Background
Limbic encephalitis involving the temporomedial lobes and amygdalae is characterized by subacute memory impairment, seizures and neuropsychiatric symptoms with variable evidence of cerebrospinal fluid inflammation, anti-neuronal antibodies and a paraneoplastic origin [
1‐
4].
The two most common targets of encephalitis associated pathogenic autoantibodies are the recently identified leucine rich glioma inactivated 1 protein (LGI1), which is extracellularly complexed with voltage-gated potassium channels (VGKC), and the subunit 1 (NR1) of the N-methyl-D-aspartate (NMDA) receptor [
5‐
8]. LGI1 antibodies associated with limbic encephalits specifically inhibited the ligand-receptor interaction between LGI1 and ADAM22 (disintegrin and metalloproteinase domain-containing protein 22) and reversibly reduced synaptic AMPA (α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid) receptor clusters in rat hippocampal neurons [
9]. Cerebrospinal fluid samples or purified immunoglobulin G (IgG) from patients with anti-NMDA receptor encephalitis led to a marked reduction of NR1 (and NR2B) surface expression and NMDA receptor mediated currents in hippocampal cultures [
10,
11] as well as to increased corticomotor hyperexcitability in rats [
12,
13].
While the anti-LGI1 syndrome predominantly reminds of a classic limbic encephalitis with amnesia, (faciobrachial dystonic) seizures and psychiatric manifestations, the anti-NMDA receptor encephalitis is characterized by memory deficits, psychiatric symptoms with psychotic and catatonic features, language disintegration, dyskinetic movements, seizures, decreased consciousness, autonomic and breathing instability that often requires intensive care treatment [
6,
7,
13‐
15]. However, there may be clinical overlaps of the two entities. A tumor association is infrequently found in the anti-LGI1 syndrome, whereas ovarian or other teratoma were diagnosed in up to 50% of anti-NMDA receptor encephalitis patients [
4,
6,
13,
16].
In order to detect underlying tumor manifestations, whole-body
18F-fluoro-2-deoxy-d-glucose positron emission tomography (FDG-PET) is performed in many patients with limbic encephalitis, whereby cerebral FDG uptake can also be determined. Some case reports have published FDG-PET data showing brain metabolic abnormalities of varying degree and localization in adult patients with anti-LGI1 [
17] or anti-NMDA receptor encephalitis [
16,
18‐
22]. Recently, an FDG-PET study revealed a frontal and temporal hypermetabolism associated with occipital hypometabolism in six patients with anti-NMDA receptor encephalitis [
23], while a hypermetabolism in the medial temporal lobes and basal ganglia was detected in ten anti-LGI1 encephalitis patients [
24]. The aim of our study was to compare cerebral FDG uptake of whole-body FDG-PET imaging in patients with anti-LGI1 and anti-NMDA receptor encephalitis for detailed analysis of brain metabolic disease patterns that may lead to an improved diagnostic accuracy.
Discussion
This retrospective study showed distinct brain metabolic patterns in FDG-PETs of anti-LGI1 and anti-NMDA receptor encephalitis patients. Analysis of the anti-LGI1 group revealed hypermetabolism in cerebellar, basal ganglia, occipital and precentral regions and hypometabolism limited to the anterior cingulate/frontomesial cortex, whereas in anti-NMDA receptor encephalitis we found hypermetabolism in frontotemporal areas and widespread hypometabolism in parietal lobes.
The imaging results for anti-NMDA receptor encephalitis patients are in agreement with a recent FDG-PET study demonstrating frontotemporal hypermetabolism and occipital hypometabolism [
23] despite considerable methodological differences. Our anti-NMDA receptor encephalitis patients with mRS ≥ 4 at the time of PET (n = 5) had to receive propofol narcosis during the whole-body tumor scan to prevent severe artifacts by involuntary movements, while Leypoldt
et al. managed imaging acquisition in patients showing similar initial mRS (median 4.5) without sedation or narcosis (F. Leypoldt, personal communication).
The tracer FDG is taken up by active brain neurons as if it was glucose and is then metabolized in the cells to FDG-6-phosphate emitting radioactivity that can be measured using PET. Uptake and metabolic trapping of FDG in the brain as FDG-6-phosphate is completed to 80–90% 32 minutes after the intravenous injection [
29]. In an FDG-PET study with healthy volunteers receiving narcosis 25 min before FDG administration and during PET scanning, the regional glucose metabolic rate was reduced during propofol anaesthesia in all brain areas to 48–66% (p < 0.01) with highest significant reductions in the occipital lobe, the lingual gyrus, parietal lobe, temporal lobe and thalamus [
30]. Although we applied FDG one hour before initiation of narcosis and PET scanning, we cannot completely rule out a minor hypometabolic propofol effect on brain metabolism particularly in temporal and parietal areas of our anti-NMDA receptor encephalitis patients. In most affected temporal regions including hippocampus and parahippocampus an underestimation of encephalitis-induced hypermetabolism due to propofol narcosis seems possible. In previous FDG-PETs without sedation or narcosis, frontotemporal hypermetabolism was more pronounced in anti-NMDA receptor encephalitis with similar mRS [
23] suggesting a minor hypometabolic propofol effect in these brain areas of our patients.
