Systemic Lupus Erythematosus (SLE) is a quintessential systemic autoimmune disease defined by a profound loss of self-tolerance, leading to the production of a wide array of autoantibodies and subsequent multi-organ inflammation and damage (1). Nervous system involvement represents one of the most severe and least understood challenges in the management of the disease.

The nomenclature used to describe this condition has evolved in parallel with our understanding of its pathology. Historically, terms such as “lupus cerebritis” or “lupus sclerosis” were employed, particularly when central nervous system (CNS) inflammation was presumed to be the primary driver (1). However, these terms are now considered outdated as they fail to capture the full mechanistic diversity of the condition. The modern, accepted term is Neuropsychiatric SLE (NPSLE), which serves as a broad umbrella for the entire spectrum of neurological and psychiatric syndromes affecting both the central and peripheral nervous systems (1). This terminological shift reflects a critical paradigm change-away from a singular focus on inflammation (“-itis”) and toward a more sophisticated appreciation of a multifactorial syndrome encompassing vascular, metabolic, and neurotoxic mechanisms that are not always overtly inflammatory. This review will use the term NPSLE throughout, specifying CNS or peripheral nervous system (PNS) involvement where appropriate. Regardless of nomenclature, NPSLE contributes significantly to disease-related morbidity, mortality, and a profoundly diminished quality of life for patients (2).

The true frequency of NPSLE has been a subject of considerable debate, with historical estimates varying dramatically from as low as 12% to as high as 95% (3). This wide range is largely a product of methodological differences between studies, including retrospective versus prospective designs, inconsistencies in case definitions, and the inclusion of both major and minor, non-specific syndromes (2). More recent, rigorous meta-analyses of prospective studies have provided a more realistic estimate, indicating that approximately 56% of systemic lupus erythematosus (SLE) patients will experience at least one neuropsychiatric event during their disease course. Among these events, common manifestations such as headache, mood disorders, and cognitive dysfunction are the most frequent contributors to this figure (4).

Critically, a neuropsychiatric event occurring in an SLE patient is not synonymous with NPSLE. The concept of attribution is paramount; rigorous clinical evaluation is required to determine whether a given event is a direct consequence of SLE-related pathogenic mechanisms or is attributable to other causes, such as medication side effects, infections, or comorbid conditions. In daily clinical practice, attribution remains largely a diagnosis of exclusion, typically relying on multidisciplinary consensus to rule out “mimics” (e.g., infections, metabolic derangements) rather than solely on research-based attribution algorithms. Large, prospective cohort studies have consistently shown that only a fraction of these events-estimated to be between 30 and 50%-can be directly attributed to SLE itself (5). This distinction is not merely academic; it is fundamental to appropriate management, as misattribution can lead to unnecessary and potentially harmful immunosuppression.

The clinical heterogeneity of NPSLE is formally captured by the 19 distinct syndromes defined by the American College of Rheumatology (ACR) in 1999, which are broadly divided into 12 affecting the CNS and seven affecting the PNS (2). These syndromes can be further categorized based on their clinical presentation as either focal or diffuse (2). Focal manifestations, such as stroke, transverse myelitis, or seizures, are characterized by neurological deficits that can be localized to a specific neuroanatomical area. In contrast, diffuse manifestations, including cognitive dysfunction, psychosis, mood disorders, and acute confusional states, involve more widespread brain dysfunction.

This clinical distinction provides a foundational framework for considering the underlying pathophysiology, which is broadly divisible into two major mechanistic pathways: ischemic/thrombotic and autoimmune/inflammatory (6). Ischemic events are frequently driven by a prothrombotic state, often in the context of antiphospholipid antibodies (aPL), leading to vasculopathy and thrombosis. The inflammatory pathway is more complex, involving the direct or indirect effects of autoantibodies, pro-inflammatory cytokines, and infiltrating immune cells on neuronal and glial function. It is now clear that these pathways are not mutually exclusive and often interact, creating a complex pathogenic landscape that underlies the diverse clinical presentations of NPSLE and poses significant diagnostic and therapeutic challenges (7). Table 1 provides an overview of key NPSLE syndromes and their primary pathogenic considerations.

