CSF neopterin and quinolinic acid are biomarkers of neuroinflammation and neurotoxicity in FIRES and other infection‐triggered encephalopathy syndromes

Abstract Objective Infection‐triggered encephalopathy syndromes (ITES) are potentially devastating neuroinflammatory conditions. Although some ITES syndromes have recognisable MRI neuroimaging phenotypes, there are otherwise few biomarkers of disease. Early detection to enable immune modulatory treatments could improve outcomes. Methods We measured CSF neopterin, quinolinic acid, kynurenine and kynurenine/tryptophan ratio using a liquid chromatography coupled to tandem mass spectrometry (LC–MS/MS) system. The CSF of 18 children with ITES were compared with acute encephalitis (n = 20), and three control groups, namely epilepsy (n = 20), status epilepticus (n = 18) and neurogenetic controls (n = 20). Results The main ITES phenotypes in 18 patients were acute encephalopathy with biphasic seizures and late restricted diffusion (AESD, n = 4), febrile infection‐related epilepsy syndrome (FIRES n = 4) and other ITES phenotypes. Influenza A was the most common infectious trigger (n = 5), and 50% of patients had a preceding notable neurodevelopmental or family history. CSF neopterin, quinolinic acid and kynurenine were elevated in ITES group compared to the three control groups (all p < 0.0002). The ROC (area under curve) for CSF neopterin (99.3%, CI 98.1–100) was significantly better than CSF pleocytosis (87.3% CI 76.4–98.2) (p = 0.028). Elevated CSF neopterin could discriminate ITES from other causes of seizures, status epilepticus and febrile status epilepticus (all p < 0.0002). The elevated CSF metabolites normalised during longitudinal testing in two patients with FIRES. Interpretation CSF neopterin and quinolinic acid are neuroinflammatory and excitotoxic metabolites. This CSF metabolomic inflammatory panel can discriminate ITES from other causes of new onset seizures or status epilepticus, and rapid results (4 h) may facilitate early immune modulatory therapy.


