August 2026 Research Update

Across moderate-strength clinical studies, no single test clearly worked for every person with suspected autoimmune encephalitis. Instead, papers highlighted a package of evidence: antibody testing, spinal fluid (cerebrospinal fluid) measures that reflect inflammation or nerve injury, and brain imaging patterns that can help separate autoimmune encephalitis from other conditions. [1] [2] [3] [4] [5]

A real-world study of 141 people referred for suspected autoimmune encephalitis reported a characteristic brain sugar-uptake pattern on FDG PET/CT, including lower activity in parts of the brain cortex with higher activity in the mesiotemporal area and other regions. This pattern helped distinguish autoimmune encephalitis from look-alike diagnoses, with diagnostic performance around an AUC of 0.82 to 0.83, and visual reading performed better in younger patients than in older patients. [1]

In a prospective multicenter study of 92 adults with definite autoimmune encephalitis (anti-NMDAR, anti-LGI1, anti-CASPR2), intrathecal kappa free light chain (KFLC) synthesis in spinal fluid was common, especially in anti-NMDAR cases (94%), but the test was not very specific when compared with a mixed set of inflammatory and non-inflammatory control groups. In the same study, neurofilament light chain (NfL) levels in spinal fluid and blood tracked with how severe symptoms were and how they changed over time, while GFAP and cytokines added little for diagnosis or prognosis. [2]

Several studies emphasized that age and clinical context can affect what shows up in testing. In a multicenter cohort, older adults with autoimmune encephalitis were less likely to have headache or spinal fluid white blood cell increases, and antibody types differed by age (more anti-NMDAR in younger adults and more anti-LGI1/anti-CASPR2 in older adults). A pediatric hospital screening study found that among children with acute encephalitis symptoms who met a possible autoimmune encephalitis picture, all laboratory-confirmed cases were anti-NMDAR antibody positive, while some children met probable autoimmune encephalitis criteria despite negative antibody tests. [4] [3]

Other work looked at additional spinal fluid measures and when autoimmune testing is less informative. In 103 people with anti-NMDAR encephalitis, an elevated spinal fluid to blood albumin ratio (a marker linked to blood-spinal fluid barrier function) lined up with higher general immune-protein measures in spinal fluid, but it did not predict antibody levels or short-term discharge outcomes. In a case-control study of visual snow syndrome, autoimmune encephalitis antibodies were not found in blood or spinal fluid, and spinal fluid results were mostly similar to controls. [6] [7]

Smaller hospital series added examples of how imaging and antibodies can vary across subtypes. In an 18-person series, anti-NMDAR disease predominated in young women, some antibody-negative cases had acute mesiotemporal MRI changes, and anti-GAD65 cases presented later with difficult-to-control seizures, showing that different autoimmune encephalitis types can have different diagnostic clues. Separately, an early-stage lab study described a sensor designed to detect anti-LGI1 antibodies in spiked human serum, but it has not yet been validated in patients. [5] [8]

For patients and caregivers, this set of studies supports the idea that autoimmune encephalitis diagnosis often relies on multiple pieces of information that may not all be present at once, and that results can look different in children versus older adults and across antibody types. Some tools (like FDG PET/CT patterns or spinal fluid KFLC and NfL measures) may add supportive evidence, but several markers were either not specific across other inflammatory illnesses or were mainly useful for tracking severity rather than confirming a diagnosis. A new anti-LGI1 blood test device is still background research because it has only been tested in the lab with spiked samples, not in real patients. [1] [2] [4] [3] [6] [7] [5] [8]

Across these studies, most evidence came from small case series or single-patient reports, so it cannot prove which treatment works best. Still, the overall pattern was that antibody-driven autoimmune encephalitis sometimes improved after additional immune therapies when first treatments were not enough, including plasma exchange and newer targeted medicines. The biggest uncertainty is how well these results generalize, because patient types, severity, and added treatments differed a lot across reports. [9] [10] [11] [12] [13] [14] [15]

