain lesions exhibited more severe thalamic atrophy than those without lesions and healthy controls (HCs) (Hyun et al., 2017). Thus, we hypothesized that brain lesions contribute to cortical atrophy in NMOSD. The present study aimed to determine whether and how cortical atrophy presents in patients with NMOSD who do not exhibit visible brain lesions. Some researchers have used quantitative MRI to explore occult damage in NMOSD, selecting normal-appearing GM or WM as a region of interest (ROI) for further investigation, regardless of the presence of other brain lesions (Jeong et al., 2017; Sun et al., 2019). In contrast, other studies included all patients with NMOSD (those with lesions and those with normal-appearing brain tissue) for further analysis (Cacciaguerra et al., 2021; Masuda et al., 2022). However, few structural MRI studies have included only those patients with NMOSD who do not have any brain lesions, and the overall sample size in these studies is limited (Duan et al., 2014; Chen et al., 2021).

Surface-based morphometry (SBM) and voxel-based morphometry (VBM) are effective methods for estimating structural abnormalities in the brain (Wu et al., 2021; Goto et al., 2022; Li et al., 2022). VBM analysis is the most common method used to identify volumetric changes. However, voxel-based registration in VBM can lead to registration artifacts. Additionally, highly variable folding patterns of the brain are not considered in VBM, which can reduce the accuracy of alignment (Liao et al., 2021). SBM analyses have several advantages over the volume-based analysis. SBM adopts an alternative method of registering by matching the sulcal and gyral geometry to an inflated spherical atlas that raises buried sulci to the surface, thus reducing potential misalignment (Yotter et al., 2011). Moreover, SBM allows for analysis of several additional metrics such as cortical thickness, gyrification, and sulcal depth (Dahnke et al., 2013).

Despite a variety of structural MRI studies investigating correlations between structural metrics and cognitive function in NMOSD, conclusions are inconsistent. Researchers have reported widespread cortical (Kim et al., 2016) or neocortical thinning (Liu et al., 2014) in patients with NMOSD, the degree of which did not significantly correlate with cognitive ability. However, others have demonstrated that atrophy of the nucleus accumbens was associated with cognitive function, including visual memory and performance on attention and information processing tasks (Kim et al., 2017). Although the findings remain scarce and inconclusive, cognitive alteration and its correlation with MRI results in patients with NMOSD continues to be a topic of interest for researchers in the field.

Based on this background, the primary goal of the present study was three-fold: (1) to perform an SBM analysis across the whole brain to measure cortical thickness, gyrification, and sulcal depth, and to explore potential abnormalities in NMOSD that presents with normal-appearing brain tissue; (2) to explore whether subgroups (based on presence or absence of clinical episodes of optic neuritis) differ in the surface parameters that are significantly altered in patients with NMOSD relative to HCs; and (3) to determine whether any cognitive functions correlate with the observed morphological changes.

Methods

Ethics statement

This case-control study was approved by the Ethics Committee of the Second Xiangya Hospital of Central South University (approval No. 061, May 26, 2021 and approval No. 086, July 9, 2019) (Additional file 1 ^{(254.6KB, pdf)} ), and all participants provided written informed consent (Additional file 2 ^{(99KB, pdf)} ). This study was reported according to the STrengthening the Reporting of OBservational Studies in Epidemiology (STROBE) guidelines (von Elm et al., 2007) (Additional file 3). The trial was performed in strict accordance with the Declaration of Helsinki.

STROBE Statement—Checklist of items that should be included in reports of case-control studies

