Article Text
PDF
Abstract
Purpose This study investigated the topological structural characteristics of systemic lupus erythematosus (SLE) with and without neuropsychiatric symptoms (NPSLE and non-NPSLE), and explore their clinical implications.
Methods We prospectively recruited 50 patients with SLE (21 non-NPSLE and 29 NPSLE) and 32 age-matched healthy controls (HCs), using MRI diffusion tensor imaging. Individual structural networks were constructed using fibre numbers between brain areas as edge weights. Global metrics (eg, small-worldness, global efficiency) and local network properties (eg, degree centrality, nodal efficiency) were computed. Group comparisons of network characteristics were conducted. Clinical correlations were assessed using partial correlation, and differentiation between non-NPSLE and NPSLE was performed using support vector classification.
Results Patients with oth non-NPSLE and NPSLE exhibited significant global and local topological alterations compared with HCs. These changes were more pronounced in NPSLE, particularly affecting the default mode and sensorimotor networks. Topological changes in patients with SLE correlated with lesion burdens and clinical parameters such as disease duration and the systemic lupus international collaborating clinics damage index. The identified topological features enabled accurate differentiation between non-NPSLE and NPSLE with 87% accuracy.
Conclusion Structural networks in patients SLE may be altered at both global and local levels, with more pronounced changes observed in NPSLE, notably affecting the default mode and sensorimotor networks. These alterations show promise as biomarkers for clinical diagnosis.
- Systemic Lupus Erythematosus
- Magnetic Resonance Imaging
- Autoimmunity
Data availability statement
Data are available upon reasonable request.
This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/.
Statistics from Altmetric.com
WHAT IS ALREADY KNOWN ON THIS TOPIC
Neuropsychiatric systemic lupus erythematosus (NPSLE) manifests with diverse neuropsychiatric symptoms. MRI techniques have been widely used in the diagnosis of SLE. There is a link between cognitive dysfunction and altered topology of the brain’s structural network in patients with SLE.
WHAT THIS STUDY ADDS
Structural networks can be significantly altered at both global and local levels in patients with SLE, particularly in those with NPSLE, prominently affecting the default mode network and sensorimotor network. Topological changes in patients with SLE correlated with lesion burdens and clinical parameters.
HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY
Structural network alterations represent promising biomarkers for clinical diagnosis and could potentially aid in evaluating treatment responses and stratifying patients for clinical trials.
Introduction
Systemic lupus erythematosus (SLE) is a chronic relapsing–remitting autoimmune disease affecting multiple organs, primarily affecting women, particularly those of childbearing age.1–4 Between 25% and 95% of patients with SLE experience involvement of the central nervous system (CNS), known as neuropsychiatric systemic lupus erythematosus (NPSLE).5–7 NPSLE manifests with diverse neuropsychiatric symptoms including aseptic meningitis, cerebrovascular diseases, demyelinating syndrome, headache, movement disorders, myelopathy, seizure disorders, acute confusional state, anxiety disorder, cognitive dysfunction, mood disorders, psychosis, acute inflammatory demyelinating polyradiculoneuropathy, autonomic disorder, mononeuropathy, myasthenia gravis, cranial neuropathy, plexopathy and polyneuropathy.8 9 These symptoms are associated with high morbidity, mortality and poor prognosis in patients with SLE, with an unclear underlying pathogenesis.8 Therefore, elucidating the pathological mechanisms of NPSLE, particularly in its early stages, is crucial for clinical diagnosis, treatment decisions and prognosis improvement.10 11
Microstructural changes preceding conventional MRI abnormalities (eg, white matter hyperintensity (WMH) lesions on fluid-attenuated inversion recovery (FLAIR) images and cerebral atrophy) have been observed in patients with SLE using advanced MRI techniques such as diffusion tensor imaging (DTI).12–17 Disruptions in the white matter network associated with working memory,15 18 attention,18 learning and memory19 and language processing functional systems14 have been demonstrated in patients with SLE, indicating a link between cognitive dysfunction and altered topology of the brain’s structural network.20 Even in patients with SLE without overt neuropsychiatric symptoms (non-NPSLE), altered structural topological organisation involving the frontal lobe, parietal lobe, temporal lobe, cingulate and subcortical nuclei has been observed.14 15 However, previous studies have shown conflicting results regarding structural networks in SLE, especially differences between NPSLE and non-NPSLE. Some studies reported significant structural and functional degenerations in NPSLE compared with non-NPSLE,21 22 while others found no structural differences, especially in the structural network characteristics revealed by DTI.13 23 Therefore, the debate continues regarding whether there are differences in the structural networks of NPSLE and non-NPSLE.