In a retrospective hospital cohort that included 10 people with autoimmune encephalitis, functional scores were lower (better) 90 days after therapeutic plasma exchange, but some patients still had severe disability and deaths occurred, making it clear results were mixed and the study cannot show cause and effect. Side effects linked to plasma exchange were reported as minor in a subset of the overall 133-patient group, with no plasma-exchange-related deaths in that center. [9]

For anti-NMDA receptor encephalitis that did not respond to first-line immunotherapy, a five-patient case series reported rapid improvement after efgartigimod, along with drops in total antibody proteins and anti-NMDA receptor antibody levels, and improvement on brain scans and brain-wave tests. No serious safety problems were reported in this small group, but some patients also received other ongoing immune treatments, so the specific role of efgartigimod remains uncertain. [10]

A single-patient report of severe anti-NMDA receptor encephalitis described seizures stopping and psychiatric symptoms improving quickly after eculizumab, a medicine that blocks part of the complement immune pathway; this may be relevant when laboratory markers suggest complement activation. Because this was one patient, it is not possible to know how often this would happen or which patients are most likely to benefit. [11]

Several individual reports and small series described improvement using commonly used immune therapies, sometimes in intensive care situations. One older man with anti-GABA-B receptor encephalitis and hard-to-control seizures improved by 3 months after steroids and plasma exchange plus anti-seizure and ICU care, though treatment effects could not be separated. A 3-year-old with myelin oligodendrocyte glycoprotein (MOG) antibody-associated disease improved markedly after plasmapheresis when steroids and immune globulin had limited effect, with MRI changes resolving by day 24. [12] [13]

Not all antibody-linked conditions had the same course, and lingering symptoms were noted even after early improvement. In a seven-person anti-LGI1 autoimmune encephalitis series, most improved early with steroids plus intravenous immune globulin, but some needed second-line immune suppression for incomplete response or relapse, and some had ongoing memory or thinking problems during follow-up. In a separate single-patient report, scheduled intravenous immune globulin was linked to long-term seizure control in anti-GAD antibody-associated refractory temporal lobe epilepsy. [15] [14]

For patients and caregivers, this set of papers mainly adds background about real-world salvage approaches used when symptoms continue after initial immune treatment, especially in anti-NMDA receptor encephalitis and other antibody-defined types. The most direct new information is limited to a five-patient experience with efgartigimod and several single-patient reports, so it is still unclear how often benefits occur, how durable they are, and what tradeoffs exist across different antibodies and severities. Larger, comparative studies are needed to clarify which rescue options are most helpful for which people and what outcomes (function, seizures, thinking) are most likely to improve.

Across these studies, the strongest and most repeatable signal was that how severe autoimmune encephalitis is at the start can shape recovery later on. Antibody test results alone did not consistently separate who did well versus poorly, but certain antibody-defined groups showed different relapse patterns and different needs for stronger immune treatments. Some work also focused on longer-term seizure outcomes, suggesting that a subset of people go on to have ongoing epilepsy after the acute illness. [16] [17] [18] [19]

In one single-hospital pediatric study, children with definite NMDA receptor antibody-positive disease arrived sicker and were more likely to need second-line immune treatment than children with probable antibody-negative disease. Relapses in that study happened only in antibody-positive children who did not receive rituximab, while none of those who received rituximab relapsed. Brain MRI changes were also more common in the antibody-positive group, and epilepsy after encephalitis was reported in both groups. [16]

A separate multicenter pediatric study from Iran did not find meaningful differences between antibody-positive and antibody-negative groups in early severity, long-term function, brain MRI or EEG results, relapse, or death. Instead, the factors most tied to poor prognosis were higher initial severity and having underlying medical conditions. Intensive care admission was common in this group. [17]

A Vietnam cohort tested whether the NEOS score (a prognostic score originally created for anti-NMDA receptor encephalitis) could help predict outcomes in children. NEOS predicted poorer outcome in antibody-negative autoimmune encephalitis but did not predict outcome in anti-NMDA receptor encephalitis, and most patients in both groups had good function by follow-up. This suggests that a single scoring tool may not fit every autoimmune encephalitis subgroup equally well. [18]

