Anti-leucine rich glioma inactivated 1 protein and anti-N-methyl-D-aspartate receptor encephalitis show distinct patterns of brain glucose metabolism in ¹⁸F-fluoro-2-deoxy-d-glucose positron emission tomography

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.

The brain FDG uptake from whole-body FDG-PETs of six anti-NMDA receptor encephalitis patients (6 females, median age 36.5, interquartile range 30.5-47.25, Table 1) and four anti-LGI1 encephalitis patients (4 males; median age 68.0, interquartile range 61–72.75, Table 2) admitted to Hannover Medical School between 2008 and 2012 was retrospectively analyzed.

Diagnostic criteria were the above mentioned typical clinical symptoms of limbic encephalitis and detection of either anti-LGI1 or anti-NMDA receptor IgG titer in serum or cerebrospinal fluid (CSF) of at least 1:10. Antibody detection was performed using Mosaik Biochips (Euroimmun, Lübeck, Germany) as described previously [25]. The clinical severity at the time of PET investigation and follow up was staged using the modified Rankin Scale (mRS). All patients with anti-LGI1 syndrome but only one patient with anti-NMDA receptor encephalitis were able to lie quietly in the PET scanner, therefore, five of the latter patients had to be investigated in propofol narcosis. The patient imaging data were compared to separate age and sex matched control groups without neuropsychiatric diseases who received PET due to an extracerebral neoplasm (anti-NMDA controls: 5 females, median age 40.0, interquartile range 37.5-46, anti-LGI1 controls: 4 males/2 females, median age 61.0, interquartile range 58.75-64.5). All controls were investigated without narcosis using the same PET acquisition protocol.

For positron emission tomography a Biograph LSO Duo PET/CT (Siemens, Erlangen, Germany) was used. One hour before scanning 350 MBq ¹⁸F-fluoro-2-deoxy-d-glucose (FDG) were injected intravenously. In patients not able to avoid movements during the scan while awake, anesthesia was started shortly before imaging, i.e. after a sufficient uptake phase of FDG. First a whole body low-dose computed tomography (CT)-scan was acquired from the mid-tight to the head using adjusted tube current (Care Dose4D). Subsequently emission scanning was performed in 3D mode typically over 8 bed positions of 4 minutes each. For the present study brain images were reconstructed separately with a matrix size of 256 × 256 and zoom factor 2 using a 2D OSEM (ordered subset expectation maximization) algorithm with 4 iterations and 8 subsets, smoothing with a 5-mm FWHM (full width at half maximum) filter and applying CT based attenuation correction.

Brain images were spatially normalized into the stereotatic standard space according to Montreal Neurological Institute using the Statistical parametric mapping 2 (SPM2) FDG brain template with default parameter settings (statistical parametric mapping, Wellcome Trust Centre for Neuroimaging, London, UK). Thereafter each data set was smoothed with an isotropic 3D Gaussian filter kernel with 10-mm FWHM. In group comparisons (e.g. patients vs. controls) proportional scaling to the cerebral global mean was employed and the two sample t-test to detect voxels/regions with either decreased or increased FDG uptake i.e. glucose metabolism. In single-subject analyses effects were assessed for each patient individually. Moreover, antibody titers and clinical scores were correlated to glucose metabolism as covariates at a group level. Statistical inferences were based on a p-value of 0.005 uncorrected for multiple comparisons and an extent threshold of 30 voxels. We additionally performed group analyses using pons and motor cortex as reference region for scaling leading to essentially the same results. For spatial assignment of significant changes in SPM analyses volume of interest (VOI) templates according to Cyceron or reflecting Brodmann areas were used. In order to detect changes in larger brain regions, small VOIs according to Cyceron were summarized to large VOIs (Additional file 1: Table S1) and statistical analyses for group differences (t-test) and correlations (linear regression analysis) were performed using JMP10 software (SAS Institute). A p-value < 0.05 was used for statistical inferences.

The demographic and clinical data of patients as well as their diagnostic procedures and therapy are summarized in Tables 1 and 2. The variability of PET time points in anti-NMDA encephalitis patients was caused mainly by late admission to Hannover Medical School in the disease course of some patients, however, 5/6 patients were considered to be at the clinical peak of encephalitis at the time of PET. They presented with typical signs and symptoms including generalized seizures which were treated successfully with antiepileptic drugs before PET investigation. At the time of PET the electroencephalography (EEG) did not record epilepsy specific potentials and the modified Rankin Scale (mRS) varied between 3 and 5 (median 4.5, interquartile range 3.75-5.0, n = 6).PET metabolic abnormalities of anti-NMDA receptor encephalitis patients did not correspond to magnet resonance imaging (MRI) lesions that were shown in 4/6 patients: Hippocampal T2-hyperintensities bilaterally with diffusion elevation (patient 1, Figure 1A-B), small single unspecific left paraventricular T2-hyperintensity without contrast enhancement or diffusion restriction (patient 3), multiple very small subcortical T2-hyperintensities with T1-contrast enhancement (patient 4) and faint T2-hyperintensities in both cingular gyri without contrast enhancement (patient 5). In patients 2 and 6 MRI did not display any lesions.