The CNS is protected by highly specialized barriers, primarily the blood–brain barrier (BBB) formed by tight junctions between endothelial cells, and the blood-cerebrospinal fluid (CSF) barrier located at the choroid plexus (8). Disruption of these barriers is a critical initiating event in many forms of NPSLE, as it permits the entry of circulating autoantibodies, inflammatory cytokines, and immune cells from the periphery into the brain microenvironment (8).

Evidence for BBB permeability in NPSLE comes from multiple sources. Analysis of CSF often reveals an elevated albumin quotient (Qalb), which indicates leakage of this high-molecular-weight serum protein into the CNS (9). More advanced neuroimaging techniques, such as dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI), provide a quantitative measure of this disruption. Studies using DCE-MRI have demonstrated increased permeability, reflected by a higher volume transfer constant (Ktrans), particularly in the hippocampus of SLE patients, even in the absence of overt neuropsychiatric symptoms (8). The integrity of these barriers can be compromised by systemic inflammation, the direct action of cytokines like IL-6, and autoantibody-mediated endothelial injury, creating a gateway for the peripheral immune system to access and damage the CNS (9).

Once the BBB is breached, the resident immune cells of the CNS-microglia and astrocytes-are not passive bystanders. They become key amplifiers and drivers of the neuroinflammatory cascade, shaping the nature and severity of the neurological deficit.

Microglia, the brain’s resident macrophages, are exquisitely sensitive to disturbances in CNS homeostasis. In NPSLE, they undergo a profound activation, polarizing towards a pro-inflammatory M1 phenotype (10). This is evidenced by increased levels of M1-associated cytokines (e.g., IL-1, IL-6, TNF-α) in the CSF of NPSLE patients and in the brains of MRL/lpr lupus mice (10). A key intracellular platform for this pro-inflammatory response is the NLRP3 inflammasome, which, upon activation in microglia, processes pro-IL-1 into its highly inflammatory mature form (11). Transcriptomic analysis of microglia from lupus models reveals the upregulation of genes associated not only with inflammation but also with phagocytosis (12–15). This enhanced phagocytic activity leads to a pathological process known as synaptic stripping, where activated microglia engulf and eliminate synapses, a mechanism directly implicated in the cognitive deficits and anxiety-like behaviors observed in animal models of NPSLE (10).

Astrocytes, the most abundant glial cells in the CNS, also play a dual role (16). Under pathological conditions like NPSLE, they can become reactive and adopt different phenotypes. Pro-inflammatory signals from activated M1 microglia, such as IL-1, TNF-α, and C1q, can induce astrocytes to polarize into a neurotoxic A1 phenotype. A1 astrocytes lose their normal neurosupportive functions and release factors that are directly toxic to neurons and oligodendrocytes (17). Evidence of astrocytic injury in NPSLE is robust, primarily through the measurement of glial fibrillary acidic protein (GFAP), an astrocyte-specific intermediate filament protein. Elevated levels of GFAP in the CSF of NPSLE patients are a reliable marker of astrocytic damage and correlate with disease activity and abnormalities on conventional MRI (18). More recently, serum GFAP has emerged as a promising, less invasive biomarker that is also elevated in active major NPSLE, reflecting this ongoing glial pathology (19).

The inflammatory cascade within the CNS is initiated and sustained by a continuous influx of pathogenic mediators from the systemic circulation.

Autoantibodies are central to this process. Several specific autoantibodies have been strongly implicated in NPSLE through direct pathogenic mechanisms:

Pro-inflammatory cytokines produced systemically can cross a compromised BBB or be produced intrathecally, further driving neuroinflammation:

Neutrophils, key cells of the innate immune system, also contribute significantly. In SLE, neutrophils are prone to forming Neutrophil Extracellular Traps (NETs), web-like structures of decondensed chromatin and granular proteins. These NETs are a potent source of autoantigens (e.g., dsDNA, histones) and pro-inflammatory mediators (e.g., neutrophil elastase). By promoting endothelial cell damage and thrombosis, NETs can contribute to both BBB disruption and the ischemic manifestations of NPSLE (23).