A two-center study of people with autoimmune encephalitis who had seizures during the acute phase looked at who later developed ongoing epilepsy related to autoimmune encephalitis. Progression was more likely when more anti-seizure medications were needed at hospital discharge, while having infection-like symptoms before encephalitis was linked to lower risk. The authors reported their prediction model had good overall performance, but it still needs testing in other settings. [19]

One study focused on a different antibody type, anti-AMPAR encephalitis, using a combined group of hospital cases plus published cases. Poorer outcomes were more likely with older age, having an associated tumor, and higher disability at diagnosis. In the same analysis, mental health symptoms and having more than one anti-neuronal antibody were linked to a higher chance of an associated tumor. [20]

For patients and caregivers, these studies mainly add context about why recovery can look different from person to person: early illness severity and other health conditions repeatedly showed up as important, and some antibody-defined groups may have different relapse patterns. The work on prediction (NEOS score and a seizure-to-epilepsy model) is most useful as background research right now, since tools can perform differently across autoimmune encephalitis subtypes and care settings. Important uncertainties remain because most data here were retrospective (looking back in records), and different hospitals and regions may see different patient mixes and follow-up practices. [16] [17] [18] [19] [20]

Across these studies, a common pattern is that recovery can look very good on broad disability scales, yet ongoing symptoms can remain, especially around memory and behavior. Brain scans in both adults and children suggest that long-term problems may relate to subtle, region-specific brain changes and less efficient connections within brain networks rather than a single obvious injury. Some work is also focused on better ways to measure ongoing symptoms in children so lingering difficulties are not missed. [21] [22] [23] [24]

In a moderate-sized adult study of post-acute anti-NMDA receptor encephalitis, reduced structural "complexity" on MRI was seen in both hippocampi and a wider fronto-cingulo-temporal network. Stronger MRI changes were linked to more persistent memory problems and psychiatric symptoms after immunotherapy, and to worse performance on verbal and visuospatial memory tests. The symptom pattern also shifted over time, from psychosis being more prominent during peak illness to mood-type symptoms being more prominent later. [21]

In a pediatric case-control MRI study, total thalamus size was similar to healthy children, but specific thalamic subregions were smaller, and the overall efficiency of the thalamic network was reduced. This supports the idea that long-term effects may involve selective parts of the brain and network-level changes that might not show up as a single large volume loss. [22]

In a multicenter retrospective pediatric anti-NMDA receptor encephalitis cohort with a median follow-up of about 5 years, most children reached very good scores on a broad disability scale, but a symptom-focused scale still captured persistent difficulties in activities of daily living, speech/communication, and sleep. The same study described real-world symptom-directed medication use and reported that these treatments were generally tolerated, while risperidone had more adverse events in this cohort. [23]

A prospective study developed the Pediatric Autoimmune Encephalitis Severity Scale (PASS) and found very high agreement between different raters and a meaningful link between the worst PASS score during illness and parent-rated quality of life at 6 months. PASS items grouped into two main dimensions, one reflecting cognitive/behavioral issues and another reflecting disinhibition/excitability, which may help clinicians describe symptom types more clearly over time. [24]

A small case series of three adults with severe bilateral hippocampal damage described an unusual, intermittent feeling of having "just awakened" from unconsciousness, alongside profound anterograde amnesia. The authors proposed a name for this experience (Stream of Consciousness Impairment) and suggested it may relate to hippocampal-limbic memory networks, but emphasized it is hypothesis-generating and may not apply broadly. [25]

For patients and caregivers, these papers reinforce that "doing well" on overall disability ratings does not always mean all symptoms have resolved, especially for memory, mood/behavior, sleep, speech, and daily-life functioning. The imaging studies add background evidence that lasting symptoms can track with changes in specific brain regions and networks, but they do not yet prove cause-and-effect or predict an individual person's outcome. The work on PASS mainly matters now as a way to more consistently describe and follow pediatric symptoms, while the rare consciousness experience described in three adults is best viewed as an uncommon possibility that needs more study.

