Evaluation of MRI-based brain oxygen extraction fraction mapping in patients with systemic lupus erythematosus

Abstract

Objective

We aim to assess the cerebral oxygen extraction fraction (OEF) changes in patients with systemic lupus erythematosus (SLE) and neuropsychiatric SLE (NPSLE) by using an MRI-based technique and examine the relationship between OEF and cognition.

Methods

43 SLE patients (18 NPSLE and 25 non-NPSLE) and 26 healthy controls (HC) were recruited. Cognitive function was assessed via the Montreal Cognitive Assessment (MoCA). OEF was calculated by quantitative susceptibility mapping plus quantitative blood oxygen level-dependent model (QQ). Whole-brain voxel-wise analysis of OEF was performed. In subcortical grey matter structures, regional OEF values were measured, and their relationship with MoCA scores was explored.

Results

Whole-brain voxel-wise analysis revealed significant changes of OEF primarily in the limbic system, including the orbitofrontal cortex and bilateral insular lobes, among HC, non-NPSLE and NPSLE groups. Regional analysis indicated reduced OEF values in subregions of the amygdala, hippocampus and caudate nucleus in non-NPSLE compared with HC, with decreasing trends observed in all selected regions of subcortical grey matter structures. In the right hippocampus, OEF values were increased in NPSLE patients compared with non-NPSLE patients. Considering all subjects in the study, OEF values in the bilateral medial amygdalae, right lateral amygdala, left rostral hippocampus and right dorsal caudate nucleus were positively correlated with MoCA scores.

Conclusion

Cerebral OEF mapping in patients with SLE is readily available using the MRI-based QQ method, which has the potential to serve as an adjunctive tool for diagnosing NPSLE and monitoring cognitive impairment in SLE.

Correspondence to Dr Shun Zhang; [email protected] ; Professor Wenzhen Zhu; [email protected]

WHAT IS ALREADY KNOWN ON THIS TOPIC

• Ischaemia and autoimmune-mediated neuroinflammation, which are principal contributors to cerebral damage in systemic lupus erythematosus (SLE), have been linked to alterations in the brain’s oxygen extraction fraction. Nevertheless, how global and regional brain oxygen metabolism changes in SLE remains unclear.

WHAT THIS STUDY ADDS

• Brain oxygen extraction fraction alterations in SLE occurred mainly in the limbic system. SLE patients with neuropsychiatric involvement exhibited higher oxygen extraction fraction values in hippocampal subregions than those without. Reduced oxygen extraction fraction in certain subcortical grey matter regions was associated with cognitive deterioration.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

• Non-invasive MRI-based cerebral oxygen extraction fraction assessment may provide evidence of pathological changes in SLE with and without neuropsychiatric involvement and has potential to be an adjunctive tool for diagnosis and management of neuropsychiatric lupus.

Introduction

Systemic lupus erythematosus (SLE) is an autoimmune disease characterised by multiorgan involvement, including the central nervous system (CNS).¹ Neuropsychiatric SLE (NPSLE) may affect up to half of patients with SLE, typically occurring within the first 3–5 years after disease onset.^{2 3} NPSLE presents with diverse clinical manifestations and is among the major causes of death in SLE.^2–4 Compared with other SLE-related complications, NPSLE exhibits a unique and complex pathogenesis and can occur independently of systemic disease activity.⁵

Brain atrophy and white matter lesions are frequently observed MRI findings in patients with NPSLE.⁶ However, it has been reported that up to 40%–50% of NPSLE patients have normal MRI findings,³ suggesting the limited sensitivity of conventional imaging for detecting neuropsychiatric (NP) involvement in SLE. Furthermore, in unselected SLE cohorts, white matter lesion burden correlates with age, global disease severity and duration rather than specifically with NP manifestations.⁷ A recent review of neuroimaging studies in NPSLE concluded that studies combining all NPSLE syndromes lack sufficient specificity to distinguish NPSLE from non-NPSLE patients.⁸ Advanced neuroimaging modalities, such as diffusion tensor imaging and resting-state or task-based functional MRI, have identified structural and functional brain network abnormalities as well as altered cortical activity in non-NPSLE patients.⁸ However, these findings may potentially reflect chronic disease-related injury and non-specific alterations. Despite technological advancements, neuroimaging specificity for NPSLE (ranging from 60% to 82%) remains limited.² Besides, current MRI techniques provide insufficient spatial resolution required to identify microvascular involvement, a pathology observed in up to 42% of SLE patients with neurological manifestations.⁹ While positron emission tomography (PET) provides objective insights into brain function, its clinical applications are often restricted by high costs and limited accessibility.^{9 10} These limitations underscore the need for more sensitive and accessible imaging biomarkers capable of detecting subtle or early cerebral metabolic disturbances in SLE.

Oxygen extraction fraction (OEF) is a critical physiological indicator that quantifies the amount of oxygen extracted by brain tissue from circulating blood.¹¹ It serves as a marker of the dynamic balance between cerebral oxygen delivery and metabolic demand.¹² Disruptions in this balance—whether due to hypoperfusion, mitochondrial dysfunction, or neuroinflammation—are increasingly recognised as contributors to neurological damage across diverse CNS diseases.^11–15 In ischaemic stroke, OEF decreased within the ischaemic core from the acute to subacute phase reflecting functional tissue changes during stroke progression.¹⁶ In multiple sclerosis (MS), reduced OEF was detected aligning with cumulative neuronal loss, while focally elevated OEF in the rim of chronically active MS lesions supports the presence of active inflammation.¹⁷ Notably, Alzheimer’s disease (AD) research revealed that cognitively intact individuals carrying the apolipoprotein-E4 allele, a genetic variant strongly linked to AD, exhibited reduced global cerebral OEF.¹⁸ As a result, cerebral OEF measurement could be a useful tool to evaluate tissue function and vulnerability.¹⁹

In SLE, NP manifestations are hypothesised to arise from two inter-related yet distinct mechanisms: ischaemic injury secondary to microvascular thrombosis and autoimmune-mediated neuroinflammation.^{2 20} This mechanistic duality is supported by brain fluorine-18 fluorodeoxyglucose PET studies in NPSLE, which demonstrated hypermetabolism (reflecting neuroinflammation) and hypometabolism (indicating metabolic suppression) in affected brain regions.²¹ Notably, both ischaemia and neuroinflammation are established modulators of OEF in other neurological conditions such as stroke, cerebral small vessel disease, MS and AD where they have been shown to cause increased OEF due to impaired oxygen delivery or increased metabolic demand.^{13 15 17} We, therefore, propose that OEF mapping may uniquely capture the cerebral oxygen metabolism changes of NPSLE.

