Anakinra usage in febrile infection related epilepsy syndrome: an international cohort

Abstract Febrile‐infection related epilepsy syndrome (FIRES) is a devastating neurological condition characterized by a febrile illness preceding new onset refractory status epilepticus (NORSE). Increasing evidence suggests innate immune dysfunction as a potential pathological mechanism. We report an international retrospective cohort of 25 children treated with anakinra, a recombinant interleukin‐1 receptor antagonist, as an immunomodulator for FIRES. Anakinra was potentially safe with only one child discontinuing therapy due to infection. Earlier anakinra initiation was associated with shorter duration of mechanical ventilation, ICU and hospital length of stay. Our retrospective data lay the groundwork for prospective consensus‐driven cohort studies of anakinra in FIRES.


Introduction
Febrile-infection related epilepsy syndrome (FIRES) is a devastating neurological condition with significant mortality and morbidity. [1][2][3] It represents a subset of children with new onset refractory status epilepticus (NORSE) in whom a febrile infection precedes the onset of seizures. 4 Emerging evidence suggests that neuroinflammation may contribute to the pathologenesis. [5][6][7] Despite improvements in pediatric intensive care unit (ICU) support and encouraging results from ketogenic diet for refractory status epilepticus, there are no immunomodulatory treatments for children with FIRES.
Experimental models of status epilepticus (SE) have implicated innate immunity as a potential etiology of seizure susceptibility. [8][9][10] Microglial activation and monocyte infiltration have been observed in the brain following SE. 9 Similarly, SE increases brain mRNA and protein levels of interleukin-1 beta (IL-1b). 8,9 Exogenous administration of IL-1b enhances seizure susceptibility 8,10 . Administration of an IL-1 receptor antagonist (IL-1ra) ameliorates the pro-convulsant effects of IL-1b, 10 suggesting that IL-1ra may represent a candidate therapy.
In support of IL-1ra as a potential target, children with febrile SE have lower serum ratios of IL-1ra levels to proinflammatory cytokines (IL-6, IL-8). 11 Hyperactive IL-1b activity and functional IL-1ra deficiency are observed in children with FIRES, which can be ameliorated with recombinant IL-1ra (anakinra). 5 Anakinra therapy appears to dampen seizures, facilitates the withdrawal of anesthetic agents, and is associated with favorable clinical outcome in two children with FIRES. 6,7 These cases raise the possibility that exogenous IL-1ra may represent a promising FIRES therapy. Here we report the safety and potential efficacy of anakinra as an immunomodulator for FIRES in an international case series of 25 children.

Methods
Children on anakinra were identified through the international McMaster Rheumatology List serve between October 2017 and February 2018; and by contacting individual institutions. Two subjects were published in previous case reports. 12 Contributing institutions received approval from the respective Institutional Review Boards or Ethics boards. Site investigators determined FIRES diagnosis according to the proposed consensus definitions 4 but excluded children with suspected autoimmune encephalitis. We collected demographics, clinical and laboratory parameters at the initial ICU admission, anakinra dosing and duration, adverse events, and functional outcomes using a standardized abstraction form (supplemental materials). The Pediatric Cerebral Performance Category (PCPC) was determined by the site investigators. Neuropsychological domain assessments were performed by the respective contributing centers and reported as normal or abnormal. In a subset of children (n = 15), local investigators determined electrographic and electroclinical seizure frequency immediately before and one week after anakinra treatment. Subsequently we dichotomized these children into those with greater than 50% seizure reduction (n = 11) and those without (n = 4). We evaluated the demographics, clinical characteristics, and outcomes using descriptive statistics. The results are reported as number (%) or median [interquartile range]. Pearson correlation was used to evaluate the association between the timing of anakinra initiation and duration of mechanical ventilation, ICU length of stay (LOS) and hospital LOS.

Demographics and Diagnostic Work-up
Most patients were male (68%) and the median age at FIRES diagnosis was 8 years [5.2-11 years] ( Table 1). The majority of initial cerebrospinal fluid and brain imaging studies were normal. Elevations in cerebrospinal fluid and serum cytokine/neopterin levels were often found in those children who underwent comprehensive testing (Table 1).