Across these mostly single-patient reports and small case series, autoimmune encephalitis did not always start with the "classic" picture. Symptoms sometimes looked like a primary psychiatric illness, a heart rhythm problem, a stroke, a movement disorder, or a balance disorder, and early brain scans could be normal. Together, the papers show that antibody testing and follow-up evaluations sometimes clarified the diagnosis only after the illness evolved. [26] [27] [28] [29] [30] [31] [32] [33]

Several reports highlighted psychiatric and perception-dominant presentations of anti-NMDA receptor encephalitis. One 34-year-old man was initially treated for schizophrenia before spinal fluid testing found anti-NMDA receptor antibodies. [26] A 20-year-old woman had prominent altered time perception with visual distortions and hallucinations, plus seizures, with low antibody levels in spinal fluid. [32]

Unusual neurologic entry points also included smell loss and cerebellar symptoms. A 59-year-old man with anti-GABA A receptor encephalitis presented with olfactory dysfunction, behavior changes, and seizures, and later scans showed recurring multifocal cortical and subcortical brain lesions. [27] Two adults with antineuronal antibodies (including anti-Zic4 and anti-NMDA receptor) had imaging patterns suggestive of cerebellar degeneration, yet no tumor was found during follow-up, raising the possibility of a non-cancer related immune process in these cases. [28]

Some papers focused on non-neurologic or stroke-like beginnings that later fit autoimmune encephalitis. Two people with anti-LGI1 encephalitis initially presented with sick sinus syndrome, a heart rhythm problem, before clearer encephalitis features emerged. [30] Another patient with probable GAD65 antibody-associated autoimmune encephalitis started with brief episodes that looked like cerebrovascular events and later developed very severe seizures, while initial spinal fluid antibody testing was negative and later blood testing detected low-titer GAD65 antibodies. [31]

In a 28-month-old child, early tests and imaging were initially unrevealing despite seizures and developmental regression, and the diagnosis was delayed until antibody testing supported anti-NMDA receptor encephalitis. The report described substantial improvement after first-line immunotherapy, with remaining mild right-hand movement issues and ongoing language recovery. [33]

For patients and caregivers, these reports are mainly background about how varied autoimmune encephalitis can look at the start, and why some people experience delays or changes in diagnosis over time. They also show that normal early scans or an initial negative antibody panel did not always end the evaluation in these cases. Because most evidence here comes from individual cases, it is uncertain how often any one unusual presentation happens or how well it predicts recovery.

These papers describe special situations where autoimmune encephalitis either appeared alongside other major conditions (pregnancy, cancer, thyroid autoimmunity) or worsened soon after a possible external trigger (chelation therapy), where a look-alike condition (porphyria in a woman with HIV) was at first suspected to be autoimmune encephalitis, and where monitoring tests did not always move in sync with symptoms. Evidence is limited to case reports and small case series, so the patterns may not apply broadly. [34] [35] [29] [36] [37] [38] [39] [40]

Pregnancy and tumor context were highlighted in anti-NMDA receptor encephalitis. A 30-year-old pregnant woman at 10-11 weeks had anti-NMDA receptor encephalitis with bilateral mature ovarian teratomas; after combined immunotherapy, intensive care, and surgical removal of both tumors, her neurological status gradually improved with near-normal cognition at 3 months but persistent memory issues. [34] Anti-NMDA receptor encephalitis was also reported after surgical resection of a right insular low-grade astrocytoma in a 34-year-old man, with improvement after steroids and intravenous immunoglobulin and no encephalitis recurrence over 4 years of follow-up. [29]

Cancer therapies and cancer-associated autoimmunity appeared in two different ways. In a five-person series, acute-to-subacute parkinsonism developed after immune checkpoint inhibitor therapy, and most patients improved over follow-up after stopping the cancer immunotherapy with or without immunomodulating treatment; in one tested person, a dopamine transporter scan abnormality later reversed. [35] Separately, a man with thymoma-associated myasthenia gravis developed anti-AMPAR encephalitis and had a fulminant course with poor neurological outcome despite multiple immunotherapies including rituximab. [40]