	Item No	Recommendation
Title and abstract	1	(a)Indicate the study’s design with a commonly used term in the title or the abstract
		(b)Provide in the abstract an informative and balanced summary of what was done and what was found
Introduction
Background/rationale	2	Explain the scientific background and rationale for the investigation being reported
Objectives	3	State specific objectives, including any prespecified hypotheses
Methods
Study design	4	Present key elements of study design early in the paper
Setting	5	Describe the setting, locations, and relevant dates, including periods of recruitment, exposure, follow-up, and data collection
Participants	6	(a)Give the eligibility criteria, and the sources and methods of case ascertainment and control selection. Give the rationale for the choice of cases and controls
		(b)For matched studies, give matching criteria and the number of controls per case
Variables	7	Clearly define all outcomes, exposures, predictors, potential confounders, and effect modifiers. Give diagnostic criteria, if applicable
Data sources/ measurement	8*	For each variable of interest, give sources of data and details of methods of assessment (measurement). Describe comparability of assessment methods if there is more than one group
Bias	9	Describe any efforts to address potential sources of bias
Study size	10	Explain how the study size was arrived at
Quantitative variables	11	Explain how quantitative variables were handled in the analyses. If applicable, describe which groupings were chosen and why
Statistical methods	12	(a)Describe all statistical methods, including those used to control for confounding
		(b)Describe any methods used to examine subgroups and interactions
		(c)Explain how missing data were addressed
		(d)If applicable, explain how matching of cases and controls was addressed
		(e)Describe any sensitivity analyses
Results
Participants	13*	(a)Report numbers of individuals at each stage of study—eg numbers potentially eligible, examined for eligibility, confirmed eligible, included in the study, completing follow-up, and analysed
		(b)Give reasons for non-participation at each stage
		(c)Consider use of a flow diagram
Descriptive data	14*	(a)Give characteristics of study participants (eg demographic, clinical, social) and information on exposures and potential confounders (b)Indicate number of participants with missing data for each variable of interest
Outcome data	15*	Report numbers in each exposure category, or summary measures of exposure
Main results	16	(a)Give unadjusted estimates and, if applicable, confounder-adjusted estimates and their precision (eg, 95% confidence interval). Make clear which confounders were adjusted for and why they were included
		(b)Report category boundaries when continuous variables were categorized
		(c)If relevant, consider translating estimates of relative risk into absolute risk for a meaningful time period
Other analyses	17	Report other analyses done—eg analyses of subgroups and interactions, and sensitivity analyse;
Discussion
Key results	18	Summarise key results with reference to study objectives
Limitations	19	Discuss limitations of the study, taking into account sources of potential bias or imprecision. Discuss both direction and magnitude of any potential bias
Interpretation	20	Give a cautious overall interpretation of results considering objectives, limitations, multiplicity of analyses, results from similar studies, and other relevant evidence
Generalisability	21	Discuss the generalisability (external validity) of the study results
Other information
Funding	22	Give the source of funding and the role of the funders for the present study and, if applicable, for the original study on which the present article is based

*Give information separately for cases and controls.

Note: An Explanation and Elaboration article discusses each checklist item and gives methodological background and published examples of transparent reporting. The STROBE checklist is best used in conjunction with this article (freely available on the Web sites of PLoS Medicine at http://www.plosmedicine.org/, Annals of Internal Medicine at http://www.annals.org/, and Epidemiology at http://www.epidem.com/). Information on the STROBE Initiative is available at http://www.strobe-statement.org

Participants

All patients with NMOSD were recruited between December 2020 and February 2022 from the Neurological Department of the Second Xiangya Hospital of Central South University. The inclusion criteria for patients were as follows: (a) diagnosis of NMOSD by two neurologists based on the 2015 revised diagnostic criteria (Wingerchuk et al., 2015); (b) at least one clinical demyelinating episode of the central nervous system (myelitis, optic neuritis, or encephalopathy) that persisted for at least 24 hours; (c) age > 18 years and < 70 years; (d) serological or cerebrospinal fluid antibody tests performed using a cell-based assay method; and (e) either anti-AQP4 antibody positivity or negativity, but myelin oligodendrocyte glycoprotein antibody-negative. The exclusion criteria were as follows: (a) history of other neurological diseases, including stroke, epilepsy, traumatic brain injury, or psychiatric problems; (b) contraindications to MRI scanning or inability to cooperate during the MRI scan; (c) visible brain lesions; and (d) excessive head motion. According to some researchers, the sample size required for the measurement model should be at least five times that of the observed variables, with ten times being most appropriate (Bentler and Chou, 1987). As our study included five recorded variables (i.e., disease duration, Expanded Disability Status Scale (EDSS), Digit Symbol Substitution Test (DSST), Trail Making Test (TMT)-A, and TMT-B scores), the sample size should be more than 25, and a sample size closer to 50 would be most appropriate. Age-, sex-, and educational background-matched HCs were recruited among volunteers in the community and met the same exclusion criteria as the patients did.