Treatment Prior to Initiation of Anakinra
Prior to the initiation of anakinra, all children were treated with anesthetic agents to achieve seizure control ( Table 2). Two children were on midazolam infusion alone; 5 on pentobarbital alone; and 18 on both infusions. All children received at least 4 additional anti-seizure medications (ASMs), with 18 children (72%) having failed 7 or more agents prior to anakinra initiation. Nineteen children (76%) were on the ketogenic diet; 7 (28%) received cannabinoids. Corticosteroids and intravenous immunoglobulin (IVIG) were used in 22 children (88%). Of these 22 children, 11 (44%) received plasmapheresis and 5 (20%) rituximab.

Anakinra therapy
Anakinra was started at a median of 20 days [14-25 days] after the onset of seizures. Initial median anakinra dose was 3.8 mg/kg per day [3-5 mg Table 2). Three children (12%) developed drug reaction with eosinophilia and systemic symptoms syndrome (DRESS), which was treated with the addition or escalation of corticosteroids. All recovered without complications. Two children (8%) developed cytopenias that eventually resolved without specific intervention. Anakinra was discontinued in only one child due to infection.

Outcomes
The  [3][4]. Earlier anakinra initiation after seizure onset was associated with shorter duration of mechanical ventilation, and ICU and hospital LOS (r = 0.46 (P = 0.03), r = 0.50 (P = 0.01) and r = 0.48 (P = 0.03), respectively). Amongst children with available seizure frequency data (n = 15), 11 exhib-ited> 50% seizure reduction at 1 week of anakinra treatment. Although there were no statistical differences due to the small sample size, several observations were notable between children with and without seizure reduction ( Three children died (12%), all from withdrawal of support due to the persistent super-refractory SE and the expected poor neurological outcome. The timing of withdrawal varied from 1 to 18 days following anakinra initiation. The median length of follow-up for the surviving children with available information (n = 17) was 321 days [219-420 days]. Six had no or minimal disability (PCPC 1-2), 6 had moderate disability (PCPC 3), and 5 had severe disability or vegetative state (PCPC 4-5). All surviving children had drug resistant epilepsy. The most prominent neuropsychological deficits at follow-up were in attention (n = 17), executive functioning (n = 15), and speech (n = 13). Twelve children returned to school, 10 required academic accommodations or special education classes.

Discussion
Here we describe the safety and potential efficacy of anakinra therapy in the largest international retrospective  cohort of FIRES patients. Infections, transaminitis, and neutropenia represent potential anakinra-associated side effects. In our cohort, the prevalence of infection before and after anakinra initiation were comparable; and only one had transaminitis. While increased eosinophil count is a known side effect of anakinra, DRESS syndrome has not been associated with anakinra in rheumatic conditions. 13 In FIRES patients, DRESS may likely reflect an underlying immune dysregulation, as evidenced by some children with FIRES meeting criteria for hemophagocytic lymphohistiocytosis characterized by pathologic immune activation. 14 None of the patients experienced adverse outcomes due to DRESS syndrome or cytopenia with conservative management. However, these complications may present more serious challenges with increasing anakinra utilization for the treatment of FIRES. Early anakinra initiation was associated with shorter mechanical ventilation days, ICU and hospital LOS, and possibly seizure reduction, suggesting that innate immunity may contribute to the pathology of FIRES. Specifically, elevated IL1-b levels have been described in the cerebrospinal fluid and the serum of three children with FIRES; and the functional IL1-ra deficiency in the cerebrospinal fluid of one children. 5 Exogenous IL-1ra administration ameliorated the functional IL1-ra deficiency ex vivo, providing a rationale for using recombinant IL-1ra (anakinra) as an immunomodulatory treatment for FIRES. 5 We found that administering anakinra to children with FIRES may indeed be beneficial, which supports hyperactive IL1-b activity and/or functional IL-1ra deficiency as significant pathological factors underlying FIRES. There are several limitations to our study that are inherent in a retrospective case series and highlight current knowledge gaps in the field. Although early anakinra initiation may be beneficial as demonstrated in this study, we were unable to ascertain the optimal therapeutic window for anakinra treatment. The duration of therapy was highly variable with 10 children receiving < 1 month of treatment. In contrast, response to anakinra in the rheumatologic conditions such as systemic juvenile arthritis has been assessed no earlier than 4-12 weeks from initiation of therapy. 15 These limitations, coupled with patient and other clinical factors, likely contributed to the lack of improved long-term neurological outcome in this study. Nevertheless, our findings provide additional support for anakinra as a potential immunomodulator for patients with FIRES. Prospective studies are necessary to understand: a) the optimal timing, dosing and duration of anakinra therapy, b) rational biologic correlates of anakinra response, and c) safety and efficacy of anakinra and ASMs.