Although PET imaging with [¹⁵O]-oxygen tracers remain the gold standard for evaluating OEF, it involves radiation exposure and the requirement of complex setup to produce ¹⁵O tracers, which have ultra-short half-life.²² Emerging MRI-based techniques provides a non-invasive alternative for estimating OEF by exploiting the magnetic property differences between oxygenated (diamagnetic) and deoxygenated (paramagnetic) haemoglobin.¹⁹ Among these, a comprehensive OEF mapping method based on MRI incorporating quantitative susceptibility mapping and quantitative blood oxygen-level dependent (QSM+qBOLD or QQ) allows for both whole brain and regional OEF estimation.²³ The QQ–OEF mapping method has been validated against ¹⁵O-PET in healthy adult brains and represents an accessible tool for assessing cerebral oxygen metabolism in clinical settings.²⁴

The present study aims to investigate whole-brain and regional changes in OEF using the QQ method in patients with SLE and to explore the association between regional OEF alterations and cognitive performance.

Methods

Participants

Patients with SLE were recruited from the Department of Rheumatology and Immunology at our hospital and healthy controls (HC) were recruited via social media. All patients met the 2019 EULAR/ACR classification criteria for SLE.²⁵ NPSLE were classified according to the 1999 American College of Rheumatology case definitions.⁴ A validated algorithm was used to help attribute NP events to SLE.²⁶ The classification and attribution of NP events were determined through multidisciplinary consensus involving specialists from rheumatology, neurology and radiology. The exclusion criteria were as follows: (1) patients with previous other neurologic diseases; (2) participants taking CNS-acting medications; (3) images with motion or imaging artefacts; (4) MRI abnormalities such as stroke or haemorrhage. Patients with mild white matter injury (Fazekas grade 1) and brain atrophy were not excluded.²⁷ We excluded three patients, two of which had a prior or present brain haemorrhage or major infarction and one was excluded due to image artefact.

Demographic characteristics, vascular risk factors and laboratory data were obtained. The duration of SLE was defined as the time from the initial onset of SLE to the brain MRI imaging. Disease activity in patients with SLE was evaluated using the Systemic Lupus Erythematosus Disease Activity Index 2000 (SLEDAI-2K).²⁸ Cognitive function was assessed within 1 week of the MRI examination using the Montreal Cognitive Assessment (MoCA), described in more detail in online supplemental appendix A of supplementary data.²⁹ Vascular risk factors including systolic and diastolic blood pressure (SBP and DBP), fasting blood glucose (FBG), low-density and high-density lipoprotein cholesterol (LDL and HDL) and triglyceride (TG) were collected. Antiphospholipid antibody including anticardiolipin, lupus anticoagulant and anti-β2 glycoprotein 1 were collected. Medications, including corticosteroids, immunosuppressants and antimalarial drugs, administered prior to the MRI scan were recorded.

MRI examination

All participants were scanned using the same 3.0-Tesla MR scanner (Discovery MR750, General Electric Medical Systems, Milwaukee, Wisconsin) with a 32-channel head coil. A standardised MRI protocol was used, which included three-dimensional (3D) T1-weighted images acquired using a brain volume sequence and a 3D multiecho gradient-echo (mGRE) sequence. The scan parameters are provided in online supplemental appendix A of supplementary data. QSM maps were reconstructed from mGRE images by applying the morphology-enabled dipole inversion with automatic uniform cerebrospinal fluid zero referenced (MEDI+0).^30–32 Conventional MR sequences, including T2, T2 fluid-attenuated inversion recovery and diffusion-weighted imaging were also obtained. Images were examined for quality control to exclude unsatisfactory scans with motion or imaging artefacts. The total volume of grey matter and white matter was computed using segmentation derived from FMRIB Software Library’s FAST algorithm on 3D T1-weighted images (https://fsl.fmrib.ox.ac.uk/fsl).

OEF mapping processing

OEF mapping was generated by applying the QSM+qBOLD method, which integrated mGRE phases and magnitudes.³³ The first is a QSM-based model that separates voxel-wise susceptibility into that of venous blood and non-blood tissue. The second is a qBOLD model, which characterises intravoxel mGRE magnitude signal decay based on the susceptibility difference between deoxygenated venous blood and surrounding brain tissue.¹⁷ The QSM+qBOLD method can eliminate the assumptions required by using either modelling alone. To ensure accurate regional OEF estimation, imaging parameters were optimised to provide reliable input data for the QQ model, including both QSM and magnitude signal decay. High spatial resolution (voxel size=0.46875×0.46875×2 mm³) prevented susceptibility underestimation associated with larger voxels in QSM.³⁴ Regarding magnitude signal decay, the QQ model distinguishes quadratic behaviour in the short echo regime from linear behaviour in the long echo regime.³⁵ Therefore, sufficient data points in both regimes are required, which is achieved by the eight echoes used in this study: three echoes in the short echo regimes and five echoes in the long echo regimes. These imaging parameters ensured reliable quantification of OEF across brain regions. To reduce noise in data fitting, temporal clustering, tissue composition and total variation (CCTV) algorithm was combined for QQ OEF mapping.³⁶ CCTV is a clustering-based machine learning approach. This clustering is based on voxel-wise signal decay similarities, rather than spatial information. CCTV effectively suppresses the spatial propagation of motion artefact onto OEF maps.³⁶ In addition, arbitrary magnitude signals in mGRE were accounted for by the QQ model parameter s0, the signal intensity at TE=0. As a result, further preprocessing steps such as intensity normalisation were not required.

3D T1-weighted images were aligned to the Montreal Neurological Institute (MNI152) space using the deformable b-spline coregistration routine in Advanced Normalisation Tools ((ANTs) http://stnava.github.io/ANTs). The transformation matrices were then used to wrap native space OEF images to MNI152 space. A flowchart of the postprocessing steps is shown in figure 1.

Figure 1. The pipeline flowchart of OEF calculation, coregistration, standardisation and analysis. The MEDI+0 method was used to construct QSM from mGRE sequence. OEF was calculated using the QQ model; T1-weighted images were aligned to the MNI space using the deformable b-spline co-registration routine in ANTs, and the same transformation matrices were then used to wrap native space OEF images to MNI space. OEF maps were then used for ROI assessment and were smoothed for whole-brain assessment. ANTs, Advanced Normalisation Tools; MEDI+0, morphology-enabled dipole inversion with automatic uniform cerebrospinal fluid zero referenced method; mGRE, multi-echo gradient echo; MNI, Montreal Neurological Institute; OEF, oxygen extraction fraction; qBOLD, quantitative blood oxygen level dependent; QSM, quantitative susceptibility mapping; QQ, QSM+qBOLD.