Cognitive tests and clinical assessment

Participants underwent two cognitive performance assessments: DSST and TMT. The DSST is used to test psychomotor performance, including processing speed, sustained attention, and working memory. During the test, the participants were shown numbers 1–9 and a key with corresponding symbols (e.g., ‘+’, ‘V’, ‘<’), and were then presented with a series of numbers (1–9), each paired with a blank box. The task was to fill in as many blank boxes as possible with the correct corresponding symbol within 90 seconds (Rosano et al., 2016). The TMT is widely used in neuropsychological evaluations and consists of two separate subscales: TMT-A and TMT-B. The TMT-A was designed as a baseline measure of visual scanning ability, graphomotor speed, and mental flexibility. The participants were presented with a group of circles numbered 1–25. The task was to connect the circles in order as quickly as possible, with a shorter time indicating better performance (Misdraji and Gass, 2010). The TMT-B evaluates letter sequencing, mental double tracking, and task switching. The participants were instructed to draw lines alternately connecting circles that contained 13 numbers (1–13) and 12 letters (A–L) in ascending order (i.e., 1-A-2-B…12-L-13) as quickly as possible (Sánchez-Cubillo et al., 2009). All participants were required to complete the DSST, TMT, and MRI scans on the same day. EDSS scores (Çinar and Yorgun, 2018) and disease duration (calculated in months from the date of first symptom onset to the scan date) were also recorded.

MRI scanning

All imaging data were acquired using a 3.0T MRI scanner (MAGNETOM Skyra; Siemens Healthcare, Erlangen, Germany) with a 32-channel head coil. During the scanning, participants maintained a supine position and wore earplugs. Foam pads were placed between the head and coil to minimize head-motion artifacts. MRI scanning sequences included T1-weighted imaging, T2-weighted imaging, fluid-attenuated inversion recovery (FLAIR), and T1-weighted three-dimensional magnetization-prepared rapid acquisition gradient echo (MP-RAGE) imaging. The MP-RAGE imaging was acquired with the following parameters: echo time = 2.45 ms, repetition time = 2000 ms, voxel size = 1 mm × 1 mm × 1 mm, flip angle = 8°, field of view = 256 mm × 256 mm, acquisition matrix = 256 × 256, and 192 axial slices.

SBM analysis

Imaging data were analyzed using the Statistical Parametric Mapping 12 (SPM12; www.fil.ion.uck.ac.uk/spm/software/spm12) and the Computational Anatomy Toolbox 12 (CAT12; www.neuro.uni-jena.de/cat) (Gaser et al., 2022) in MATLAB (R2020b, The MathWorks, Inc., Natick, MA, USA). The high-resolution T1-weighted images were spatially normalized to Montreal Neurologic Institute space using affine and non-linear registration, and were automatically segmented into GM, WM, and cerebrospinal fluid (Rajapakse et al., 1997). The imaging data were then visually inspected for potential segmentation and registration errors.

Cortical thickness estimation and reconstruction of the central cortical surface was carried out using an automated projection-based thickness method in the CAT12 toolbox (Dahnke et al., 2013). The central cortical layer was defined as the layer lying in the geometric center of the cortex. Compared with the inner (WM/GM interface) and outer (GM/cerebrospinal fluid interface) cortical surfaces, the surface defined on the central layer can provide cortical geometry that is better for detecting changes in atrophied brains (Liu et al., 2008; Dahnke et al., 2013). Automated cortical parcellation was performed using the Desikan-Killiany atlas (Desikan et al., 2006). CAT12 uses tissue segmentation to measure the WM distance and then uses a neighboring relationship described by the WM distance to project the local maxima (equal to the cortical thickness) to other GM voxels (Dahnke et al., 2013). Two other morphometric metrics, gyrification (Luders et al., 2006) and sulcal depth (Sheffield et al., 2021), were also calculated.

Surface data were resampled to the template space at a 32k mesh resolution. Following the recommendations of the CAT12 toolbox manual, cortical thickness data were smoothed using a Gaussian kernel with a full-width at half-maximum of 15 mm, and gyrification and sulcal depth data were smoothed using a full-width at half-maximum of 20 mm. Detailed procedures for the use of software are present in Additional file 4 ^{(101.4KB, pdf)} .

