Systemic lupus erythematosus (SLE) is a chronic multi-organ autoimmune disease with alternating periods of relapse and remission. A major complication is neuropsychiatric lupus (NPSLE), which mainly affects the central nervous system (CNS) and poses a diagnostic challenge due to the low sensitivity and specificity of available tests. Magnetic resonance imaging (MRI) is the standard diagnostic technique, although it has limitations, as about 60% of patients with NPSLE may have normal MRI results. Advanced studies have identified volumetric changes in the brain, which could facilitate early diagnosis and improve clinical outcomes. This study focused on characterizing MRI findings using automated brain volumetric analysis in patients with and without NPSLE at a high-complexity university hospital in Cali, Colombia.
MethodsMRI scans of patients with a diagnosis of NPSLE over the last 8 years were reviewed. A total of 43 patients were included, divided into two groups: 26 with NPSLE and 17 without such manifestations. A control group of healthy subjects was selected via propensity score matching to control for sociodemographic differences. Brain volumes were normalized as volumetric indices (VI), and cortical thickness was measured in 34 regions according to the Desikan–Killiany atlas. Differences between groups were statistically analyzed using Student's t or Mann–Whitney U tests.
ResultsNo statistically significant differences in age and sex were observed between healthy controls and NPSLE and non-NPSLE patients. In the NPSLE group, the most frequent manifestations were seizures (34.6%), headache (30.8%), and cerebrovascular disease (23.1%), while in the non-NPSLE group, cranial neuropathy was observed in 17.6%. NPSLE patients showed significant reductions in cortical grey matter, white matter (WM), hippocampus, amygdala, and cerebellar cortex volumes compared to healthy controls, together with an increase in the VI of WM hypointensities. In addition, cortical analysis revealed reductions in the thickness of frontal and temporal regions. In the non-NPSLE group, significant reductions in hippocampal, cerebellar cortex, and ventricular volumes were also observed compared with healthy controls, accompanied by an increase in the VI of WM hypointensities. In addition, supratentorial volume, excluding ventricles, was significantly lower in NPSLE than in non-NPSLE.
ConclusionsA significant reduction in brain volume was identified in patients with NPSLE compared to healthy controls, mainly affecting the temporal and frontal lobes, as well as the hippocampus and cerebellar cortex. These findings, associated with disease duration, neuropsychiatric manifestations, and cumulative corticosteroid doses, suggest a multifactorial etiology and reinforce the value of MRI in detecting predictors of CNS involvement.
El lupus eritematoso sistémico (LES) es una enfermedad autoinmune crónica multiorgánica con periodos alternos de recaída y remisión. Una complicación importante es el lupus neuropsiquiátrico (LESNP), que afecta principalmente al sistema nervioso central (SNC) y plantea un reto diagnóstico debido a la baja sensibilidad y especificidad de las pruebas disponibles. La resonancia magnética (RM) es la técnica diagnóstica estándar, aunque presenta limitaciones, ya que alrededor del 60% de los pacientes con LESNP pueden presentar imágenes normales. Estudios avanzados han identificado cambios volumétricos en el cerebro, lo que podría facilitar el diagnóstico precoz y mejorar los resultados clínicos. Este estudio se enfocó en caracterizar los hallazgos en RM mediante técnicas automatizadas de análisis volumétrico cerebral en pacientes con y sin LESNP en un hospital universitario de alta complejidad en Cali, Colombia.
MétodosSe revisaron las RM de pacientes con LESNP diagnosticado en los últimos 8 años. Se incluyeron 43 pacientes, divididos en dos grupos: 26 con LESNP y 17 sin manifestaciones. Se seleccionó un grupo de control de sujetos sanos mediante emparejamiento por puntuación de propensión, a fin de evitar diferencias sociodemográficas. Los volúmenes cerebrales se normalizaron como índices volumétricos (IV) y se midió el grosor cortical en 34 regiones según el atlas de Desikan-Killiany. Las diferencias entre grupos se analizaron estadísticamente mediante pruebas t de Student o U de Mann-Whitney.
