Serum HMGB1 in febrile seizures

The role of high-mobility group box 1 (HMGB1) in the pathogenesis of febrile seizures (FSs) is unclear. In our controlled follow-up study, we compared serum levels of HMGB1 (s-HMGB1) in the same individuals after the first FS, during febrile episodes without a FS, after recurrent FS, during healthy periods after FS, and between patients and controls. In all, 122 patients with FSs were included in the final analysis, including 18 with recurrent FSs with a complete follow-up protocol. We recruited 30 febrile children and 18 matched febrile children without seizures as controls. S-HMGB1 was lower in patients with recurrent FSs after the first FS than that in matched febrile control children (median 1.12 μ g/L (0.14 – 2.95) vs 1.79 μ g/L (0.33 – 47.90), P < 0.04). We did not find any other differences in s-HMGB1 between the groups. S-HMGB1 did not differ in different types of FSs. We updated a meta-analysis of s-HMGB1 in patients with FSs and found that the differences were significant only in the studies conducted in East Asian populations. We conclude that S-HMGB1 does not seem to be a key factor in the pathogenesis of FSs but differences in HMGB1 concentrations could explain some of the ethnicity related susceptibility to FSs.


Introduction
Febrile seizures (FSs) are seizures associated with fever without evidence of intracranial infection or other defined reason.FSs occur in 2-5 % of children 6 months to 6 years of age (Nelson and Ellenberg, 1978).Recurrences occur in 20-30 % of these patients in subsequent febrile episodes.The exact mechanisms remain unknown.Virus-induced excessive production of proinflammatory cytokines has been suggested to be one of the key factors in the pathogenesis of FSs, as it causes hyperexcitability of neurons which can lead to seizures.
High mobility group box 1 (HMGB1) is one of the cytokines known to connect inflammatory response and neuroendocrine systems (Andrassy et al., 2008).HMGB1 triggers inflammation by activating the inflammatory cascade via Receptor for advanced glycation end products (RAGE) and Toll-like receptor (TLR), which promote the production of cytokines such as interleukin (IL)-1β (Kang et al., 2014, Paudel et al., 2018).Serum HMGB1 seems to be more stable and less affected by preanalytical conditions than most cytokines, which increases its appeal as a biomarker (Ottestad et al., 2019, Lehner et al., 2012).In experimental studies, HMGB1 inhibitors and antibodies have been used to reduce viral replication and inflammatory tissue injury, and blocking of the HMGB1 pathway has shown antiepileptic effects, making it a potential target for the prevention of inflammation-induced diseases such as FS (Ding et al., 2021, Maroso et al., 2010).Elevated levels of HMGB1 have been detected in the cerebro-spinal fluid (CSF) and serum of patients immediately after seizure (Choi et al., 2011, Kaya et al., 2021, Mahmoud et al., 2018).Differences in serum HMGB1 levels in different types and durations of FSs have also been demonstrated (Li and Pan, 2021;Li et al., 2020).
Previously, we have shown that in patients with FSs, the IL-1-axis is excessively activated only in febrile episodes leading to seizures (Hautala et al., 2023).HMGB1 might be responsible for this activation.
Considering this, we hypothesized that serum HMGB1 levels could differ in patients with FSs in different situations and compared to febrile children without seizures, and that serum HMGB1 level could correlate with IL-1RA level, a marker of IL-1 axis activation.In addition, we wanted to see whether HMGB1 levels are different in simple versus complex FSs and in single versus recurrent FSs and whether HMGB1 levels depend on the duration of the seizure.