Voxel-wise whole-brain analysis of OEF

Grey and white matter probability masks were generated using the Statistical Parametric Mapping V.12 (SPM12) software. Those probability masks were normalised to the MNI space. A threshold of 0.6 was used to exclude non-parenchymal voxels in the normalised OEF maps, thereby generating the final brain mask. The normalised OEF images were then smoothed with a 3D isotropic Gaussian kernel of 8 mm full width at half maximum (FWHM). OEF group comparisons were established in SPM12 by one-way analysis of variance (ANOVA), with age, education years and sex included as covariates. The brain mask was used to limit the statistical calculation to brain regions only. For multiple comparison corrections, a significance threshold of p<0.001 (uncorrected at the voxel level) was set, with the family-wise error (FWE) correction at p < 0.05 at the cluster level. The results of significant clusters were reported and visualised using XjView toolbox (http://www.alivelearn.net/xjview).

Regions-of-interest analysis

Subcortical grey matter including the basal ganglia, hippocampus and amygdala was chosen as regions-of-interest (ROIs) in our study because of their vulnerability against hypoxia and possible role in the pathogenesis of NPSLE, as indicated by previous imaging studies.^{21 37} The ROIs were delineated using the BNA246 atlas (http://stnava.github.io/ANTs) in the MNI space. The ROIs were visually inspected and manually adjusted by an experienced neuroradiologist (ShunZ, 9 years of experience) with reference to coregistered MNI-standard space QSM images using ITK-SNAP (V.4.2, www.itksnap.org). Subsequently, the OEF values for the selected ROIs were extracted for each subject.

Statistical analysis

SPSS (V.26) and R software (V.4.4.1) were used for the statistical analyses. The age, gender, years of education and MoCA score between the three groups were compared using ANOVA, Fisher’s exact test or Kruskal-Wallis test. When significant, further multiple comparisons were performed, and p value was adjusted by Bonferroni method. Disease duration, SLEDAI and vascular risk factors were compared between non-NPSLE and NPSLE by Student’s t-test or Mann-Whitney U test. Regional OEF differences between the groups were assessed using ANOVA (model 1). To determine the influence of age, gender, education on OEF measurements, we include these covariates as confounding factors using analysis of covariance (ANCOVA) in model 2. Model 3 included additional adjustment for total brain volume to address the potential impact of brain atrophy. Model 4 further incorporated medication use, specifically corticosteroids, immunosuppressive agents and antimalarial drugs. Finally, model 5 included additional adjustment for vascular risk factors, including SBP, DBP, TG, HDL, LDL and FBG. Partial Spearman rank correlation tests were used to evaluate the associations of OEF measures with MoCA adjusted for age, sex and years of education.

Due to the exploratory nature of this imaging-based study and the lack of prior effect size estimates for QQ OEF metrics in SLE, prestudy sample size estimation was not feasible. To assess the adequacy of our sample size, we performed a post hoc power analysis using a simulation-based method (10 000 repetitions). Simulated datasets were generated based on the adjusted means and SD obtained from the ANCOVA analyses. Effect sizes (Cohen’s f) were computed for between-group differences in OEF values across brain ROIs. The results of the post hoc power analysis are summarised in online supplemental appendix table S1. Three regions (the left medial amygdala, left rostral hippocampus and right rostral hippocampus) demonstrated power values ≥0.80, indicating that the sample size was sufficient to detect moderate-to-large group effects. Five regions showed power ranging from 0.71 to 0.77, while the remaining regions showed power below 0.70, suggesting limited power for detecting statistically significant differences in those areas. Additionally, another simulation using hypothetical balanced scenario (n=23 per group, the average of the actual group sizes) showed slight deterioration in power (summarised in online supplemental appendix table S2).

Results

Clinical and demographic information

The participants (mean age, 30.59±9.37 years; 60 women and 9 men) included 26 HC (mean age, 31.00±9.18 years; 22 women and 4 men), 25 non-NPSLE patients (mean age, 30.24±9.44 years; 21 women and 4 men) and 18 NPSLE patients (mean age, 30.50±10.05 years; 17 women and 1 man). No significant differences were found in age, sex or years of education among the HC, non-NPSLE and NPSLE groups. There were also no significant differences in the disease duration, vascular risk factors (including SBP, DBP, FBG, TG, HDL and LDL), antiphospholipid antibody status or medications between non-NPSLE and NPSLE patients. In terms of SLEDAI, patients with NP scored higher than those without NP (p<0.001). Significant differences in total MoCA scores (p=0.004) and domain-specific scores including visuospatial (p=0.003), attention (p=0.03) and language (p=0.04) were found between the three groups. An overview of the participants’ demographics is found in table 1. The detailed types of NP involvement for all 18 patients are provided in online supplemental appendix table S3.

Table 1

Demographic and clinical information of the participants

	HC (n=26)	SLE (n=43)		P
		Non-NPSLE (n=25)	NPSLE (n=18)
Age (years)	31.0±9.2	30.2±9.4	30.5±10.1	0.959*
Gender (male/female)	4/22	4/21	1/17	0.668†
Education (years)	16 (14.3–16)	16 (12–16)	13.5 (11.3–16)	0.103‡
MoCA score	28 (26–29)§	25 (24–28.5)	24 (21.5–26) §	0.004‡
Visuospatial	5 (4-5)§	4 (3-5)§	3 (2-4)§	0.003‡
Naming	3 (3-3)	3 (2-3)	3 (2-3)	0.099‡
Attention	6 (6-6)	6 (5.5–6)	6 (5.3–6)	0.03‡
Language	2 (2-3)§	2 (2-3)	2 (1-2.8)§	0.04‡
Abstraction	2 (2-2)	2 (1-2)	2 (1.3–2)	0.189‡
Memory	4 (3-5)	4 (3-5)	2.5 (1-4)	0.078‡
Orientation	6 (6-6)	6 (6-6)	6 (6-6)	0.344‡
Disease duration (months)		1 (1-10)	9 (1-45)	0.106¶
SLEDAI		10 (5–13)	14 (13–18.3)	<0.001¶
SBP (mm Hg)		118.8±15.5	121.1±22.4	0.696**
DBP (mm Hg)		86.0 (77.5–94.0)	81.5 (73.8–95.5)	0.613¶
TG (mmol/L)		2.03±0.96	2.04±0.96	0.98**
HDL (mmol/L)		0.92±0.27	1.04±0.36	0.232**
LDL (mmol/L)		2.49±0.69	2.53±0.96	0.877**
FBG (mmol/L)		5.32 (4.91–6.64)	5.43 (5.19–6.78)	0.427¶
Antiphospholipid antibody status
ACL positive (%)		3/25 (12)	2/16 (12.5)	1†
LA positive (%)		7/24 (29.2)	4/16 (25)	1†
Aβ2GPI positive (%)		5/25 (20)	5/16 (31.3)	0.472†
Medications
Corticosteroids (%)		6/25 (24)	8/18 (44.4)	0.198†
Immunosuppressants (%)		4/25 (16)	4/18 (22.2)	0.701†
Antimalarials (%)		6/25 (24)	5/18 (27.8)	1†

Continuous variables were given as mean±SD for normal distribution or as median with IQR for deviated distribution.

