1 INTRODUCTION

Part of a group of severe but treatable neurological diseases, anti-gamma-aminobutyric-acid B (GABAB) receptor encephalitis is a novel form of autoimmune encephalitis (AE) associated with antibody to GABAB receptor of cell surface.1, 2 The majority of patients with anti-GABAB receptor encephalitis present with new-onset seizures, cognitive deficits, and mental and behavioral disorders, all with or without the presence of underlying small cell lung cancer (SCLC).3 Early recognition of anti-GABAB encephalitis is of vital importance because most patients respond well to timely treatment.4-6 Current diagnostic criteria depend highly on positive GABAB antibody in serum or cerebrospinal fluid (CSF), which can lead to false negatives or unavailable results, causing diagnostic difficulties and treatment delay.7 Thus, it is necessary to consider an additional novel biomarker in early diagnosis and prognostic evaluation for anti-GABAB receptor in encephalitis patients. Although neuroimaging plays a key role in the routine evaluation of neurological diseases,8 it has received little attention for anti-GABAB receptor encephalitis. Some prior studies have shown that magnetic resonance imaging (MRI) in these patients have mainly presented with hyperintensity signals in medial temporal lobe (MTL), but only 18%–50% patients have abnormal and specific results.3, 9, 1018F-fluoro-2-deoxy-d-glucose positron emission tomography (18F-FDG-PET) is frequently used for whole-body tumor screening, but recently, it has been reported to demonstrate MTL hypermetabolism in patients with anti-GABAB receptor encephalitis, especially when MRI was negative.11 Nevertheless, only a limited number of isolated cases have suggested that 18F-FDG-PET might be useful for early diagnosis in subjects with anti-GABAB receptor encephalitis, and there are no current data available to determine the sensitivity of brain 18F-FDG-PET in encephalitis associated with GABAB receptor antibody.11-14 Moreover, the majority of previous studies of 18F-FDG-PET in anti-GABAB receptor encephalitis have been restricted to qualitative characterization of FDG-PET findings.15 In addition, to our knowledge, there is no systematically relevant study to evaluate the correlation between PET index and clinical prognosis in patients with anti-GABAB receptor encephalitis. Overall, the metabolic pattern and prognostic role of 18F-FDG-PET in patients with anti-GABAB receptor encephalitis are still not well described.

To address these unclear questions, we conducted a semi-quantitative study reviewing the 18F-FDG-PET data of 21 patients with a definite diagnosis of anti-GABAB receptor encephalitis. One aim of this study was to identify the 18F-FDG-PET pattern of anti-GABAB receptor encephalitis, especially in those with unremarkable MRI changes. Further, this study also sought to find an imaging prognostic predictor by evaluating the association between metabolic signature and prognosis in anti-GABAB receptor encephalitis.

2 METHODS 2.1 Study participants

The study was approved by the ethics committee of the Beijing Tiantan hospital that was affiliated to the Capital Medical University of the People's Republic of China. The study was conducted in accordance with the Declaration of Helsinki, and all patients and controls provided informed consent for the use of their medical records.

A total of 21 patients with GABAB receptor encephalitis were retrospectively reviewed between March 2015 and December 2020 at the Department of Neurology in the Beijing Tiantan Hospital of the Capital Medical University. All patients underwent MRI and 18F-FDG-PET at first hospitalization and fulfilled the included clinical diagnostic criteria based on representative clinical symptoms and the presence of purely positive GABAB receptor antibodies in the serum and cerebrospinal fluid (CSF). The neuroimaging data (MRI and PET) carried out in the acute stage of encephalitis after symptom onset were included. The demographic, clinical presentation, laboratory testing, and electroencephalograph (EEG) were collected by searching the electronic medical records.

For individual or group analysis of 18F-FDG-PET, the clinical cohort was matched for age and gender with 30 controls (19 males, 11 females; median age, 52 years; interquartile range [IQR], 43–56 years; range, 33–66 years) who had normal neurological assessment and functional metabolism from Beijing Tiantan Hospital of the Capital Medical University.