Study population and study design
This was a controlled follow-up study.All patients visiting the hospital's paediatric ER with their first FS were invited to join the study.We included patients aged between 6 months and 6 years.The exclusion criteria for patients with FSs were a history of previous afebrile seizures or epilepsy, any syndrome or cerebral malformation that increased the risk of epilepsy, an evident central nervous system infection, a significant electrolyte imbalance, a body temperature <38 • C, or psoriasis.We only included patients with at least one native Finnish-speaking parent to avoid communication problems.The study clinician/nurse contacted the parents of each patient after the first FS.The study protocol included one visit for the first FS and a second for a febrile infection without FSs.If there were no recurrent seizures, the follow-up ended.If the patient had recurrent FSs, a third visit was scheduled for a recurrent FS, and a fourth while the patient was healthy (Fig. 1).The order of the revisits varied because some patients had more FS episodes before having a febrile episode without a FS.Samples taken while the patient with FSs was healthy were programmed to be collected 3 weeks after the recurrent FS.In some cases, collection was postponed due to recurrent infections.Altogether, 122 patients with FSs were included in the final analysis, including 18 patients with recurrent FSs with a complete follow-up protocol.
We recruited both febrile control children and matched febrile control children, matched with the index patient by age (±6 months) and sex.We recruited febrile control children in 1:4 ratio, i.e. one febrile control child after every fourth patient with their first FS.For the controlled follow-up study, we wanted to rule out the effect of age and sex on the cytokine production by selecting one matched febrile control child for each patient with a recurrent FSs.We recruited the controls from hospital patients with febrile infections with body temperatures >38 • C. The exclusion criteria for all the controls were a history of FSs and otherwise the same as for patients with FSs.
We collected a blood sample of 5 ml from a peripheral vein from each patient for HMGB1 measurements simultaneously when blood samples were taken for clinical purposes, usually to measure C-reactive protein (CRP) level, at each study visit.For control children, we obtained the study samples when samples were taken for clinical use for diagnostics or follow-ups.The blood samples were centrifugated within 1 hour of collection, and the sera were stored at − 80 • C until the analysis.For the FS-associated visits, we obtained the samples within 24 hours postseizure.
Patient data were systematically collected from medical records and through interviews with the parents using a structured questionnaire.We contacted all the parents in January-March 2022 in order to find possible seizure recurrences left unreported by the families.For each patient, the collected data included demographic information, family history of FSs, and clinical details concerning the current febrile episode and the FS episode (Table 1).Complex FSs were defined as focal seizures, seizures lasting more than 15 minutes, or recurrent seizures during a 24-hour period.
The Ethics Committee (Institutional Review Board) of Northern Ostrobothnia Hospital District, Finland, found the study plan acceptable (24.01.2011, diary number 4/2011).Informed consent was obtained from the guardians of the participating children.The study was conducted according to the guidelines of the Declaration of Helsinki.The study was conducted in Oulu University Hospital, Finland from January 2012 to May 2019.

HMGB1 measurement
HMGB1 concentrations were measured using a commercially available HMGB1 ELISA kit E-ELH1554 according to the manufacturer's instructions (eLabscience, Houston, Texas).

Outcomes
The serum HMGB1 levels of patients with FSs were compared after the first FS and during febrile infection without FS (Fig. 1).For patients with recurrent FSs, we also compared HMGB1 levels after a recurrent FS, and while the patient was healthy after a FS.In these comparisons, patients with FSs served as their own controls.All patients with FSs were compared to febrile control children, and we also made a separate comparison of the different visits of patients with recurrent FSs and a complete follow-up protocol to matched febrile control children.
We studied the correlation of serum HMGB1 levels in patients with recurrent FSs and a complete follow-up protocol to the serum pro-and anti-inflammatory cytokine levels available from the previous study as well as the correlation of serum HMGB1 in all patients with FSs after the first FS to the duration of the seizure, CRP level, and leukocyte level (Hautala et al., 2023).
We compared the HMGB1 levels in patients with different types of FSs and in patients with clinical diagnoses of either bacterial or nonbacterial infections.We used the values of all patients with FSs after the first FS in the comparisons.
The recent meta-analysis regarding serum HMGB1 and FSs was updated by adding our data to this meta-analysis to see if our results changed the outcome (Li et al., 2023).