*One-way analysis of variance.

†Fisher’s exact test.

‡Kruskal-Wallis rank sum test.

§Denote significant differences between the two groups marked with the same letter by post hoc pairwise comparisons using the Wilcoxon rank sum test.

¶Mann-Whitney U test.

**t-test.

ACL, anticardiolipin; Aβ2GPI, anti-β2 glycoprotein 1; DBP, diastolic blood pressure; FBG, fasting blood glucose; HC, healthy controls; HDL, high-density lipoprotein cholesterol; LA, lupus anticoagulant; LDL, low-density lipoprotein cholesterol; MoCA, Montreal Cognitive Assessment; NPSLE, neuropsychiatric lupus; SBP, systolic blood pressure; SLE, systemic lupus erythematosus; SLEDAI, Systemic Lupus Erythematosus Disease Activity Index; TG, triglyceride.

Whole-brain and regional OEF assessment

Voxel-wise whole-brain analysis showed significant changes of OEF predominantly in the orbitofrontal cortex and bilateral insular lobes among HC, non-NPSLE and NPSLE patients (ANOVA; uncorrected voxel-level p<0.001, FWE-corrected cluster-level p<0.05; see table 2 and figure 2).

Table 2

Significant clusters of voxel-wised comparisons in OEF among HC, non-NPSLE and NPSLE groups

	Voxel number	P FWE-corr	Peak			Location
			X	Y	Z	Regions	Voxels
Cluster 1	9754	0.002	11	36	−28	Rectal gyrus	3274
						Medial frontal gyrus	1878
						Inferior frontal gyrus	1440
						Orbital gyrus	1433
						Middle frontal gyrus	227
						Subcallosal gyrus	120
						Superior frontal gyrus	91
Cluster 2	5874	0.018	−43	−12	-4	Left insula	2676
						Left superior temporal gyrus	1921
						Left inferior frontal gyrus	563
						Left precentral gyrus	209
						Left transverse temporal gyrus	173
Cluster 3	6707	0.011	40	8	5	Right insula	3843
						Right postcentral gyrus	735
						Right inferior frontal gyrus	707
						Right superior temporal gyrus	384
						Right precentral gyrus	378
						Right transverse temporal gyrus	204

FWE, family-wise error; HC, healthy controls; NPSLE, neuropsychiatric lupus; OEF, oxygen extraction fraction.

Figure 2. Group differences in OEF among HC, non-NPSLE and NPSLE at the voxel level. ANOVA was used to assess significant clusters (voxel-level p<0.001 uncorrected and cluster-level FWE-corrected p<0.05), controlling for age, sex and education. ANOVA, analysis of covariance; FWE, family-wise error; HC, healthy controls; NPSLE, neuropsychiatric lupus; OEF, oxygen extraction fraction.

In addition to the voxel-wise whole-brain assessment, we performed the region-wise analysis in subcortical grey matter because of their high oxygen consumption and possible role in the pathogenesis of NPSLE. The regional OEF values of the 20 ROIs are listed in online supplemental appendix table S4. In the unadjusted model, OEF values were significantly lower in non-NPSLE patients compared with HC in the bilateral medial amygdalae (left: p=0.017; right: p=0.034), bilateral rostral hippocampi (left: p=0.004; right: p=0.023) and bilateral dorsal caudate nuclei (left: p=0.033; right: p=0.034). There was a general trend of decreased OEF in other ROIs in non-NPSLE patients compared with HC. Except in the left medial amygdala, where OEF values were significantly lower in NPSLE patients than HC (p=0.01), no difference in OEF values was found in other ROIs comparing the two groups, while NPSLE group had significantly higher OEF values than non-NPSLE group in bilateral rostral hippocampi (left: p=0.047; right: p=0.040) and right caudal hippocampus (p=0.038). In model 2, age, sex and education were set as covariates (figure 3). OEF values remained significantly lower in non-NPSLE group than HC in bilateral medial amygdalae (left: p=0.017; right: p=0.035), rostral hippocampi (left: p=0.004; right: p=0.022) and dorsal caudate nucleus (left: p=0.035; right: p=0.045). In left medial amygdala, OEF values were also significantly lower in NPSLE patients than HC (p=0.016). Significantly higher OEF values were consistently observed in the right rostral and caudal hippocampus (rostral: p=0.049; caudal: p=0.039) of NPSLE group than non-NPSLE group. In model 3, which included total brain volume as an additional covariate to model 2, OEF values remained significantly lower in non-NPSLE patients than HC in the left medial amygdala (p=0.025) and bilateral rostral hippocampi (left: p=0.009; right: p=0.045). The NPSLE group continued to show higher OEF values than the non-NPSLE group in the bilateral rostral hippocampi (left: p=0.043; right: p=0.025) and the right caudal hippocampus (p=0.030). Model 4 was further adjusted for the use of corticosteroids, immunosuppressive agents and antimalarial drugs. Model 5 included additional adjustment for vascular risk factors, including SBP, DBP, TG, HDL, LDL and FBG. Even after these adjustments, significant group differences in OEF values were consistently observed in the bilateral rostral hippocampi (model 4: left p=0.020, right p=0.004; model 5: left p=0.015, right p=0.004) and right caudal hippocampus (model 4: p=0.005; model 5: p=0.005).

Figure 3. Comparisons of OEF values in subcortical grey matter structures between HC, non-NPSLE patients and NPSLE patients (ANOVA adjusted for age, sex, and years of education, post-hoc between-group analysis corrected using Bonferroni method, *p<0.05, **p<0.01). OEF values in bilateral medial amygdalae (a and b), bilateral rostral hippocampi (c and d), right caudal hippocampus (e) and bilateral dorsal caudates (g and h) were significantly lower in non-NPSLE patients than in HC. In right rostral and caudal hippocampus (d and e), OEF values were significantly higher in NPSLE patients than that in non-NPSLE patients. In the left ventral caudate (f), although adjusted ANOVA indicated a significant group effect (p = 0.039), no significance was found in post-hoc between-group comparison. ANOVA, analysis of covariance; HC, healthy controls; NPSLE, neuropsychiatric lupus; OEF, oxygen extraction fraction.

Correlation analyses

Considering the NPSLE and non-NPSLE groups separately or combined, OEF values in the subcortical grey matter ROIs did not have significant correlation with MoCA scores. However, when all subjects in the study were included, the MoCA scores were positively correlated with OEF value in bilateral medial amygdalae (left medial: r=0.335, p=0.006; right medial: r=0.285, p=0.02), right lateral amygdala (r=0.251, p=0.04), left rostral hippocampus (r=0.254, p=0.04) and right dorsal caudate (r=0.266, p=0.03) after adjusting for age, sex and years of education (figure 4).