2.2 Antibody test

All patients underwent serum and CSF antibody detection, including classical paraneoplastic antibodies (Hu, Yo, Ri, Ma2, CV2, amphiphysin), N-methyl-D-aspartate receptor (NMDAR), leucine-rich glioma inactivated-1 (LGI1), contactin-associated protein-2 (CASPR2), GABAB receptor, α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid receptor (AMPAR), and glutamic acid decarboxylase 65 (GAD65). Serum and CSF samples were tested for the presence of isolated GABAB receptor antibodies, using both cell-based assays (Euroimmun, Lübeck, Germany) and immunohistochemical analyses in the neuroimmunology laboratory of the Peking Union Medical College Hospital.

2.3 MRI

All MRI examinations were acquired using a 3.0 Tesla Siemens Trio MRI scanner (Siemens Healthcare). The standard MRI protocols included T1-weighted imaging [T1WI, repetition time (TR) = 2000ms, echo time (TE) = 20 ms, field of view (FOV) = 250 mm × 220 mm, matrix = 400 × 250], T2-weighted imaging (T2WI, TR = 6,200 ms, TE = 90 ms, FOV = 240 mm × 220 mm, matrix = 3,850 × 385), fluid-attenuated inversion recovery (FLAIR, TR = 11,000 ms, TE = 120 ms, FOV = 250 mm × 221 mm, matrix = 240 × 160), and diffusion-weighted imaging (DWI, TR = 2,600 ms, TE = 60 ms, FOV = 230 mm × 230 mm, matrix = 140 × 130). MRI results were independently evaluated by two experienced neurologists and radiologists (Qun Wang, Lin Ai). In case of obvious discordance in their initial evaluations, an informed consensus statement was reached.

2.4 18F-FDG-PET acquisition and statistical parametric mapping (SPM) analysis

18F-FDG-PET was performed according to previously published methods.16 PET images were acquired using a PET/CT scanner (Elite Discovery, GE HealthCare). All patients fasted for at least 6 h, and fasting blood glucose levels could not exceed 8 mmol/L. No patients received neuroleptic drugs before undergoing FDG-PET. 18F-FDG was intravenously injected at a dose of 3.7–5.0 MBq/kg within 1 min, and subsequent uptake required that patients be in a quiet resting status for 1 h prior to scanning in a dedicated room after 18F-FDG injection. First, a low-dose CT scan (120 kV, 60–180 mA/s, slice thickness 3.75 mm) was performed. The PET scan was subsequently performed, with a whole-body (including brain region) FDG-PET scanning acquired for approximately 30–35 min. The brain imaging data were reconstructed into trans-axial slices with a matrix size of 128 × 128 and a slice thickness of 3.3 mm.

After acquisition of PET images, all data were preprocessed by SPM12 software implanted in a MATLAB 2018a environment (MathWorks Inc.). The pre-processing steps were as follows: first, the PET images were segmented and spatially normalized into a common Montreal Neurological Institute (MNI) atlas anatomical space following a 12-parameter affine transformation and non-linear transformations, yielding images composed of 2 mm × 2 mm × 2 mm voxels. Then, default SPM smoothing was applied using 12-mm Gaussian kernel to increase the signal-to-noise ratio. For single subject analysis, the statistical basic models were performed between individual patient and controls using two-sample t test model with age and gender as the nuisance variables. For groupwise analysis, FDG uptake was compared voxel-by-voxel between patients and controls group using a two-sample t-test of SPM. Significant results were viewed at the height threshold (p < 0.001) and corrected for multiple comparisons (familywise error [FWE] corrected or false discovery rate [FDR] corrected, p < 0.05). If significant clusters were not found, the more liberal threshold was considered (p < 0.001, uncorrected; extent threshold, k = 300).17

2.5 Follow-up and prognosis analysis based on SPM

The follow-up and clinical outcome information was obtained from outpatient visits and telephone interviews with patients or relatives. The modified Rankin Scale (mRS) was used to assess neurological disability at the last follow-up in patients with anti-GABAB receptor encephalitis3, 18, 19; patients were considered to have a good outcome if mRS score was ≤2 and poor outcome was defined as mRS score >2. In order to assess the potential imaging predictors on 18F-FDG-PET that might influence the long-term outcome in subjects with anti-GABAB receptor encephalitis, we performed group comparisons of 18F-FDG-PET data between patients with poor outcomes (n = 8) and good outcomes (n = 13) by two-sample t-test model of SPM12. The statistically significant and corrected method of multiple comparisons were same as aforementioned standards.