Statistical analyses
The study protocol and the sample size calculations are described in our previous study (Hautala et al., 2023).Our calculations resulted in a sample size of at least 16 patients for each group.
The differences in serum HMGB1 between the different visits of the patients with FSs as well as those between patients with FSs and febrile control children were tested a priori using the Kruskal-Wallis test with the Mann-Whitney U test as a post-hoc test.The differences in serum HMGB1 between the different visits of the patients with recurrent FSs as well as those between patients with FSs and matched febrile control children were tested a priori using the Friedman test with the Wilcoxon signed-rank test as a post-hoc test.The HMGB1 levels are given as means and SDs as well as medians and minimum and maximum values.We used Spearman's rank correlation to study the correlation of serum HMGB1 and other cytokines as well as seizure duration, leukocyte levels, and CRP levels.
To compare our results to those of a current meta-analysis regarding HMGB1 and FSs, we combined our data with the data of the metaanalysis (Li et al., 2023).We used a random effects model because of the notable heterogeneity of the results.We extracted the necessary data of the study by Choi J et al. from a figure in the article and the data of the study by Issac M et al. from another recent meta-analysis (Chen et al., 2023, Choi et al., 2019, Issac et al., 2015).The thesis paper by Zhou J. was published locally and was therefore inaccessible for us; thus, it had to be disregarded (Zhou, 2013).
All analyses were performed using IBM Statistics for Windows version 27 (IBM Corp., Armonk, New York) and StatsDirect statistical software version 3 (StatsDirect Ltd., Birkenhead, UK).Figures were drawn using GraphPad Prism version 9 (Graphpad, San Diego, California) and SmartDraw version 2008 (SmartDraw Software, LLC, The Woodlands, Texas).

Results
In total, 599 patients with FSs visited the paediatric ER during the study period.Of them, 306 patients were invited to join the study, and 250 accepted the invitation.In all, 122 of them were included in the final analysis (Fig. 1, Table 1), including 18 patients with recurrent FSs that completed the follow-up (Fig. 1, Table 1); 30 febrile children served as controls for all patients with FSs and 18 matched febrile children for the patients with recurrent FSs.
There were no statistically significant differences in serum HMGB1 levels following first FSs and during a febrile period without seizures.There were no statistically significant differences in serum HMGB1 levels between these two visits of patients with FSs and febrile control children (Fig. 2).
The serum HMGB1 levels were significantly lower following the first FS in patients with recurrent FSs compared with the levels in matched febrile control children (Fig. 3).We did not find any other statistically significant differences in the serum HMGB1 levels between the different visits of patients with recurrent FSs, i.e. following first FS, during a febrile period without seizure, following a recurrent FS, or during a healthy period after FS (Fig. 3).
We did not find any correlations of serum HMGB1 to IL-1RA in  9) 11 ( 9) 3 ( 17) 3 ( 17) 3 ( 10) patients with recurrent FSs.Serum HMGB1 and anti-inflammatory IL-10 showed a positive correlation during a febrile episode without FSs, and serum HMGB1 and proinflammatory interferon-γ showed a positive correlation after recurrent FSs (Table 2).During healthy periods, there were positive correlations of serum HMGB1 to proinflammatory IL-1β, IL-6, IL12-p70, interferon-γ, tumour necrosis factor-α, as well as tumour necrosis factor-β.There were also positive correlations during healthy periods to anti-inflammatory IL-4 and IL-10 (Table 2).The levels of HMGB1 after the first FS in all patients with FSs did not correlate with leukocyte counts or CRP.We did not find any statistically significant differences regarding serum HMGB1 in different types of FSs.We used the serum HMGB1 concentrations measured after the first FS of all patients with FSs in the comparisons.We compared patients with simple FSs to patients with complex FSs, patients known to have had a single FS episode to patients with recurrent FSs in the future, patients with a seizure duration of under 15 min to patients with a seizure duration of over 15 min, patients with or without a family history of FSs, and finally, patients with or without a clinical diagnosis of bacterial infection.
We updated the recent meta-analysis by Li et al. (2023).The applied meta-analysis comprised six studies including 790 patients and 292 controls, and it produced a pooled effect size of 0.897 (95 % CI, 0.283-1.511,P<0.01) (Fig. 4).Patients with FSs had a higher serum HMGB1 compared to febrile control children without seizures, but there were two studies with contrary results (Fig. 4).After the addition of our study with contradictive results, the asymmetry of the funnel plot suggests that two of the studies differed from the others (Fig. 5).