VGKC-complex positive encephalitis patients without LGI1 antibodies compose a more heterogenous group than patients with LGI1 encephalitis. In a recent PET study [
31], only 1/7 VGKC-positive patients also showed autoantibodies against LGI1 leading to hypometabolism in the association cortices. In the remaining six patients, two scans were rated as normal and four showed different findings mostly involving the basal ganglia [
31] suggesting a heterogeneous pattern of brain glucose metabolism in the VGKC-positive subtypes of limbic encephalitis. In previous case reports, the FDG-PETs in encephalitis patients with autoantibodies against the VGKC-complex – LGI1 antibodies were not determined – indicated bilateral temporomesial hypermetabolism and/or temporal hypometabolism depending on the course of the disease [
32‐
35].
The glycoprotein LGI1 is secreted from presynaptic terminals and associates with synaptic Kv1 VGKCs [
36]. LGI1 is highly expressed in the hippocampus and the neocortex, and mutations in the LGI1 gene cause autosomal dominant lateral temporal lobe epilepsy [
13,
37,
38]. LGI1 antibodies have been detected predominantly in limbic encephalitis, epilepsy and few patients with Morvan’s disease [
5,
13,
36]. Furthermore, LGI1 antibodies were detected in a glutamic acid decarboxylase antibody-positive patient suffering from progressive encephalomyelitis with rigidity and myoclonus (PERM) [
39].
Recent case reports demonstrated hypermetabolism in the basal ganglia as well as in the left hippocampus and amygdala in two patients 3–8 months after clinical onset of encephalitis associated with LGI1-antibodies [
17,
40]. Very lately Shin
et al. demonstrated temporal and bilateral basal ganglia hypermetabolism in 7/10 anti-LGI1 encephalitis patients (3 days to 2 years between symptom onset and diagnosis), which had not been compared to a matched control group [
24], suggesting that the anti-LGI1-induced brain metabolic pattern may depend on the disease course, time point of diagnosis/therapy initiation and treatment regimen.
Our anti-LGI1 patients were investigated without propofol narcosis by FDG-PET 1–12 months after clinical onset showing only minor hypermetabolism in temporal areas, whereas hypermetabolic activity was most pronounced in cerebellar, basal ganglia, occipital and precentral regions. Interestingly, the precentral hypermetabolism correlated positively with the mRS at follow up suggesting that increased metabolism in this area may represent a negative prognostic factor in LGI1 patients. The anti-LGI1 metabolic pattern contrasts the PET results of patients with anti-NMDA receptor encephalitis underlining these two etiologically different subtypes of autoimmune limbic encephalitis. In addition to the autoantibody detection as most important diagnostic feature, FDG-PET may prove to be a useful imaging tool, besides whole-body tumor search, for differentiating these subtypes of limbic encephalitis.
Two distinct brain metabolic patterns in immunologically mostly undefined (7/9) limbic encephalitis patients have been described in a recent FDG-PET study [
41]. Five younger patients including the two NMDA-receptor antibody positive cases displayed a mixed metabolic pattern characterized by hypermetabolism in the temporal and orbitofrontal cortex as well as occipital hypometabolism. While this pattern in younger patients reminds of brain metabolic changes seen in limbic encephalitis induced by NMDA receptor IgG-antibodies [
23], four older patients presented with subacute cognitive decline showing diffuse cortical hypometabolism closely resembling Alzheimer’s disease or dementia with Lewy Bodies [
41]. This neurodegenerative-like hypometabolism is neither likely to reflect the cognitive decline induced by NMDA receptor IgA-antibodies leading to occipital hypermetabolism [
42] nor the anti-LGI1 syndrome where Shin
et al. and we found predominantly hypermetabolic changes.
Conclusions
In conclusion, our retrospective FDG-PET study provides novel evidence for distinct brain metabolic patterns in patients with anti-LGI1 and anti-NMDA receptor encephalitis. In anti-NMDA receptor encephalitis the regionally limited hypermetabolism in frontotemporal areas contrasted extensive hypometabolism in parietal lobes, whereas the anti-LGI1 syndrome is characterized by hypermetabolism in cerebellar, basal ganglia, occipital and precentral areas and minor frontomesial hypometabolism.
Ethical standard
We obtained an approval from the local Ethics Committee of Hannover Medical School (No. 1625–2012) and patients or their carers gave their written informed consent.
Competing interests
The authors declare that they have no competing interest.
Authors’ contributions
FWe, FWi, PR, CH, GB, EN: study design and data interpretation. FWe, FWi, PR, SBT, CH, RRR, GB, EN: data analysis. FWe, SBT, ALB, CT, MS, EV, CS, FL, JA, AL: patient selection and treatment. FWe, FWi, PR, RD, LG, FMB, GB, EN: manuscript drafting. All authors read and approved the final manuscript.