The Clinical Outcomes of Tocilizumab Usage in Cryptogenic Febrile Infection-Related Epilepsy Syndrome at Both Acute and Chronic Phase: A Retrospective Study

Introduction

Febrile infection-related epilepsy syndrome (FIRES), according to the consensus definition, is regarded as a subcategory of new-onset refractory status epilepticus (NORSE) rather than as an independent entity, as previously proposed.¹ Its management remains highly challenging, characterized by a high mortality rate during the acute phase, with approximately one-third of patients succumbing to uncontrollable refractory status epilepticus.² Survivors often experience long-term neurological sequelae, including persistent cognitive deficits and drug-resistant epilepsy. The etiology of FIRES remains elusive; however, emerging research indicates a potential link to immune-mediated encephalitis.³ Inflammatory cascades involving microglia activation, T lymphocytes, and autoantibodies are implicated, particularly with intrathecal overproduction of pro-epileptogenic cytokines such as IL-6 and IL-1.^4–6 Innate immunity indicators such as pro-inflammatory cytokines in both serum and cerebrospinal fluid (CSF) and neutrophil-to-lymphocyte ratios positively correlated with worse outcome in patients.^7,8

Anti-inflammatory immunotherapy has been a research hotspot in recent years. First-line treatments include intravenous steroids and immunoglobulins (IVIG), while second-line options comprise agents such as anakinra and tocilizumab (TCZ).⁹ TCZ (a humanized monoclonal antibody against the IL-6 receptor) was reported effective in stopping status epilepticus (SE) or reducing seizure frequency for 70% of patients in the acute phase.¹⁰ Despite ongoing interest in TCZ for pediatric FIRES, clinical data remain limited, particularly regarding its utility in the chronic phase.¹¹ Current international guidelines recommend initiating TCZ within 7 days after onset.^12,13 However, as a tertiary provincial referral center, our hospital receives most patients from county and municipal hospitals, resulting in typical TCZ initiation beyond the 7-day window. Nevertheless, some of these late-treated patients also benefited from TCZ. In this case series, we summarized and presented long-term follow-up data from pediatric FIRES patients treated with TCZ in both acute and chronic phases, providing more observational evidence for its potential efficacy even when administered beyond the recommended early window.

Materials and Methods

Participants

This study was a retrospective analysis of patients diagnosed with FIRES. We reviewed medical records of pediatric patients (aged from 1 to 18 years old) who were diagnosed with FIRES in the Pediatric Department of Xiangya Hospital, Central South University. Twenty-two FIRES patients treated with TCZ from 2022 to 2025 were enrolled. All participants had a clinical diagnosis of FIRES by more than 2 pediatric board-certified neurologists, based on the proposed consensus definitions. Due to lack rigorous standard for acute phrase and chronic phrase, therefore, the acute phase was defined as the period within 4 weeks after seizure onset, based on a prior retrospective study. And chronic phrase was defined as the period following the acute phrase.¹⁴ This study design was approved by Xiangya Hospital Ethics Committee in accordance with the Helsinki declaration (No. 202307148-2). Informed consent for participant and publication was obtained from the parents or guardians of each subject (individuals included in table).

Data Collection

Demographics, clinical and laboratory parameters, cranial magnetic resonance imaging (MRI) images, electroencephalogram (EEG) recordings, genetic tests results, dosage and duration of TCZ, and adverse events, including infections, hypersensitivity reactions, liver damage, neutropenia, and thrombocytopenia, were collected. Specially, cytokines in cerebrospinal fluid (CSF) and serum, including IL-1, IL-6, and TNF-α and IL-10, were documented. Electrodes were placed according to the international 10–20 system. EEG recordings were reviewed by a group consist of 3 board-certified neuroelectrophysiological technicians and 2 pediatric board-certified neurologists. The MRI images were also reviewed by 2 pediatric board-certified neurologists. The course of illness was documented, including the duration of seizures prior to treatment and the timing of TCZ administration relative to seizure onset. The usage of anti-seizure medications (ASMs) was recorded, including the number of ASMs prescribed before and after the initiation of TCZ treatment.