In this study, our aim is to further investigate the structural characteristics of NPSLE and compare them with non-NPSLE, in order to enhance our understanding of the underlying substrates associated with NPSLE. Additionally, we explore the clinical significance of these structural characteristics.
Materials and methods
Participants
All participants were informed about the study’s purpose and provided written consent.
Between January 2017 and December 2018, 74 female patients diagnosed with SLE were recruited from the Rheumatology Department of Peking Union Medical College Hospital. This cohort included 40 patients with NPSLE and 34 without neuropsychiatric manifestations (non-NPSLE). Additionally, 36 age-matched healthy female controls were enrolled for comparison.
Inclusion criteria were as follows: (1) Diagnosis of SLE based on the American College of Rheumatology (ACR) classification criteria24 25; (2) primary CNS NPSLE manifestations as per the ACR and Systemic Lupus International Collaborating Clinics model B criteria26 27; (3) age between 18 and 65 years; (4) right-handedness; (5) MRI scan conducted within 7 days of clinical evaluation.
Exclusion criteria were: (1) Use of psychoactive medications or history of substance abuse; (2) history of primary mental illness; (3) secondary NPSLE due to other causes (eg, infections, electrolyte disturbances, hypertension); (4) healthy controls (HCs) with brain lesions; (5) poor MRI image quality.
Clinical information collected included sex, age, age at onset, disease duration, Systemic Lupus Erythematosus Disease Activity Index 2000 (SLEDAI-2k),28 and Systemic Lupus International Collaborating Clinics/American College of Rheumatology damage index (SDI).29 All patients had received treatment with steroids and immunosuppressors.
Image acquisition
MRI scanning was conducted using a 3T MR scanner (GE MR750, GE Healthcare) equipped with a 32-channel head coil. Imaging protocols included conventional two-dimensional axial T2-weighted (T2w) and FLAIR sequences with 4 mm slice thickness, and three-dimensional (3D) sagittal T1-weighted (T1w) images with isotropic 1 mm voxel size. Additionally, DTI was performed with the following parameters: axial acquisition using a spin echo-echo planar imaging (SE-EPI) sequence, time of repetition=7522 ms, time of echo=81 ms, flip angle=90°, slice thickness=2 mm with no slice gap, in-plane resolution of 2 mm×2 mm, matrix size=112×112, 70 slices and acquisition of two b values (0 and 1000 s/mm2) with 64 motion-sensitive gradient directions. MRI scans of patients with SLE were conducted at least 4 weeks after their last relapse and treatment to minimise confounding effects.
Lesion segmentation
WMH lesions were initially segmented using the Lesion Segmentation Toolbox (LST, https://www.applied-statistics.de/lst.html) based on FLAIR and 3D T1 images. Subsequently, segmentation accuracy was verified and manually adjusted by a senior neuroradiologist (YD, with 12 years of experience in neuroradiology). Lesion volumes were quantified. Lesion probability maps for patients with non-NPSLE and NPSLE were generated in the Montreal Neurological Institute (MNI) space using a two-step co-registration approach with Statistical Parametric Mapping software (SPM12, https://www.fil.ion.ucl.ac.uk/spm/): (1) FLAIR images were co-registered to individual 3D T1 images, and the resulting registration parameters were applied to transform the lesion mask into 3D T1 space; (2) The 3D T1 images were normalised to MNI space, and normalisation parameters were used to transform the lesion mask into MNI space. Finally, normalised lesion masks were summed and averaged across all included patients.
DTI image processing
DTI image processing was carried out using DiffusionKit software (https://diffusionkit.readthedocs.io/en/latest/). The preprocessing steps included: (1) Eddy current correction to mitigate EPI distortion and head motion artefacts; (2) Rewriting of Bvec files using transformation parameters from eddy current correction; (3) Extraction of B0 volumes, averaging, and removal of skull artefacts; (4) Calculation of DTI tensor and fractional anisotropy (FA) through least squares fitting; (5) Deterministic fibre tracking performed at the whole-brain level.
Structural network construction
Structural nodes were defined according to the Anatomical Automatic Labelling (AAL) atlas, with 90 supratentorial partitions. Structural network construction used a two-step registration process based on SPM12 to warp the AAL atlas into individual DTI image space. First, the skull-stripped individual 3D T1 image was registered to the MNI152 template, which was in the same space as the AAL atlas. Second, the skull-stripped individual 3D T1 image was registered to the corresponding individual averaged B0 volume for the same subject. Third, the AAL atlas was warped into the individual DTI image space by combining the transformation parameters from these two registrations (backward and forward transformations).
