This study provides new insight into the typical metabolism pattern of anti-NMDAR encephalitis patients in the acute or subacute phase, especially focusing on revealing the metabolic characteristics of the parietal lobes which are remaining controversial. In addition, the correlation of 18F-FDG PET findings with different T2 FLAIR features was also highlighted to provide objective evidence for early diagnosis.
The initial symptoms of anti-NMDAR encephalitis range greatly between studies. Psychiatric symptoms[23], cognitive impairment[24] and seizure[25] are the most common clinical features to emerge. The clinical symptoms are complex and varied, making the diagnosis even more challenging in the initial stage of the disease. Patients with anti-NMDAR encephalitis mistaken with acute psychosis may worsen after treating with antipsychotics[26, 27] and patients misdiagnosed as having degenerative dementia may miss the therapeutic opportunity of reversible cognitive decline of anti-NMDAR encephalitis patients[28]. Therefore, early and accurate diagnosis of anti-NMDAR encephalitis is crucial.
MRI is one of the most commonly used and reliable neuroimaging modalities, which has an important role in excluding differential diagnoses as the first-line method. A systematic review reported that among 1167 patients in the acute phase assessed, 440 abnormal MRIs (37.7%, 35.0-40.5 95%CI) were found[3]. Our present study showed a relatively higher diagnostic capacity of MRI with a sensitivity of 75.0% in acute or subacute phase patients, likely due to a better understanding of MRI characteristics of anti-NMDAR encephalitis at present than before. Otherwise, the presence of any type of abnormal T2 Flair findings was considered abnormal MRI, making more patients identified. According to previous studies, T2 FLAIR hyperintensity was the most commonly reported abnormalities[29], and DWI changes were usually accompanied by T2 FLAIR abnormalities, which was consistent with our research findings. We didn’t find atrophy shown on MRI in these cases, which was previously described as a common sign in a general diffuse pattern, likely due to the limitation of enrolled patients in the acute or subacute phase. Contrarily, we found that shallow of sulci and swelling of brain tissue and increased signal in the sulci were also commonly seen in the present study, besides increased signal on brain gray matter, adjacent white matter, brain linear structure or the basal ganglia. If taking shallow sulci and swelling of brain tissue as abnormal indicators into comprehensive consideration, the diagnostic sensitivity of MRI can be greatly improved. 15 patients showed shallow sulci and swelling of brain tissue, and 3 of them showed simple brain tissue swelling with no signal abnormalities. Therefore, we suggested that radiologists and neurologists should take encephalitis as a diagnosis into account, even if only simple swelling of brain tissue is found on MRI. In the current study, PET outperformed MRI with a higher diagnostic sensitivity (95.83% vs. 75.00%), and there was only one patient with negative PET and MRI results. Most of the abnormalities on T2 FLARI descript above demonstrated hypermetabolism on FDG PET, except alterations of brain linear structure and basal ganglia on T2 FLAIR with hypometabolism on FDG PET. 18F-FDG PET can reveal positive findings in the absence of MRI abnormalities, which have been reported by previous studies[30].
Nowadays, 18F-FDG PET was suggested as a valuable modality in the work-up of suspected anti-NMDAR encephalitis patients, especially in those patients with negative MRI findings. Some characteristic metabolic patterns were identified to provide important clues to improve clinical decision-making. Anteroposterior glucose metabolism gradient was a widely recognized metabolism pattern in this disorder[18–23, 31]. Most research revealed hypermetabolism in frontal and temporal lobes and hypometabolism in bilateral occipital lobes. Different distribution of NMDA receptors, electrophysiology, both regulation of ligand and voltage-gated channels, and different glucose utilization in different locations of the human brain may contribute to the unknown pathophysiological basis of this metabolic signature of anti-NMDAR encephalitis[32, 33]. The autoantibody against the GluN1 subunit of the NMDA receptor was ubiquitously located in the brain, but at higher density in frontotemporal areas and the hippocampus[34]. Jing Yuan et al proposed a hypothesis that abnormal metabolic patterns might be associated with disrupted NMDAR signaling, for ketamine, an NMDAR antagonist, showed similar symptoms and frontal-to-occipital gradient of glucose metabolism detected by FDG PET in healthy human subjects as shown in anti-NMDAR encephalitis patients[22].