Outcome Evaluation

The modified Rankin Scale (mRS) was used to evaluated the neurological state of each subject. EEG and MRI changes, seizure frequency, number of ASMs, and mRS were used to evaluate the TCZ response.

Neuropsychological domain assessments, seizure frequency, mRS, current immune therapies or ketogenic diet (KD), and VNS usage, and being able to rectum to school were used to evaluate as the long-term outcome. ICU length of stay, hospital length of stay, mechanical ventilation duration, and days of status epilepticus (SE) were assessed as the short-term outcome. The definition of SE was described in our previous study.¹⁵

The assessments were undertaken when patients returned to the outpatient or inpatient unit by pediatric neurologists.

Reporting of Studies Conducted Using Observational Routinely Collected Data (RECORD) Guidelines

This study uses and adherences to RECORD guidelines (https://www.record-statement.org/checklist.php). RECORD checklist was attached as supplementary materials.

Statistics

Descriptive statistics were used to show the demographics, clinical characteristics, and outcomes. The results are reported as number (%) or median [interquartile range]. Wilcoxon signed-rank sum test for paired samples, Mann–Whitney test, and Fisher Exact test were used to estimate differences among comparison groups. All statistical analyses were conducted on SPSS 20. Differences with P < 0.05 were considered statistically significant.

Results

Demographic Characteristics and Etiology

The study included 22 patients with a median age of 9 years, ranging from 2 to 18 years old, consisted of 15 males and 7 females. No encephalitis-related antibodies were detected in the cerebrospinal fluid (CSF) or serum, and bacterial cultures for CSF and serum were negative (Supplementary Table 1). Besides, whole-exome sequence (WES) was conducted on 8 patients, no pathogenic or likely pathogenic variants were found (Table 1). All patients were considered cryptogenic febrile infection-related epilepsy syndrome.

Table 1 Patient Demographics and Clinical Characteristics of 22 FIRES Patients Treated with TCZ

Clinical Characteristics

The median age at onset for these 22 cases was 6 [4, 8] years (Table 1). The observation period was 17 [10, 27] months after the first dose of TCZ. Focal seizures (22/22, 100%) occurred on every case. Six patients had migrant focal seizures. EEG of patent 9 demonstrated focal seizures (Figure 1A and B), which shifted from one hemisphere to the contralateral (Figure 1C and D). Two had focal to bilateral tonic-clonic seizures (FBTCS). During disease course, three patients had tonic, epileptic spasm, and absence seizures and one patient had atypical absence seizures. The majority of initial cerebrospinal fluid (CSF) were normal, except that six patients (6/21, 28.6%) had elevated WBC. Seven patients (7/21, 33.3%) had elevated protein levels and 6 patients (6/21, 28.6%) had abnormal glucose levels. Notably, C-reactive protein (CRP) or erythrocyte sedimentation rate (ESR) was elevated in 15 patients (15/18, 83.3%), and thyroid antibody (A-TG or A-TPO) elevation was noted in 9 cases (9/17, 52.9%). Twelve patients had tested cytokines in CSF, and two patients showed normal level of cytokines. The median time of first lumber puncture to assess cytokines in CSF was 21 [5.5, 171.5] days. Elevated IL-6 and IL-1β were found in 10 patients (10/12, 83.3%) and 5 patients (5/12, 41.7%), respectively. TNF-α were increased in 2 (2/12, 16.7%) cases. Serum cytokines were tested in each subject. IL-6 and TNF-α were the most frequently increased cytokines, found in 17 (17/22, 77.3%) and 18 cases (18/22, 81.8%), separately. IL-1β was above the normal levels in 12 patients (12/22, 54.5%). IL-10 was elevated in only two cases (2/22, 9.1%). There are a total of nine individuals with elevated levels of three or more inflammatory factors. (9/22, 40.9%) (Table 1).

