Article Text
PDF
Abstract
Background Cognitive dysfunction (CD) is highly prevalent in systemic lupus erythematosus (SLE), yet the underlying mechanisms are poorly understood. Neuroimaging utilising advanced MRI metrics may yield mechanistic insights. We conducted a systematic review of neuroimaging studies to investigate the relationship between structural and diffusion MRI metrics and CD in SLE.
Methods We systematically searched several databases between January 2000 and October 2023 according to Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines. Retrospective and prospective studies were screened for search criteria keywords (including structural or diffusion MRI, cognitive function and SLE) to identify peer-reviewed articles reporting advanced structural MRI metrics and evaluating CD in human patients with SLE.
Results Eighteen studies (8 structural MRI, 9 diffusion MRI and 1 with both modalities) were included; sample sizes ranged from 11 to 120 participants with SLE. Neurocognitive assessments and neuroimaging techniques, parameters and processing differed across articles. The most frequently affected cognitive domains were memory, psychomotor speed and attention; while abnormal structural and/or diffusion MRI metrics were found more consistently in the hippocampus, corpus callosum and frontal cortex of patients with SLE, with and without clinically diagnosed central nervous system involvement.
Conclusion Advanced structural MRI analysis can identify total and regional brain abnormalities associated with CD in patients with SLE, with potential to enhance clinical assessment. Future collaborative, longitudinal studies of neuroimaging in SLE are needed to better characterise CD, with focus on harmonised neurocognitive assessments, neuroimaging acquisitions and postprocessing analyses and improved clinical characterisation of SLE cohorts.
- Systemic Lupus Erythematosus
- Magnetic Resonance Imaging
- Psychology
- Prevalence
- Health-Related Quality Of Life
Data availability statement
All data relevant to the study are included in the article or uploaded as supplementary information.
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
- Systemic Lupus Erythematosus
- Magnetic Resonance Imaging
- Psychology
- Prevalence
- Health-Related Quality Of Life
WHAT IS ALREADY KNOWN ON THIS TOPIC
Cognitive dysfunction is a common symptom for patients with systemic lupus erythematosus (SLE), yet often difficult to attribute to neuropsychiatric SLE (NPSLE) using existing clinical tools such as conventional MRI. Utilising advanced neuroimaging to examine structural brain abnormalities that may increase our understanding of the mechanisms underlying cognitive dysfunction in SLE.
WHAT THIS STUDY ADDS
The results of this literature review examining structural MRI abnormalities in SLE and their associations to cognitive dysfunction indicate that memory and attention are the most consistently impaired cognitive domains, associated with abnormalities in the hippocampus and corpus callosum. Additionally, longer disease duration and higher cumulative glucocorticoid doses were reported in association with abnormal brain structural metrics in patients with SLE.
HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY
The study identifies current gaps in the application or neuroimaging to understand cognitive dysfunction in NPSLE, and highlights opportunities to optimise utilisation of advanced neuroimaging through harmonised protocols and pipelines for collaborative research in NPSLE. It also highlights the need for more longitudinal studies, and additional research including children with SLE.