Whole-brain fibres were pruned for each pair of brain areas (network nodes), and the number of fibres connecting each pair of brain areas was used as the edge weights for the structural network.
Calculation of network characteristics
Global network characteristics (including small-worldness index, global efficiency, averaged local efficiency, averaged characteristic shortest path length and averaged cluster coefficient) and nodal-level network properties (including degree centrality, betweenness centrality, nodal efficiency, nodal local efficiency and nodal cluster coefficient) were computed using GRETNA software30 based on weighted structural networks for each subject.
Statistical analyses
Statistical analyses were conducted using SPSS (V.25.0 for Windows, IBM, Armonk, New York, USA) and the statistics toolbox in Matlab (V.2019a). Data were presented as mean and SD for normally distributed measurement data, and median and IQR for variables with non-normal distributed measurement data. For normally distributed data, one-way analysis of variance with post-hoc comparisons or Student’s t-test was applied. For data with non-normal distribution, the Kruskal-Wallis test with post-hoc analysis or the Mann-Whitney U test was used. The categorical data were expressed as n (%), and the differences between the two groups were examined by χ2 analysis or Fisher’s exact test. Multiple comparisons were adjusted using Bonferroni correction. A p value<0.05 was considered statistically significant.
Edge-wise and node-wise statistical analyses were conducted using a general linear model for group comparisons, with age as a covariate. Significance was determined at p<0.05 after false discovery rate (FDR) correction.
Partial correlation analysis was performed to examine associations between MRI and clinical features, adjusting for age as a covariate. A significance level of p<0.005 was used.
Support vector machine for differentiation between non-NPSLE and NPSLE
Support vector machine (SVM) was employed to identify patients and distinguish between different types of patients with SLE based on their structural network features. Feature selection was initially performed using Lasso regression. The SVM model was trained and evaluated using leave-one-out cross-validation. Classification performance was assessed based on accuracy, sensitivity, specificity, positive predictive value, and negative predictive value.
Results
Demographic and clinical features
24 patients with SLE and 4 HCs were excluded due to: (1) incomplete medical history records (n=6); (2) secondary NPSLE due to other causes (n=5); (3) remittent mild headache as the sole neuropsychiatric symptom (n=3); (4) poor MRI data quality (n=14). The final study cohort included 29 patients with NPSLE with classified neuropsychiatric symptoms, 21 patients with non-NPSLE and 32 HCs. There were no significant differences in age, age at onset, SLEDAI or cumulative steroid dose between patients with non-NPSLE and NPSLE. However, patients with NPSLE had longer disease duration (6 (1.1–14) vs 2.3 (0.9–4.5) years; p=0.048) and higher SDI (1 (0–1) vs 0 (0–0); p<0.001) compared with patients with non-NPSLE. As for the types of syndromes in patients with NPSLE, among the 29 patients with NPSLE, 13 (44.83%) presented with seizure disorders, 12 (41.38%) presented with cognitive dysfunction, 9 (31.03%) presented with mood disorders and 9 (31.03%) presented with demyelinating syndrome. 8 (27.59%) presented with headache, 6 (20.69%) with psychosis, 5 (17.24%) with cerebrovascular diseases, 5 (17.24%) with acute confusional state and 1 (3.45%) with autonomic disorder. There were significant differences in some of the WM DTI measurements among the three groups, such as FA, mean diffusivity, radial diffusivity (p<0.05). In addition, we report medication and other comorbidities in the NPSLE and non-NPSLE groups, showing differences in the use of hydroxychloroquine and comorbidity of liver disorder (both p<0.05) (table 1).
Demographic and clinical features of the recruited subjects
Characteristics of brain lesions
WMH lesions were predominantly located in periventricular areas, frontal, parietal and occipital lobes, as well as subcortical regions in patients with both non-NPSLE and NPSLE (figure 1). However, patients with NPSLE exhibited a higher frequency and more widespread distribution of lesions compared with patients with non-NPSLE. The volume of lesions was also larger in patients with NPSLE compared with patients with non-NPSLE (0.25 (0.074–3.09) mL vs 0.13 (0.053–0.80) mL; p=0.032).