ResultadosNo se observaron diferencias estadísticamente significativas en edad y sexo entre los controles sanos y los pacientes con y sin LESNP. En el grupo con LESNP, las manifestaciones más frecuentes fueron convulsiones (34,6%), cefalea (30,8%) y enfermedad cerebrovascular (23,1%), mientras que en el grupo sin LESNP se observó neuropatía craneal en el 17,6%. Los pacientes con LESNP mostraron reducciones significativas en los volúmenes de materia gris cortical, materia blanca (MB), hipocampo, amígdala y corteza cerebelosa en comparación con los controles sanos, junto con un aumento del IV de hipointensidades de la MB. Además, el análisis cortical reveló reducciones en el grosor de las regiones frontal y temporal. En el grupo sin LESNP, también se observaron reducciones significativas en los volúmenes del hipocampo, la corteza cerebelosa y el ventrículo en comparación con los controles sanos, junto con un aumento del VI de hipointensidades de la MB. Además, el volumen supratentorial, excluyendo los ventrículos, fue significativamente menor en el grupo con LESNP que en el grupo sin LESNP.
ConclusionesSe identificó una reducción significativa del volumen cerebral en pacientes con LESNP en comparación con controles sanos, que afectó principalmente a los lóbulos temporal y frontal, así como al hipocampo y a la corteza cerebelosa. Estos hallazgos, asociados a la duración de la enfermedad, las manifestaciones neuropsiquiátricas y las dosis acumuladas de corticosteroides, sugieren una etiología multifactorial y refuerzan el valor de la RM para detectar factores predictores de la afectación del SNC.
Systemic lupus erythematosus (SLE) is a chronic autoimmune disease that involves multiple organs and is characterized by periods of relapse and remission. Its incidence and prevalence are higher among young women aged 20–30 years, with men outnumbering at approximately 15:1 [1]. The mortality rate of patients with SLE is 2–5 times higher than that of the general population [2,3].
Neuropsychiatric SLE (NPSLE), also known as “neurolupus”, has a broad incidence, with multiple case series reporting rates of 12–95% and cumulative rates of 30–40%. Approximately 40–50% of these events occur at disease onset and within 1–2 years following diagnosis, and 90% are associated with the central nervous system (CNS). However, <40% of these events are directly attributed to SLE [4,5]. To date, no case series of NPSLE has been reported in Colombia. Therefore, the spectrum of this clinical presentation in the Colombian population is currently unknown.
The neuropsychiatric manifestations of SLE represent a diagnostic challenge, as current diagnostic tests, both laboratory and imaging, have low sensitivity and specificity, making it difficult to differentiate NPSLE from other clinical conditions such as infections, primary CNS diseases, and drug toxicities [6–8]. Various studies have suggested an association between certain autoantibodies and NPSLE. These autoantibodies include anti-N-methyl-d-aspartate receptor, which is associated with autoimmune encephalitis, and anti-ribosomal P, which has been associated with neuropsychiatric manifestations of SLE, especially psychosis [9]. Marín et al., in their 2022 study, sought to determine the association between anti-ribosomal P and various manifestations of SLE. They found no association between NPSLE and anti-P antibodies (p>0.05); their results were limited by the low frequency of specific neuropsychiatric manifestations in the study population (<7%) and a small sample size [9].
The first diagnostic approach for treating NPSLE involves neuroimaging, with magnetic resonance imaging (MRI) being the preferred technique to rule out other differential diagnoses [6,10]. The European League Against Rheumatism recommendations suggest an imaging protocol that includes T1- and T2-weighted sequences, fluid-attenuated inversion recovery (FLAIR), gradient-echo, and diffusion-weighted imaging, and T1-weighted sequences following contrast medium administration [11,12].
Multiple non-specific anatomical abnormalities have been described in MRI, such as the hyperintensity of the white matter (WM) and gray matter (GM) in T2 sequences, diffuse or focal cerebral atrophy, large infarctions, lacunar infarctions, inflammatory injuries, myelitis, intraparenchymal hemorrhages, subarachnoid hemorrhages, brain microhemorrhages, and posterior reversible encephalopathy syndrome [10,13–15]. However, these imaging alterations may be observed in approximately 50% of patients with SLE without neuropsychiatric manifestations [8,10,14], and approximately 60% of patients with NPSLE may show normal imaging results [15].
