Inflammatory Biomarkers in Febrile Seizure: A Comprehensive Bibliometric, Review and Visualization Analysis

Background: Inflammatory markers association with many diseases is the subject of many articles and reviews. This study presents a comprehensive bibliometric review and visualization analysis of inflammatory biomarkers (IB) in the context of febrile seizure (FS) patients. Methods: The study focused on IB in FS using (1) bibliometric analysis specific indicators and maps in order to analyze and present the network of authors, journals, universities, and countries, and (2) automated literature screening and unsupervised clustering approach for filtering and topic cluster identification. Results: We conducted a literature mining search on FS research, specifically IB in the context of FS, using the automated tools VOSviewer and Bibliometrix. Indices of the inflammatory response (in the context of febrile seizures) identified by the literature mining are (pro/anti-inflammatory) cytokines, such as interleukin IL-1β, IL-6, IL-8, I-10, IL-22, tumor necrosis factor (TNF-α), neutrophil-to-lymphocyte ratio (NLR), mean platelet volume (MPV), platelet count (PLT), and red blood cell distribution width (RDW). The resulted bibliometric maps and topic clusters offer a comprehensive overview, the status and leading trends on existing research of inflammatory biomarkers in FS. Conclusion: The analysis using bibliometrics and review with graphical presentations can be useful in investigating and (better) understanding the relationship between FS and IB.


Introduction
Febrile seizures triggered by infections can occur in children up to 5 years of age. Febrile seizures' pathogenesis is based on cytokines release (pro-and anti-inflammatory) and genetic susceptibility to enhanced inflammation consisting of genetic variants of the pro-inflammatory and anti-inflammatory cytokines [1,2]. The link between inflammation and febrile seizures has been studied for more than 20 years [2]. The most relevant prospective case-control studies highlight the role of elevated levels of pro-inflammatory cytokines, triggering a rapidly rising fever, neuronal hyperexcitability and eventually the seizure event [2]. Among pro-inflammatory cytokines (released by activated microglia in the central nervous system or by monocytes, macrophages or T-lymphocytes in plasma) interleukin 1-beta (IL-1β), IL-6, and tumor necrosis factor (TNF)-α were found to be significantly elevated in children with febrile seizures [2]. The anti-inflammatory cytokines, such as IL-1 receptor antagonist (IL-1RA) and IL-10 along with anti-inflammatory cholinergic signals from the efferent vagus nerve, provide a negative feedback on the inflammation. Nevertheless, in susceptible patients, it seems that the negative feedback control is lost, and the elevated systemic levels of pro-inflammatory cytokine elicit the seizure event [1,2]. In clinical trials involving children with febrile seizures, it is expensive and not always available to validate specific patterns of the relevant pro-inflammatory cytokines. However, recent research reports highlight the importance of other inflammatory low-cost biomarkers, such as the neutrophil-to-lymphocyte ratio (NLR), platelet count (PLT) ratio, mean platelet volume (MPV) and platelet count, and red blood cell distribution width (RDW). It seems that NLR and MPV might synergistically determine the FS occurrence [2].
The cumulative number of research documents in many research fields, in general, and particularly in the febrile seizures (FS) field, is continuously increasing. Automated, interactive, flexible tools could be used by researchers to perform a systematic literature review and bibliometric analysis. Bibliometric analysis offers a quantitative and qualitative analysis of the analyzed publications. It is used in medical research but also in others research areas [3][4][5][6][7][8].
The current research aimed to perform a literature mining analysis on inflammatory biomarkers (IB) in the context of febrile seizure (FS). By analyzing the published documents and their citation and co-citation data, the current research themes presented in inflammatory biomarkers in FS research were identified. The specific objectives of this analysis were the following: (i) to assess the range of research topics; (ii) to identify which are the inflammatory biomarkers associated with febrile seizure discussed in the published literature; (iii) to identify who has driven this research; (iv) to assess what we can learn from this research; and (v) to guide investigators potential research directions and potential collaboration partners.