Figure 1 EEG recordings of patient 9. Focal-onset seizures ((A and B) black arrow), shifting from one hemisphere to the contralateral ((C and D) red arrow).

The median time of first assessment of brain MRI in all patients was 5 [3, 9] days. Half of those cases (11/22, 50.0%) had normal brain MRI at the beginning of the course. Cytotoxic cerebral edema was observed on one case. Four cases displayed single lesion and 6 had two or more lesions on the first cranial MRI. One patient was considered febrile infection-related epilepsy syndrome with claustrum lesion (FIRES-C) due to abnormal signals in bilateral claustrum. The median time of last assessment of brain MRI in all patients was 473.5 [275.5, 1010] days. Only two patients had normal cranial MRI at the first time did not undertake MRI examination again during follow-up. Two and more times of MRI were done on the remaining 20 cases. As the progression of disease, 3 cases still showed normal MRI. MRI of 9 cases got worse when compared to the first MRI. After drug intervention, lesions of 13 subjects ameliorated, and lesions completely disappeared in 12 patients. Twelve cases remained brain atrophy and 5 cases left hippocampus atrophy (Table 1).

Treatments in Addition to TCZ

At acute phase, to stop the super-refractory seizures, midazolam infusion was used on each child and three children were on midazolam along with dexmedetomidine (Table 2). Anesthetic agents like fentanyl, propofol, and esketamine were used on 9 cases. Immune therapies included methylprednisolone and IVIG (22/22, 100%), with some patients (8/22, 36.4%) receiving plasmapheresis, continuous renal replacement therapy (CRRT), or semi whole blood replacement. The median time of first methylprednisolone and IVIG was 4 [2, 5] and 2 [1, 3] days, respectively. All cases received 2 to 7 ASMs, and 17 cases (17/22, 77.3%) had KD. Two patients tried vagal nerve stimulation (VNS) and one patient had anakinra at the acute phase.

Table 2 Additional Treatments of 22 FIRES Pediatric Patients

In chronic phrase, one patient was lost to follow up. Of the remaining 21 patients, 11 patients needed immune therapies. Eight patients took Anakinra, 10 cases had IVIG, and 8 one had methylprednisolone. Specially, one patient had ACTH and three patients used methylprednisolone due to hypsarrhythmia and epileptic spasm as the disease progressed.

TCZ Therapy and Treatment Response

In acute group, the first dose of TCZ was used from 1 day to 28 days after seizures onset. Over the course of disease, 11 patients had 2 to 13 times of TCZ. Patient 12 had only one dose of TCZ. The dosage of TCZ varied from 7 to 12 mg/kg. Treatment responses, including the seizure frequency, mRS score, and number of ASMs were evaluated within 1 month before or after the administration of TCZ. Cranial MRI and EEG, however, were not carried out so frequently, especially during chronic phrase. All patients had over 50% reduction of seizure frequency after TCZ administration (median time: 15 [7.5, 18.5] days), and 10 patients got seizure controlled during follow-up (Table 3). But seizures re-occurred in 4 patients, and 6 cases were seizure free for more than 6 months at last follow-up. The median dosage of TCZ to achieve status epilepticus resolution was 1 [1, 1]. Nine patients had reduced number of antiseizure medications (ASMs) after using TCZ, and two of them later had to add ASMs due to the increased seizure frequency. Eleven patients had improvements of EEG results. Attenuation of internal discharges (IEDs) was observed on 11 patients (11/12, 91.7%). Ten had reduced slow waves, but 2 of them got worse with disease progression. Specially, absent slow waves and IEDs were monitored on 6 patients (6/12), although 2 of them had slow waves and IEDs later. Alleviation on cranial MRI was found on 7 patients. MRI of patient 3 revealed no abnormalities at disease onset (Figure 2A and B). Subsequent imaging demonstrated the development of multifocal cortical lesions involving the frontal, temporal, occipital, and insular lobes (Figure 2C and D), which later resolved following treatment with tocilizumab (TCZ) (Figure 2E and F). No better performance was obtained on one patient (P10). No lesion was observed on more than two times of cranial MRI in two patients. The mRS scores were decreased by 1 to 5 on those patients. And the scores were significantly lower after the final dose of TCZ (Z = – 3.195, P = 0.001). Serum IL-6 and CSF IL-6 were monitored on 9 and 4 patients during the treatment of TCZ, respectively. The median time of last lumber puncture to assess cytokines in CSF was 26 [24, 109.5] days. Only data that tested before using anakinra were analyzed. Decreased CSF IL-6 was observed in 4 patients. Five patients had elevated serum IL-6 levels after TCZ, but IL-6 levels then decreased after the subsequent dose of TCZ in two of them (P3 and P11). The full information of times and dosage of TCZ on each subject, results of EEG and cranial MRI as well as mRS score and number of ASMs after and before every dose of TCZ can be seen in Supplementary Table 2.