Introduction
Infection-triggered encephalopathy syndromes (ITES) are a group of clinical and radiological phenotypes, that are due to a neuroinflammatory response after common childhood infections. 1 A genetic vulnerability is presumed and can be demonstrated in some patients (such as RANBP2 in acute necrotising encephalopathy), but the majority of patients with ITES do not harbour genetic variants that are measurable in current clinical practise. 2 ITES can occur as a recognisable 'clinico-radiological phenotype', such as acute encephalopathy with biphasic seizures and late restricted diffusion (AESD), febrile infection-related epilepsy syndrome (FIRES), acute necrotising encephalopathy of childhood (ANE), acute fulminant cerebral oedema 3 (AFCE) and mild encephalopathy with reversible splenial lesion (MERS). 1,4 Less common ITES syndromes are acute infantile encephalopathy predominantly affecting the frontal lobes (AIEF) which is a variant of AESD, 5 and hemiplegia hemiconvulsion epilepsy syndrome (HHE). 6 Given the similarity of FIRES with other ITES phenotypes (fever and infection provocation, acute encephalopathy, distinct clinico-radiological phenotypes, evidence of neuroinflammation without encephalitis), we present these syndromes all together under the umbrella term of ITES.
Patients with ITES present with an acute febrile encephalopathy, similar to acute encephalitis. Acute encephalitis is due to direct invasion of the central nervous system (CNS) by a neurotropic virus/microorganism with secondary innate and adaptive immune cell inflammation affecting the brain parenchyma (due to T cells, B cells and autoantibodies). Alternatively, acute encephalitis is due to infiltration of the brain by peripheral innate or adaptive immune cells and autoantibodies as part of an autoimmune or inflammatory process. Both viral and autoimmune encephalitis can result in secondary CNS resident immune (glia) activation. 7 By contrast, in ITES, there is no definable autoimmune process, and no infiltration of the brain by peripheral immune cells. Instead, the inflammation of ITES is thought to be due to an indirect activation of the immune system within the brain (such as microglia and astrocytes), secondary to infection occurring outside of the CNS. 2 This separation of ITES from encephalitis is important not only from a diagnostic perspective, 7 but also from a therapeutic perspective. There is now great interest in the use of immune therapies which target innate immune activation molecules, such as interleukin-1 and interleukin-6, which are pro-inflammatory pleiotropic cytokines that are pro-epileptic as well as proinflammatory. For example, anakinra, an interleukin-1 receptor antagonist, and tocilizumab, an IL-6 receptor antagonist, are potentially valuable therapeutic agents for ITES. 8,9 Despite this progress, there remain major challenges in making a positive diagnosis and discriminating ITES from encephalitis and other causes of new onset seizures (epilepsy) or new onset encephalopathy.
However, some of the syndromes do not have typical radiological biomarkers, such as FIRES, 10 which is hard to discriminate from acute encephalitis or other causes of refractory epilepsy in the early days of the illness. Likewise, there are a number of ITES which lack a radiological phenotype (normal MRI imaging) but are clearly infection provoked encephalopathy syndromes (including . Although cerebrospinal fluid (CSF) cytokines are potentially measurable and have been shown to be elevated in patients with ITES and FIRES, 11 these cytokines are typically only used for research purposes, and due to the batching nature of the platforms, are often not translatable into clinical settings at this time.
Initially using untargeted CSF metabolomics, followed by targeted metabolomics, we have created a validated and reproducible CSF inflammatory metabolite panel for use in clinical service. We have found that CSF neopterin, quinolinic acid, kynurenine and kynurenine/tryptophan ratio to be useful in detecting neuroinflammation. 12 The panel uses a liquid chromatography coupled to tandem mass spectrometry (LC-MS/MS) platform, and generates results for patient samples in 4 h, with the capability of individualised testing for clinical urgency. 13 Given the fact that many ITES lack useful biomarkers, we explored the utility of the CSF inflammatory metabolite panel in a cohort of ITES, and compared the findings to controls that are relevant to the differential diagnosis facing clinicians.

Cohort
Between 2016 and 2022, 18 patients with ITES were seen at the Children's hospital at Westmead (n = 12) or referring hospitals (n = 6) and had stored frozen CSF available for analysis, as described in detail below. Regarding therapeutics, intravenous methylprednisolone pulses were 30 mg/kg/day for 3-5 days (maximum 1 g), intravenous immunoglobulin was 2 g/kg for 2-5 days for induction dose and 1 g/kg for subsequent doses when given. Anakinra starting dose was 3-5 mg/kg/day subcutaneously (daily) and final dose was up to 10 mg/kg/day, and tocilizumab was 8-12 mg/kg given intravenously once and repeated every 2 weeks as required. Modified Rankin score (MRS) was generated retrospectively by RCD on review of the clinical histories and discussion with treating clinicians.
To compare the findings in ITES patients, we used control groups that are important differential diagnoses for ITES presentations, namely encephalitis (n = 20), epilepsy (n = 20), status epilepticus (febrile and afebrile) (n = 18) and non-inflammatory and non-epileptic neurological controls (neurogenetic, n = 20). The ages were not significantly different in any control group compared to the ITES group (Mann-Whitney U test) (Table 1). The patient with influenza-associated encephalitis had MRI inflammatory changes compatible with encephalitis (rather than an ITES).