Discussion
In our controlled follow-up study, patients with recurrent FSs had significantly lower serum HMGB1 levels during the first FS episode compared to those of matched febrile control children.This is contradictory to earlier findings Li et al. (2023).We did not find any other significant differences in serum HMGB1 between the different visits of patients with FSs or compared to febrile control children.Neither did we find correlations of serum HMGB1 levels to IL-1RA in patients with recurrent FSs.In our previous study, patients with FSs had higher levels of IL-1RA following the first FS than during episodes without FSs as well as following the first and recurrent FSs compared to those of matched febrile control children (Hautala et al., 2023).In our updated meta-analysis, the differences in serum HMGB1 levels were significant only in those studies done in East Asian populations.The two studies done in non-East Asian (i.e.present and Egyptian) populations did not reveal significant differences in the levels of serum HMGB1 (Issac et al., 2015).
FSs are considered to be a provoked temporary malfunction of developing normal neural networks whereas epileptogenesis is a gradual process of abnormally functioning neural networks.Cytokine-induced increased excitability of neurons is considered to be important in both of them.Signs of activation of the HMGB1-related pathway have been repeatedly found from brain biopsies and serum samples of patients with M.K. Hautala et al. epilepsy, as well as in experimental seizure models in animals (Zurolo et al., 2011, Walker et al., 2014).Maroso M. et al. (2010) showed that HMGB1 levels increased in mice with kainic acid-induced acute and chronic seizures and that antagonists of HMGB1 and its receptor TLR4 prolonged seizure precipitation and decreased acute and chronic seizure recurrence.In our study, the patients with recurrent FSs had lower levels of serum HMGB1 compared to the matched febrile control children, suggesting that HMGB1 is not a determining factor in the pathomechanism of FSs.This does not rule out a modifying role of HMGB1.FS can be an end-result of various disturbances in cytokine network.Also, we think that our finding adds on the evidence of the different pathomechanisms of epilepsy and FSs.Our study populations filled the sample size calculations decreasing the risk of an incidental finding.
In terms of availability and ethicalness, we decided to measure HMGB1 only from serum samples.However, cytokines produced both peripherally and in the central nervous system (CNS) are known to directly modify neuronal hyperexcitability.In addition to that, peripheral cytokines are thought to stimulate the nervous system via the vagal nerve (Han and Han, 2023;Goehler et al., 2000).This means that studies considering serum cytokine levels explain only part of this complex chain of events.Some studies have described correlations of HMGB1 with other cytokines in patients with FSs.HMGB1 is thought to act by launching a Fig. 3. Serum HMGB1 concentrations in with recurrent FSs during the first FS, during febrile episodes without FSs, during recurrent FSs, and during a healthy period after FSs and in matched febrile control children.M.K. Hautala et al. cascade of other cytokines, mainly from IL-1 axis.In our previous study, we found signs of IL-1 axis activation in patients with FSs only in febrile episodes leading to FSs.In case of a positive correlation, HMGB1 might be responsible for the activation of the cytokine in question.However, we did not find any clinically significant correlations of HMGB1 to other cytokines that would have indicated a connection.FSs are a condition affecting children globally at a certain developmental stage of the CNS and immune system.FSs are prevalent in all populations, but the incidence varies according to ethnicity.The highest incidences are found in Asian and Pacific Island countries (Okubo and Handa, 2017).In a study done in the United Kingdom, Hamdy et al. (2007) detected a 2.3-times higher incidence of FSs in South Asians compared to an indigenous British population.According to the meta-analysis by Li et al. (2023), HMGB1 seems to have some role in the pathogenesis of FSs.However, the addition of our results to this meta-analysis shows that the significance of HMGB1 seems to be dependent on the ethnicity of the individual, as there was significance in the East Asian studies but not in the present study in a Finnish population or in the other in an Egyptian population.Considering this ethnicity-related difference and the universal nature of FSs, we think that serum HMGB1 cannot be a key factor in the pathomechanism of FSs.However, differences in HMGB1 concentrations could explain some of the ethnicity-related susceptibility to FSs.