Introduction
Neuropsychiatric systemic lupus erythematosus (NPSLE) remains one of the most challenging manifestations of SLE due to the broad spectrum of symptoms (up to 19 clinical syndromes affecting either the central—CNS or the peripheral nervous system) and limited understanding of the underlying disease neurobiology.1 These syndromes can be categorised into focal or diffuse, with the latter manifestations showing highest prevalence, mainly as cognitive dysfunction (CD, ~80%)2 3 and mood disturbances (~65%).1 3 In addition, neuropsychiatric involvement occurs more frequently in childhood-onset SLE (cSLE, up to 95%) than in adult-onset SLE (aSLE, 11%–81%), and it strikes during a critical period of neurodevelopment that may potentially lead to irreversible negative impact in cognitive function.4
CD in SLE has been defined by the American College of Rheumatology (ACR) as a significant deficit in one or several of the following domains: attention, reasoning, executive skills, memory, visual-spatial processing, language and psychomotor speed,5 with attention and memory among the most regularly affected cognitive domains.6 7 CD due to NPSLE could be a consequence of several pathologic mechanisms related to vascular involvement, blood–brain barrier breach and cell-mediated inflammation.8 Yet, NPSLE pathogenesis is still poorly understood, making diagnosis and monitoring particularly challenging. CD may also be due to other factors and, therefore difficult to attribute to NPSLE.7 8
MRI is the gold-standard neuroimaging tool to diagnose and monitor NPSLE.1 9 However, conventional structural MRI abnormalities, such as white matter (WM) hyperintensities, gross brain tissue atrophy and ventricular enlargement in response to this atrophy are not always observed in patients with NPSLE.10 In contrast, more advanced postprocessing techniques enable the quantification of structural brain metrics beyond total tissue volumes from standard T1-weighted MRI, such as regional volumes, surface area and cortical grey matter (GM) thickness.11 12 These metrics can be semiautomatically calculated from human brain atlases and can reveal subtle NPSLE-related brain abnormalities (eg, frontal cortex atrophy), not apparent with more conventional clinical tools.10–12
In addition to these postprocessing methods, other less conventional structural MRI sequences, specifically diffusion MRI, have become a subject of interest in current NPSLE clinical research.10 13 Diffusion MRI measures the random motion of water molecules (diffusion coefficient D) while they interact with tissue boundaries, cell membranes and other biological barriers, yielding structural metrics linked to axonal loss, inflammation and demyelination, particularly in the WM14; thus, it can probe the status of brain tissue microstructure in relationship to neurological symptoms. Common metrics include the mean (MD), axial (AD) and radial (RD) diffusivities, which respectively characterise water diffusion in bulk, parallel and perpendicular to a WM tract, and fractional anisotropy (FA), which quantifies the degree of diffusion anisotropy or directionality in a voxel.15 16 Additionally, diffusion-weighted imaging (DWI) can be used to weight the strength of WM connections by quantifying properties of brain-wide structural networks (eg, node strength, density),17 and to evaluate intravoxel incoherent motion (IVIM), where the microcirculation of blood–water in the capillary network (D*), and tissue D and perfusion can be estimated.18
An increasing number of studies are utilising advanced MRI to investigate NPSLE.10 Altered tissue microstructure has been reported in several brain regions of patients with and without NPSLE diagnosis when compared with healthy controls,11 19–21 and they have correlated with higher CD.20 Overall, the presence of these associations, even in the absence of clinical NPSLE diagnosis, suggests that brain involvement could be underdetected in SLE. However, existing advanced neuroimaging research in SLE has been limited in generalisability and interpretation due to small cohorts and often incomplete characterisation of clinical features. In response to these knowledge gaps, we conducted a systematic review to evaluate (1) the effect of SLE on brain structure, (2) the neuroimaging correlates of CD in SLE and (3) potential disease-related contributors including but not limited to disease activity, duration and glucocorticoid exposure. A better understanding of these associations will help inform attribution of CD to SLE, characterisation of domain-specific CD trajectories, and possible new therapeutic strategies for protection of cognitive function in children and adults diagnosed with SLE.
Methods
Search strategy
This systematic review was conducted in agreement with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines.22 The search was performed in PubMed, MEDLINE, Embase, Web of Science, Google Scholar and Cochrane databases, and it was aided by the web-based literature review manager Covidence. It included the following terms: ‘systemic lupus erythematosus (SLE)’, OR ‘neuropsychiatric lupus (NPSLE)’, OR ‘central nervous system (CNS) lupus’, OR ‘antiphospholipid syndrome SLE’; AND ‘magnetic resonance imaging (MRI, structural MRI)’; OR ‘diffusion MRI (diffusion tensor imaging—DTI, DWI)’. More details on the search strategy are included in online supplemental file 1. The systematic review was registered into the Registry of Systematic Reviews/Meta-Analyses in Research Registry (https://www.researchregistry.com/) with Review Registry Unique Identifying Number ‘reviewregistry1844’.