2.6 Statistical analysis

SPSS 22.0 software package for Windows (IBM Corp.) and Prism 7 (GraphPad software) were used for statistical analyses. The normality of data distribution of continuous variables was tested by one-sample Shapiro-Wilk test. Continuous variables with a normal distribution were presented as the mean ± standard deviation, and non-normal variables were expressed as the median (interquartile range [IQR]). Continuous variables were compared using the t-test or non-parametric Mann–Whitney U-test. Categorical variables were compared and analyzed by Fisher's exact test. With the exception of SPM analysis, a two-tailed p value less than 0.05 (p < 0.05) was considered statistically significant.

3 RESULTS 3.1 Patient characteristics

The clinical details of 21 patients (15 male, 6 female) with anti-GABAB receptor encephalitis are shown in Table 1. The median age of disease onset was 50 years (IQR, 46–64; range, 28–69 years), and seizures (21/21, 100%) were the most common clinical symptoms. In the population, 50% of patients showed more than one seizure types, and the main seizure classification was as follows: focal onset with awareness (14%), focal onset with impaired awareness (48%), generalized onset seizures (95%), and status epilepticus (SE, 14%). Other clinical presentations mainly included memory loss (n = 19, 90%), psychosis and abnormal behaviors (n = 13, 62%), sleep disturbances (n = 10, 48%), speech disorders (n = 2, 10%), and reduced level of consciousness (n = 2, 10%). Tumors were identified in 7 of 21 patients (33%). Both serum and CSF GABAB receptor antibody were observed in 16 patients (76%); the remaining 5 patients had GABAB receptor antibodies either in the serum (2/21, 10%) or in the CSF (3/21, 14%). All serum and CSF samples were negative for other neuronal cell-surface antibodies (NMDAR, LGI1, CASPR2, AMPAR), whereas Hu antibodies were found in 3 patients with GABAB antibodies; other intracellular antibodies (Yo, Ri, Ma2, CV2, GAD65, and amphiphysin) were negative. The CSF pleocytosis was observed in 16 patients (76%), and protein concentration was elevated in 11 of 21 patients (52%). 71% of patients had positive oligoclonal bands. EEG features of all anti-GABAB receptor encephalitis patients during the ictal and interictal phase were reviewed; a total of 16 patients (76%) experienced EEG abnormalities, which mainly included slow wave activities and epileptic discharges in the temporal regions.

TABLE 1. Patient characteristics (n = 21) Clinical variables Values Age at onset, median (IQR, range), year 50 (46–64, 28–69) Male, n (%) 15 (71%) Clinical symptoms during disease course, n (%) Seizures 21 (100%) Generalized seizures 20 (95%) Focal seizures with impaired awareness 10 (48%) Focal seizures without impaired awareness 3 (14%) Status epilepticus 3 (14%) Memory loss 19 (90%) Psychosis and abnormal behaviors 13 (62%) Sleep disturbances 10 (48%) Speech disorders 2 (10%) Reduced level of consciousness 2 (10%) Tumors, n (%) 7 (33%) GABAB receptor antibody, n (%) 21 (100%) Both in serum and CSF 16 (76%) Only in serum 2 (10%) Only in CSF 3 (14%) Associated paraneoplastic antibodies, n (%) Hu 3 (14%) CSF abnormalities at onset, n (%) Pleocytosis 16 (76%) Elevated protein levels 11 (52%) Oligoclonal bands 15 (71%) EEG abnormalitiesa, n (%) 16 (76%) Initial MRI results, n (%) Normal 12 (57%) MTL T2/FLAIR hyperintensities 9 (43%) Unilateral 5 (24%) Bilateral 4 (19%) Treatment, n (%) Immunotherapy 21 (100%) Tumor removal or chemotherapy 7 (33%) Follow-up and outcome Follow-up time from onset, median (IQR), m 33 (16–52) mRS at the last follow-up, median (IQR) 2 (1–4) Good outcome (mRS 0–2), n (%) 13 (62%) Poor outcome (mRS 3–6), n (%) 8 (38%) Abbreviations: CSF, cerebrospinal fluid; EEG, electroencephalogram; FLAIR, fluid-attenuated inversion recovery; GABAB, gamma-aminobutyric-acid B; IQR, interquartile range; MRI, magnetic resonance imaging; mRS, modified Rankin Scale; MTL, medial temporal lobe. a Temporal area slow waves or epileptiform discharges were considered abnormal. 3.2 Comparisons of MRI and 18F-FDG-PET findings