There have been implications that HMGB1 might be related to the generalization and recurrence of seizures.This has been modelled in experimental animal studies of induced seizures (Maroso et al., 2010).In clinical studies, the highest levels of HMGB1 have been found in patients with drug-resistant epilepsy (Walker et al., 2014).A Turkish study by Kaya et al. (2021) found that patients with complex FSs had the highest levels of CSF HMGB1.They also found a correlation between seizure duration and CSF HMGB1.Two studies of serum HMGB1 done in East Asian populations showed similar results (Li and Pan, 2021;Li et al., 2020).On the other hand, Issac et al. (2015) did not find differences between different types of FSs in an Egyptian population.Our cohort did not exhibit a significant difference in serum HMGB1 levels in patients with simple versus complex FSs.The serum levels of HMGB1 during the first FS episode were similar in patients that had only one FS episode during the follow-up period compared to patients that had recurrent FSs later.Finally, we did not find a correlation of serum HMGB1 with seizure duration.HMGB1 does not seem to affect seizure recurrence, type, or duration in Caucasian patients.Li and Pan (2021) and Li et al. (2020) found that HMGB1 levels correlated with leukocyte counts and CRP in patients with FSs.Interestingly, we did not find correlations between HMGB1 levels and leukocyte counts, CRP levels, nor the clinical diagnosis of bacterial infection.
In our study, HMGB1 sampling in the controls took place later from the fever rise than in the patients with FSs.The clinical need for blood sampling in infectious patients usually happens later, after fever rise, and we avoided taking samples only for scientific purposes for ethical reasons.In most previous studies, an exact sampling time or even a time limit for the sampling from the fever rise was not reported, so a profound comparison between studies was not possible, but as HMGB1 appears later and sustains longer than most proinflammatory cytokines, sampling time does not seem to be a critical factor affecting the concentrations of HMGB1.In an experimental study, HMGB1 started its rise at 8 hours after the initiation of inflammation and continued to increase up to 16-32 hours (Wang et al., 1999).In a study with adult epilepsy patients, serum HMGB1 levels rose immediately and were still significantly increased 6 hours after seizure (Nass et al., 2020).In our study, the samples from patients with FSs were taken at the presumed peak time of HMGB1 whereas samples from the control children were taken well after peak time.Yet the serum HMGB1 levels were lower in patients with FSs.We think it is unlikely that sampling time had an impact on the values detected.
The HMGB1 levels detected in our study were generally low.Looking at the previous studies, the use of similar ELISA kits seems to have reproduced results of the same magnitude.Considering the wellrecognized problems of wide variance and low comparability of cytokine measurements in separate studies, we believe that we must concentrate on the differences between study populations instead of direct comparisons of HMGB1 levels.
Our study represents a sample of unselected Caucasian patients with FSs.Our versatile study protocol with a follow-up period after first FSs allowed examination of HMGB1 levels in different situations in the same individuals, comparisons of different types of FSs, as well as the exclusion of patients with epilepsy.The addition of our results to the previous meta-analysis by Li et al. (2023) indicates differences in the role of HMGB1 in FSs according to the ethnicity of the patient population.
We conclude that according to our results, the only difference in the serum levels of HMGB1 was that the patients with recurrent FSs actually had lower levels of HMGB1 after the first FSs compared to matched febrile control children.We did not find a correlation of HMGB1 with IL-1RA.We did not detect significant differences in the serum HMGB1 levels in different types of FSs.Adding our results to those of a recent meta-analysis suggests that ethnicity may explain the published controversial results of HMGB1 differences in patients with FSs.Serum HMGB1 does not seem to be a key factor in the pathogenesis of FSs.

Fig. 1 .
Fig. 1.Flow chart of the inclusion and statistical comparisons of patients and controls in the study.

Fig. 2 .
Fig. 2. Serum concentrations in patients with FSs during the first FS, during febrile episodes without FSs, and in febrile control children.

Fig. 4 .
Fig. 4. Forest plot of the difference in serum HMGB1 concentrations between patients with FSs and febrile control children after the addition of our results to the meta-analysis (Li S et al. 2023).

Fig. 5 .
Fig.5.Funnel plot of the difference of serum HMGB1 concentrations between patients with FSs and febrile control children after the addition of our results to the meta-analysis(Li S et al. 2023).

Table 1
Clinical characteristics of patients with FSs after the first FSs, during febrile episodes without FSs, after recurrent FSs, and during a healthy period after FSs and of febrile control children.

Table 2
Correlations of serum HMGB1 concentrations to other cytokine levels in patients with recurrent FSs during the first FSs, during febrile episodes without FSs, during recurrent FSs, and during a healthy period after FSs and in matched febrile control children.