Advanced techniques, such as MRI spectroscopy, have revealed that patients with NPSLE may exhibit metabolic abnormalities in the posterior cingulate gyrus and dorsal thalamus, with normal lenticular nuclei and WM [16,17]. Furthermore, brain volumetric analysis is a widely used technique for studying CNS pathologies, such as epilepsy, multiple sclerosis, and dementias, because it can detect brain changes before symptom onset and plays an important role in disease follow-up and prognosis. Performing volumetric analysis in patients with SLE could reveal chronic changes in the brain, which, when associated with neuropsychiatric symptoms, make it more likely to predict that the patient is suffering from NPSLE. Compared with patients without neurological symptoms, patients with NPSLE have been shown to have reduced volumes of the corpus callosum and the right thalamus [8,16,17]. In addition, patients with NPSLE show a reduction in GM volume in the occipital lobes compared with those without neuropsychiatric manifestations; moreover, compared with healthy individuals, patients with NPSLE show a reduction in GM volume in the frontal and temporal lobes [16,17].
Identifying acute, subacute, or chronic changes that facilitate convenient diagnosis and treatment may significantly influence patient outcomes. Thus, this study aimed to characterize MRI results using automated techniques for brain volumetric analysis in patients with SLE who presented with and without neuropsychiatric manifestations at a high-complexity university hospital in Cali, Colombia.
Materials and methodsSubjectsMRI scans of patients with diagnostic criteria for SLE according to the American College of Rheumatology (ACR) and NPSLE from the last 8 years were reviewed; only those that allowed adequate measurement of brain volumes were considered. A total of 100 patients were recruited. After excluding those without a complete medical history, 43 patients were included divided into two groups: 26 patients with SLE who presented with neuropsychiatric manifestations according to the ACR classification (24 women, 2 men; average age: 30.8±13.3 years) and 17 patients with SLE who presented without neuropsychiatric manifestations (15 women, 2 men; average age: 35.3±15.4 years). A control group of healthy subjects was included using data from an institutional database, matched via propensity score to avoid statistically significant differences in sociodemographic characteristics between subjects and patients. This study was approved by the ethics committee.
Acquisition protocolStudies were performed using a 1.5-T MRI scanner (Avanto, SIEMENS, Germany). Patients were placed in the supine position, ensuring head immobilization using pads. In the conventional brain MRI protocol, a T1-weighted multiplanar acquisition was performed from the vertex to the posterior fossa of the brain (field of view=256mm, echo time=3.37ms, repetition time=1900ms, angle=15°, and thickness=1mm). Imaging studies associated with head motion, patients with intracranial space-occupying lesions, or MRI imaging performed using a different acquisition protocol were excluded.
Processing – brain volumetric analysisCortical reconstruction and volumetric segmentation were performed using the FreeSurfer v5.3.0 imaging analysis package installed in an iMac computer with an Intel Core i5 3.4GHz processor, 32GB of RAM, and OS X 10.9.5 operating system. FreeSurfer is documented and available free of charge (http://surfer.nmr.mgh.harvard.edu/). The technical details of these procedures have already been described in several publications [18–35]. These processes include the correction of head movement; the elimination of non-brain tissue; the segmentation of WM and of the subcortical structures of GM, including hippocampus and amygdala; intensity normalization; the tessellation of the limit between GM and WM; automatic topological correction; and the deformation of surfaces to determine GM–WM and GM–cerebrospinal fluid borders.
Once the cortical models were completed, a series of deformation procedures was performed for further analysis, including surface inflation, alignment to the spherical atlas, brain cortex parcellation, and the generation of data from surfaces, such as maps showing sulcal curves and sulcal depths. The procedures used for measuring cortical thickness have been validated using histological analyses and manual measurements [30,34,36]. The automatic subcortical segmentation (ASEG) atlas included in the software package, along with the Desikan–Killiany atlas, was used for cortical parcellation.
Quantitative analysisVolumetric values were normalized using the volumetric index (VI), defined as the ratio of the volume of the brain segment of interest and the total brain volume multiplied by 100. For each subject, VI was calculated for each of the 37 brain segments included in the ASEG atlas. Cortical thickness was assessed in millimeters for the 34 cortical regions of interest identified using the Desikan–Killiany atlas.