Table 3 The Therapeutic Effect of TCZ of 22 FIRES Patients

Figure 2 The MRI of patient 3 before and after TCZ (Axial T2 Flair). (A and B) Normal MRI at the beginning of course. (C and D) Multifocal cortical lesions presented on frontal, temporal, occipital and insular lobes (white arrows). (E and F) Lesions disappeared after the 8th time of TCZ.

In chronic group, patients had TCZ 46 days to 6 years after disease onset. Two patients initiated 6–8 weeks post-onset, one patient started 9 months post-onset, and most of them (7/10, 70%) started TCZ administration 1 to 6 years after FIRES onset. Three to 10 times of TCZ (7~12 mg/kg) were administrated in those patients. Five patients had over 50% reduction of seizure frequency and two had seizure controlled during the course. However, seizures recurred on those two patients. One patient had once reduced number of ASMs during TCZ usage but later resumed the previous number of ASMs. Four patients had better mRS score after the administration of TCZ, and the mRS scores were decreased significantly after the final dose of TCZ (Z = – 2.000, P = 0.045). Reduced slow waves and IEDs on EEG were found in three patients, when two patients showed ameliorations on cranial MRI. Specially, of the 7 patients treated after more than one year, 3 experienced reduced seizure frequency, 3 showed decrease in mRS score, 2 demonstrated EEG improvements, and 1 had radiographic improvement. Serum IL-6 was tested in 8 patients before and after TCZ. One patient used anakinra was not analyzed. Six patients had elevated serum IL-6, and one of them (P21) had decreased IL-6 levels after the subsequent dose of TCZ. CSF IL-6 was not repeated after TCZ.

No patient experienced severe adverse events. Nine (9/22, 40.9%) patients, including 5 in acute group and 4 in chronic group, had neutropenia after TCZ. Two cases (2/22, 9.1%) (P5 and P9) had elevated transaminase levels within 2 folds. Overall, 59.1% (13/22) patients did not suffer from any adverse side effects.

Outcomes

We counted the days of ICU length, hospital length, SE duration, and mechanical ventilation duration at the first hospitalization, as well as number of ASMs at SRSE to evaluated the short-term outcomes of those 22 cases, and then compared within the acute and chronic group to figure out the efficacy of TCZ at acute stage of FRIES (Table 4). Results showed that the median days of ICU length of stay, hospital length of stay, mechanical ventilation, and SE duration were shorter in acute phase. However, only SE duration was significantly decreased in acute group, indicating the potential effectiveness of TCZ in stopping SE. The number of ASMs at SRSE between two groups had no statistical difference, either.