The traditional dichotomy between ischemic and inflammatory NPSLE is an oversimplification that fails to capture the intricate interplay between these pathways. Systemic inflammation, driven by cytokines and activated innate immune cells, creates a prothrombotic state by activating endothelial cells and platelets (15). Conversely, vascular damage and ischemia can further disrupt the BBB, creating a feed-forward loop that facilitates the entry of more inflammatory mediators and immune cells into the CNS (6). This integration of mechanisms explains why some patients present with “mixed” phenotypes and underscores the need for therapies that can address both arms of NPSLE pathogenesis. Table 2 summarizes the key immune cell populations involved in this integrated model.

T lymphocytes are critical orchestrators of the autoimmune response in SLE, and their infiltration into the CNS is a key feature of inflammatory NPSLE (7). A fundamental aspect of this pathology is an imbalance between effector and regulatory T cell populations.

Regulatory T cells (Tregs), which are essential for maintaining self-tolerance, are often numerically deficient or functionally impaired in SLE, allowing autoreactive T cells to escape control and proliferate (24). Concurrently, pro-inflammatory T helper subsets, including IFN-γ-producing Th1 cells and IL-17-producing Th17 cells, are expanded and contribute to tissue damage and BBB breakdown (25).

Animal models have been instrumental in dissecting the heterogeneity of pathogenic T cells within the CNS. In the MRL/lpr mouse model, two distinct populations have been identified as key contributors:

The presence of autoantibodies in the CSF of NPSLE patients has long implicated B cells in the disease process. It is now increasingly clear that the CNS is not merely a passive recipient of peripherally produced antibodies. Instead, it can become an active site of B cell responses, supporting local differentiation, clonal expansion, and intrathecal antibody production (1).

Studies of CSF from patients with various neuroinflammatory disorders have consistently shown an enrichment of B cells and, most notably, antibody-secreting cells (ASCs) like plasma cells and plasmablasts (26). These intrathecal B cell populations are often clonally expanded and have undergone immunoglobulin class-switching and somatic hypermutation, hallmarks of a mature, antigen-driven immune response occurring within the CNS compartment (26). This local immune activity is fostered by a supportive microenvironment, which includes the production of the potent B cell chemoattractant CXCL13 and B cell survival factors such as BAFF and APRIL (27). Pathogenic B cell subsets that are expanded systemically in SLE, such as Age-Associated B Cells (ABCs), are also likely contributors to the pool of autoreactive cells that can enter and function within the CNS (28).

Flow cytometric studies of CSF from patients with neuropsychiatric connective tissue diseases (N-CTD), including NPSLE, have provided important diagnostic context. While one key study by Heming et al. found no single CSF parameter that could reliably distinguish N-CTD from non-neuropsychiatric CTD, it made a crucial observation: an expansion of plasma cells in the peripheral blood was highly effective at differentiating N-CTD patients from those with multiple sclerosis (MS) (29). This highlights that while intrathecal B cell activity is a feature of neuroinflammation, specific signatures in the peripheral blood may hold greater differential diagnostic power, reflecting the systemic nature of the disease driving CNS pathology.

Spontaneous mouse models of lupus have been invaluable for mechanistic studies, though it is crucial to recognize that no single model perfectly recapitulates the human disease. A balanced perspective requires considering insights from multiple strains.

The MRL/lpr mouse is perhaps the most widely studied model for NPSLE. These mice develop an aggressive, accelerated autoimmune disease due to a mutation in the Fas gene, leading to defective lymphocyte apoptosis. They exhibit a prominent neurobehavioral phenotype, including depression- and anxiety-like behaviors and cognitive deficits, which often appear early in the disease course (30). As discussed, their CNS pathology is characterized by early and robust T cell infiltration into the choroid plexus and widespread microglial activation (10).