Search Methodology
In October 2020, we conducted a literature search on the Web of Science Core Collection (WoS) online databases to identify scientific contributions regarding febrile seizures and their association with inflammatory biomarkers. The search strategy included the terms "febrile seizures" alone and in combination with terms "marker" or "biomarker" and "inflamm*". The asterisk was used to retrieve related/derivative words (inflammation, inflammatory, etc.). The search identified publications that contain the mentioned terms in their title or abstract or keywords. RC and BMN review the pool of documents selected by the automated review to reach a consensus on the inclusion of the topic-relevant ones. Discrepancies were sorted out with discussion. Papers that contained some of the search keywords but for which the major focus of the document was not related to inflammatory biomarkers in febrile seizure were excluded: studies on seizures with fever onset in epilepsy context; acute symptomatic seizures defined by (i) CNS (central nervous systems) infections (meningitis, encephalitis) and post infectious autoimmune encephalitis, (ii) dyselectrolytemia, (iii) TBI (traumatic brain injury), and (iv) others-sepsis, acute intoxications, seizures not respecting age criteria, FIRES (febrile infection-related epilepsy syndrome), irrelevant studies (psychiatric diseases, multiple sclerosis, stroke, TORCH infection, typhus, celiac disease, asthma, others) ( Figure 1). No other restrictions/filters were considered for the research model (people/animal, article/review/letter/, English/French, etc.).

Data Extraction
Characteristics of each publication identified from the search include, besides the publication title, abstract and keyword, the following: authorship, document type, publication year, journal title, language, journal category, number of total citations. Data were extracted from WoS databases and exported as "tab-delimited text file", "excel file", "plain text" for additional processing.
The VOSviewer (visualizations of similarities, van Eck and Waltman, Leiden University, Leiden, Netherlands [9]) software was used to create network visualization maps (visualize and map bibliometric indicators). We analyzed the list of retrieved documents and bibliometric indicators for which ranking countries, institutions, journals, authors, keywords were presented [9][10][11][12][13][14]. We conducted term/items co-occurrence (maps) analysis based on the text data from the title and/or abstract and/or keyword of the publications, using the binary counting method (binary counting method vs. full counting method, the presence or absence of a term in a document matter vs. number of occurrences of a term in a document matter). In the term co-occurrence map, terms are represented by bubbles/circles, and some of them are labeled (to avoid overlapping). The number of publications in which the term was found is represented by the size of the circle. The relatedness of the terms is represented by the distance between the terms (the closer two terms are located to each other, the smaller the distance between the two terms; the larger the number of co-occurrences of the two terms-terms co-appear often, the stronger their relatedness) and curved lines between the terms (the larger the number of publications in which two terms were both found-co-occurrence, the stronger the relation/link between the terms and the thicker the line that links the two terms). Groups of terms which are strongly related to each other are represented using the same color (red, blue, green, etc.). When using the VOSviewer instrument, the following thresholds were set: a minimum of 5 occurrences for terms encountered in author keywords, all terms occurrences for terms encountered in the title. The historiographical citation network, co-citations network statistics, authors' production over time, three-field plots were created using the Bibliometrix 3.1 package (Aria and Cuccurullo, University of Naples and University of Campania's Luigi Vanvitelli, Italy, [15]) and RStudio 1.4.1717 environment (CRAN, https://cran.r-project.org/).

Results
The literature search using VOSViewer followed by document screening resulted in 71 publications indexed in WoS, the earliest published in 1996. Contributions in the field of febrile seizures and their association with inflammatory biomarkers came from 114 institutions located in 18 countries/regions. The most common type of retrieved documents were research articles (58; 82%) followed by reviews papers (8; 11%). The analysis of research areas of the retrieved documents showed that 53 articles were in neuroscience and neurology, 21 in pediatrics, 7 in immunology, 3 in pharmacology and pharmacy and in psychiatry, while 2 were in areas such as science and technology, internal medicine, allergy and others.
Using Bibliometrix, a historiograph of publications is displayed in Figure 8, and network statistics (in terms of centrality, cohesion, impact) are presented in Table 4. The centrality measure-betweenness centrality (based on leading eigenvalues clustering method)-highlighted as main "influencers" (have high betweenness centrality) the documents written by Dube [36], Virta [37,44], and Haspolat [24]. These publications are core/pivotal nodes that make connections to other publications within the network. The historiograph from Figure 8 provides an overview of the trends and evolution of the field by the flow of links between the cited publications (from the left, the older one) and the citing publications (from the right, the recent one).    Inflammation in rat pups subjected to short hyperthermic seizures enhances brain long-term excitability Epilepsy Research 31 2 When analyzing the relationship between three publications meta-data-countries/institutions, keywords, references ( Figure 9)-it is possible to observe that researchers from particular countries/institutions used/analyzed particular keywords/biomarkers and considered particular reference sources (citations traceability/patterns in relation with terms/biomarkers).