Statistical analysis

Statistical analysis was performed using SPSS software (version 26.0; IBM Corp., Armonk, NY, USA) and GraphPad Prism version 8.3.0 (GraphPad Software, San Diego, CA, USA, www.graphpad.com). The Shapiro-Wilk test was used to evaluate normality. The two-sample t-test (for normally distributed data) or Mann-Whitney U test (for non-normally distributed data) was used to compare the differences in age, education levels, and DSST, TMT-A, and TMT-B scores between groups. Fisher’s exact test was used to compare sex differences between groups. Two-sample t-tests in the SPM12 and CAT12 toolboxes were used to detect any significant differences in cortical thickness, sulcal depth, and gyrification index between HCs and the patients. Statistical maps were assigned thresholds at a cluster-level false discovery rate (FDR)-corrected P-value < 0.001. Surface parameters (i.e., cortical thickness, sulcal depth, and gyrification index) exhibiting significant differences between the NMOSD group and the HCs were further explored between subgroups of patients with NMOSD, which were based on the presence or absence of clinical optic neuritis episodes. Statistical maps were assigned thresholds at the cluster-level FDR-corrected P < 0.01.

Brain regions for which surface parameters were observed to differ significantly between the NMOSD and HC groups were defined as ROIs. The Desikan-Killiany atlas (Desikan et al., 2006) was used to define the cortical regions resulting from the between-group analyses and extract the ROI-based surface values. The mean values inside the corresponding ROIs for each participant were calculated using the ROI tools in CAT12. Post hoc t-tests were conducted to analyze the between-group differences in surface values in each ROI, and the results are presented as boxplots.

The ROI-based surface values and clinical parameters were checked for normal distribution, and Pearson’s (for normally distributed data) and Spearman’s (for non-normally distributed data) correlation analyses were applied to examine the correlation between the mean surface values in the ROIs and the five clinical parameters (DSST, TMT-A and TMT-B scores, EDSS score, and disease duration). The threshold of significance was set at P < 0.05, adjusted using FDR correction.

Results

Demographic and clinical characteristics

Forty-three patients diagnosed with NMOSD who had normal-appearing brain tissue were included in this study, along with 45 HCs. Figure 1 shows the flowchart of the selection process. The mean age of the patients was 46.47 ± 11.14 years and 97.67% of them were female. Four patients were anti-AQP4 antibody-negative and the rest were anti-AQP4 antibody-positive. The mean EDSS score at the time of the MRI was 2.17 ± 1.82. The mean disease duration was 38.83 ± 51.31 months. There were no statistically significant differences between the patients and HCs in terms of age, sex, or years of education (all P > 0.05). Among all participants, seven patients failed to complete the cognitive tests, because of visual impairment or upper limb weakness. Performance on the DSST, TMT-A, and TMT-B was significantly worse in the remaining patients than in the HCs (P < 0.001). Nineteen patients with NMOSD had a history or ongoing clinical episodes of optic neuritis (NMOSD-ON) and 24 had not (non-NMOSD-ON). The demographic characteristics of the NMOSD and HCs groups are summarized in Table 1, and those of the NMOSD-ON and non-NMOSD-ON groups are shown in Table 2.

Figure 1.

Flow chart of the selection process for included participants.

AQP4: Aquaporin-4; HC: healthy control; MOG: myelin oligodendrocyte glycoprotein; MRI: magnetic resonance imaging; NMOSD: neuromyelitis optica spectrum disorder.

Table 1.