Table 4 Comparison Between Patients Treated with and without TCZ at Acute Phase

The seizure frequency, number of ASMs, mRS ranks, and other treatments during the last 6 months, like immune therapies, KD and VNS, and neuropsychological domain assessment at last follow-up, were regarded as long-term outcomes. The observation period was 17 [10, 27] months after the first dose of TCZ. One patient (P16) was lost to follow up, who was not included in long-term outcome evaluation. Overall, there were 15 patients (15/21, 71.4%) had more than 50% reduction of seizure frequency, and 5 patients (5/21, 23.8%) were completely controlled. In acute group, 100.0% patients gained more than 50% reduction, and 41.7% of them were seizure free, both of which were significantly higher than chronic group (Table 5). The number of ASMs in acute group ranged from one to five. Overall, 50% patients in acute group needed more than 3 kinds of AMSs, statistically lower than that in chronic group (100%). The mRS ranks in those two groups were not significantly different. Then, five different neuropsychological domains were assessed, and it turned out that memory deficit was the most frequently sequela (14/21, 66.7%), followed by attention deficit (11/21, 52.4%). Speech and motor deficit were present in 38.1% (8/21) and 28.6% (6/21) patients separately. The comparison between two groups demonstrated that patients in chronic group had higher probability to have speech and motor deficit (P< 0.05). Except for three preschoolers, a total of 9 patients (9/18, 50.0%) returned to school, and patients in acute group were more likely to go back to school. Current therapies, like IVIG, steroid, and Anakinra, were undertook in 4~6 patients during the last 6 months. Three patients were on KD, and 4 patients had VNS at the last visit. No significant difference was discovered between two groups.

Table 5 Clinical Outcomes of 21 FIRES Patients Treated with TCZ

Discussion

In this study, we demonstrated the potential effectiveness of tocilizumab (TCZ) in children with FIRES during both acute and chronic disease phases. In recent years, immunomodulatory therapies have garnered increasing attention for the treatment of various immune-related disorders. Among these, TCZ has emerged as a promising therapeutic option for FIRES, showing unique potential in the treatment of FIRES.² Recent reports suggest its effectiveness in treating both new-onset refractory status epilepticus (NORSE) and FIRES, particularly in cases with elevated IL-6 levels.^11,16–18 By blocking IL-6 signaling, TCZ can rapidly alleviate frequent epileptic seizures and demonstrate significant therapeutic effects. Previous case reports and series involving over 40 patients have documented its use.^10,18 For instance, He et al described 23 pediatric FIRES patients treated primarily during the acute phase, among whom 88.89% developed epilepsy one-year post-discharge.¹⁸

Limited literatures reported that TCZ may facilitate recovery of consciousness and neurological function, especially when administered during the acute phase.^16,18 Cognition improvement was observed in both acute and chronic phase patients in our study. mRS scores decreased in most patients, including those from the chronic group. At last follow-up, 76.2% patients have an mRS ≤ 2, consistent with the 66.67% of TCZ-treated patients in He et al’s study who attained a Pediatric Cerebral Performance Category (PCPC) score ≤ 3.¹⁸ Neuropsychological evaluation gives us hints on the deficits patients remained. Patients may have memory impairment and attention deficits. Speech and motor impairments appeared to be less frequent—a profile consistent with reports of FIRES patients treated with anakinra.¹⁹

Current international consensus recommends the initiation of TCZ within 7 days after disease onset.^1,2 However, since most patients in our cohort were referrals from other hospitals, only 4 received their first TCZ dose within this 7-day window. Due to the limited sample size, no statistical comparison was performed between these 4 patients and those treated later. Nonetheless, clinical benefits appeared to be observed even in patients who started TCZ beyond 7 days—including some more than 1 year after onset. Improvements seemed to be present in seizure frequency, mRS scores, EEG, and cranial MRI.

For children at chronic phase, our experience suggested that if they still have refractory seizures, cognitive impairment, or discontented mRS improvement after other treatments, which were not attributed to structural abnormalities on cranial MRI, TCZ can be an effective attempt. It is worth noting that the use of TCZ in pediatric FIRES remains off-label and clinical experience is still limited. Thorough discussion with guardians and informed consent are therefore essential prior to treatment.