Figure 4. Relationship between OEF values and MoCA scores in subregions of the amygdala, hippocampus and caudate nucleus. In the bilateral medial amygdalae (a and b), right lateral amygdala (c), left rostral hippocampus (d) and right dorsal caudate nucleus (e), OEF values were positively correlated with MoCA scores in the entire cohort (adjusted for age, sex and years of education, p<0.05). MoCA, Montreal Cognitive Assessment; OEF, oxygen extraction fraction.

Discussion

In our study, the QSM+qBOLD technique was used to evaluate cerebral OEF alteration in patients with SLE. Our main findings were (1) voxel-wise whole-brain analysis revealed significant differences of OEF values primarily in the limbic system including the orbitofrontal cortex and bilateral insular lobes, across HC, non-NPSLE and NPSLE groups. (2) Regional analysis showed that OEF values were significantly lower in the bilateral medial amygdalae, rostral hippocampi, dorsal caudates and right caudal hippocampus in non-NPSLE patients compared with HC, while NPSLE patients have significantly higher OEF in the right hippocampus than non-NPSLE patients. (3) OEF values in the bilateral medial amygdalae, right lateral amygdala, left rostral hippocampus and right dorsal caudate nucleus were positively correlated with MoCA scores.

To the best of our knowledge, only one prior study has examined cerebral OEF changes in SLE patients. In that study, cerebral OEF, measured by the MRI-based QSM method, was significantly higher in patients with SLE than HC and was positively correlated with SLEDAI and NP symptom scores.³⁸ The results appear to be contrary to ours. However, it should be noted that the OEF they evaluated was the mean OEF calculated by averaging measures from 18 ROIs at the level of centrum semiovale, which had a lack of spatial specificity. In addition, the older age of participants in the previous study may partly account for the inconsistent results, as cerebral OEF tends to increase with age, particularly in individuals with higher vascular risks, such as those with SLE.^{11 39} Besides, in the previous study, the SLE group included both non-NPSLE and NPSLE patients, which may have opposite changes of OEF, indicating within-group heterogeneity.

Our voxel-wise whole-brain analysis revealed OEF differences primarily in the orbitofrontal cortex and bilateral insular lobes, which are part of the limbic system. The limbic system, composed of highly interconnected cortical and subcortical structures, helps support brain functions such as emotion and memory.⁴⁰ Previous studies using structural, diffusion and functional MRI have confirmed the involvement of the limbic system in SLE.^41–43 Our findings provide additional metabolic evidence to support the limbic system as a key structure frequently involved in SLE.

In the regional OEF analysis, we observed lower OEF values in the non-NPSLE group in subregions of the subcortical grey matter compared with HC, with decreasing trends across all selected ROIs. The reduced OEF in non-NPSLE patients suggests that CNS damage may occur before the onset of NP symptoms. The lower OEF values observed in non-NPSLE patients may be attributed to tissue degradation and apoptosis, which have been proposed as contributing factors to regional brain hypometabolism seen in early and advanced SLE patients, as seen in previous PET and calibrated functional MRI studies.^44–46 Decreased OEF in non-NPSLE patients may represent reduced metabolic demands in the brain of non-NPSLE patients.

Our results indicate that, even after adjusting for confounding factors such as age, sex, education years, brain volume, medication use and vascular risk factors (model 5), the NPSLE group continued to show higher OEF values than the non-NPSLE group in the subregions of the hippocampus. The observed increase in OEF in the NPSLE group may be attributed to the combined effects of autoimmune-mediated neuroinflammation and ischaemic injury secondary to microvascular thrombosis.^{2 20} Brain histopathological findings in NPSLE demonstrated a spectrum of vascular damage, including complement-associated microthrombi and diffuse vasculopathy.⁹ The ischaemic conditions resulting from microvascular damage in NPSLE may drive elevated OEF, as seen in cerebral small vessel disease,¹³ where impaired perfusion prompts increased oxygen extraction to sustain metabolic demands. In addition, the hippocampus is a critical site for neuroinflammation in NPSLE. It has been shown that the hippocampus is one of the main targets for brain-reactive autoantibodies such as anti-N-methyl-D-aspartate receptor antibodies and anti-ribosomal P protein antibodies in NPSLE.^47–49 Kowal et al proposed that the hippocampus is particularly vulnerable to autoantibody-mediated injury in SLE, with circulating antibodies preferentially binding to hippocampal neurons, thereby disrupting hippocampal metabolism.⁵⁰ Despite the pseudo-normalisation of OEF in the NPSLE group, neuronal damage in the hippocampus may still persist. A PET study of mice with brain-reactive autoantibodies demonstrated reduced hippocampal metabolism during the first 14 days after blood-brain barrier damage, consistent with neuronal toxicity and cell death observed histologically.^{51 52} Although hippocampal metabolism gradually increased over the following 9 weeks, hippocampal neuronal loss continued, possibly due to compensatory mechanisms or inflammatory-driven processes like microglial recruitment to necrotic areas.

Cognitive dysfunction (CD), often referred to as ‘brain fog’, is highly prevalent in patients with SLE.⁵³ Studies have reported that up to 80% of individuals with SLE experience CD.^{51 53} Despite its high prevalence and impact on quality of life, CD remains under-recognised in rheumatology practice, as routine cognitive screening is rarely performed. In our study, no significant difference of MoCA score was found between the non-NPSLE and NPSLE groups. This lack of difference may be due to the limited sample size, the presence of subclinical cognitive impairment in non-NPSLE patients and the clinical heterogeneity within the NPSLE group. Our study showed that OEF values in several subcortical grey matter regions were positively correlated with MoCA scores considering the whole cohort. These findings extend existing evidence linking cerebral oxygen metabolism with cognition. Previous studies have reported an association between reduced global cerebral OEF and cognitive decline. For example, Jiang et al demonstrated that reduced global cerebral OEF was associated with poorer cognitive performance across HCs, individuals with mild cognitive impairment and those with dementia.¹⁴ Similarly, another study reported that in APOE4 carriers, lower global cerebral OEF was associated with poorer executive cognitive function.¹⁸ Unlike previous studies that focused on global OEF, our study used a QQ-based OEF mapping method, which enabled the calculation of regional OEF values. Using this approach, we found that regional OEF in several subcortical grey matter regions was associated with cognitive performance. Specifically, subregions of the hippocampus—a critical region for memory encoding and storage⁵⁴—as well as the amygdala and caudate nucleus, which are implicated in cognitive flexibility,^{55 56} demonstrated positive correlations with MoCA scores. These findings suggest that altered oxygen metabolism in specific brain regions may underlie CD in SLE and highlight the value of regional OEF mapping in detecting functional changes. Although our current sample size limits the ability to define clinically meaningful thresholds for cognitive impairment risk, these findings support the potential utility of regional OEF as a functional biomarker.^54–56