Demographic and clinical characteristics in NMOSD patients with normal-appearing brain tissue and HCs

	NMOSD (n = 43)	HCs (n = 45)	t/Z	P-value
Age (yr)	46.47±11.14	44.31±9.63	0.972	0.334a
Gender (male/female)	1/42	1/44	–	1.000b
Education level (yr)	10.21±2.82	11.01±3.35	–1.004	0.315c
Right handedness	43	45		–
AQP4 antibody (positive/negative)	39/4	–	–	–
EDSS score at the time of MRI	2.17±1.82	–	–	–
Disease duration (mon)	38.83±51.31	–	–	–
DSST	40.61±12.78	52.40±11.63	–3.975	< 0.001c
TMT-A	66.86±27.46	44.92±13.88	–4.03	< 0.001c
TMT-B	162.67±61.89	110.29±40.25	–4.02	< 0.001c

Among all participants, seven patients with NMOSD failed to complete the cognitive tests including the DSST, TMT-A, and TMT-B. Data were analyzed by two-sample t-test (a), Fisher’s exact test (b), or Mann-Whitney U test (c). AQP4: Aquaporin-4; DSST: Digit Symbol Substitution Test; EDSS: Expanded Disability Status Scale; HC: healthy control; MRI, magnetic resonance imaging; NMOSD: neuromyelitis optica spectrum disorder; TMT-A: Trail Making Test Subscale A; TMT-B: Trail Making Test Subscale B.

Table 2.

Demographic and clinical characteristics in NMOSD-ON and non-NMOSD-ON patients

	NMOSD-ON (n = 19)	non-NMOSD-ON (n = 24)	t/Z	P-value
Age (yr)	48.95±11.46	44.50±10.71	1.312	0.197a
Gender (male/female)	1/18	0/24	–	0.442b
Education level (yr)	9.95±2.74	10.42±2.93	–0.365	0.715c
Right handedness	19	24		–
AQP4 antibody (positive/negative)	16/3	23/1	–	0.306b
EDSS score at the time of MRI	2.34±1.93	2.04±1.75	–0.659	0.510c
Disease duration (mon)	48.66±58.77	31.04±44.29	–0.722	0.470c
DSST	38.86±12.05	41.73±13.39	–0.617	0.537c
TMT-A	70.38±19.06	64.62±31.91	–1.363	0.173c
TMT-B	171.48±51.38	157.06±68.29	0.676	0.504a

Seven patients with NMOSD failed to complete the cognitive tests including the DSST, TMT-A, and TMT-B. Data were analyzed by two-sample t-test (a), Fisher’s exact test (b), or Mann-Whitney U test (c). AQP4: Aquaporin-4; DSST: Digit Symbol Substitution Test; EDSS: Expanded Disability Status Scale; MRI: magnetic resonance imaging; NMOSD: neuromyelitis optica spectrum disorder; NMOSD-ON: NMOSD patients who experienced at least one clinical episode of optic neuritis; non-NMOSD-ON: NMOSD patients without clinical episodes of optic neuritis; TMT-A: Trail Making Test Subscale A; TMT-B: Trail Making Test Subscale B.

SBM results

Compared with the HCs, the NMOSD group exhibited significantly thinner cortex in the bilateral rostral middle frontal gyrus (rMFG) and left superior frontal gyrus (SFG) (P < 0.001, FDR-corrected at the cluster level; Figure 2A). The cluster sizes and peak values in the brain regions with significantly lower cortical thickness are shown in Table 3. Boxplots for the ROI-based between-group analysis of cortical thickness are shown in Figure 2B. The comparisons that post-hoc t-tests indicated were significantly different. Cortical gyrification indices and sulcal depth did not significantly differ between the HCs and these patients with NMOSD who had normal-appearing brain tissue.

Figure 2.

Brain regions with cortical thicknesses that differed significantly between the patients with NMOSD who had normal-appearing brain tissue and the healthy controls.

(A) The NMOSD group showed significantly thinner cortex in the bilateral rostral middle frontal gyri (rMFG_L and rMFG_R) and left superior frontal gyrus (SFG_L) compared with the HC group (P < 0.001, false discovery rate-corrected at the cluster level). These brain regions were defined as ROIs. (B) Mean cortical thicknesses for each group and each ROI. *P < 0.05, ***P < 0.001 (two-sample t-test). HC: Healthy control; L: left; LH: left hemisphere; NMOSD: neuromyelitis optica spectrum disorder; R: right; RH: right hemisphere; rMFG: rostral middle frontal gyrus; SFG: superior frontal gyrus.

Table 3.