The MRI and 18F-FDG-PET findings in subjects with anti-GABAB receptor encephalitis are summarized in Table 2. The brain MRIs were performed at a median of 34 days (IQR: 14–54 days) after disease onset and a median of 42 days (IQR: 34–63 days) from 18F-FDG-PET. There was no difference in the duration of symptoms to imaging between MRI and PET (p = 0.107). The cerebral MRI showed abnormal findings in 9 of 21 patients (42.9%), whereas abnormal metabolic patterns on 18F-FDG-PET were seen in 17 of 21 patients (81% vs. 42.9%, p = 0.025). Importantly, 18F-FDG-PET was 100% positive when MRI was positive, and 18F-FDG-PET was 67% positive even if there were no associated abnormalities on MRI (Figure 1A). For 12 patients with normal MRI, 5 patients (5/12, 41.7%) showed MTL hypermetabolism on 18F-FDG-PET (Table 2, Figure 1B, C). Of the 17 subjects with abnormal PET metabolism, the alter limbic lobe glucose metabolism (mostly hypermetabolism) was observed in 14 patients (14/21, 66.7%), of whom 10 (10/14, 71.4%) demonstrated hypermetabolism in the medial temporal lobe (MTL). 18F-FDG-PET demonstrated isolated cerebral hypermetabolism in 4 of 21 patients (19%), including brain regions in MTL, inferior temporal gyrus, basal ganglia, and cingulate gyrus. Isolated hypometabolism was noted in four patient (19%), which was related to cingulate gyrus and thalamus. Of the 21 patients, 9 (42.9%) revealed mixed metabolism patterns with hypermetabolism in combination with hypometabolism. Of those 9 patients, 7 (77.8%) with mixed metabolism mainly presented with a relatively common metabolic pattern consisting of MTL hypermetabolism and relative cortex hypometabolism (frontal and parietal lobe) (FWE corrected, p < 0.05; Figure 2A). In addition, groupwise analysis also confirmed MTL hypermetabolism in association with relative hypometabolism in frontal or parietal lobe, extending to cingulate gyrus, was a general metabolic pattern in subjects with anti-GABAB receptor encephalitis (FDR corrected, p < 0.05; Figure 2B).

TABLE 2. Detailed MRI and 18F-FDG-PET features in patients with anti-GABAB receptor encephalitis (n = 21) Patient Sex/Age Time from onset to MRI (days) MRI results Time from onset to 18F-FDG-PET (days) PET results of SPM analysis T2/FLAIR hyperintensities Hypermetabolism Hypometabolism 1 M/62 31 Normal 35 MTL Parietal lobe 2 F/69 54 MTL 35 Middle Temporal Gyrus, Parietal lobe Frontal lobe 3 M/62 65 Normal 43 MTL —— 4 M/45 35 MTL 37 MTL Parietal lobe 5 F/54 6 Normal 9 —— —— 6 M/69 15 Normal 48 MTL Frontal lobe, parietal lobe, transverse temporal gyrus, cingulate gyrus 7 F/32 3 Normal 25 —— Cingulate gyrus 8 M/49 12 Normal 39 MTL, Inferior temporal gyrus —— 9 F/50 29 Normal 42 —— —— 10 M/28 27 MTL 65 —— Cingulate gyrus 11 M/50 68 MTL 50 MTL Parietal lobe, cingulate gyrus 12 M/62 1 Normal 41 —— —— 13 F/67 4 Normal 25 MTL Frontal lobe, parietal lobe 14 M/44 35 Normal 25 —— Thalamus 15 M/47 80 MTL 70 MTL —— 16 M/45 86 MTL 112 Basal ganglia Parietal lobe 17 M/47 74 MTL 76 —— Cingulate gyrus 18 M/57 53 Normal 68 —— —— 19 M/65 23 MTL 46 MTL Frontal lobe 20 M/69 45 MTL 61 MTL, Frontal lobe Parietal lobe, inferior temporal gyrus 21 F/50 35 Normal 33 Basal ganglia, cingulate gyrus —— Abbreviations: 18F-FDG-PET, 18F-fluoro-2-deoxy-d-glucose positron emission tomography; F, female; FLAIR, fluid-attenuated inversion recovery; GABAB, gamma-aminobutyric-acid B; M, male; MRI, magnetic resonance imaging; MTL, medial temporal lobe; SPM, statistical parametric mapping.