CSF metabolomics
All patient CSF was sent to the Children's Hospital at Westmead Neurochemistry laboratory for routine diagnostic services in multiple aliquots. All samples are frozen at À80 degrees within 1 h. An aliquot that had not been previously thawed was used for this research assay. For the ITES, acute encephalitis and status epilepticus, the CSF samples were taken in the first 3 days of admission (as part of initial diagnostic work up) and before immune therapy, but not necessarily before symptomatic therapy (e.g. anti-epileptic drugs). For epilepsy and neurogenetic control groups, the CSF was taken during the chronic but symptomatic phase of the illness. The selected metabolites of interest in the panel are associated with neuroinflammation and were reported as statistically significant in an untargeted metabolomics study. The measurement of neopterin and tryptophankynurenine pathway metabolites (tryptophan, kynurenine, quinolinic acid and kynurenine/tryptophan ratio) was conducted in accordance to the method of Yan et al., 13 and reference ranges were used, as reported previously. 14 In brief, human CSF was deproteinised using metaphosphoric acid/ethylenediaminetetraacetic acid solution, vortexed and centrifuged in Nanosep 0.2 lM centrifugal devices. The supernatant was collected and analysed on the Waters ACQUITY UPLC I-Class System UPLC system coupled to a Xevo TQ-XS triple quadrupole mass spectrometer. The Acquity UPLC BEH C18 column (2.1 mm 9 150 mm 1.7 lm particle size) was used to achieve chromatographic separation of the metabolites and detected on the mass spectrometer using the multiple reaction monitoring mode.

Statistics
Statistical analyses were performed using GraphPad Prism 8 (GraphPad, San Diego, USA) and R (The R Foundation for Statistical Computing, version 4.2.2). For the metabolomics data, pairwise non-parametric tests (Mann-Whitney U test) were performed, and GraphPad Prism was used for analyses and figure generation. We compared the inflammation group with the other four groups (encephalitis, epilepsy, status epilepticus and neurogenetic controls), therefore allowing for 16 pairwise tests, the Bonferroni correction was 0.0031. The pROC R package was used to generate ROC curves and areas under the curves (AUCs). 15 The 95% CI of the AUC was computed using the pROC ci.auc function and DeLong method. Comparisons of two correlated (paired) ROC curves were computed using the pROC roc.test function and DeLong method. 16

Ethics
The Sydney Children's Hospitals Network Ethics Committee approved this study (LNR/14/SCHN/275; 2019/ ETH06182), including informed consent from parents and/or guardians, as per ethics protocol.

Results
Infection-triggered encephalopathy syndrome ( Table 2) Demographics and previous history 18 children with infection-associated encephalopathy syndrome (ITES) had available stored CSF for analysis after routine testing. 12 of the 18 had a recognised clinicoradiological syndrome: Four patients had AESD (including one with AIEF), four patients had FIRES, two patients had ANE and one patient each with HHE, AFCE and MERS. A further five patients had ITES but normal imaging: These patients lacked features of a clinico-radiological syndrome, all had an encephalopathy syndrome associated with preceding or concurrent infection but lacked features of encephalitis or meningitis.
Nine of the 18 patients were female and the median age was 4.9 (range 0.7-17.3). The AESD patients tended to be younger than the FIRES patients ( Table 2).
Nine of the 18 patients had a preceding personal history or notable family history; a neurodevelopmental syndrome was present in six patients, including autism in two and Dravet syndrome (SCN1A positive) in one patient. Notable immune history was present in two patients (JIA and Type 1 diabetes). One patient with ANE had a pathogenic RANBP2 mutation, as did an unaffected parent.

Infection triggers
All patients had preceding and often ongoing fever and/or infection at the time of neurological symptom onset ( Table 2). The mean duration of fever was 3 days (1-7), with AESD, ANE, MERS, HHE and AFCE patients tending to have a shorter fever duration than the FIRES patients ( Table 2). A specific infectious agent was identified in 12, with a higher yield in AESD, ANEC and ITES, and lower yield in FIRES. Influenza A was found on nasopharyngeal aspirate in five, rhinovirus in three and SARS-CoV2 in two patients ( Table 2). Three patients had two infectious agents identified (Cases 1 and 17). Other notable infectious triggers of ITES included, Salmonella typhi found in blood in resident returning from Pakistan   The grey shade helps to visualise the different subgroups of the cohort. 1 With normal neuroimaging. who developed encephalopathy 7 days into typhoid fever illness; Escherichia coli septicaemia in a child presenting a week in to an admission with diabetic ketoacidosis who developed encephalopathy in the absence of meningitis or biochemical explanation.