Overview of bilateral areas of cluster-level significant effects of cortical thickness reduction in patients with NMOSD with normal-appearing brain tissue compared to healthy controls

Overlay of atlas region	Hemisphere	Cluster size	Peak MNI coordinates (mm)			FDR-corrected P-value
			x	y	z
79% rostralmiddlefrontal	L	243	–21	63	2	0.014
19% superiorfrontal
2% lateralorbitofrontal
98% superiorfrontal	L	180	–9	53	39	0.045
2% rostralmiddlefrontal
86% rostralmiddlefrontal	R	553	26	58	7	< 0.001
10% superiorfrontal
4% frontalpole

Atlas labeling was conducted according to the Desikan-Killiany atlas. FDR: False discovery rate; L: left; MNI: Montreal Neurological Institute; NMOSD: neuromyelitis optica spectrum disorder; R: right.

Because cortical thickness differed significantly between HCs and the NMOSD group, we further explored whether cortical thickness differed between the two patient subgroups. Compared with the non-NMOSD-ON group, we found that the cortex was significantly thinner in the bilateral cuneus, superior parietal cortex, and pericalcarine cortex of the NMOSD-ON group (P < 0.01, FDR-corrected at the cluster level; Table 4 and Figure 3).

Table 4.

Overview of bilateral areas of cluster-level significant effects of cortical thickness reduction in NMOSD-ON compared to non-NMOSD-ON

Overlay of atlas region	Hemisphere	Cluster size	Peak MNI coordinates (mm)			FDR-corrected P-value
			x	y	z
38% superiorparietal	L	757	–3	–78	21	< 0.001
29% cuneus
20% pericalcarine
6% lateraloccipical
5% precuneus
58% cuneus	R	443	5	–77	19	0.012
24% superiorparietal
15% pericalcarine
3% lateraloccipical

Atlas labeling was conducted according to the Desikan-Killiany atlas. FDR: False discovery rate; L: left; NMOSD: neuromyelitis optica spectrum disorder; NMOSD-ON: NMOSD patients who experienced at least one clinical episode of optic neuritis; non-NMOSD-ON: NMOSD patients without clinical episodes of optic neuritis; R: right.

Figure 3.

Brain regions with cortical thicknesses that differed significantly between the NMOSD-ON group and the non-NMOSD-ON group.

The NMOSD-ON group exhibited significantly thinner cortex in the bilateral cuneus, superior parietal cortex, and pericalcarine cortex compared with the non-NMOSD-ON group (P < 0.01, false discovery rate-corrected at the cluster level). LH: Left hemisphere; NMOSD: neuromyelitis optica spectrum disorder; NMOSD-ON: NMOSD patients who experienced at least one clinical episode of optic neuritis; non-NMOSD-ON: NMOSD patients without clinical episodes of optic neuritis; RH: right hemisphere.

Correlation analyses of SBM results and clinical parameters

As shown in Figure 4, mean cortical thickness in the left rMFG positively correlated with DSST scores (r = 0.469, P = 0.004, P_FDR = 0.060) and negatively correlated with TMT-A (r = –0.380, P = 0.022, P_FDR = 0.110) and EDSS (r = –0.318, P = 0.037, P_FDR = 0.093) scores. In the right rMFG, mean cortical thickness positively correlated with DSST scores (r = 0.457, P = 0.005, P_FDR = 0.038) and negatively correlated with TMT-A (r = –0.356, P = 0.033, P_FDR = 0.099), TMT-B (r = –0.340, P = 0.042, P_FDR = 0.09), and EDSS (r = –0.336, P = 0.028, P_FDR = 0.105) scores. However, mean cortical thickness in the left SFG did not correlate with clinical parameters, and no significant correlations were observed between mean cortical thickness in the identified brain regions and disease duration (P > 0.05).

Figure 4.

Correlation analyses between the mean cortical thickness and clinical parameters, including disease durations, EDSS, and cognitive scale scores.

