Identification of Distinct Biological Groups of Patients With Cryptogenic NORSE via Inflammatory Profiling

Abstract

Background and objectives: Emerging evidence suggests that immune dysregulation plays a pivotal role in triggering cryptogenic new-onset refractory status epilepticus (c-NORSE), prompting a consensus on early initiation of immunotherapy. However, despite similar timing of administration, responses to immunotherapies have been varied and unpredictable, suggesting the presence of heterogeneous underlying mechanisms. The aim of this study was to identify distinct inflammatory response subtypes in patients with c-NORSE by analyzing their cytokine profiles. Insights into underlying mechanisms were sought to understand the pathophysiology and guide personalized therapies to improve patient outcomes.

Methods: Sixty-two patients with c-NORSE were included. A comprehensive panel of 96 cytokines was analyzed in serum samples. Patients were clustered based on their cytokine profiles using the Louvain algorithm, an unsupervised graph-based clustering method. The identified clusters of patients were compared regarding cytokine levels and clinical features. Protein pathway analysis was used to explore the biological relevance of the inflammatory markers within each cluster. Patients with c-NORSE were compared with control patients (n = 18) and patients with other forms of refractory SE (n = 45).

Results: Compared with controls, patients with c-NORSE exhibited significant differences in 33 cytokines. Pathway analysis revealed dysregulations in chemotaxis and neutrophil recruitment and migration, highlighting the importance of innate immunity in patients with c-NORSE. Within the c-NORSE cohort, 3 clusters of patients emerged: cluster A, lacking specific inflammatory markers; cluster B, with a much stronger innate-immunity cytokine-driven inflammatory response compared with clusters A and C; and cluster C, defined by dysregulated autoimmune processes. Notably, patients in cluster B showed a statistically significant elevation of innate immune-related proinflammatory cytokines associated with leukocyte recruitment and degranulation. By contrast, those in cluster C showed activation of Janus kinase signal transducer and activator of transcription (JAK-STAT) pathways, suggesting autoimmune mechanisms. Patients in clusters B and C demonstrated varied responses to immunotherapies, with cluster C patients showing favorable outcomes after multiple immunotherapies.

Discussion: The identification of distinct inflammatory subgroups in c-NORSE suggests that variations in the underlying immune mechanisms contribute to differential treatment responses. These findings underscore the importance of personalized therapeutic strategies, potentially targeting specific inflammatory pathways, to optimize clinical outcomes in this challenging condition.

Figure 1. Different Groups of Biological Features

(A) The Spearman correlation matrix identifies 3 distinct groups of biological features. (B) A complete list of the biological features in group 1 (green), group 2 (red), and group 3 (blue) is provided. The graph illustrates the correlation between the various markers, with bars representing the contribution of each marker in the principal component analysis (PCA), which was performed to condense all features within a group into a single composite feature.

Figure 2. Different Clusters of Patients With c-NORSE

(A) The connectivity graph was extracted from the patients' connectivity matrix, and the 8 subclusters of patients with c-NORSE, identified using the Louvain algorithm, were condensed into 4 clusters of patients. (B) The diagram illustrates the 8 subclusters of patients in the center, followed by the 4 clusters, and finally, the levels of all groups of biological markers (1, 2, and 3), represented by color bars (with reddish colors indicating higher levels). Connections between patients are shown in the center of the circle, where black links represent inter-subcluster connections and colored links indicate intra-subcluster connections. c-NORSE = cryptogenic new-onset refractory status epilepticus.

Figure 3. Network Analyses of Protein Interaction and Implicated Biological and Pathologic Processes Involved for the Clusters of Patients With c-NORSE

The Venn diagram highlights the pathways shared by patients in clusters A, B, and C, as well as those unique to specific clusters. A protein pathway analysis was performed, focusing on inflammation-related pathways specific to each cluster. None of the pathways overexpressed in cluster A were related to inflammation. For clusters B and C, cellular processes are represented by yellow rectangles (e.g., leukocyte migration), disease states by blue rectangles (e.g., autoimmunity), and functional pathways by orange hexagons (e.g., JAK). Markers are illustrated in either intracellular or extracellular spaces, based on their expression location. Relationships are shown as lines and arrows. c-NORSE = cryptogenic new-onset refractory status epilepticus.

Figure 4. Distribution of Patients With a RSE of Known Etiology

4A. Patients with a RSE of known etiology were mapped onto the connectivity graph. 4B. These patients are represented in light colors, with various etiologies indicated by icons, as shown in the legend box. The diagram illustrates the homogeneity of biological profiles within the different etiology groups. RSE = refractory status epilepticus.