Neurological syndrome
Seizures (n = 12) and encephalopathy (n = 6, drowsy, coma, behavioural change) were the presenting features. Investigations (Table 3) Full blood count showed slightly reduced white blood count in four patients, and reduced platelet count in six patients. Mild or more severe transaminitis was common in the first days of neurological onset (present in 16 of 18). CSF pleocytosis (defined as >5 monocytes) was present in 7 of 16 patients with available data and elevated CSF protein >0.4 g/dl was evident in only 4 of 15 patients with available data.
MRI neuroimaging (Table 3 and Fig. 1) In AESD, the first MRI when performed in the first days were typically normal, and the AESD pattern was evident at Days 5-6, with volume loss evident on follow-up scans. In FIRES, MRI was often normal, although showed hippocampal swelling or brightness in two patients in the first or second week, and atrophy in the patient with sequential imaging.
The ANE, HHE and MERS patients showed the characteristic imaging patterns as per Table 3 and Figure 1. By definition the MRI in the other ITES without clinicoradiological syndromes were normal. 1 Electroencephalography EEG was diffusely slow or focally slow in all patients and showed electrical seizures in one AESD patients and all four FIRES patients. There was evidence of bihemispheric shifting seizures and PLEDs in three of four FIRES patients. 17

Therapeutics and outcome
Sixteen ITES patients received immunotherapy: intravenous methylprednisolone (IVMP) or dexamethasone was given in 16 patients and IVIG in eleven. Anakinra was given to four patients (all FIRES), tocilizumab to four (three FIRES and one ANE) and rituximab to one patient (FIRES). Midazolam infusions and ketamine infusions were needed for the FIRES cases, and ketogenic diet was given to three FIRES cases. Symptomatic AED were given as required.
Regarding outcome, two patients died, one during the hyperacute course of COVID-19-induced encephalopathy (AFCE), and one patient with FIRES who survived the acute episode but died unexpectedly after discharge of possible sudden unexpected death in epilepsy (SUDEP). In the surviving patients, at mean 1.88 years follow-up (range 0.3-6.5 years), 7 of 16 had a good outcome (defined as MRS 0-2) and 9 of 16 had poor outcome (MRS 3-5) ( Table 3). Specific deficits included intellectual disability (or global developmental delay) in seven, attention deficit hyperactivity disorder (ADHD) or emotional dysregulation (executive dysfunction) in five, acquired motor disability in three and ongoing epilepsy in six. In survivors, the patients with ITES (no clinico-radiological syndrome) and MERS had the better outcomes, and the patients with AESD, FIRES and ANE had worse outcomes (Table 3). One patient with ITES had a relapse 3 months after the first event (Case 17).

CSF metabolomics
CSF neopterin (NEO), quinolinic acid (QUIN), kynurenine (KYN) and kynurenine/tryptophan (KYN/TRP) ratio were significantly elevated in ITES compared to the epilepsy, status epilepticus and neurogenetic control groups (except ITES vs. status epileptics for KYN/TRP ratio), as per Figure 2 and Table 4. The CSF metabolites were generally more elevated in the ITES group compared to the encephalitis group, but not statistically significantly ( Fig. 2 and Table 4).
The raw values for all the cases are presented in Table S1, and the cohort values including the minimum, quartiles, median and maximum are presented for all groups are presented in Table S2. In Table 5, using the 95th centile from a large cohort of non-inflammatory neurological controls, 14 we list the percentage of patients in each group above the threshold.   Fig. 3A). On paired correlation of ROC curves, only the CSF neopterin AUC was significantly superior to the CSF pleocytosis AUC (p = 0.028).