Comparisons between MRI and 18F-FDG-PET in patients with anti-GABAB receptor encephalitis. (A) Neuroimaging testing plays an essential role in the diagnosis of anti-GABAB receptor encephalitis. The sensitivity of 18F-FDG-PET is higher than that of MRI; (B) representative images in one patient with anti-GABAB receptor encephalitis (patient #3): 18F-FDG-PET is positive (c and d) even if MRI is negative (a and b); (C) brain hypermetabolism limited to MTL structures in same representative case based on semi-quantitative statistical parametric mapping method. Abbreviations: 18F-FDG-PET, 18F-fluorodeoxy-glucose positron emission tomography; GABAB, gamma-aminobutyric-acid B receptor; MTL, medial temporal lobe; MRI, magnetic resonance imaging

The results of 18F-FDG-PET pattern in anti-GABAB receptor encephalitis. (A) Representative images of abnormal metabolism in individual patient (patient #13). Hypermetabolism pattern: MTL; hypometabolism pattern: frontal and parietal lobe (FWE corrected, p p https://www.nitrc.org/projects/bnv/). Abbreviations: 18F-FDG-PET, 18F-fluorodeoxy-glucose positron emission tomography; FDR, false discovery rate; FEW, familywise error; GABAB, gamma-aminobutyric-acid B receptor; Hyper, hypermetabolism; Hypo, hypometabolism; MTL, medial temporal lobe 3.3 Correlation analysis between long-term outcome and 18F-FDG-PET based on SPM

All patients received first-line immunotherapy, including intravenous immunoglobulin and steroid pulse therapy, and 7 patients with tumors were additionally treated with chemotherapy or tumor removal (Table 1). After a median follow-up of 33 months (IQR, 16–52 months) since symptom onset, 13 of 21 patients (62%) showed a good outcome, whereas a poor outcome was observed in 8 patients (38%, Table 1). Three patients (14.3%) died of cancer malignancy progression. The clinical relapses in anti-GABAB receptor encephalitis were not common; only 1 patient developed recurrence after 46 months from onset.

The baseline comparison between patients with good and poor outcomes is shown in Table 3. Tumors were more frequently observed in patients with poor outcomes than those with good outcomes (p = 0.003). However, there were no other differences between patients with good outcomes (n = 13) and poor outcomes (n = 8), including age, sex, specific clinical features (seizures, cognitive deficits, psychosis and change of behaviors, sleep disorders), CSF findings, EEG, and MRI, treatments, although the power was limited due to the sample size. In addition, the median interval from symptoms onset to PET between prognostic subgroups did not reach statistical significance (p = 0.203). Further group comparisons of 18F-FDG-PET pattern based on voxel-based analysis of SPM showed patients with poor outcomes demonstrated relatively increased metabolism in the MTL compared to those with good outcomes in anti-GABAB receptor encephalitis (uncorrected, p < 0.001, Figure 3).