Supplemental material
Inclusion and exclusion criteria
Inclusion criteria: (1) peer-reviewed articles, limited to human research, and published between January 2000 and October 2023, including observational, case series, cross-sectional, longitudinal, retrospective or prospective study designs of SLE populations; (2) neuroimaging studies that used structural (T1-weighted) and/or diffusion MRI; (3) evaluation of cognitive function/performance in SLE.
Exclusion criteria: (1) reviews, meta-analyses and manuscripts that do not refer to MRI data directly collected from SLE cohorts; (2) studies solely reporting conventional T1-weighted MRI metrics, such as total brain volumes (and not regional volumes), lateral ventricles volume and/or WM hyperintensities numbers/volumes.
Identification of eligible studies
Title and abstracts were reviewed for eligibility by six team members (DVC, TET, IM, SEA, SF and JL). A full-text review of potentially eligible articles according to inclusion and exclusion criteria was undertaken independently by two members (DVC and TET), and afterwards final articles were selected by consensus.
Data extraction and risk of bias assessment
Information extracted from studies included main publication details (first author, year, country), study design (cross-sectional, longitudinal), cohort demographics (sample size, age, sex, ethnicity), clinical variables (NPSLE clinical diagnosis, disease duration, SLE Disease Activity Index—SLEDAI scores, Systemic Lupus International Collaborating Clinic Damage Index – SDI scores, glucocorticoid use), CD assessments (cognitive domains, neuropsychological tests, CD definitions), MRI technical details (magnetic fields, voxel sizes, b-values, diffusion directions), protocol type (structural or diffusion MRI sequences) and structural (total GM, WM and regional volumes; cortical thickness) and diffusion MRI metrics (FA, MD, AD RD). Two reviewers (DVC and TET) individually extracted data from the included articles regarding associations between atypical structural and/or diffusion brain MRI metrics and CD in SLE. Associations between brain MRI abnormalities and other clinical variables were summarised when available. Risk of bias assessment was performed by two independent raters (DVC and TET) in agreement with The National Heart, Lung, and Blood Institute Study Quality Assessment Tools for Observational Cohort and Cross-Sectional Studies (https://www.nhlbi.nih.gov/health-topics/study-quality-assessment-tools) and consensus was reached after discussions including the senior author (AMK) in case of major disagreements.
Results
A total of 18 articles that evaluated the effect of SLE on brain structure and their links with CD were included in this review (online supplemental file S1).12 19–21 23–36 From these studies, eight used T1-weighted MRI,12 28 29 31–33 35 36 nine focused on diffusion MRI19–21 24–27 30 34 and one reported brain metrics from both modalities.23 Most studies (15) were cross-sectional in design, while three studies assessed longitudinal changes in brain structure in their SLE cohorts at two different time points.25 31 35 In terms of methodological quality and risk of bias assessment, half of the studied articles (9) were rated with an overall fair quality (unclear risk of bias), one-third (6) were rated as poor quality (high risk of bias), and three studies were rated as good quality (low risk of bias). Four of the six articles at high risk of bias had measuring bias (statistical analyses not adjusted for key potential confounding variables) while two articles had possible cohort selection bias (healthy and SLE cohorts with different demographic characteristics). Risk of bias assessments for each article are summarised in online supplemental figure S2.
Sample sizes, demographics and clinical features
An overview of demographic and clinical characteristics of the SLE cohorts is presented in table 1. SLE sample sizes ranged from 11 to 120 with a mean of 50 participants, and in all cohorts at least 80% were women (17 out of 18 papers reported biological sex as part of patient’s demographics), which is in line with the highest prevalence of SLE in women.37 Three manuscripts reported information regarding race/ethnicity, and in two of them over half of the patients were African-American individuals (55% and 70% of their respective cohorts).25 27 Cohort ages ranged 14.7–48.9 years on average with a pooled mean of 35.5 years; 16 studies examined adult patients with SLE, 2 included patients with cSLE27 28 and one compared patients with aSLE versus cSLE.26 Healthy controls with age and sex distributions comparable to their respective SLE cohorts were used for group comparisons in 16 studies. Only three studies longitudinally evaluated 56%, 65% and 100% of their SLE cohorts at follow-up time points that ranged 12–19 months from the date of their respective baseline MRI scans.25 31 35
Demographic and clinical characteristics of SLE cohorts (least to most recent publication date)
The number of patients with clinical NPSLE diagnosis was reported in 16 cohorts and it was quite variable (ranged 0%–100% of their respective cohorts), with half of these cohorts describing proportions over 50%, and three studies mentioning CD as a previous clinical NPSLE symptom. SDI ranged from 0 to 2.3 and it was the most frequently reported clinical variable (12 studies). Nine articles reported SLEDAI ranging from 2.0 to 15.3, whereas one manuscript used the systemic lupus activity measure (9.9±4.9). Disease duration was stated in 16 papers and spanned 0.5–38 years since SLE symptom onset. Additionally, information regarding glucocorticoid exposure, retrieved from either current or cumulative use, was provided in 14 papers, and in 10 of them over 75% of patients were exposed to some form of glucocorticoids.