Clinical scenario of a child presenting with new onset seizures
The prevailing hypothesis is that early treatment in ITES and other neuroinflammatory disorders will influence outcomes. Therefore, recognising ITES syndromes at initial presentations, such as with a cluster of seizures or with status epilepticus, is important. We therefore compared the CSF neopterin in patients with status epilepticus who transpired to have ITES with those patients with status epilepticus who did not have ITES. ITES patients presenting with status epilepticus had higher neopterin that those patients with non-ITES status epilepticus (Fig. 3B, p < 0.0001), and febrile status epilepticus only (Fig. 3C, p = 0.0002). A CSF neopterin value of 82 nmol/ L was the highest febrile non-ITES status epilepticus value, and the lowest ITES status epilepticus value. Likewise, children presenting with any seizure (including status epilepticus) who transpired to have ITES had higher neopterin than those children with seizures of other causes (including non-ITES status epilepticus) (Fig. 3D, p < 0.0001). Figure 4 separates the four metabolites by syndrome, with comparison to the combined control data (n = 58). No exploratory statistics were performed. As can be seen, the FIRES subgroup generally had higher neopterin and quinolinic acid compared to the other groups. There was no correlation between CSF metabolites and MRS outcome (not significant using Pearson correlation). samples were significantly elevated, and then normalised over time.