Discussion and Conclusions
Although pathogenic mechanisms of febrile seizures remain unclear, experimental studies demonstrate that inflammation and inflammatory mediators are the main causes and propagators of febrile seizures [36]. New trends are targeting cytokines as more sensitive, yet more expensive biomarkers, in exploring febrile seizures, as evidenced by our bibliometric tool search.
In this study, we used the bibliometric methodology to perform a literature-driven analysis in order to identify research directions and trends in our study context. U.S.A. appeared in the top positions in citations, bibliographic coupling and co-citation analysis, indicating the impact and high-ranking level of this country. The country's bibliographic coupling analysis, reflecting how they are connected in terms of the bibliographic common literature, indicate two clusters: the first cluster includes seven countries (colored with green): U.S.A., Japan, China, Italy, Canada, France and Australia. The second cluster includes also seven countries (colored with red): Turkey, Iran, South Korea, England, Egypt, Germany and India. Epilepsia, the most productive and co-cited journal, was the influential journal in the field. Besides this journal, Pediatric Neurology was also among the top productive and co-cited journals.
The most frequently encountered anti-inflammatory cytokines in the studied papers were IL-1RA, IL-10 and IFN-β. IL-1RA [18] counteracts IL-1 β, IL-10 inhibits TNF-α and IFN-γ, while IFN-β is a marker of Toll-like receptor-3 activation, suggesting a host response to viruses. It modulates the expression of both pro-and anti-inflammatory agents in the brain, reducing the blood-brain barrier permeability to inflammatory cells. It is acknowledged in the current literature that therapy with IFN-β alleviates neuroinflammation. In febrile seizures however, it did not show a statistical significance compared to the control group, according to the reports of Sahin S. et al. [54]. The genetic variants for anti-inflammatory cytokines mentioned in the literature are mainly related to IL-1RA [31] and IL-10 [32].
The bibliometric analysis provided little evidence of studies related to common inflammatory biomarkers sensitivity and specificity in febrile seizures (e.g., total leukocytes count, neutrophils, lymphocytes and monocytes count). However, inflammatory indices, such as NLR, PLT, the NLT/PLT ratio, MPV, and RDW, seem to be useful biomarkers in this respect. These indices are easier and less expensive to evaluate in clinical practice than cytokines [38][39][40]. Recent research papers explore this possibility. It seems that NLR and MPV might synergistically determine the FS occurrence [2]. Moreover, NLR and MPV might differentiate simple versus complex febrile seizures [41,42].
Considering the network visualization of terms and network visualization of publications, we can highlight the fact that, in relation to inflammatory biomarkers in febrile seizures, there are two clusters of publications: the first one includes publications that address the relationship between cytokines/interleukins (expensive, not always available biomarkers) and febrile seizure, while the second one includes publications which address the relationship between biomarkers from blood (NLR, MPV, PLT, RDW-low-cost, available biomarkers) and febrile seizures. Among the most cited publications in the researched field are those referring to cytokines [31,32,54], and only a small number of publications have attempted to address the relationship between the aforementioned indices from blood and febrile seizures [41,42]. Moreover, there are studies in the literature that show IL's action on the liver, thus leading to increased C-reactive protein [56]. Due to this fact, the testing of this anti-inflammatory (available) marker could be useful in febrile seizures. In addition, the synthesis of the abovementioned interleukin is triggered by certain biological stimuli, represented by the endotoxins of Gram-negative bacteria, often involved in urinary tract infections [57].
The analysis of publications, keywords, and references offers an overview of the trends and popular topics on the state of febrile seizure and inflammatory biomarkers. Using bibliometric analysis could guide researchers toward journals and authors associated with the field of their research interest. Furthermore, it could provide clues and facilitate links to other research centers with the same field of interest, as we identified five main institutions (Tehran Univ. of Medical Sciences, Mario Negri Institute for Pharmacological Research, University of California, Irvine (UCI), Ehime Univ., and University of Social Welfare and Rehabilitation Sciences) with activity related to inflammation and febrile seizures. Among the limitations of the current research, we mention the fact that we analyzed a sample of related publications (not claiming to cover all related literature), only from the WoS database. The results are based on a keyword (from title, abstract, keyword plus) analysis, and not on the publications full-text analysis. Researchers should also consider combining different bibliometric, review and meta-analysis tools [58][59][60][61][62][63][64][65] for their literature mining research. Going deeply with bibliometric analysis, considering network and cluster metrics, multivariate methods, comparisons between networks generated using different keywords, developing content analysis, meta-analysis, etc., could offer better insights in understanding the research topics.
Author Contributions: I.M. and G.M. conceived the work, performed data collection and analysis, and drafted the manuscript. R.C. and B.M.N. participated in data preprocessing, analysis and drafting of the manuscript. All authors critically revised the manuscript and approved the submission of the manuscript. All authors have read and agreed to the published version of the manuscript.