Neurocognitive assessments
Studies assessed cognitive performance of SLE patients with diverse approaches classified into the following categories: neuropsychological batteries, computerised batteries, screening tests and incomplete/mixed designs.38 A full neurocognitive characterisation of the reviewed SLE cohorts is available in table 2.
Neurocognitive characterisation, cognitive domains, neuropsychological assessments and cut-offs for cognitive dysfunction (CD) classification in SLE cohorts
Comprehensive neuropsychological batteries comprised at least four tests sensitive to specific cognitive domains (≥2 assessed domains) that have been validated in SLE (including the ACR-SLE, and the modified version endorsed by the Childhood Arthritis & Rheumatology Research Alliance).38 These included, for example: Wechsler Intelligence Scales (Adult—WAIS-3 and WAIS-4, Children—WISC-4 and Abbreviated—WASI in 7 studies, Continuous Performance Test (simple attention, 8 studies), Stroop Color-Word Test (attention and working memory, 5 studies), California Verbal Learning Test and Rey Auditory Verbal Learning Test (RAVLT)—verbal memory and learning, 3 studies each, Trail Making Test (visual-spatial processing, psychomotor speed and memory, 3 studies), Hayling Test (executive skills, specifically response initiation and inhibition, 3 studies).
Six studies included neuropsychological batteries.23 24 27 28 30 36 While the above standard neuropsychological batteries were still the most frequently employed category, they require lengthy assessments that must be administered by clinical psychologists.8 Thus, computerised batteries, such as CNS Vital Signs (3 studies) and Automated Neuropsychological Assessment Metrics (2 studies), have become more popular in SLE and they were used in five manuscripts.19 21 25 29 33 These computerised batteries are shorter than traditional neuropsychological batteries, have demonstrated high sensitivity and specificity with the ACR-SLE battery and can be administered by less clinically specialised staff.38 39
Screening tests are recurrently used to evaluate multiple domains and provide a quick global estimate of cognitive function. These screening tools were applied in six studies: Mini Mental State Examination (6 studies),20 30–32 34 35 Montreal Cognitive Assessment (3 studies),20 26 34 Addenbrooke’s Cognitive Examination (3 studies),20 32 34 and National Adult Reading Test (2 studies).20 34 The use of screening tests has increased during the last decade as they might complement subjective assessments and guide additional neuropsychological tests to evaluate specific cognitive domains of interest in patients with SLE. Mixed/incomplete designs that included a combination of these screening tests and/or less than four domain-specific neuropsychological tests were reported in three papers.12 31 35
Lacking a generally accepted definition of CD in SLE, cut-offs for related deficits were selected in seven studies by choosing standardised z-scores −1.0 to −2.0 SD below the normative mean, either in at least two individual cognitive domains (if cut-off was lower than −1 SD)27 28 or in only one domain (if cut-off was lower than −1.5 or −2 SD).12 27 28 30 31 35 36 Two of these seven studies exclusively focused on verbal memory (RAVLT).12 30
MRI technical details and brain structural findings