Lesion probability maps of patients with non-NPSLE and NPSLE. WMH lesions were predominantly located in periventricular areas, frontal, parietal and occipital lobes, as well as subcortical regions in patients with both non-NPSLE and NPSLE. Patients with NPSLE exhibited a higher frequency and more widespread distribution of lesions compared with patients with non-NPSLE. NPSLE, neuropsychiatric systemic lupus erythematosus; non-NPSLE, systemic lupus erythematosus without neuropsychiatric manifestations; WMH, white matter hyperintensity.
Global and local network measures
As shown in figure 2, patients with both non-NPSLE and NPSLE demonstrated significant alterations in global and nodal-level structural network characteristics. Both types of patients with SLE showed decreased global efficiency and averaged local efficiency, as well as increased averaged characteristic path length compared with HCs. Additionally, patients with NPSLE exhibited a higher averaged cluster coefficient compared with both HCs and non-NPLSE.
Structural network characteristics in non-NPSLE and NPSLE. All patients with SLE demonstrated significant alterations in global and local structural network characteristics. (A) Global network alterations included decreased global efficiency, decreased averaged local efficiency and increased averaged characteristic path length. Additionally, patients with NPSLE exhibited a higher averaged cluster coefficient compared with both HCs and non-NPLSE. (B–C) Local network alterations in non-NPSLE and NPSLE compared with HCs. For C, blue to red colour of the edges represents the statistical T value between groups (values in first group – values in second group). HCs, healthy controls; non-NPSLE, systemic lupus erythematosus without neuropsychiatric manifestations; NPSLE, neuropsychiatric systemic lupus erythematosus. The full names of brain region abbreviations are listed in online supplemental table S1.
Supplemental material
In patients with non-NPSLE, decreased values of degree centrality were observed in the right middle frontal gyrus (MFG.R), while nodal efficiency was reduced in the right superior frontal orbital part (ORBsup.R) and right insula (INS.R) and nodal local efficiency decreased in the left precentral gyrus (PreCG.L), all compared with HCs. In contrast, patients with NPSLE showed diffuse decreases in degree centrality, nodal efficiency and nodal local efficiency, predominantly affecting frontal regions (bilateral dorsal/superior frontal gyrus (SFGdor), MFG, supplemental motor area (SMA), PreCG, left olfactory (OLF.L)), parietal regions (eg, left precuneus (PCUN.L), bilateral postcentral gyrus (PoCG), bilateral angular (ANG)) and cingulate and subcortical nuclei (eg, bilateral pallidum (PAL), left putamen (PUT.L)). Patients with NPSLE also exhibited diffuse increases in nodal cluster coefficient in frontal regions (eg, left medial/superior frontal gyrus (SFGmed.L), bilateral right inferior frontal triangular part (IFGtriang.R), OLF.R, bilateral INS), temporal regions (eg, bilateral superior temporal pole (TPOsup), bilateral Rolandic operculum (ROL), left Heschl gyrus (HES.L)), occipital regions (eg, left middle occipital gyrus (MOG.L), right inferior occipital gyrus (IOG.R), right calcarine (CAL.R), right lingual (LING.R)) and subcortical nuclei (eg, bilateral parahippocampus (PHG), right thalamus (THA.R), right amygdala (AMYG.R)), compared with HCs.
In terms of structural connectivity, patients with non-NPSLE demonstrated decreased connectivity between brain regions, particularly involving frontal areas (eg, SFGmed.L, SFGdor.R, IFGtriang.R, MFG.R, OLF.R, parietal regions (eg, PCUN.L), occipital lobes (eg, IOG.R), cingulate, and subcortical nuclei (eg, PUT.R), compared with HCs. On the other hand, patients with NPSLE exhibited diffuse decreases in structural connectivity involving frontal regions (eg, SFGmed.L, SFGdor, IFGtriang.R, MFG, OLF.L, PreCG, bilateral=SMA), parietal regions (eg, PoCG, PCUN, SPG.R, ANG.R), temporal regions (eg, HES.R, STG.R) and cingulate and subcortical nuclei (eg, PUT.L).
Furthermore, compared with patients with non-NPSLE, patients with NPSLE showed diffuse decreases in structural connectivity involving frontal regions (eg, SFGdor, MFG.L, OLF.L, PreCG.R, bilateral SMA) and parietal regions (eg, PoCG.R), as well as cingulate and subcortical nuclei (eg, PUT.R).