Several limitations should be noted. First, since the participants in our study were prospectively enrolled based on strict inclusion and exclusion criteria—particularly for the NPSLE group as we exclude patients with stroke or haemorrhage—the sample size was relatively limited. The current study should follow an exploratory framework. While post hoc power analysis indicated moderate power for detecting OEF differences in some brain regions, caution is warranted when interpreting findings in regions with lower power due to the potential for type II error, which may limit detection of significant differences in those areas. Additionally, despite the use of covariate-adjusted ANCOVA, the imbalance in group sizes may still affect the precision and robustness of estimates. Nevertheless, the observed effect sizes from this study can provide a basis for future prospective designs. Future studies with larger sample sizes and multicentre recruitment are needed to confirm these findings and enhance generalisability. Second, according to the Systemic Lupus International Collaborating Clinics attribution model, the temporal proximity between SLE onset and NP events strengthens causal inference.^{26 57} To enhance the attribution certainty of NP manifestations to SLE, about 80% of the patients included had a disease duration of less than 1 year. This approach helped minimise confounding effects from long-term medication use, chronic comorbidities or age-related brain pathologies. However, the relatively short disease duration may limit the extrapolation of findings to patients with longstanding SLE. Future work should investigate cerebral oxygen metabolism in patients with longer disease durations and incorporate longitudinal designs to monitor dynamic OEF changes over time. Although the effects of corticosteroids and immunosuppressive therapies, commonly used in the treatment of SLE, on brain OEF are not yet fully understood, these therapies could potentially influence OEF. Corticosteroids are known to affect brain metabolism,⁵⁸ while immunosuppressive therapies have been associated with neurotoxicity,⁵⁹ both of which may alter cerebral oxygen metabolism. As such, future studies involving patients with longer disease durations must take into account the treatment duration and dosage to clarify the impact of these therapies on OEF. Third, in addition to OEF, cerebral blood flow (CBF) and cerebral metabolic rate of oxygen (CMRO₂) are also important indicators for brain metabolism, as they further indicate oxygen supply and demand. Future studies should incorporate these additional parameters to gain a more thorough understanding of oxygen metabolic alterations in SLE.

Despite these limitations, our findings provide clues into cerebral oxygen metabolism alterations in SLE, particularly in relation to NP involvement and cognitive function. By using this non-invasive, MRI-based OEF mapping technique, our study reveals region-specific metabolic disruptions in both NPSLE and non-NPSLE patients that are not detectable by conventional MRI. The association between regional OEF and MoCA scores further supports the potential of OEF as a functional biomarker for cognitive vulnerability in SLE. Mechanistic studies are needed to clarify the biological basis of OEF alterations in SLE. To establish the clinical utility of OEF mapping, subsequent studies should confirm these findings in larger, multicentre populations with a broader range of disease durations and clinical phenotypes. Longitudinal designs will be particularly important for assessing the predictive value of OEF in tracking NP events or cognitive decline. Additionally, technical standardisation of MRI acquisition and postprocessing is necessary to ensure reproducibility. Incorporating physiological parameters such as CBF and CMRO₂ will also provide a more comprehensive understanding of cerebral metabolic alterations in SLE and help refine the role of OEF within a multiparametric neuroimaging framework.

We extend our thanks to Dr. Qiao Huang (Zhongnan Hospital of Wuhan University) for or his advice concerning statistical methods.

Data availability statement

Data are available upon reasonable request.

Ethics statements

Patient consent for publication

Not applicable.

Ethics approval

This study involves human participants and was approved by the Medical Ethics Committee of Tongji Hospital, Tongji Medical College, Huazhong University of Science and Technology (TJ-IRB20230455). Participants gave informed consent to participate in the study before taking part.

Footnote

ShuoqiZ, JM and SW contributed equally.

Contributors ShuoqiZ and ShunZ contributed to the design of the study. JM and LD recruited the participants. ShuoqiZ and SW contributed to the acquisition of imaging data. JM and ZH contributed to the acquisition of clinical data. JM contributed to the neuropsychological assessments. ShuoqiZ, ShunZ, SY, JC and YW processed and analysed the imaging data. ShuoqiZ drafted the manuscript. WZ is responsible for the overall content of the study as guarantor. All authors contributed to the review and editing of the manuscript. All authors approved the final manuscript.

Funding This work was supported by Hubei Provincial Natural Science Foundation Joint Fund for Innovation and Development Project (Number 2023AFD046, 2024AFD108), Hubei Provincial Natural Science Foundation of China (Number 2021BCA123) and the National Key Research and Development Program of China (Number 2022YFC2406903).

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.

References

¹ Barber MRW, Drenkard C, Falasinnu T, et al. Global epidemiology of systemic lupus erythematosus. Nat Rev Rheumatol 2021; 17: 515–32. doi:10.1038/s41584-021-00668-1

² Schwartz N, Stock AD, Putterman C. Neuropsychiatric lupus: new mechanistic insights and future treatment directions. Nat Rev Rheumatol 2019; 15: 137–52. doi:10.1038/s41584-018-0156-8

³ Govoni M, Hanly JG. The management of neuropsychiatric lupus in the 21st century: still so many unmet needs? Rheumatology (Oxford) 2020; 59: v52–62. doi:10.1093/rheumatology/keaa404

⁴ The American College of Rheumatology nomenclature and case definitions for neuropsychiatric lupus syndromes. Arthritis Rheum 1999; 42: 599–608. doi:10.1002/1529-0131(199904)42:4<599::AID-ANR2>3.0.CO;2-F

⁵ Gelb S, Stock AD, Anzi S, et al. Mechanisms of neuropsychiatric lupus: The relative roles of the blood-cerebrospinal fluid barrier versus blood-brain barrier. J Autoimmun 2018; 91: 34–44. doi:10.1016/j.jaut.2018.03.001

⁶ Ota Y, Srinivasan A, Capizzano AA, et al. Central Nervous System Systemic Lupus Erythematosus: Pathophysiologic, Clinical, and Imaging Features. Radiographics 2022; 42: 212–32. doi:10.1148/rg.210045

⁷ Hanly JG, Kozora E, Beyea SD, et al. Review: Nervous System Disease in Systemic Lupus Erythematosus: Current Status and Future Directions. Arthritis Rheumatol 2019; 71: 33–42. doi:10.1002/art.40591

⁸ Mackay M, Tang CC, Vo A. Advanced neuroimaging in neuropsychiatric systemic lupus erythematosus. Curr Opin Neurol 2020; 33: 353–61. doi:10.1097/WCO.0000000000000822