Technical parameters, postprocessing and group differences in MRI metrics are available in online supplemental table S1. In 10 studies, brain MRI data were acquired with 3T scanners, while remaining studies were acquired at lower magnetic fields (2 studies at 2 T and 6 studies at 1.5 T). Regional volume was the most commonly evaluated structural MRI metric (8 out of 9 studies) and it was mainly calculated from automated segmentations (4 studies) and voxel-based morphometry (3 studies). Both postprocessing techniques employ semiautomated algorithms for volume quantification, reducing measurement susceptibility to operator skill. Additionally, segmentation methods (manual and automated) were targeted to specific brain regions in three studies (two in hippocampus, one in hippocampus and corpus callosum), therefore these studies did not evaluate potential abnormalities across the entire brain.29 32 35 Diffusion MRI metrics from 9 DTI studies included FA (8 studies), mean and directional diffusivities (MD, RD, AD—4 studies) as well as FA-weighted global and local brain structural connectivity metrics (1 study).34 They were mainly computed with tractography, tract-based spatial statistics or voxel-wise analyses. One tractography study focused on the corpus callosum, cingulum and uncinate fasciculus tracts19 while another study solely retrieved metrics from corpus callosum automated segmentations.26 IVIM-derived diffusion-perfusion metrics were examined in one study.27
Supplemental material
Structural MRI: a summary of the structural MRI abnormalities in SLE reported across the studies is depicted in figure 1, including T1-weighted MRI segmentation maps of GM and WM structures and atlas-based parcellations of cortical structures. Lower total GM volume was reported in two SLE cohorts when compared with healthy controls. The hippocampus was the most frequently affected structure bilaterally, with smaller volumes reported in patients with SLE relative to controls (2 studies),32 35 and worse hippocampus atrophy observed in patients with NPSLE (1 study)29 or CD (1 study).36 The next most frequent abnormalities were smaller frontal (2 studies) and temporal (2 studies) GM regions in adults and children with SLE and CD.28 31 Additionally, regions within the frontal, temporal and parietal cortices of patients with SLE with memory deficits were thinner when compared with patients without memory deficits and controls.12 Regarding longitudinal assessments, one study showed that the percentage of the SLE cohort with hippocampus atrophy increased by 23% when follow-up versus baseline MRI volumes were compared (from 47 out of 107 patients with hippocampus atrophy at baseline to 40 out of 60 patients at follow-up),35 while lower corpus callosum, frontal, dorsolateral and medial temporal cortical volumes were reported in patients relative to controls during over a year follow-up period in another longitudinal study.31
Shown are T1-weighted MRI segmentation maps of GM and WM structures (A), and atlas-based parcellations of cortical structures (B) summarising findings from the reviewed articles. Compared to controls, lower total white matter, grey matter, cerebellum and hippocampus (hippo) volumes (A), and lower cortical volumes (particularly frontal cortex) and thickness (B) were commonly observed in patients with SLE (sagittal views). Structural MRI metrics were displayed from automatic segmentations and parcellations labelled from a human brain atlas (Desikan-Killiany). Beyond general cognitive function, memory, psychomotor speed and attention were the most frequently affected cognitive domains found to be associated with structural MRI findings in SLE (C). GM, grey matter; SLE, systemic lupus erythematosus; WM, white matter.