Possible external triggers and look-alike conditions were also described. Three adults with chronic mercury exposure and anti-LGI1 and/or anti-Caspr2 autoimmune encephalitis worsened within 1-3 days after starting DMPS chelation, but recovered after prompt immunotherapy and had favorable long-term outcomes. [36] In another report, a 32-year-old woman with HIV had seizures, psychosis, and neuropathy with imaging that suggested autoimmune encephalitis, but testing showed variegate porphyria and she became clinically stable only after ritonavir was withdrawn from her antiretroviral regimen. [38]

Monitoring tests sometimes did not align neatly with symptom changes. In a 61-year-old man with anti-GABA A receptor encephalitis, symptoms and EEG improved quickly after immunotherapy and spinal fluid antibodies became negative, while MRI changes improved more slowly and an inflammation-focused brain scan (TSPO-PET) stayed positive despite clinical recovery. [39] In a 70-year-old woman with steroid-responsive encephalopathy associated with autoimmune thyroiditis, confusion persisted after her thyroid levels and other metabolic problems had been corrected, then improved within 48 hours of prednisone, helping distinguish this immune encephalopathy from severe hypothyroidism alone in that case. [37]

For patients, this theme is mostly about context: autoimmune encephalitis can appear alongside pregnancy and tumors, after brain surgery, or during evaluation for other problems that mimic encephalitis (such as porphyria). It also shows that different tests (symptoms, EEG, antibody levels, MRI, inflammation imaging) may improve at different speeds in some individuals. Because these are small reports, it remains uncertain which triggers matter most, how often they occur, and which monitoring tests best reflect recovery in everyday care.

Across this report, the overall message is that building cognitive and emotional screening into routine neuroimmunology care can uncover problems that might otherwise be missed and can speed access to specialized evaluation. The evidence comes from a single, small retrospective clinic experience in children with several neuroimmune brain and spinal cord conditions, including a few with NMDA receptor encephalitis. [41]

In this clinic model, children completed targeted neuropsychology screening before seeing the neuroimmunology team, and cognitive impairment was detected in 71% of patients. Anxiety or adjustment disorders were also noted in 36%, highlighting that emotional and coping challenges often appear alongside thinking and learning changes. [41]

The clinic approach was also linked to faster access to neuropsychology services, with reported wait time dropping from 15 months to 1.9 months. The authors reported that in 65% of patients, cognitive issues were identified and addressed during the same clinic visit, with results used to inform individualized care planning. [41]

For patients and caregivers, this study mainly adds background on how care may be organized to better recognize and document cognitive and emotional effects of neuroimmune conditions, including autoimmune encephalitis. It suggests that combining brief cognitive testing with specialist visits may help teams notice problems earlier and reduce delays to more detailed evaluation, but it does not show whether this model improves long-term school, daily functioning, or recovery outcomes. It also remains uncertain how well these results apply outside pediatric centers or to adults, since the report described a small group at one clinic. [41]

Across this report, the main pattern is that an autoimmune signal in the blood (a brain-targeting antibody) appeared alongside repeated seizures and behavior changes in a cat, while standard brain scans and spinal fluid tests were normal. Because it is a single case and immune treatment was not used, it remains uncertain whether the antibodies caused the symptoms or were a bystander. [42]

In a 5.5-month-old domestic shorthair cat with recurrent seizures, restlessness, and pica (eating non-food items), testing found metabotropic glutamate receptor 1 (mGluR1) antibodies in the blood at a titer of 1:80, while the spinal fluid was negative and other tested antibodies were negative. The brain MRI and spinal fluid routine results were unremarkable, and infectious testing was negative, which made the antibody result stand out as a possible clue. [42]

After antiseizure medicines were optimized, seizures decreased (from about every 2 days to a single seizure over the follow-up period) and pica resolved, and the blood mGluR1 antibody level dropped to 1:20 after 4 weeks. The cat did not receive immune-suppressing treatment, so the report cannot show whether immune therapy would have changed the course. [42]

For patients and caregivers, this animal report mainly serves as background research showing that autoimmune encephalitis-like antibody signals have also been described in veterinary medicine, including in very young animals with seizures and behavioral changes. Right now, it does not change human care, and even for animals it is unclear how often this antibody pattern occurs or how strongly it is tied to symptoms. More cases and longer follow-up would be needed to understand whether blood-only antibody results reliably track with disease activity.