Background

Pediatric encephalitis is inflammation of the brain parenchyma that is often complicated by significant morbidity and mortality. Encephalitis is accompanied by global dysfunction and encephalopathy, and inflammation manifesting as fever, seizures, cerebral spinal fluid (CSF) pleocytosis, and neuroradiologic abnormalities. A specific etiology is identified in only half of cases, with viral and autoimmune being most common [1]. In the remainder, encephalitis is often unexplained [1, 2]. While genetic variation is cited as contributing to the development of unexplained encephalitis, few specific loci have been implicated and their prevalence unknown [3].

While typically considered rare, neuroinflammatory disorders are impacted by genetics. For example, hemophagocytic lymphohistiocytosis (HLH) is a monogenic disorder characterized by a dysregulated hyperinflammatory cytotoxic NK and T cell response, where CNS involvement is a poor prognostic indicator [4]. In complex genetic disorders such as systemic lupus erythematosus (SLE), specific loci are associated with CNS involvement [5]. Additionally, TLR3 mutations have been associated with severe herpes simplex encephalitis [6]. Subsequently, we hypothesize that children with unexplained encephalitis may carry variants associated with neuroimmunologic conditions predisposing to CNS inflammation.

Methods

The study was approved by the University of Pittsburgh’s Institutional Review Board (#20010099). Written informed consent was obtained from one or more parents/guardians for minors, and from participants over 18 years of age. For minor participants, child assent was garnered when able. Patients admitted to the Children’s Hospital of Pittsburgh Intensive Care Unit between June 2018–November 2021 with a diagnosis of acute encephalitis were eligible if they displayed altered consciousness, cognition, personality, or behavior for >24 h and had ≥2 of the following: (1) fever; (2) CSF WBC > 4 cells/ul; (3) neuroradiologic evidence of inflammation; (4) seizures not attributable to known diagnoses; (5) focal neurologic signs; and (6) encephalopathy on EEG. Enrollees were a limited convenience sample of eligible participants. Ten participants had unexplained encephalitis and one subject with anti-NMDA receptor encephalitis was included.

Genetic sequencing

Whole genome sequencing (WGS) was performed on DNA extracted from whole blood at the University of Pittsburgh Institute for Precision Medicine on Illumina’s NovaSeq 6000 with a mean coverage of 43.7x. FASTQ files were aligned to homo sapiens reference sequence GRCh38. Resultant VCF files were analyzed in the Fabric Genomics Opal 5.2.2 software [7] to identify missense, nonsense or frameshift mutations. Variants were filtered for coverage >10, PHRED score >30.

We limited candidate variants to a 515 genes list, combining monogenic inborn errors of immunity classified by the International Union of Immunologic Societies [8] and a validated sequencing panel for pediatric neuroinflammation [9]. Next, candidates were restricted to rare variants with a minor allele frequency <5% (MAF < 0.05) in the ExAC [10] database. All rare variants were evaluated in Qiagen’s Professional Human Genetic Mutation Database (HGMD), which classifies variants’ pathogenicity based on peer-reviewed reports of the variant in human disease [11]. These reports were manually reviewed for inheritance pattern and presence of the corresponding neurologic/inflammatory phenotype consistent with Online Mendelian Inheritance in Man [12]. Thus, only variants classified as deleterious or potentially deleterious with respect to the particular phenotype of interest were included.

Candidate variants were classified into three groups. In group 1, variants were limited to those reported as deleterious (DM) or potentially deleterious (DM?) in the HGMD professional database [11] and found in a disease-consistent inheritance pattern (one variant for autosomal dominant (AD) or X-linked (XL) disorders in males, and two variants for autosomal recessive (AR) disorders). The DM? designation indicates an uncertain linkage of variant and phenotype, and represents a potential rather than definitive association. Group 2 included heterozygous DM or DM? AR variants, and variants of unknown significance (VUS) if another DM or DM? AR variant was identified at the same locus. Finally, group 3 variants were disease risk polymorphisms in HGMD.

Results

Ten of 11 subjects were previously healthy prior to admission and one had a history of epilepsy (Table 1). Participants were between 10 months to 18 years of age. Clinical diagnoses included anti-NMDA receptor, parainfectious, limbic, and acute necrotizing encephalitis, and febrile illness-related epilepsy syndrome. CSF WBC was elevated in 8 of 11 (73%) participants. In total, 64% had seizures and 100% showed EEG slowing consistent with encephalopathy. MRI findings consistent with CNS inflammation were found in 91% of participants. Notably, 3 of 11 (27%) had poor neurologic outcome, with Pediatric Cerebral Performance Score ≥ 4 (severe disability, coma/vegetative state or death) at between 4 months to 8 years of follow-up.

Table 1 Clinical characteristics of study participants.
Full size table

In total, six patients (55%) had a DM or DM? variant with an AD inheritance (Table 2, Group 1). These included atypical hemolytic uremic syndrome (aHUS): CFI p.Pro64Leu, CD46 p.Ala353Val, and CFHR5 p.Gly145Glu; common variable immunodeficiency (CVID): TCF3 p.Lys101Arg; congenital neutropenia: ELANE p.Pro257Leu; aplastic anemia: TERT p.His412Tyr; autoimmune lymphoproliferative syndrome (ALPS): CASP10 p.Pro501Leu; and familial Mediterranean fever: CARD14 p.Q422K. Five patients (45%) were compound or synergistic heterozygotes for an AR condition (Table 2, Group 2). These included CVID: LRBA p.Met467Val, SKIV2L p.Arg324Trp, and PSMB9 p.Arg173Cys; HLH: UNC13D p.Arg928Cys and p.Met795Thr (VUS); Primary immunodeficiency: IL21R p.Gly345Ser and p.Leu329Val (VUS); Cerebrotendinous xanthomatosis: CYP27A1 p.Pro384Leu; and Neuronal ceroid lipofuscinosis: CLN6 p.Gly259Ser. Six subjects (55%) carried a risk polymorphism (Group 3), including two related to SLE: DNASE1 p.Arg2Ser and p.Gly127Arg. Other risk polymorphisms included Crohn’s disease: NOD2 p.Leu248Arg; FCN3 deficiency: FCN3 p.Leu117SerfsTer65; Increased IL-6, TNF-ɑ, Ig levels: CD40 p.Pro227Ala; MASP2 deficiency: MASP2 p.Asp120Gly; Reduced apoptotic function: CASP10 p.Tyr446Cys; and leukoencephalopathy with brainstem and spinal cord involvement and lactate elevation: DARS2 p.Gly338Gln.