(A) Heatmap of the correlational r-values between the mean cortical thickness in the brain regions showing significant differences from controls, and clinical parameters including disease durations, the EDSS scores, and cognitive scale scores (i.e., DSST, TMT-A, TMT-B). (B, C) In the left rMFG, mean cortical thickness positively correlated with DSST scores and negatively correlated with TMT-A and EDSS scores. (D, E) In the right rMFG, mean cortical thickness positively correlated with the DSST scores and negatively correlated with TMT-A, TMT-B, and EDSS scores. Pearson’s (ROI-based surface values vs. TMT-B scores) and Spearman’s (ROI-based surface values vs. disease durations, EDSS, DSST, and TMT-A scores) correlation analyses were applied. *P < 0.05, **P < 0.01. DSST: Digit Symbol Substitution Test; EDSS: Expanded Disability Status Scale; L: left; R: right; rMFG: rostral middle frontal gyrus; SFG: superior frontal gyrus; TMT-A: Trail Making Test (subscale A); TMT-B: Trail Making Test (subscale B).

Discussion

In this study, we used SBM analysis to quantify whole-brain differences in cortical thickness between HCs and patients with NMOSD who had normal-appearing brain structure. Because the cortex has low AQP4 expression, differences in cortical thickness of the frontal cortex in NMOSD are not thought to be mediated by AQP4. Rather, these differences are thought to be a secondary degeneration resulting from lesions of the spinal cord or optic nerve (Sun et al., 2019). In particular, occult cortical atrophy in the bilateral rMFG of patients with NMOSD was correlated with clinical disability and cognitive function, including attention and executive performance. Thus, our study provides insight into NMOSD with normal-appearing brain tissue, based on structural imaging patterns and their clinical correlates.

In general, the characteristics of NMOSD that are revealed on conventional MRI include optic nerve lesions extending over half the optic nerve length or involving the optic chiasm, and intramedullary lesions extending over three contiguous segments (Wingerchuk et al., 2015). Abnormal brain regions are typically located in areas associated with high AQP4 expression, such as the periependymal surface of the corpus callosum, third and fourth ventricle, hypothalamus, and area postrema. However, some patients with NMOSD exhibit brain MRI that appears normal or have non-specific dots and patches in the subcortical and deep WM (Kim et al., 2015; Wingerchuk et al., 2015; Dutra et al., 2018).

Many researchers have explored occult damage in NMOSD by applying quantitative MRI, with some selecting the normal-appearing GM or WM for ROI-based analysis regardless of the presence of other brain lesions (Jeong et al., 2017; Sun et al., 2019) and others pooling patients with or without lesions for further analysis (Masuda et al., 2022). There are limited studies of structural MRI that only include patients with NMOSD that do not have brain lesions, and the overall sample size of these studies is limited (Duan et al., 2014; Chen et al., 2021). Cortical atrophy has been observed in NMOSD (Kim et al., 2016). As cortical thinning in multiple sclerosis has been shown to correlate with WM lesion load, and cortical atrophy has been proposed to result from WM lesions that influence the axon and highly connected regions (Charil et al., 2007; Treaba et al., 2021), it is plausible that brain lesions in patients with NMOSD might contribute to cortical atrophy. To determine whether and how cortical atrophy might occur in NMOSD with normal-appearing brain tissue, in the current study, we only included patients who had NMOSD without visible brain lesions.

We found a significant reduction in cortical thickness in the left SFG, which has been shown to have a crucial effect on the integrative architecture of the brain’s cognitive framework (Zhu et al., 2017). Anatomically, the SFG is located at the superior part of the prefrontal cortex. It is connected to several critical brain network nodes, playing a role in cognitive control and execution, motor control, and sensorimotor-related cognition (Li et al., 2013). du Boisgueheneuc et al. (2006) conducted a comparative study between HCs and patients with either left prefrontal lesions restricted to the SFG or focal brain lesions sparing the SFG, and demonstrated that the posterior portion of the left SFG was a key component of a neural network for working memory, that is triggered by the highest level of executive processing. Collectively, because deficits in memory and executive function have been reported in patients with NMOSD, structural abnormalities in the SFG may be important anatomical substrates and potential imaging biomarkers for NMOSD.

Moreover, the bilateral rMFG, locating in the dorsolateral prefrontal cortex, was identified as another important brain region with structural differences in NMOSD in our study. The bilateral MFG is thought to be involved in dorsal and ventral attention networks (Corbetta et al., 2008) which is a highly interconnected cortical structure involved in multiple tasks such as working memory and comprehension (Briggs et al., 2021).