Statistical analysisTo compare brain volume and cortical thickness between patients and the control group, the Student's t-test or the Mann–Whitney U test was used, depending on whether the Lilliefors test met the normality assumptions. A p-value of <0.05 was considered to indicate a statistically significant difference. The analysis was performed using MATLAB (MATLAB and Statistics Toolbox Release 2017a, The MathWorks, Inc., Natick, Massachusetts, United States).
ResultsNo statistically significant differences were observed regarding age and sex between the control group and patients with SLE who presented with and without neuropsychiatric manifestations (Table 1). In the NPSLE group, the most common CNS manifestation was seizure (n=9, 34.6%), followed by headache (n=8, 30.8%) and cerebrovascular disease (n=6, 23.1%). Other observed manifestations were myelopathy, mood disorders, and acute delirium (n=3, 11.5%); demyelination syndrome, movement disorders, and aseptic meningitis (n=2, 7.7%); and psychosis (n=1, 3.8%). Moreover, in the SLE group, 17.64% of patients (3/17) showed cranial neuropathy.
At the time of MRI, no statistically significant differences were observed in disease duration, complement levels (C3 and C4), or SLEDAI-2k index score between patients with and without NPSLE. However, patients with NPSLE tended to have higher SLEDAI-2K scores (Table 2).
Clinical characteristics of the study population.
| Variable | Patients with NPSLE(n=26) | Non-NPSLE(n=17) | p |
|---|---|---|---|
| Duration of disease (months) | 48 (0–300) | 36 (0–312) | 0.954 |
| C3 level (mg/dL) | 70.01 (±31.85) | 82.99 (±29.45) | 0.195 |
| C4 level (mg/dL) | 15.15 (±9.29) | 14.77 (±7.58) | 0.949 |
| SLEDAI-2k | 20.92 (±10.20) | 15.76 (±12.27) | 0.161 |
| Sjögren's syndrome | 4 (15.38%) | 1 (5.88%) | – |
| Antiphospholipid syndrome | 3 (11.53%) | 2 (11.76%) | – |
| Rheumatoid arthritis | 2 (7.69%) | 3 (17.64%) | – |
Lumbar punctures were performed on those patients who met the medical criteria for this procedure (fever, altered state of consciousness, focal symptoms). A total of 13 lumbar punctures were performed: 10 in patients with NPSLE and 3 in patients with SLE. Pleocytosis and elevated protein levels in cerebrospinal fluid were observed in the NPSLE group. No abnormalities were observed in the SLE group (Table 3). Additionally, lupus nephritis was reported in 16 patients (10 with NPSLE and 6 with SLE), with class IV being the most common type observed in both groups (Table 4).
Results of cerebrospinal fluid analysis based on the results of needle lung biopsy performed in the study population.
| Variable | Patients with NPSLE(n=26) | Non-NPSLE(n=17) | p |
|---|---|---|---|
| Lumbar puncture | 10 (38.46) | 3 (17.64%) | 0.146 |
| CSF white blood cell (cell/mm3) | 14.50 (1–1000) | 0.00 (0.00–4.00) | 0.06 |
| CSF glucose (mg/dL) | 54.65 (24–104.60) | 55.00 (0.00–547.00) | 1.00 |
| CSF proteins (mg/dL) | 39.0 (21.6–221.2) | 21.00 (20.00–77.00) | 0.15 |
| CSF LDH (u/L) | 29 (0–80) | 18.00 (0.00–49.80) | 0.61 |
Classification of lupus nephritis cases reported in the study population.
| Variable | Patients with NPSLE(n=26) | Non-NPSLE(n=17) | p |
|---|---|---|---|
| Lupus nephritis | 10 (38.46%) | 6 (35.29%) | 0.83 |
| Class II | 1 (3.84%) | 0 (0%) | |
| Class III | 1 (3.84%) | 0 (0%) | |
| Class IV | 5 (19.23%) | 4 (23.53%) | |
| Class V | 1 (3.84%) | 1 (5.88%) | |
| Class VI | 2 (7.69%) | 1 (5.88%) |
In the NPSLE group, 88.46% (23/26) of patients received oral steroids at a median dose of 13.5mg (range: 2.5–60) of prednisone at the time of MRI. Furthermore, 57.69% (15/26) of patients received a high-dose methylprednisolone regimen (250mg/day×3 days), 53.85% (14/26) received cyclophosphamide pulse therapy, and 26.92% (7/26) received plasma exchange therapy.