Lesion association with structural network features
As indicated in table 2, lesion volumes exerted significant associations with both global and local network parameters. Specifically, they exhibited a negative correlation with local network parameters such as network edges, nodal degree centrality and efficiency in parietal areas (eg, SMA), while demonstrating positive correlations with cluster coefficients in specific areas including frontal, occipital and subcortical regions.
Clinical correlations with structural network features
Clinical correlation of structural network features
Table 2 illustrates the associations between structural network characteristics and clinical variables. Age at onset exhibited weak negative correlations with the structural network edge (fibre numbers) in PCUN.L and PCUN.R (r=−0.36; p=0.003), nodal cluster coefficients in ROL.L (r=−0.40; p=0.003) and INS.R (r=−0.52; p<0.001), as well as the averaged cluster coefficient (r=−0.43; p=0.002). Conversely, it showed a weak positive correlation with nodal efficiency in SMA.R (r=0.41; p=0.003).
Disease duration was weak negatively correlated with the structural network edge in PCUN.L and PCUN.R (r=−0.36; p=0.002), nodal efficiencies in SMA.L (r=−0.41; p=0.004) and SMA.R (r=−0.43; p=0.002), and weak positively correlated with nodal efficiencies in ROL.L (r=0.41; p=0.003), INS.R (r=0.53; p<0.001) and PHG.R (r=0.40; p=0.004), as well as the averaged cluster coefficient (r=0.44; p=0.002).
SDI showed weak negative correlations with the structural network edge in DCG.L and PCG.L (r=−0.43; p=0.002), degree centrality in SFGdor.R (r=−0.40; p=0.005), weak positive correlations in ROL.R and HES.R (r=0.49; p=0.001), nodal efficiency in IFGtriang.R (r=0.41; p=0.004), ACG.R (r=0.45; p=0.001), PHG.R (r=0.47; p<0.001), CAL.R (r=0.45; p=0.001), MOG.L (r=0.41; p=0.004) as well as the averaged cluster coefficient (r=0.45; p=0.001) and moderate positive with the network edge in INS.R (r=0.52; p<0.001) and TPOsup.R (r=0.50; p<0.001). Cumulative steroid dose exhibited a weak negative correlation with degree centrality in SFGdor.R (r=−0.41; p=0.004), and a weak positive correlation with nodal efficiency in INS.R (r=0.43; p=0.002).
Identification of SLE by structural network features
According to table 3, using structural network features (including structural network edges, global and local network features), the identification of non-NPSLE and NPSLE from HCs achieved accuracies of 81% and 96%, respectively. The identification of NPSLE from non-NPSLE achieved an accuracy of 87%, with a sensitivity of 88% and specificity of 86%.
Identification of SLE by structural network characteristics using SVM classification
Discussion
In this study, we investigate the topological structural characteristics in patients with SLE with and without neuropsychiatric symptoms. The main findings can be summarised as follows: (1) Patients with both non-NPSLE and NPSLE exhibited global and local topological alterations compared with HCs, including decreased global and local network efficiency, reduced nodal degree centrality and structural connections involving default mode network (DMN) and sensorimotor network (SMN) related brain areas. Notably, patients with NPSLE showed significantly decreased structural connections in the prefrontal gyrus (eg, SFGdor and MFG) and cingulate and motor areas (eg, SMA and PreCG) compared with patients with non-NPSLE. (2) The observed topological alterations in patients with SLE correlated with clinical measures such as age at onset, disease duration, SDI score and cumulative steroids dose. (3) These topological features in patients with SLE contributed to distinguishing NPSLE from non-NPSLE with an accuracy of 87%.