Diffusion MRI: a summary of the microstructural brain abnormalities on diffusion MRI in SLE reported across the studies in depicted in figure 2, including WM atlas-based segmentations superimposed in a colour-encoded FA map and an example of a 3D-rendered tract. The corpus callosum was the WM pathway most frequently damaged in SLE (6 studies), with lower FA (5 studies), higher MD and RD (2 studies) and higher AD and free-water (1 study) in patients with SLE when compared with controls.19 20 23 25 26 30 Other frequently affected WM pathways were the cingulum19–21 25 and fascicles connecting the frontal cortex in four studies (inferior fronto-occipital fasciculus, inferior and superior longitudinal fasciculus, uncinate fasciculus),21 23 25 30 and projection tracts in three studies (thalamic radiation, internal and external capsule, corticospinal tract, corona radiata).20 23 30 One study evaluated diffusion metrics in GM and reported abnormalities in patients with cSLE, mainly higher D and D* in the precuneus/cuneus, occipital, postcingulate and parietal regions.27 Within SLE subgroups, one study reported large affected areas in patients with SLE with and without memory deficits versus controls (lower FA and higher MD/RD).30 Another study evaluated patients with cSLE versus aSLE and reported lower FA and higher diffusivities in the corpus callosum in the former relative to the latter subgroup,26 while no diffusion differences were observed between NPSLE and non-NPSLE patients in two manuscripts.19 24 No changes in diffusion metrics in patients with SLE after the follow-up period were observed in the DTI publication with longitudinal data.25
Shown are brain maps (sagittal, coronal and axial views) depicting reported abnormal widespread white matter microstructure (A). Compared to controls, lower fractional anisotropy and higher diffusivities were reported in patients with SLE, with a predilection for the corpus callosum (CC) and periventricular/frontal tracts. Diffusion MRI metrics were mostly calculated with tractography (3D-rendered genu of the corpus callosum viewed from the top, A) or from WM tract labels with tract-based spatial statistics. Beyond general cognitive function, memory, psychomotor speed and attention were the most frequently affected cognitive domains found to be associated with diffusion MRI findings in SLE (B). Cg, cingulum; CR, corona radiata; IC/EC, internal/external capsula, IFOF, inferior fronto-occipital fasciculus; ILF/SLF, inferior/superior longitudinal fasciculus; WM, white matter.
Cognitive function and relationships between brain structural metrics
Poor overall cognition function was the most frequently reported metric of CD (8 studies). The cognitive domains most consistently impaired in SLE were: memory (visual, verbal, working, episodic, composite) in five articles,12 24 25 31 35 followed by psychomotor speed and attention (simple, complex, sustained) in four19 24 27 33 and three studies,21 24 31 respectively. One longitudinal study reported that the prevalence of CD after the follow-up period remained the same,25 while it increased by 34% (5 patients) after the follow-up period in another longitudinal study.35 Brain structures/regions implicated in these cognitive domains, effect measures of these associations and the direction of the effects are summarised in table 3, and their anatomic locations are shown in figures 1 and 2.
Associations between brain structural MRI metrics, cognitive performance and other clinical features in SLE cohorts of previous studies
With regard to structural brain MRI metrics, lower GM and WM volumes were associated with worse overall cognitive function and a greater number of impaired cognitive domains in patients with SLE.31 Smaller volumes in the temporal and frontal lobes were associated with worse composite memory, while smaller volumes in the parietal lobe correlated with worse attention.31 Abnormal hippocampus metrics frequently related to CD (3 studies), with smaller volumes linked with worse overall cognitive function,32 35 memory (composite and verbal) and recall.35 Other cortical and subcortical areas showing associations with CD were frontal and precentral cortex (lower thickness with lower episodic memory) and the cerebellum (lower volumes and slower psychomotor speed).12 33
With regard to diffusion MRI metrics, for total WM, higher MD was associated with lower overall cognitive function.20 The corpus callosum was the WM structure that was most recurrently associated with CD (3 studies), with higher MD26 and lower FA24 correlating with worse overall cognitive function, lower FA24 and higher free water21 being associated with poor attention, and lower FA with visual memory and psychomotor speed.24 Lower FA in the cingulum associated with slower psychomotor speed and worse cognitive flexibility (executive functioning domain),19 while higher free water was associated with worse attention.21 Lower FA in anterior portions of the corona radiata (2 studies), thalamic radiation (1 study) and right external capsula (1 study) correlated with lower overall cognitive function24 and executive skills,23 with anterior thalamic radiation also correlating with worse processing speed.24 Lower FA in the superior longitudinal fasciculus (2 studies) was associated with lower overall cognitive function, visual memory, psychomotor speed, attention24 and executive skills.23 Lower FA in the inferior fronto-occipital and longitudinal fasciculus, respectively, correlated with lower visual and verbal memory in one study.24 Additionally lower parahippocampal FA related to worse spatial memory,25 higher node strength in the frontal cortex and in caudal/lingual regions, respectively, correlated to greater overall cognitive function and lower episodic memory,32 and higher IVIM-derived perfusion in the precuneus associated with slower psychomotor speed and worse visual spatial processing.25
Associations between clinical variables and abnormal brain structure (n=10) and CD (n=1)