In the SLE group, 88.24% (15/17) of patients received oral steroids at a median dose of 5.0mg (range: 2.5–60) at the time of MRI. Moreover, 41.18% (7/17) of patients received methylprednisolone pulse therapy, 35.29% (6/17) received cyclophosphamide pulse therapy, and 17.65% (3/17) received plasma exchange therapy.
Compared with the control group, patients with NPSLE had a statistically significant reduction in the volume of cortical GM (NPSLE=28.34±3.45; control group=29.73±1.96; p=0.03), WM (NPSLE=29.66±2.70; control group=31.10±1.81; p=0.04), the left cerebellar cortex (NPSLE=3.27±0.34; control group=3.53±0.28; p<0.01), and the right cerebellar cortex (NPSLE=3.35±0.38; control group=3.62±0.31; p<0.01). A statistically significant reduction in the volume of the ventricular system (VS) was also reported (NPSLE=1.42 [0.38–5.09]; control group=0.81 [0.44–1.94]; p<0.01).
The subcortical structure analysis revealed that compared with the control group, patients with NPSLE had a significantly reduced volume of the left hippocampus (NPSLE=0.25±0.06; control group=0.29±0.03; p<0.01), right hippocampus (NPSLE=0.25±0.08; control group=0.29±0.03; p=0.02), left putamen (NPSLE=0.33±0.05; control group=0.35±0.04; p=0.02), right putamen (NPSLE=0.31±0.04; control group=0.34±0.04; p=0.02), and left amygdala (NPSLE=0.08±0.01; control group=0.09±0.01; p=0.01). However, the VI of WM hypointensity was significantly higher in patients with NPSLE than in the control group (NPSLE=0.11 [0.06–0.94]; control group=0.07 [0.04–0.14]; p<0.01). The cortical analysis revealed a statistically significant reduction in the thickness of different regions of the frontal and temporal lobes (Table 5).
Results of brain volumetric analysis in patients with NPSLE vs. healthy controls.
| Cortical region | Healthycontrols(n=44) | Patients withNPSLE(n=26) | p |
|---|---|---|---|
| Entorhinal cortex (mm) | |||
| Right hemisphere | 3.63±0.36 | 3.10±0.64 | <0.01 |
| Left hemisphere | 3.28±0.34 | 2.97±0.52 | 0.01 |
| Fusiform cortex (mm) | |||
| Right hemisphere | 2.66±0.16 | 2.48±0.27 | <0.01 |
| Left hemisphere | 2.61±0.13 | 2.44±0.26 | <0.01 |
| Parahippocampal cortex (mm) | |||
| Right hemisphere | 2.67±0.34 | 2.15±0.36 | <0.01 |
| Left hemisphere | 2.71±0.36 | 2.17±0.39 | <0.01 |
| Superior temporal cortex (mm) | |||
| Right hemisphere | 2.78±0.24 | 2.64±0.27 | 0.04 |
| Left hemisphere | 2.80±0.20 | 2.61±0.25 | <0.01 |
| Superior frontal cortex (mm) | |||
| Right hemisphere | 2.87±0.17 | 2.64±0.32 | <0.01 |
| Left hemisphere | 2.88±0.21 | 2.69±0.36 | 0.02 |
| Rostral anterior cingulate cortex (mm) | |||
| Right hemisphere | 2.91±0.25 | 2.75±0.29 | 0.04 |
| Left hemisphere | 2.88±0.23 | 2.70±0.26 | 0.01 |
| Cingulate cortex (isthmus) (mm) | |||
| Right hemisphere | 2.43±0.17 | 2.20±0.20 | <0.01 |
| Left hemisphere | 2.48±0.20 | 2.26±0.26 | <0.01 |
Compared with the control group, patients with SLE had a significantly reduced volume of left (SLE=3.35±0.30; control group=3.53±0.28; p=0.03) and right cerebellar cortex (SLE=3.43±0.34; control group=3.62±0.31; p<0.05). A statistically significant reduction in the volume of ventricles was also observed (SLE=1.17 [0.55–4.55]; control group=0.81 [0.44–1.94]; p<0.01).