Discussion
Infection-triggered encephalopathy syndromes (ITES) are an important group of conditions, which have concerning outcomes. Only MERS has a generally good prognosis, whereas the other clinico-radiological syndromes generally have poorer outcomes with epilepsy, neurodevelopmental and motor disability. 1 Although some syndromes have typical radiological phenotypes, other cases lacked a demonstrable radiological phenotype. It is possible that the five patients with ITES presented in this cohort who lacked a clinical or radiological phenotype was because sequential imaging was not performed, and a latent radiological biomarker as seen in AESD was missed. Notably, the outcome of these five patients was generally good. Taken together, the cohort of 18 patients have features that have been previously described. Some ITES syndromes affect younger children (AESD, ANE and AFCE), whereas other ITES syndromes affect school age children (FIRES and MERS). Infectious associations in our cohort were broad, although influenza A was the most common infectious association. 18 It should be noted that although viruses were the most common triggers, ITES was not exclusively triggered by viruses, indeed two patients in this cohort had clear bacterial triggers as part of an encephalopathic syndrome (Salmonella typhi and E. coli).
The clinical phenotypes in the ITES patients were dominated by encephalopathy with seizures, although dyskinetic movements were common in AESD. The hospital admissions were often complicated by prolonged admissions (apart from MERS), with high rates of intensive care admission, and multiple symptomatic agents (particularly for FIRES, and to a lesser extent ANE and AESD). Other than radiological phenotypes, blood testing was not differentiating from other encephalopathy syndromes, although alterations in full blood count (leukopenia and thrombocytopenia) and elevation of liver enzymes were commonly observed. The speed of diagnosis for encephalitis has greatly improved, and it is now possible to confirm or rule out infectious and antibody-mediated encephalitis within 24-48 hours of ICU admission, via multiplex PCR panel for pathogens and relevant antibody panel testing. FIRES or other ITES syndromes will be the likely diagnosis in children with a severe febrile encephalopathy who do not have pathogen-or autoantibody-associated encephalitis, in the absence of an underlying genetic or toxic/metabolic disorder. Without the aid of validated biomarkers for ITES, clinicians have a dilemma about the whether to use neuroprotective strategies only or to treat patients with immune or biologic therapy. This window of opportunity to introduce disease and outcome modifying interventions is short, and time sensitive. Characteristic neuroimaging is an important biomarker for ITES diagnosis but is only useful in patients with ANE and AESD, and not in patients with FIRES. When present, diffusion restriction often indicates non-reversible brain injury is occurring, 2 apart from in MERS.
ITES has a complex biology, and a 'cytokine storm' is a postulated, but yet unproven, mechanism of CNS injury. 2 Our findings present that elevated NEO and QUIN is significantly elevated in ITES compared to controls but not different to patients with known encephalitis and lends credence to an immune-directed pathobiology in FIRES and other ITES syndromes. CSF pleocytosis was often absent in our ITES cohort, and this is a clear demonstration of the need for novel biomarkers. Only 7 of 16 ITES patients had CSF pleocytosis, and excluding FIRES, only 4 of 12 non-FIRES ITES patients had CSF pleocytosis.
One of the primary findings of our study is that CSF neopterin is a valuable biomarker in ITES and shows superiority over CSF pleocytosis according to the ROC calculations. In addition, we showed that CSF neopterin has the potential of alerting clinicians to the possibility of ITES (or encephalitis) in the child presenting with new onset encephalopathy with or without seizures, and in differentiating ITES from other causes of febrile status  epilepticus or febrile seizures. NEO is produced due to activation of indoleamine dioxygenase-1 (IDO1), an enzyme which is activated by interferon species. 19 The inflammatory chemokines, CXCL9 and CXCL10 are also induced by interferon, 20 and have been reported to be elevated in the CSF of FIRES patients. 11 Furthermore, time course measurements suggest that NEO changes in parallel with CXCL9, CXCL10 and interferon, 21 supporting the suitability of NEO as a surrogate biomarker of interferon-related inflammation. The other major finding was the utility of quinolinic acid (QUIN) as a biomarker, which was elevated in ITES. QUIN is produced by activated infiltrating macrophages or by activated resident microglia, so is a potential biomarker of glia activation. QUIN is also an N-methyl-Daspartate receptor agonist and therefore neurotoxic. 22 Neopterin nmol/l Figure 3. ROC AUC showing superiority of CSF neopterin and CSF quinolinic acid over CSF pleocytosis (Fig. A). CSF neopterin has potential to discriminate status epilepticus due to ITES, compared to febrile and afebrile status epilepticus (Fig. B), status epilepticus due to ITES compared to febrile status epilepticus (C) and seizures due to ITES compared to non-ITES seizures (D) (all p < 0.0002). Indeed, QUIN has accepted pro-inflammatory, neurotoxic, gliotoxic and pro-oxidant properties, and also can affect blood-brain barrier integrity, so its measurement in acute encephalopathy syndromes is potentially relevant to aid neuroprotective interventions. 23 In recent years, there has been interest in the use of blood, urine and CSF neopterin and quinolinic acid in paediatric diagnostics. Neopterin can be diagnostically useful in infantile encephalopathy, 24 Aicardi-Gouti eres syndrome, 25 hemophagocytic lymphohistiocytosis syndrome, 26 encephalitis, meningitis and infection. 19,[27][28][29] Quinolinic acid has also been reported to play an important role in autism spectrum disorders, 30,31 attention deficit/hyperactivity disorder, 32 juvenile idiopathic inflammatory myopathies 33 and acute neuroinflammatory disorders. 