The New Zealand Black/White (NZB/W) F1 mouse represents another key spontaneous model. In contrast to MRL/lpr mice, NZB/W F1 mice develop a more indolent disease with a later onset of neuropsychiatric symptoms (31). Their behavioral phenotype is also distinct, primarily characterized by impaired learning and memory and increased anxiety, rather than the prominent depressive-like features of the MRL/lpr strain (31). The neuropathology in NZB/W F1 mice often involves diffuse lymphoproliferative infiltrates in the meninges, choroid plexus, and perivascular spaces, along with evidence of immune complex deposition (32).

The differences between these models are instructive. The MRL/lpr strain, with its acute, T-cell-driven inflammation at the blood-CSF barrier, may better model the acute inflammatory or psychotic syndromes seen in human NPSLE. In contrast, the NZB/W F1 model, with its later onset and more diffuse, chronic immune infiltration, might be more representative of the progressive cognitive decline associated with widespread, smoldering neuroinflammation. Critically evaluating findings from both models provides a more comprehensive picture of the potential pathogenic pathways active in human disease (30). However, caution is warranted when extrapolating these findings to human disease, as murine models often possess specific genetic mutations (e.g., Fas) and display simplified behavioral phenotypes that do not fully capture the complexity of human NPSLE.

Single-cell RNA sequencing (scRNA-seq) has revolutionized immunology by enabling the transcriptomic profiling of thousands of individual cells simultaneously, revealing cellular identity, activation states, and functional potential (33). Multimodal approaches like Cellular Indexing of Transcriptomes and Epitopes by Sequencing (CITE-seq) further enhance this by concurrently measuring surface protein expression, providing a more robust classification of immune cell subsets (34). These technologies are indispensable for deconstructing a clinically and pathologically complex disease like NPSLE (7).

However, it is crucial to acknowledge the current landscape of this research accurately. Direct scRNA-seq studies on brain tissue or CSF from human NPSLE patients are exceptionally rare due to the invasive nature of sample collection. Therefore, our current high-resolution understanding is a mosaic, built from detailed mechanistic insights from animal models and inferences drawn from human CSF studies in related, and often more common, neuroinflammatory diseases like MS (26). Synthesizing these disparate data sources allows for the generation of robust, testable hypotheses for NPSLE.

The application of scRNA-seq to the MRL/lpr mouse brain has provided a detailed catalog of the T cell landscape in lupus-associated neuroinflammation, identifying the senescent/exhausted Eomes+ DNTs and other conventional T cell subsets (7). To understand the potential relevance of these findings to human disease, we can turn to large-scale single-cell atlases of human CSF. These studies, while not specific to NPSLE, reveal common features of T cell responses during neuroinflammation. They consistently show that CNS-infiltrating T cells shift towards memory and effector phenotypes. Clonally expanded T cells, often cytotoxic CD8 + subsets, upregulate genes associated with cytotoxicity (e.g., granzymes, perforin), CNS trafficking (e.g., CXCR3), and markers of senescence or exhaustion (26). This convergence of findings-particularly the theme of T cell exhaustion in both the lupus mouse brain and inflamed human CSF-suggests that this may be a shared terminal differentiation state for T cells operating within the chronic inflammatory milieu of the CNS.

A particularly powerful application of single-cell technology is the combination of scRNA-seq with single-cell B cell receptor sequencing (scBCR-seq). This dual-omics approach makes it possible to directly link a B cell’s complete transcriptional profile-revealing its activation state, functional potential, and interaction pathways-to the precise sequence of its B cell receptor (35).

For NPSLE, this technology holds transformative potential. By applying it to B cells isolated from the CSF, researchers could definitively identify the antigen specificities of the autoreactive B cell clones operating within the CNS. Furthermore, it would enable the tracking of clonal lineages between the CSF and peripheral blood, clarifying trafficking patterns and cellular origins. This approach could ultimately reveal whether distinct NPSLE clinical subtypes (e.g., psychosis vs. cognitive dysfunction) are driven by unique CNS B cell receptor repertoires or are associated with specific B cell functional states, thereby pinpointing the “culprit” clones and their pathogenic mechanisms (36).