The subcortical structure analysis revealed that compared with the control group, patients with SLE had a significant reduction in the volume of the left (SLE=0.24±0.04; control group=0.29±0.03; p<0.01) and right hippocampus (SLE=0.26±0.03; control group=0.29±0.03; p=0.01). However, a significant increase in the VI of WM hypointensities was observed (SLE=0.86 [0.04–0.23]; control group=0.07 [0.04–0.14]; p<0.01). The cortical analysis revealed a statistically significant reduction in the thickness of different regions of the temporal lobes (Table 6).
Results of brain volumetric analysis in patients with SLE vs. healthy controls.
| Cortical region | Healthycontrols | Patients withSLE | p |
|---|---|---|---|
| Entorhinal cortex (mm) | |||
| Right hemisphere | 3.62±0.38 | 3.24±0.56 | 0.04 |
| Parahippocampal cortex (mm) | |||
| Right hemisphere | 2.66±0.30 | 2.35±0.47 | 0.03 |
| Left hemisphere | 2.71±0.34 | 2.32±0.36 | <0.01 |
| Posterior cingulate cortex (mm) | |||
| Left hemisphere | 2.42±0.17 | 2.28±0.20 | 0.03 |
| Cingulate cortex (isthmus) (mm) | |||
| Right hemisphere | 2.43±0.23 | 2.25±0.17 | 0.02 |
| Left hemisphere | 2.49±0.24 | 2.30±0.27 | 0.04 |
There was a statistically significant reduction in the supratentorial volume excluding ventricles in patients with NPSLE compared with those with SLE (NPSLE=851,785.86±61,530.07; SLE=922,670.08±92,238.71; p<0.01).
In terms of volumes, statistically significant differences were found for the cerebellar cortex and WM in both the left and right hemispheres (Table 7). Moreover, a statistically significant increase in the VI of WM hypointensity was observed in the NPSLE group (NPSLE=0.12±0.26; SLE=0.09±0.05; p<0.05).
Results of brain volumetric analysis in patients with NPSLE vs. those with SLE.
| Brain structure | NPSLE (VI) | Non-NPSLE (VI) | p |
|---|---|---|---|
| Cerebellar cortex | |||
| Right hemisphere | 45.946±5.132 | 49.851±6.386 | 0.03 |
| Left hemisphere | 44.951±5.056 | 48.795±6.884 | 0.04 |
| WM | |||
| Right hemisphere | 206.363±21.870 | 225.858±26.258 | 0.01 |
| Left hemisphere | 203.181±21.081 | 223.801±26.931 | <0.01 |
Currently, the diagnosis of NPSLE poses a challenge, as existing evidence reports low sensitivity and specificity for current diagnostic techniques. Using structural MRI and automatic segmentation methods, we analyzed changes in the global and regional volumes of GM and WM in patients with SLE who presented with and without neuropsychiatric manifestations, as well as in the control group (Fig. 1). To the best of our knowledge, this is the first case series to clarify and describe the spectrum of clinical presentation and imaging characteristics of SLE in the Colombian population.
Brain volume in non-NPSLE (A) and NPSLE (B). T2-weighted coronal sequences. (A) A 45-year-old female patient shows slight prominence of the sulci and gyri without alteration in signal intensity; volume 1.504.062mm3. (B) A 30-year-old female patient with loss of brain parenchyma volume; volume: 1.143.445mm3.
In this study, no statistically significant differences in SLEDAI-2K scores were observed between patients with NPSLE and those with non-NPSLE. This finding may be explained by the severity of the clinical presentation in both patient groups. Our health institution is highly complex and serves as a regional reference center, receiving patients with SLE with severe lung, hematological, and renal manifestations, where differences in parameters such as disease activity scales (SLEDAI-2K) or serum complement levels may not be determined. An association with Sjögren's syndrome and antiphospholipid syndrome, which may be independently associated with neurological effects, was primarily found in patients with NPSLE and may represent a high morbidity and physiopathological load for CNS involvement in these patients.