Topological alterations in patients with SLE, particularly in NPSLE, are consistent with previous findings.9 13 15 18 20 Contradictory findings regarding non-NPSLE have been reported in some studies, showing no differences between non-NPSLE and HCs or between non-NPSLE and NPSLE, potentially due to different structural network weighting strategies such as FA.13 23 31 The distributed structural topologies observed in both non-NPSLE and NPSLE can be attributed to underlying white matter demyelination and axonal loss,23 which result from pathophysiological substrates including microvasculopathy, autoantibodies and inflammatory mediators observed in both patient groups.32–35 Additionally, disruptions in distributed normal-appearing white matter integrity, possibly due to Wallerian degeneration secondary to brain lesions, contribute to altered structural networks, particularly evident in patients with NPSLE.23 36 As a result, structural connections between/within cortical and subcortical regions (eg, frontal, parietal, temporal areas, cingulate, subcortical nuclei) degrade, leading to functional deficits reported in previous SLE studies, including working memory degeneration, learning deficits, attention, executive function impairment, language processing difficulties and motor dysfunction.13 20 In NPSLE, significant global and local structural topological changes (eg, decreased fibre connections between distributed brain regions, reduced global and local network efficiency) involving the DMN and SMN likely contribute to various clinical neuropsychiatric manifestations.14 18 However, it is noteworthy that nodal clustering coefficients increased in patients with NPSLE, particularly in lesion-free brain areas (eg, SFGmed, IFGtriang, OLF, TPOsup, ROL, HES, MOG.L, IOG.R, CAL.R, LING.R, PHG, THA, AMYG), which appears contradictory to some previous findings13 15 20 37 and warrants further validation. For patients with non-NPSLE, mild structural topological changes may account for subtle functional deficits without obvious clinical manifestations, supporting the hypothesis that local changes may not reach a significant threshold to trigger more pronounced functional alterations.13 Nevertheless, patients with non-NPSLE still exhibited global deficits (eg, in global efficiency), indicating underlying whole-brain structural and functional impairments even in the absence of clinical symptoms.15 These specific topological features in patients with SLE, beyond brain lesions, hold promise as biomarkers for objective clinical evaluation and contribute to a deeper understanding of the underlying pathological mechanisms of SLE.
In this study, clinical associations of the aforementioned topological structural alterations in patients with SLE were investigated. Disease duration and SDI score showed significant correlations with both global and local structural network measures (eg, averaged cluster coefficient, degree centrality and nodal efficiency in SMA), suggesting that network properties could serve as markers for monitoring disease progression and damage severity.13 Additionally, age at onset and cumulative steroids dose exhibited some associations with structural network measures in specific brain regions (eg, INS), indicating potential effects of disease onset age and treatment on brain structural topology. These findings underscore the identification of specific neural biomarkers for SLE, which could be used for disease monitoring and as potential targets for therapy.
Of note, the topological structural features in SLE contribute to the objective evaluation and identification of NPSLE, potentially offering advantages over clinical subjective judgement.8 In this study, we achieved a classification accuracy of 87% in objectively identifying patients with NPSLE from non-NPSLE, which is comparable to results reported in previous studies.5 38 39 These findings could assist clinicians in making informed clinical decisions such as diagnosis and determining appropriate treatments.
However, there are several limitations to this work. First, despite adopting FDR correction for multiple comparisons, the small sample size in this study may reduce the confidence in the current findings. The smaller sample size showed differences in some of the baseline clinical features (eg, disease duration, medications and comorbidities) between the two groups, which may increase bias. Second, incorporating multimodal MRI techniques (eg, functional MRI, arterial spin labelling, quantitative susceptibility mapping) could provide additional imaging biomarkers related to function, perfusion and metabolic alterations (iron-related and phospholipid-related), which would enhance the characterisation of brain alterations in patients with SLE. Lastly, this study was cross-sectional; therefore, further longitudinal follow-up studies are needed to validate these imaging biomarkers for disease monitoring and prognosis.
Conclusion
Structural networks can be significantly altered at both global and local levels in patients with SLE, particularly in those with NPSLE, prominently affecting the DMN and SMN. These alterations represent promising biomarkers for clinical diagnosis and could potentially aid in evaluating treatment responses and stratifying patients for clinical trials.
Data availability statement
Data are available upon reasonable request.
Ethics statements
Patient consent for publication
Ethics approval
Approval was obtained from the Human and Animal Ethics Committee of the Peking Union Medical College Hospital. The procedures used in this study are in accordance with the ethical standards of the 1964 Declaration of Helsinki and its later amendments. Participants gave informed consent to participate in the study before taking part.
References
Supplementary materials
Supplementary Data
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Footnotes
XZ, ZZ and YL are joint senior authors.
Contributors FA was responsible for the statistical analyses and manuscript drafting. LS was responsible for the study design, patient recruit, data acquisition and manuscript editing. YD and JX help the data acquisition and manuscript editing. JH and XQ help the MRI data acquisition. XZ, ZZ and YL were responsible for the study design, manuscript review and final approval of this manuscript. YL is responsible for the overall content as guarantor.
Funding National Science Foundation of China, Grant/Award Numbers: 81571597, 81571631, 81870958; The Chinese National Key Research R&D Program, Grant/Award Numbers: 2017YFC0907601, 2017YFC0907602, 2017YFC090760.
Competing interests None declared.
Provenance and peer review Not commissioned; externally peer reviewed.
Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.