Longer disease duration was related to lower total GM and WM volumes,31 weaker network connectivity metrics in the whole brain34 and lower FA in the corpus callosum.19 Greater SLE damage was associated with lower nodal strength in caudate and in several cortical regions in all brain lobes,34 higher disease activity with higher water diffusion and lower blood–water fraction in precuneus,27 higher levels of DNRAb serum titres with lower FA in parahippocampal areas25 and greater number of CNS manifestations with lower hippocampus volume35 (table 3). Cumulative glucocorticoid dose was linked to lower GM volumes31 and higher free water in the corpus callosum, cingulum and WM tracts connecting to the frontal cortex.21 Greater fatigue was linked to higher MD in total WM20 and lower FA in the corpus callosum of patients with SLE, and to higher MD in the cingulum of patients with NPSLE.19 One study reported worse overall cognitive function negatively associated with greater disease duration, higher expression of the inflammatory cytokine interleukin-6 and higher level of endothelial dysfunction antigens and activity (von Willebrand factor).20
Discussion
To our knowledge, this is the first systematic review of neuroimaging literature on structural MRI abnormalities in SLE in relationship to CD. Our work included 18 peer-reviewed manuscripts and summarised their results in terms of clinical characterisations of SLE cohorts, neurocognitive assessments and CD definitions, structural brain MRI abnormalities and their links to CD and other related clinical factors in patients with SLE. We found that memory and attention as well as psychomotor speed were the most consistently impaired cognitive domains in SLE when evaluated in relationship with structural brain alterations. CD in these domains correlated with abnormal MRI metrics (low volumes, abnormal microstructure), particularly in hippocampus and corpus callosum. Longer disease duration, higher cumulative glucocorticoid doses and fatigue were disease factors often linked to regional brain structure abnormalities.
CD was associated with injury in several areas, with a particular emphasis in periventricular and frontal WM pathways (eg, the corpus callosum), cortical frontal and parahippocampal regions and certain subcortical GM structures (eg, hippocampus) that are known to be involved in cognitive processes frequently affected in patients with SLE. Lower volumes and cortical thinning were observed in frontotemporal and hippocampal/parahippocampal regions in relationship to impairments in all cognitive domains in patients with SLE. Lower FA and higher diffusivities in WM tracts such as the corpus callosum and in subcortical structures, specifically the hippocampus, were used as indicators of microstructural brain alterations in NPSLE and non-NPSLE patients, and metrics in these regions were linked to poorer attention and impaired visual and working memory. Global and regional brain abnormalities in GM and WM were related to longer disease duration, NPSLE diagnosis, higher fatigue, greater disease activity and higher cumulative glucocorticoid use, which suggests that brain damage, specifically in periventricular regions, could worsen with progressive pathology directly or indirectly caused by SLE. These results aligned with a longitudinal study that showed that hippocampus volumes, although affected early in SLE, further decreased with time and in relationship to factors such as greater total glucocorticoid dose, CD and number of CNS manifestations.35
The hippocampus proper and interconnected parahippocampal and periventricular regions are critical for memory and executive skills.40 Neuronal injury in the hippocampus could extend to neighbouring WM tracts as a consequence of anterograde or retrograde axonal degeneration and this mechanism has been proposed as a mediator of CD in SLE7 and MS.41 Microstructural degeneration of these regions could be also related to their location in the brain, adjacent to cerebrospinal fluid and vascular spaces. This location makes them particularly vulnerable to SLE pathology and treatment, including but not limited to microglial activation, as reported from in vitro and mouse studies42 and glucocorticoid use.43 Microstructural alterations in these regions could precede regional and global brain atrophy, and both microstructural and macrostructural abnormalities could even lead CD in SLE. However, to our knowledge, there have been only three MRI studies longitudinally evaluating structural brain metrics,10 31 35 and one cross-sectional study that has combined both advanced structural and diffusion metrics in relationship to CD in SLE.23 Additionally, only two studies evaluated CD in paediatric SLE populations.27 28 Children with cSLE are at higher risk for developing CNS manifestations due to NPSLE, and they represent an opportunity to investigate the effects of SLE on the brain with little presence of comorbid conditions.44 Studies of CD utilising neuroimaging in cSLE may, therefore, provide particular insight into the mechanisms underlying the impact of SLE on the brain. In addition to structural brain abnormalities, factors that directly correlated to CD in patients with SLE included long disease duration and high expression of inflammatory cytokines. However, relationships between these factors and CD were only evaluated in one study.20
It is important to note that several methodological issues limited our interpretation of the findings. These include (1) inadequately described demographic and disease-related features in heterogeneous SLE cohorts; (2) infrequent accounting of potential confounders such as glucocorticoid use, mood disorders, fatigue, disease activity and duration; large variability in (3) neuropsychological assessments utilised to evaluate CD; (4) technical details in MRI scanners and acquisitions in studies in patients with SLE; (5) lack of harmonised neuroimaging analyses that combine structural MRI metrics from different modalities to evaluate both brain tissue morphology and microstructure and (6) lack of longitudinal data.