Conversely, a statistically significant reduction in brain volume, both in GM and WM, was observed in patients with NPSLE compared with the control group with similar sociodemographic characteristics, whereas no statistically significant alterations in supratentorial volumes were observed in patients with SLE. This GM atrophy pattern observed in NPSLE, previously described as generalized and diffuse by several authors [17,37,38], is associated with disease duration, the number of CNS manifestations, and cumulative corticosteroid doses. In relation to the findings previously reported by other studies [17], our study shows that the temporal lobe is primarily affected, involving areas such as the entorhinal, fusiform gyrus, parahippocampal, and superior temporal cortices, as well as the frontal lobe, including the superior frontal gyrus and the anterior cingulate cortex. Longitudinal studies have shown that this atrophy pattern is not only progressive but also closely associated with cumulative corticosteroid doses and the secondary manifestation of cognitive deterioration. Furthermore, a statistically significant atrophy of cerebellar GM in both hemispheres and an increase in the volume of WM hypointensity were observed in our study. Although these findings have not been previously reported in patients with SLE, some authors have demonstrated similar atrophy patterns in patients with NPSLE [39], suggesting that this pattern may be a predictive factor for the conversion or onset of neurological symptoms.
The subcortical structure analysis in patients with SLE and NPSLE revealed a statistically significant reduction in bilateral hippocampal volume, consistent with previously reported findings [39,40]. The atrophy pattern in hippocampal structures, widely described in patients with SLE who present with neuropsychiatric manifestations, has been reported as a secondary finding in association with alterations in signal intensity in the head of the hippocampus and has been suggested as an early indicator of the transition from SLE to NPSLE [41].
These imaging findings in patients with SLE and those with NPSLE may suggest a multifactorial etiology where the presence and interaction between different physiological alterations such as vasculitis, vasculopathy, the duration of disease (cumulative damage), the direct effects of antibodies on neural tissue, and the adverse effects of drug products (especially of the cumulative doses of steroids) may play an important role.
The lack of statistically significant differences in clinical variables between the study groups, along with the complexity of diagnosing NPLES and evidence of structural brain changes on MRI, suggests that brain imaging, including morphometry analysis, should be considered in patients with a diagnosis of SLE. These imaging techniques allow identification of predictors of CNS involvement, thereby enabling early therapeutic interventions or strict clinical follow-up and decreasing morbidity and mortality associated with SLE.
This study has certain limitations. First, this is a retrospective study in which gaps in clinical and paraclinical variables were identified, limiting statistical analysis. For instance, as no specific information was available regarding the cumulative doses of steroids administered to patients, it was not possible to establish a relationship between doses and morphological changes, as described in the existing literature. Second, even though our sample size may be sufficient to fulfill the study's main purpose, the number of patients with associated syndromes did not allow the use of an appropriate methodological approach to establish the specific patterns of cortical-subcortical atrophy in each clinical stage. Finally, neuropsychiatric manifestations, such as cognitive dysfunction, were not considered in this study because data on this topic were unavailable; some authors have reported significant relationships between these symptoms and brain volume in patients with NPSLE [17,42].
ConclusionsThis study, the first in the Colombian population, identified, using structural MRI and automatic segmentation, a significant reduction in brain volume in patients with NPSLE compared to healthy controls, mainly affecting the temporal and frontal lobes, as well as the hippocampus and cerebellar cortex. These findings, associated with disease duration, neuropsychiatric manifestations, and cumulative corticosteroid doses, suggest a multifactorial etiology and reinforce the value of MRI in detecting predictors of CNS involvement. However, the absence of significant clinical differences between groups and methodological limitations, such as retrospective data and small sample size, highlight the need for prospective studies and comprehensive analyses to optimize early diagnosis and management of NPSLE.
Authors’ contributionsAll authors contributed to the creation of this manuscript.
Ethical considerationsThis study was approved by the institution's ethics committee.
Informed consentThis study does not include informed consent as it is retrospective in nature.
Declaration of generative AI and AI-assisted technologies in the writing processNo form of AI was used in the preparation of this manuscript.
FundingThis research received no external funding.
Conflicts of interestAll authors declare that they have no conflicts of interest.