Although ovarian PET and ultrasound were unsuspicious, an exploratory oophorectomy [6, 26, 27] was initiated in 5/6 patients (mRS ≥ 4) after the consent of the caring person had been obtained. Histopathologically a microscopic mature ovarian teratoma containing neuronal tissue was detected in 2/5 patients. Clinical outcome at the last follow up after 8–40 months (mRS median 1.0, interquartile range 0.75-4.0, n = 6) was good with mRS 0–1 (4/6) or showed improvement to a moderate degree (1/6), while one patient did not respond to treatment.

Patients with anti-LGI1 syndrome presented with symptoms of a classic limbic encephalitis including cognitive deficits and focal seizures. Two patients showed faciobrachial dystonic seizures that were treated successfully with methylprednisolone and antiepileptic therapy [28]. Routine EEG at the time of PET did not record epilepsy specific potentials. PET tumor screening was negative for neoplasia within a range of 1–12 months after clinical onset, which corresponds to the time from onset to treatment initiation. Again this variability of time points was caused by late admission of two patients to Hannover Medical School, however, 3/4 patients were at the clinical peak of the anti-LGI1 encephalitis at the time of PET with a mRS of 2.PET abnormalities of anti-LGI1 syndrome patients did not correspond to their MRI lesions that were not enhancing after gadolinium application: Multiple microangiopathic T2-hyperintensities in the subcortical white matter (patient 1) and bitemporal T2-hyperintensities (patient 2, Figure 1C-D). At the time of PET, the MRI of patient 3 had not displayed any lesions yet, while we detected a right hippocampal T2-hyperintensity 4 months later corresponding to progressive cognitive deficits (memory, executive functions) and generalized seizures, patient 4 did not show MRI lesions.

After first line treatments, two patients received a continuous oral steroid therapy, while patient 3 was treated with cyclophosphamide at mRS of 3. In most patients clinical outcome at the last follow up after 9–31 months (mRS median 0.5, interquartile range 0.25-2.5, n = 4) was good with mRS 0–1 (3/4).

In comparison to the matched reference group FDG-PET of patients with anti-NMDA receptor encephalitis (Figures 2A, 3A) displayed a regionally limited hypermetabolism of bitemporal areas including hippocampus and parahippocampus (99–439 voxels within Brodmann areas 20, 21, 22, 30, 35, 36, 48, Table 3) as well as an extensive hypometabolism in the precuneus, the pre- and postcentral, the parietal and posterior cingulate cortex (503–1821 voxels within Brodmann areas 2, 3, 4, 6, 7, 17, 23, 26, 29, 30, 40, 43, Table 4). In patients with anti-LGI1 encephalitis (Figures 2B, 3B) we observed hypermetabolism in the cerebellum, basal ganglia as well as in precentral and occipital regions (191–2286 voxels in the cerebellum, putamen, pallidum and within Brodmann areas 6, 8, 17, 18, 19, Table 4) and hypometabolism limited to the anterior cingulate/frontomesial cortex (232 voxels within Brodmann areas 10, 11, 24, 32, Table 4). Brain regions and Brodmann areas showing significant abnormalities of glucose metabolism in patients with anti-LGI1 and anti-NMDA receptor encephalitis are summarized in Table 3 and Table 4. The results of SPM analysis for each individual patient are displayed in the supplemental material (Additional files 2 and 3: Figures S1 and S2 and Additional file 4: Table S2).

Direct comparison between the above mentioned types of encephalitis likewise revealed distinct patterns with a relative hypometabolism in the precuneus, precentral, parietal, occipital and cingulate cortex (338–2616 voxels within Brodmann areas 2, 3, 4, 5, 6, 7, 10, 19, 20, 23, 26, 30, 39, 40) as well as regionally limited hypermetabolism in frontotemporal areas (40–449 voxels within Brodmann areas 10, 11, 20, 21, 22, 25, 30, 36, 38, 47, 48) of anti-NMDA receptor encephalitis patients compared to the anti-LGI1 associated syndrome (Figures 2C, 3C).

In LGI1-patients the mRS at follow up showed a positive correlation with the precentral metabolism (i.e. increasing mRS values with increasing frontal metabolism left, r = 0.95, p = 0.048). Other assessments for correlations between brain metabolism and NMDA-/LGI1-antibody titers in serum/CSF or the mRS at follow up did not reveal consistent results of SPM and VOI analyses.

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.