⁹ Cohen D, Rijnink EC, Nabuurs RJA, et al. Brain histopathology in patients with systemic lupus erythematosus: identification of lesions associated with clinical neuropsychiatric lupus syndromes and the role of complement. Rheumatology (Oxford) 2017; 56: 77–86. doi:10.1093/rheumatology/kew341

¹⁰ Curiel R, Akin EA, Beaulieu G, et al. PET/CT imaging in systemic lupus erythematosus. Ann N Y Acad Sci 2011; 1228: 71–80. doi:10.1111/j.1749-6632.2011.06076.x

¹¹ Lin Z, Lim C, Jiang D, et al. Longitudinal changes in brain oxygen extraction fraction (OEF) in older adults: Relationship to markers of vascular and Alzheimer’s pathology. Alzheimers Dement 2023; 19: 569–77. doi:10.1002/alz.12727

¹² Jiang D, Lu H. Cerebral oxygen extraction fraction MRI: Techniques and applications. Magn Reson Med 2022; 88: 575–600. doi:10.1002/mrm.29272

¹³ Kang P, Ying C, Chen Y, et al. Oxygen Metabolic Stress and White Matter Injury in Patients With Cerebral Small Vessel Disease. Stroke 2022; 53: 1570–9. doi:10.1161/STROKEAHA.121.035674

¹⁴ Jiang D, Lin Z, Liu P, et al. Brain Oxygen Extraction Is Differentially Altered by Alzheimer’s and Vascular Diseases. J Magn Reson Imaging 2020; 52: 1829–37. doi:10.1002/jmri.27264

¹⁵ Liu C, Cárdenas-Rivera A, Teitelbaum S, et al. Neuroinflammation increases oxygen extraction in a mouse model of Alzheimer’s disease. Alzheimers Res Ther 2024; 16: 78. doi:10.1186/s13195-024-01444-5

¹⁶ Zhang S, Cho J, Nguyen TD, et al. Initial Experience of Challenge-Free MRI-Based Oxygen Extraction Fraction Mapping of Ischemic Stroke at Various Stages: Comparison With Perfusion and Diffusion Mapping. Front Neurosci 2020; 14: 535441. doi:10.3389/fnins.2020.535441

¹⁷ Cho J, Nguyen TD, Huang W, et al. Brain oxygen extraction fraction mapping in patients with multiple sclerosis. J Cereb Blood Flow Metab 2022; 42: 338–48. doi:10.1177/0271678X211048031

¹⁸ Lin Z, Sur S, Soldan A, et al. Brain Oxygen Extraction by Using MRI in Older Individuals: Relationship to Apolipoprotein E Genotype and Amyloid Burden. Radiology 2019; 292: 140–8. doi:10.1148/radiol.2019182726

¹⁹ Biondetti E, Cho J, Lee H. Cerebral oxygen metabolism from MRI susceptibility. Neuroimage 2023; 276. doi:10.1016/j.neuroimage.2023.120189

²⁰ Nikolopoulos D, Fanouriakis A, Boumpas DT. Update on the pathogenesis of central nervous system lupus. Curr Opin Rheumatol 2019; 31: 669–77. doi:10.1097/BOR.0000000000000655

²¹ Berndorfler BD, Warwick JM, Doruyter AGG. Role of F-18 FDG PET-CT in neuropsychiatric systemic lupus erythematosus. Compr Psychiatry 2024; 132: 152480. doi:10.1016/j.comppsych.2024.152480

²² Fan AP, An H, Moradi F, et al. Quantification of brain oxygen extraction and metabolism with [¹⁵O]-gas PET: A technical review in the era of PET/MRI. Neuroimage 2020; 220: 117136. doi:10.1016/j.neuroimage.2020.117136

²³ Cho J, Zhang S, Kee Y, et al. Cluster analysis of time evolution (CAT) for quantitative susceptibility mapping (QSM) and quantitative blood oxygen level-dependent magnitude (qBOLD)-based oxygen extraction fraction (OEF) and cerebral metabolic rate of oxygen (CMRO 2) mapping. Magn Reson Med 2020; 83: 844–57. doi:10.1002/mrm.27967

²⁴ Cho J, Lee J, An H, et al. Cerebral oxygen extraction fraction (OEF): Comparison of challenge-free gradient echo QSM+qBOLD (QQ) with ¹⁵O PET in healthy adults. J Cereb Blood Flow Metab 2021; 41: 1658–68. doi:10.1177/0271678X20973951

²⁵ Aringer M, Costenbader K, Daikh D, et al. 2019 European League Against Rheumatism/American College of Rheumatology classification criteria for systemic lupus erythematosus. Ann Rheum Dis 2019; 78: 1151–9. doi:10.1136/annrheumdis-2018-214819

²⁶ Bortoluzzi A, Scirè CA, Bombardieri S, et al. Development and validation of a new algorithm for attribution of neuropsychiatric events in systemic lupus erythematosus. Rheumatology (Oxford) 2015; 54: 891–8. doi:10.1093/rheumatology/keu384

²⁷ Fazekas F, Chawluk JB, Alavi A, et al. MR signal abnormalities at 1.5 T in Alzheimer’s dementia and normal aging. AJR Am J Roentgenol 1987; 149: 351–6. doi:10.2214/ajr.149.2.351

²⁸ Gladman DD, Ibañez D, Urowitz MB. Systemic lupus erythematosus disease activity index 2000. J Rheumatol 2002; 29: 288–91.

²⁹ Raghunath S, Glikmann-Johnston Y, Morand E, et al. Evaluation of the Montreal Cognitive Assessment as a screening tool for cognitive dysfunction in SLE. Lupus Sci Med 2021; 8: e000580. doi:10.1136/lupus-2021-000580

³⁰ Liu Z, Spincemaille P, Yao Y, et al. MEDI+0: Morphology enabled dipole inversion with automatic uniform cerebrospinal fluid zero reference for quantitative susceptibility mapping. Magn Reson Med 2018; 79: 2795–803. doi:10.1002/mrm.26946

³¹ Kee Y, Liu Z, Zhou L, et al. Quantitative Susceptibility Mapping (QSM) Algorithms: Mathematical Rationale and Computational Implementations. IEEE Trans Biomed Eng 2017; 64: 2531–45. doi:10.1109/TBME.2017.2749298

³² Dimov AV, Nguyen TD, Spincemaille P, et al. Global cerebrospinal fluid as a zero-reference regularization for brain quantitative susceptibility mapping. J Neuroimaging 2022; 32: 141–7. doi:10.1111/jon.12923

³³ Cho J, Kee Y, Spincemaille P, et al. Cerebral metabolic rate of oxygen (CMRO 2) mapping by combining quantitative susceptibility mapping (QSM) and quantitative blood oxygenation level-dependent imaging (qBOLD). Magn Reson Med 2018; 80: 1595–604. doi:10.1002/mrm.27135