Importantly, the above limitations indicate a critical need for collaboration, consensus and coordination of research on CD in SLE, and for NPSLE overall. Our review highlights the current inconsistent use of neuropsychological tools, MRI protocols and consideration of demographic and disease-related variables potentially impacting brain function. Future efforts are needed to develop consensus recommendations and guidelines regarding relevant demographic, clinical and cognitive function measures and suitable technical MRI parameters and processing/postprocessing pipelines. Additionally, developing a standard for clinical collection of harmonised neuropsychological measures and neuroimaging protocols would allow for multicentre studies utilising clinically collected data. Furthermore, consideration and inclusion of paediatric cSLE populations will be highly valuable to understand the impact of SLE on the developing brain and across the lifespan. Such recommendations would enable larger, collaborative studies in SLE on neuroimaging, CD and NPSLE across the globe, utilising common data elements. These efforts will require multidisciplinary teams inclusive of adult and paediatric rheumatologists, neuropsychologists, neuroimaging scientists, neurologists, neuroimmunologists and others with relevant expertise.
In conclusion, the results collected in this systematic review suggest that advanced structural MRI metrics can identify CNS abnormalities in patients with SLE and CD. Specifically, this systematic review indicates that despite the limitations, the existing literature shows structural and microstructural changes in brain regions and networks that are known to be involved in many aspects of cognition. The use of these advanced structural MRI metrics might enhance the knowledge of the underlying causes of CD and other neuropsychiatric symptoms such as mood disorders in SLE, and they could help to develop more objective biomarkers for the attribution of specific NPSLE syndromes. Together with functional and metabolic neuroimaging tools, these structural metrics could also serve as complementary diagnostic tools of NPSLE as well as outcome measures in clinical trials focusing on therapeutic interventions and neuroprotection and preserving cognitive function in SLE. However, improved characterisation of SLE cohorts, guidelines for neuroimaging acquisitions and analyses and more longitudinal studies are needed to further confirm the diagnostic and predictive ability of these metrics in SLE-related CD.
Data availability statement
All data relevant to the study are included in the article or uploaded as supplementary information.
Ethics statements
Patient consent for publication
Ethics approval
Not applicable.
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
DVC and TET are joint first authors.
Contributors AMK is responsible for the overall content (as guarantor). DVC, TET and IM designed the search strategy. All authors (DVC, TET, IM, SEA, SF, JL, AMK) participated on the identification and screening of all the relevant articles. DVC and TET performed the full data extraction of eligible articles and the risk of bias assessment. DVC, TET and AMK wrote the first draft of the paper. All authors reviewed, contributed and approved the final manuscript.
Funding Funding for this project included a Lupus Foundation of America Career Development Award (TET), Lupus Research Alliance Administrative Supplement to Promote Diversity in Lupus Research (SEA), Canada Research Chair Tier 2 in Mental Health and Chronic Disease of Childhood (AMK) and Lupus Research Alliance Career Development Award to Promote Diversity in Lupus Research (AMK).
Competing interests None declared.
Patient and public involvement Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research.
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.