14 A major advantage of the mass spectrometry approach for quantification is the ability to generate results rapidly. Given the strengths of using the mass spectrometry approach in an acute setting, individualised testing and rapid turnaround times will require diagnostic laboratories to have high throughput of samples and associated staffing support. Although measuring CSF cytokines is important, 34 and we hope to be able to incorporate CSF cytokines into the inflammation panel in the future, at present there are some limitations to cytokine analysis. The first limitation is that most cytokine approaches use research kits that are not licenced for use in clinical care, and the second limitation is the need for batching in some ELISA or multiplex platforms. However, there is potential that key cytokines can be detected and quantified using CSF proteomic approaches using mass spectrometry in the future, 35 and this is particularly relevant for cytokines that are targetable with therapy, such as interleukin-1, interleukin-6, tumour necrosis factor-alpha and interferon-alpha. The severe phenotypes in our cohort were treated empirically with biologics and monoclonal antibodies targeting immune molecules, including anakinra and tocilizumab. Given the rarity of ITES syndromes, it seems unlikely and possibly unethical to perform prospective high-quality clinical trial data exploring whether these agents change natural history and improve outcomes. To date, the clinical therapeutic data using these biological therapeutics in ITES is restricted to case reports and retrospective cohorts. 8,9,36 However, international registries with standardised datasets to record clinical and therapeutic data, as have occurred so successfully for multiple sclerosis using MSBase, would be a useful opportunity for the international community to collect prospective data in these children. To date, the proposed therapeutic agents target cytokines involved in innate and adaptive inflammatory cascades (interleukin-1 and interleukin-6), and this is logical given the fact the pathogenesis in ITES is suspected to be innate immune activation of the resident cells of the CNS (glia), rather than due to the peripheral adaptive immune system. NEO is produced due to interferon activation of IDO1, and future therapeutic approaches could involve modulation of interferon signalling. JAK inhibitors modulate cell signalling of many pro-inflammatory cytokines and are being used in genetic autoinflammatory disorders 37 and could be rationalised for use in FIRES and related disorders.
However, it should be noted that rituximab (a B-cell therapy) was used in one patient with FIRES in our cohort, and the clinician felt rituximab was effective in terminating the status epilepticus after 2 months on intensive care; therefore even though the current opinion is that anakinra and tocilizumab are more logical for treating ITES, rather than rituximab and other adaptive immune drugs, the community should keep an open mind to future therapeutics.
The other therapeutically relevant finding from our study is that we should think beyond cytokines when considering the treatment of ITES. Anti-inflammatory approaches and neuroprotective strategies such as ketogenic diet should be further explored, 38 and the effects of other novel therapeutic strategies that are antiinflammatory should be considered. Re-purposing old, cheap and moderately safe drugs that have antiinflammatory properties should also be considered, and 'neuroprotective and anti-inflammatory cocktails' may be the future of clinical care for ITES, once scientific rationale is demonstrated. 39 As shown in Figure 5, these biomarkers are potentially helpful for therapeutic monitoring in very sick children who need repeat CSF study. A further therapeutic consideration is the need for rapid intervention due to the cascade of neuroinflammation, excitotoxicity and neurotoxicity that can progress rapidly. 2 Therefore, there is a need to consider these diagnostic entities and intervene as early as is safely possible, within reason.
Finally, although we have focused on the inflammatory and biomarker aspects of this ITES cohort, there are other interesting clinical features in this cohort worthy of discussion. For example, the cohort had a higher rate of preceding neurological or immunological vulnerability, with autism, Dravet syndrome, previous febrile convulsion and a history of autoimmunity found at higher rates than would be expected by chance in 18 children. This leads to the hypothesis that these patients have some innate vulnerability (genetic, epigenetic or neurodevelopmental, etc.) that is relevant to disease expression. 2 Vulnerability genes such as in carnitine palmitoyl transferase 2 (CPT2) and adenosine A2A receptor (ADORA2A) have been associated with ITES syndromes in Japanese cohorts and support the notion of innate vulnerability. 2 Limitations of this study include the retrospective nature of patient recruitment, the non-standardised timing of CSF sampling (although all acute syndromes performed CSF sampling in the first 3 days of admission) and the absence of data on whether children received specific antipyretic drugs (e.g. diclofenac, mephenamic acid) that have been proposed to trigger ITES syndromes in the past, 2 although many of these drugs are not used in Australia to treat childhood fever so are unlikely to be relevant in this ITES cohort. A further major limitation is the invasive nature of CSF testing-although CSF is without doubt the best way of measuring the 'CNS environment', CSF testing is still invasive. Peripheral blood biomarkers that can detect similar brain processes would be preferable, if less likely to be as sensitive as CSF.
In conclusion, we provide data to show that CSF metabolites have potential utility in the diagnosis and therapeutic monitoring of ITES syndromes and have superiority over CSF pleocytosis as a biomarker.