³⁴ Zhou D, Cho J, Zhang J, et al. Susceptibility underestimation in a high‐susceptibility phantom: Dependence on imaging resolution, magnitude contrast, and other parameters. Magn Reson Med 2017; 78: 1080–6. doi:10.1002/mrm.26475

³⁵ Yablonskiy DA, Sukstanskii AL, He X. Blood oxygenation level-dependent (BOLD)-based techniques for the quantification of brain hemodynamic and metabolic properties - theoretical models and experimental approaches. NMR Biomed 2013; 26: 963–86. doi:10.1002/nbm.2839

³⁶ Cho J, Spincemaille P, Nguyen TD, et al. Temporal clustering, tissue composition, and total variation for mapping oxygen extraction fraction using QSM and quantitative BOLD. Magn Reson Med 2021; 86: 2635–46. doi:10.1002/mrm.28875

³⁷ Yang A, Zhuang H, Du L, et al. Evaluation of whole-brain oxygen metabolism in Alzheimer’s disease using QSM and quantitative BOLD. Neuroimage 2023; 282. doi:10.1016/j.neuroimage.2023.120381

³⁸ Miyata M, Kakeda S, Kudo K, et al. Evaluation of oxygen extraction fraction in systemic lupus erythematosus patients using quantitative susceptibility mapping. J Cereb Blood Flow Metab 2019; 39: 1648–58. doi:10.1177/0271678X18764829

³⁹ Le LNN, Wheeler GJ, Fletcher E, et al. Assessment of brain oxygen extraction fraction between normal cognition and mild cognitive impairment in an elderly cohort using quantitative BOLD. Alzheimers Dement 2023; 19. doi:10.1002/alz.081821

⁴⁰ Catani M, Dell’acqua F, Thiebaut de Schotten M. A revised limbic system model for memory, emotion and behaviour. Neurosci Biobehav Rev 2013; 37: 1724–37. doi:10.1016/j.neubiorev.2013.07.001

⁴¹ Valdés Cabrera D, El Tal T, Mohamed I, et al. Effects of systemic lupus erythematosus on the brain: a systematic review of structural MRI findings and their relationships with cognitive dysfunction. Lupus Sci Med 2024; 11: 1–15. doi:10.1136/lupus-2024-001214

⁴² Papadaki E, Kavroulakis E, Bertsias G, et al. Regional cerebral perfusion correlates with anxiety in neuropsychiatric SLE: evidence for a mechanism distinct from depression. Lupus 2019; 28: 1678–89. doi:10.1177/0961203319887793

⁴³ Wang L, Han K, Huang Q, et al. Systemic lupus erythematosus-related brain abnormalities in the default mode network and the limbic system: A resting-state fMRI meta-analysis. J Affect Disord 2024; 355: 190–9. doi:10.1016/j.jad.2024.03.121

⁴⁴ Ramage AE, Fox PT, Brey RL, et al. Neuroimaging evidence of white matter inflammation in newly diagnosed systemic lupus erythematosus. Arthritis Rheum 2011; 63: 3048–57. doi:10.1002/art.30458

⁴⁵ Zardi EM, Taccone A, Marigliano B, et al. Neuropsychiatric systemic lupus erythematosus: tools for the diagnosis. Autoimmun Rev 2014; 13: 831–9. doi:10.1016/j.autrev.2014.04.002

⁴⁶ Lee S-W, Park M‐. C, Lee S-K, et al. The efficacy of brain F-fluorodeoxyglucose positron emission tomography in neuropsychiatric lupus patients with normal brain magnetic resonance imaging findings. Lupus 2012; 21: 1531–7. doi:10.1177/0961203312459104

⁴⁷ Moscavitch SD, Szyper-Kravitz M, Shoenfeld Y. Autoimmune pathology accounts for common manifestations in a wide range of neuro-psychiatric disorders: the olfactory and immune system interrelationship. Clin Immunol 2009; 130: 235–43. doi:10.1016/j.clim.2008.10.010

⁴⁸ Katzav A, Solodeev I, Brodsky O, et al. Induction of autoimmune depression in mice by anti-ribosomal P antibodies via the limbic system. Arthritis Rheum 2007; 56: 938–48. doi:10.1002/art.22419

⁴⁹ Kowal C, DeGiorgio LA, Lee JY, et al. Human lupus autoantibodies against NMDA receptors mediate cognitive impairment. Proc Natl Acad Sci USA 2006; 103: 19854–9. doi:10.1073/pnas.0608397104

⁵⁰ Kowal C, DeGiorgio LA, Nakaoka T, et al. Cognition and immunity; antibody impairs memory. Immunity 2004; 21: 179–88. doi:10.1016/j.immuni.2004.07.011

⁵¹ Mackay M, Vo A, Tang CC, et al. Metabolic and microstructural alterations in the SLE brain correlate with cognitive impairment. JCI Insight 2019; 4: e124002. doi:10.1172/jci.insight.124002

⁵² Vo A, Volpe BT, Tang CC, et al. Regional brain metabolism in a murine systemic lupus erythematosus model. J Cereb Blood Flow Metab 2014; 34: 1315–20. doi:10.1038/jcbfm.2014.85

⁵³ Mizrachi M, Anderson E, Carroll KR, et al. Cognitive dysfunction in SLE: An understudied clinical manifestation. J Autoimmun 2022; 132: 102911. doi:10.1016/j.jaut.2022.102911

⁵⁴ Yadav N, Toader A, Rajasethupathy P. Beyond hippocampus: Thalamic and prefrontal contributions to an evolving memory. Neuron 2024; 112: 1045–59. doi:10.1016/j.neuron.2023.12.021

⁵⁵ Gangopadhyay P, Chawla M, Dal Monte O, et al. Prefrontal-amygdala circuits in social decision-making. Nat Neurosci 2021; 24: 5–18. doi:10.1038/s41593-020-00738-9

⁵⁶ Doi T, Fan Y, Gold JI, et al. The caudate nucleus contributes causally to decisions that balance reward and uncertain visual information. Elife 2020; 9: e56694. doi:10.7554/eLife.56694

⁵⁷ Hanly JG, Urowitz MB, Su L, et al. Short-term outcome of neuropsychiatric events in systemic lupus erythematosus upon enrollment into an international inception cohort study. Arthritis Rheum 2008; 59: 721–9. doi:10.1002/art.23566

⁵⁸ Joëls M. Corticosteroids and the brain. J Endocrinol 2018; 238: R121–30. doi:10.1530/JOE-18-0226

⁵⁹ Zhang W, Egashira N, Masuda S. Recent Topics on The Mechanisms of Immunosuppressive Therapy-Related Neurotoxicities. Int J Mol Sci 2019; 20: 3210. doi:10.3390/ijms20133210