Considering the importance of these cognitive functions, being able to identify changes in the gyri might contribute greatly to our understanding of the pathophysiology of NMOSD. Several resting-state functional MRI studies of NMOSD revealed functional changes in the amplitude of low-frequency fluctuation (Liu et al., 2020), functional connectivity (Han et al., 2020), and activation intensity of olfactory-related brain networks (He et al., 2021) in the left/bilateral MFG, which is in line with our finding of reduced cortical thickness in the bilateral rMFG of NMOSD patients with normal-appearing brain tissue. However, to our knowledge, no previous study has examined the relationship between the MFG and cognitive function in NMOSD. Thus, our findings provide new evidence that the pathogenesis of NMOSD is related to deficits in attentional reorientation and working memory.

Cortical thinning in local brain regions and cognitive decline in NMOSD has been shown in previous studies (Moghadasi et al., 2021). We also observed a decline in cognitive function in our patients with NMOSD. Indeed, cognitive dysfunction is one of the most disabling symptoms of NMOSD and cognitive deficits in NMOSD have attracted substantial attention. A recent meta-analysis showed a pooled prevalence of 44% for cognitive impairments in patients with NMOSD (Moghadasi et al., 2021). In a review of the cognitive impairment profile in NMOSD, deficits in memory, attention, processing speed, and executive function were considered to be the most frequent (Czarnecka et al., 2020). The DSST and TMT are typically adopted to evaluate these aspects of cognition (Sánchez-Cubillo et al., 2009; Misdraji and Gass, 2010; Rosano et al., 2016). In the current study, we observed significantly poorer performance on the DSST and TMT in patients with NMOSD than in HCs. Additionally, cognitive scale scores in the patients correlated with mean cortical thickness of the bilateral rMFG. However, the mean cortical thickness of the bilateral rMFG correlated with cognitive function and clinical disability only at exploratory statistical thresholds, and except for a positive correlation between the mean cortical thickness of the right rMFG and DSST scores (PFDR = 0.038), significance disappeared after FDR correction for multiple comparisons.

Our subgroup analysis further revealed less cortical thickness on both sides of the cuneus, superior parietal cortex, and pericalcarine cortex in patients with NMOSD-ON than in those with non-NMOSD-ON. Interestingly, the cuneus is known to be involved in basic and high-level visual processing (Palejwala et al., 2021), which could explain the long-term visual disturbances observed in a proportion of patients with NMOSD-ON. In addition, the cuneus supports non-visual functions, including memory (Palejwala et al., 2021). This finding is expected to motivate future research on NMOSD-ON in terms of its clinical correlates and underlying brain networks.

This study had some limitations. First, the cross-sectional nature of the study prevented our evaluation of brain alterations and features in different stages of NMOSD. Second, the limited sample size prevented us from grouping patients into finer categories according to different statuses, e.g., cognitive functions, disease durations, and the presence or absence of brain lesions. Therefore, a multi-modal study with a larger sample size and long-term follow-up is needed to determine the neural mechanisms underlying NMOSD. It is worth noting that the seven patients who did not complete the DSST and TMT-A/B were excluded from the correlation analyses between mean cortical thickness of the bilateral rMFG and left SFG, and scores on DSST and the TMT-A/B, but were included in the analyses between mean cortical thickness and EDSS scores and disease duration.

In summary, our study revealed significantly thinner cortex in the bilateral regional frontal lobes of patients with NMOSD who have normal-appearing brain tissue than in HCs, and demonstrated correlations with clinical characteristics, especially cognitive functions. Our findings will help elucidate the microstructural changes and provide a better understanding of the imaging pattern of NMOSD with normal-appearing brain tissue.

Additional files:

Additional file 1 ^{(254.6KB, pdf)} : Hospital ethics approval (Chinese).

Additional file 1

Hospital ethics approval (Chinese).

Additional file 2 ^{(99KB, pdf)} : Informed consent form (Chinese).

Additional file 2

Informed consent form (Chinese).

Additional file 3: STROBE checklist.

Additional file 4 ^{(101.4KB, pdf)} : Detailed procedures for the use of Computational Anatomy Toolbox 12 (CAT12).

Additional file 4

Detailed procedures for the use of Computational Anatomy Toolbox 12 (CAT12).