This present study certified this typical metabolism pattern with a relatively large scale patient number. The cortical visual analysis per patient revealed that frontal lobes were the most prone to involvement with hypermetabolism, followed by temporal lobes and parietal lobes. Bilateral hypometabolism in the occipital lobe is considered a typical metabolic pattern used to help the physician start empirical treatment before antibody-positive confirmation. Some previous studies results showed that bilateral decreased metabolism was demonstrated at acute and subacute phases [18, 22], and marked medial occipital lobe hypometabolism may serve as an early biomarker for discriminating anti-NMDAR encephalitis from other AE[18]. This is consistent with our findings. Both visual analysis and “independent t-test” demonstrated bilateral occipital hypometabolism in anti-NMDAR encephalitis at acute and subacute phase, which commonly involved bilateral lingual and superior, middle and inferior occipital lobes. Some studies reported the possible association of the glucose hypermetabolism of frontal and temporal with occipital hypometabolism[20, 22].
We also noticed an interesting phenomenon that, when taking unilateral or bilateral affected sites into account, frontal lobes were still the most prone to bilateral involvement with hypermetabolism, but the number of bilateral involved parietal lobes was more than that of the involved temporal lobes with hypermetabolism. These were verified by the following cortical SUVR comparison. Hypermetabolism on the anterior portion (supramarginal gyrus) and hypometabolism on the posterior portion (precuneus and cuneus) of parietal lobes in the anti-NMDAR encephalitis group were proved to be statistically significantly different from the control group, as well as cingular cortex with hypermetabolism on anterior portion and hypometabolism on posterior portion, indicating that parietal lobe and cingular cortex also showed an anteroposterior glucose metabolism gradient rather than simply increase or decrease. Although previous studies discussed the metabolic change of parietal lobes, its metabolism pattern remained controversial. Most studies reported hypometabolism of parietal lobes[3, 21, 35, 36], but some others demonstrated hypermetabolism[22]. Our research with a relatively large number of patients highlighted that previously described anteroposterior gradient of activity of the parietal lobe or cingular cortex may be a potential indicator for the diagnosis of anti-NMDAR encephalitis at acute and subacute phase. Therefore, the metabolic change of the parietal lobe should be investigated and interpreted carefully, rather than simply categorized as hypometabolism or hypermetabolism. Otherwise, when interpreting PET findings, T2 FLAIR-MRI played an important role in helping to determine hypermetabolism of brain regions or hypometabolism of contralateral brain regions which sometimes were very difficult to make a correct judgment. Hypometabolism regions on PET were all negative on T2 FLAIR, but abnormalities on T2 FLAIR were found, indicating hypermetabolism of the corresponding brain regions.
Besides cortical metabolism patterns, some research was also explored to disclose the likely metabolic pattern of basal ganglia nuclei in anti-NMDAR encephalitis. A study revealed bilateral focal hypermetabolism in the caudate, putamen and anterior cingulum[21]. Jan Novy et al found a significant FDG uptake increase in the caudate nuclei in episodes of varying intensity and delay from the onset of the symptoms[31]. Another study identified selective caudate nucleus hypermetabolism along the previously described gradient of activity[31]. Putamen, caudate and thalamus were identified as hypermetabolism brain regions with visual analysis in the present study, but only the SUVR of putamen showed a significant statistical difference from that of the control group. Cerebellum involvement is rare. The present study also revealed a left cerebellum-involved case.
Several limitations of our work should be highlighted. First, there may be a potential selection bias towards races and regions due to this study only including patients from one single center. Second, analyses of subgroups with variable clinical symptoms were not conducted, which needs further investigation with a larger sample size.