A key limitation of standard scRNA-seq is that the process of tissue dissociation results in the loss of all spatial context. Spatial transcriptomics is an emerging technology that overcomes this by measuring gene expression within the intact tissue architecture (7). This allows identified cell types and states to be mapped back to their precise locations, revealing cellular neighborhoods and interactions.

Recent pioneering studies have applied this technology to the brains of lupus mouse models. These investigations have revealed that the Type I IFN gene signature, a key pathogenic driver in NPSLE, is not uniformly distributed but is enriched in spatially distinct patches within the brain parenchyma, particularly in subcortical areas (37). This finding provides strong evidence against a model of passive diffusion of peripheral IFN into the brain and instead points toward localized production or response. By integrating scRNA-seq data with these spatial maps, it becomes possible to understand how different cell types (e.g., IFN-producing glial cells and IFN-responding neurons) are organized and interact within these inflammatory foci, providing unprecedented insight into the structural basis of neuroinflammation in NPSLE. Table 3 summarizes the key findings from these and other relevant high-parameter studies, providing a comparative overview of the insights gained from different models and patient cohorts.

The current ACR classification system for NPSLE is descriptive and does not always align with underlying biology. Single-cell data offers the potential to move beyond this syndromic classification towards a more precise, mechanism-based stratification based on “immune endophenotypes” (distinct biological subgroups defined by specific immune mechanisms rather than just clinical symptoms). These endophenotypes would be defined by specific T and B cell transcriptional signatures, surface marker profiles, or unique immune receptor repertoires identified in CSF or blood. Such a biologically grounded re-classification could align more closely with pathogenic mechanisms. For example, one might hypothesize the existence of:

It is important to emphasize that these “immune endophenotypes” are currently hypothetical constructs derived largely from animal models. Their clinical validity remains to be established through “reverse translation” studies—for instance, by correlating single-cell profiles from patient biospecimens with specific clinical syndromes—to determine if they represent actionable clinical entities.

There is a critical and unmet need for reliable biomarkers to improve the diagnosis of NPSLE, which currently relies on a combination of clinical judgment and the exclusion of other conditions (19).

Fluid biomarkers from CSF and blood are a major focus of this effort. Markers of neuronal and glial injury, such as neurofilament light chain (NfL) and GFAP, have shown significant promise. Studies have demonstrated that both CSF and serum levels of NfL and GFAP are elevated in patients with active major NPSLE, and serum NfL in particular shows excellent discriminatory power in distinguishing NPSLE from non-NPSLE controls (19). Other promising candidates include the B cell chemoattractant CXCL13, which is elevated in the CSF during neuroinflammation and may reflect intrathecal B cell activity, and novel markers identified through unbiased proteomics screens of CSF (39). However, widespread clinical adoption faces practical hurdles. While potentially discriminatory, these markers currently lack standardized commercial assays with defined, age-adjusted cutoffs for SLE populations. Furthermore, as markers of general neuronal injury rather than specific lupus pathology, they must be interpreted cautiously alongside clinical features.

Advanced neuroimaging offers a non-invasive window into brain pathology. Beyond conventional MRI, which is often normal or shows non-specific findings, advanced techniques can detect subtle microstructural and functional changes:

A more granular understanding of NPSLE pathogenesis is essential for moving beyond broad immunosuppression towards targeted therapies (43). While conventional immunosuppressants like cyclophosphamide and mycophenolate mofetil (MMF) remain important for severe, inflammatory NPSLE, their use is based on broad efficacy in SLE rather than specific evidence in NPSLE (44).

The advent of biologic therapies offers the potential to target specific pathogenic pathways identified in NPSLE:

A significant hurdle in validating these targeted therapies is the historical exclusion of patients with active, severe CNS involvement from major randomized clinical trials, resulting in a scarcity of high-level evidence for this specific population. Future therapeutic development could target even more specific mechanisms highlighted by recent research, such as inhibitors of microglial activation, antagonists of the IL-6 receptor or IL-6 trans-signaling, or agents that selectively target pathogenic T cell subsets or restore Treg function.