To further evaluate the optimal timing of TCZ administration, patients who started TCZ within 4 weeks of onset were divided into an acute group, and who began treatment later into a chronic group. Comparative analysis revealed that while TCZ contributed to termination of status epilepticus (SE), it did not significantly reduce ICU or hospital length of stay, duration of mechanical ventilation, or number of antiseizure medications (ASMs) used during super-refractory SE (SRSE). These findings are consistent with previous reports, including a study in which 85.7% (6/7) of NORSE patients responded to TCZ for SE cessation,²⁰ and another which also found no significant reduction in ICU stay, hospitalization, or ventilator dependence.¹⁸ Superior long-term outcomes were associated with earlier TCZ administration. Patients in the acute group exhibited better seizure control and were less likely to require more than three ASMs (P < 0.05). Although not statistically significant, more patients in the acute group also achieved good functional recovery (mRS ≤ 2). Importantly, early TCZ use was associated with a lower incidence of motor and speech deficits (P < 0.05), and a greater likelihood of returning to school.

TCZ is a humanized monoclonal antibody directed against the interleukin-6 (IL-6) receptor.^21,22 IL-6 plays a key role in central nervous system inflammatory responses, and its levels are frequently elevated in the CSF of patients with FIRES. In this study, elevated IL-6 levels were observed in the CSF of 83.3% patients and in the serum of 77.3% patients. A previous investigation tracked IL-6 dynamics before and after TCZ administration,¹⁸ reporting decreased CSF IL-6 in 80% of patients and increased serum IL-6 in every patient post-TCZ. Similarly, our data showed increased serum IL-6 in 68.8% of patients after TCZ, though three subsequently exhibited reduced levels with further dosing. CSF IL-6 decreased in all 4 patients with available paired samples. Although elevated serum IL-6 following TCZ has been empirically linked to clinical improvement,^17,18,20,23 direct evidence supporting this correlation remains limited. In the present study, we compared outcomes, including seizure frequency, mRS scores, and EEG improvement between patients with decreased (n=8) and increased (n=5) serum IL-6 after TCZ (3 excluded due to fluctuating values). No significant differences were found between these two groups. Moreover, symptomatic and prognostic improvements appeared to be present in patients with normal CSF and serum IL-6 levels at baseline. These findings suggest that TCZ treatment need not be guided solely by IL-6 levels. However, small size and retrospective nature in our study limited further investigations and reliability.

TCZ-related infections or infusion-related reactions are generally infrequent.^18,24 In our cohort, which included multiple TCZ administrations, no severe adverse reactions occurred. Transient transaminase elevation (within 2-fold upper limit) or neutropenia was observed in 40.9% of patients. Overall, TCZ demonstrated an acceptable safety profile, though close monitoring for infections, hypersensitivity, hepatotoxicity, neutropenia, and thrombocytopenia remains advisable.

Limitations

First, small sample size, retrospective design (which leads to incomplete data and irregular follow-up schedules) and lack of control group (which may lead to biased estimation of TCZ effect) may compromise the quality of the evidence which call for careful interpretation. Secondly, heterogeneity in TCZ treatment (dosing, time of administration) and lack of validation of mRS in FIRES may hinder interpretability and reproducibility. Third, some concepts like acute phrase and chronic phrase lacked rigorous standard.

Conclusion

Our experience provided evidence of the therapeutic potential of TCZ in both acute and chronic phases of FIRES. TCZ treatment was associated with reduced seizure frequency, improved EEG and MRI findings, decreased reliance on antiseizure medications, and functional improvement as measured by mRS. Initiation of treatment during the acute phase may be associated with a shortened duration of status epilepticus and, compared to chronic-phase intervention, appears to correlate with more favorable long-term outcomes. Furthermore, treatment response did not show a clear dependence on baseline IL-6 levels in either serum or CSF.