Table 2 Identified variants according to pathogenicity classification.
Full size table

Discussion

In this pediatric encephalitis study, DM and DM? immunologic variants affected 64% of participants, possibly suggesting a genetic contribution to CNS inflammation. Several study participants had variants that impact adaptive immunity, suggesting either an autoimmune or increased susceptibility mechanism. Our data also suggest complex genetic risk, with 64% carrying multiple variants at diverse immunologic loci. However, It must be emphasized that currently, the link between immunogenetic variation and pediatric encephalitis is only associative. Further efforts to establish causality will require additional research at the population and variant level. For some variants, these follow up studies will likely demonstrate that variants are incidental findings. Others may represent true causal relationships, emphasizing the need for genome wide sequencing to categorize the full landscape of genetic risk.

For example, sequencing of Patient 6—who has anti-NMDA receptor encephalitis, caused by autoantibodies to the glutamate receptor—demonstrated risk variants for SLE, another autoantibody-mediated disease. Patient 6 also had a DM? complement variant. Classic activation of the complement innate immune pathway can be triggered by autoantibodies [13]. In CNS SLE, terminal complement complexes are elevated in CSF [14] and murine models suggest that blood brain barrier dysfunction can be inhibited by complement antagonists [15]. Alternatively, complement variants may impair immune complex clearance, a mechanism previously implicated in a case of refractory anti-NMDA receptor encephalitis with genetic complement abnormalities [16]. However, while anti-NMDA antibodies activate complement in vitro, CNS complement deposition has not been demonstrated in vivo [17, 18]. These direct relationships lend biologic plausibility to our findings in other cases of unexplained encephalitis with less well-characterized molecular pathology.

Patient 2 also had variants in CFI and CD46, complement pathway downregulators. In aHUS, gain of function variants in complement activators, or loss of function variants in regulators lead to hyperactivation, endothelial damage, and organ injury—most commonly in the kidneys—but affecting the CNS in 25% of individuals [19, 20]. A 10 year old with recurrent hemorrhagic leukoencephalitis and CFI p.Pro64Leu, the same variant found in patient 2 in our study, had undetectable CFI, low C3, and low AP50 levels [21]. However, this individual also carried CFI p.Gln88Lys. CFI variants have also been described in sterile encephalitis, associated with C3 activation and terminal complement complex deposition on brain biopsy [22]. Together, this leads us to hypothesize that genetic risk for inappropriate humoral immune response and subsequent complement activation may contribute to unexplained encephalitis in a subset of children, however this will require future functional confirmation.

Patients 7 and 10 also showed shared genetic risk, both carrying CASP10 variants, a gene which promotes lymphocyte apoptosis and, in ALPS, leads to uncontrolled proliferation and autoimmunity. We were unable to identify studies linking CASP10 or ALPS with encephalitis. However, for both patient 7 and patient 10 and patient 2 and patient 6, shared genetic risk in this small cohort is raises a question of possible shared pathobiology that warrants further study.

Additionally, unique genetic findings were encountered. Patient 2 was compound heterozygous for an UNC13D DM? and a VUS variant, potentially consistent with HLH—an AR multisystem inflammatory disorder due to impaired cytotoxic killing. HLH has been hypothesized to manifest with isolated CNS involvement [4], as seen in our patient. In the literature, a 3 month old with complex genetics involving multiple variants in PRF1, UNC13D, STXBP2 and XIAP had elevated CNS protein and neuroimaging findings consistent with acute necrotizing encephalitis [23], as observed in patient 2. Other unique findings include patient 5 who suffered devastating neurologic injury following nasopharyngeal adenoviral infection, who was compound heterozygous for an IL21R DM? and a VUS variant. Biallelic IL21R mutations cause immunodeficiency affecting both T and B cell compartments, with impaired immunoglobulin synthesis, T and NK cell dysfunction, and recurrent viral infections [24, 25].

Our study’s main limitation is that identified variants cannot be equated with immunodeficiency, as literature-based stratification may misclassify pathogenicity. Further, genotype/phenotype correlations in the cohort may be atypical where CNS manifestations are primary, possibly due to variable penetrance, environmental factors and genetic background influencing phenotype expression. Additionally, it is difficult to estimate the frequency of implicated variants in healthy controls to determine if they are overrepresented in encephalitis. However, frequency limits and reports of pathogenicity guard against false positives. Another limitation is the lack of parental sampling which prevents determination of cis and trans positioning. Our study also did not perform confirmatory Sanger sequencing, and the filter also fails to identify regulatory, structural and copy number variants which may contribute to disease. Lastly, as sequencing was performed retrospectively, it was not possible to perform additional functional and immunologic testing on participants.

In this case series, we used WGS to identify immunogenetic risk in 8 of 11 children with unexplained CNS inflammation. As a small exploratory study, this report is hypothesis-generating and warrants larger studies that include functional testing to understand the prevalence and impact of immunogenetic variation in unexplained pediatric encephalitis.