TABLE 3. Baseline comparison between patients with good outcomes and poor outcomes Patients with good outcomes (mRS 0–2, n = 13) Patients with poor outcome (mRS 3–6, n = 8) Comparison (p Value) Median age (IQR), year 50 (46–60) 64 (46–69) 0.215 Sex, n (%) Male 8 (62%) 7 (88%) 0.336 Main symptoms of encephalitis, n (%) Seizures 13 (100%) 8 (100%) >0.05 Cognitive deficits 11 (85%) 8 (100%) 0.505 Psychosis and change of behaviors 6 (46%) 7 (88%) 0.085 Sleep disorders 4 (31%) 6 (75%) 0.080 Tumors, n (%) 1 (8%) 6 (75%) 0.003* CSF findings, n (%) Pleocytosis 8 (62%) 8 (100%) 0.111 Elevated protein levels 5 (38%) 6 (75%) 0.183 EEG abnormalitiesa, n (%) 9 (69%) 7 (88%) 0.606 MRI results, n (%) MTL T2/FLAIR hyperintensities 4 (31%) 5 (63%) 0.203 Immunotherapy, n (%) 13 (100%) 8 (100%) >0.05 Median interval between symptoms onset and 18F-FDG-PET, IQR, days 39 (29–58) 47 (39–68) 0.203 Abbreviations: 18F-FDG-PET, 18F-fluoro-2-deoxy-d-glucose positron emission tomography; CSF, cerebrospinal fluid; EEG, electroencephalogram; FLAIR, fluid-attenuated inversion recovery; IQR, interquartile range; MRI, magnetic resonance imaging; mRS, modified Rankin Scale; MTL, medial temporal lobe. a Temporal area slow waves or epileptiform discharges were considered abnormal. * p < 0.05.

Correlation analysis between long-term outcome and 18F-FDG-PET based on statistical parametric mapping method in patients with anti-GABAB receptor encephalitis. Compared to patients with good outcome, patients with poor outcome demonstrated relatively increased metabolism in the MTL (uncorrected, p < 0.001). Abbreviations: 18F-FDG-PET, 18F-fluorodeoxy-glucose positron emission tomography; GABAB, gamma-aminobutyric-acid B receptor; MTL, medial temporal lobe

4 DISCUSSION

Our study has three major highlights and clinical implications. First, we revealed highly pronounced MTL hypermetabolism in association with relative frontal and parietal hypometabolism detected by semi-quantitative brain 18F-FDG-PET in patients with anti-GABAB receptor encephalitis compared to controls. Second, this study also confirmed that 18F-FDG-PET may be relatively sensitive in the early diagnosis of anti-GABAB receptor encephalitis, especially in the absence of any abnormal MRI findings; thus, 18F-FDG-PET should be considered when patients with normal MRI for suspected anti-GABAB receptor encephalitis; Third, we also found that patients with poor outcomes had more significant hypermetabolism in the MTL compared with controls and patients with good outcomes, suggesting that pronounced MTL hypermetabolism may be an imaging biomarker of poor prognosis in patients with anti-GABAB receptor encephalitis. Overall, the clinical application of 18F-FDG-PET may not only help in early diagnosing anti-GABAB receptor encephalitis, but also in predicting long-term outcomes.

Similar to prior reports, the majority of anti-GABAB receptor encephalitis patients presented with symptoms of limbic encephalitis, such as seizures, memory loss, and mental and behavioral disorders.4, 15 Additionally, around two thirds of patients with anti-GABAB receptor encephalitis indicated abnormal metabolism located in the limbic lobe (mostly in MTL) on 18F-FDG-PET in our study. Thus, these clinical and imaging findings hint toward predominant involvement of the limbic system in anti-GABAB receptor encephalitis. In addition to metabolic lesions in MTL, our study also showed functional metabolic changes on FDG-PET in the extra-limbic structures, including basal ganglia, thalamus, frontal, temporal, and parietal lobe cortex.1, 2, 20 MTL hypermetabolism was the most prominent PET feature of anti-GABAB receptor encephalitis as previously described,11 and only approximately 10% patients showed involvement of basal ganglia. In encephalitis associated with LGI1 antibody, MTL and basal ganglia hypermetabolism have been systematically described.16, 21 In addition, patients with anti-NMDAR encephalitis usually show a relative hypermetabolism in the frontal, temporal lobe and basal ganglia, hypometabolism in the parietal and occipital lobe.22-25 Hence, compared with anti-LGI1 and anti-NMDAR encephalitis, patients with anti-GABAB receptor encephalitis usually have limited involvement of the basal ganglia, which suggests that, in addition to whole-body tumor screening, 18F-FDG-PET may also be helpful in differentiating antibodies subtypes of AE. However, the sample size is relatively small and the pathophysiological mechanism of this metabolic signature, which might be related to the distribution of GABAB or other receptors in the brain, is still unclear. Recently, many studies have shown some specific PET tracers, such as [18F] cEFQ and [18F]GE-179, that have been developed for mapping excitatory receptor activations in the brain.