Autoimmune Encephalitis Misdiagnosis  



Eoin Flanagan, M.B., B.Ch.

Eoin Flanagan, M.B., B.Ch. 

Professor of Neurology
Division of Clinical Microbiology
Department of Neurology
Mayo Clinic, Rochester, Minnesota



Well, hello everybody. My name is Eoin Flanagan, and I'm a neurologist at the Mayo Clinic in Rochester, Minnesota. And today I'm excited to talk to you about this "Hot Topic" on autoimmune encephalitis misdiagnosis. 

Learning objectives

The learning objectives today are to recognize autoimmune encephalitis misdiagnosis as an important issue; to identify some of the red flags that can be useful in clinical practice to digest alternative diagnoses; and to highlight some more problematic antibodies that can sometimes cause confusion.


These are the disclosures for my talk, which is mostly focused on diagnosis, so not too relevant. And there will be some mention of off-label immunotherapy use. 

Autoimmune encephalitis misdiagnosis

Autoimmune encephalitis is a rare condition, and it's important to remember that many of the mimics are much more common. For example, toxic metabolic encephalopathy, psychiatric disease, neurodegenerative disorders, and even infectious disorders can be similar or more frequent than autoimmune encephalitis. Or even brain fog and non-encephalopathy sometimes get mixed up as being autoimmune encephalitis. It's important that we consider a broad differential diagnosis.

There's been many studies on patients in whom the initial diagnosis of autoimmune encephalitis was overlooked, and they eventually had a true diagnosis of autoimmune encephalitis. But the inverse where patients were given a diagnosis of autoimmune encephalitis but actually had something else has not been well studied, and that was the idea of our study. We encounter this reasonably commonly in our clinics, and there are patients who may be exposed to immunotherapy for a long period of time, and that has risks of course, in the setting of the pandemic and risks of other infectious complications or other complications. And also, there can be a diagnostic delay, so if a patient is misdiagnosed, you're not treating the actual condition that they do have. So there are some problems with autoimmune encephalitis misdiagnosis. 

International multicenter study on autoimmune encephalitis misdiagnosis

We set out to do a multicenter study, retrospective, on autoimmune encephalitis misdiagnosis, and this involved Mayo Clinic, the University of Oxford in the United Kingdom, University of Utah, University of California-San Francisco, UT Southwestern, and Washington University in St. Louis.


We focused this study on adults. And our inclusion criteria were really patients who had previously received a diagnosis of autoimmune encephalitis. Many of these had previously received treatments for autoimmune encephalitis. And then subsequently they had an alternative diagnosis assigned. Sometimes that was confirmed, for example, if they had a brain biopsy showing a glioma, or other times it was presumed. For example, if a patient has primary psychiatric disease, we don't have a definitive single test that can confirm that, but that seemed to make the most sense.

We collected the numbers that were misdiagnosed against the numbers that were truly diagnosed with autoimmune encephalitis to see how common this was. And then we looked at some of the clinical MRI, EEG, neuropsychological testing data, as well as CSF and serology features of these patients.

Approximately 25% evaluated for AE were misdiagnosed

Overall, we found that autoimmune encephalitis misdiagnosis occurred in just over a quarter of the referrals, and the median age was 48 years, and a slight majority were female. 

Results: Clinical features, investigations, and diagnostic criteria

It was interesting — the investigations and the clinical features that we found. Many of these patients had an insidious onset, while we usually think of an acute or subacute onset with autoimmune encephalitis, so that was a potential red flag. Only a small proportion of patients had seizures, and this would be expected to be higher in patients with true autoimmune encephalitis. Not many patients had evidence of inflammation on the spinal fluid or on the MRI brain. And indeed, almost a third of patients had completely normal neuropsychological testing, which would really not be in keeping with typical autoimmune encephalitis.

Results: Just 28% fulfilled diagnostic criteria for possible AE

We looked at the diagnostic criteria for autoimmune encephalitis. To meet autoimmune encephalitis diagnostic criteria, you have to fulfill what we term here, the possible autoimmune encephalitis diagnostic criteria, and then there's an algorithm where you may fit into other categories. But you have to fulfill these requirements. Here you can see a subacute onset, new focal CNS findings, seizures, CSF inflammation, or MRI inflammation are some of the characteristics of these features. And in our study, just 28% fulfilled these diagnostic criteria. Many patients either had a slow onset or did not have much evidence of inflammation on their testing. 

Results: Disorders misdiagnosed as AE

What disorders did we find? We found two major categories of disorders. The first was non-neurologic conditions. Functional neurologic disorder was a common mimic, also primary psychiatric disease, so that might be depression with psychosis, for example, was one disorder that could present like this. And then patients who had brain fog and non-encephalopathy. For example, they might have fibromyalgia, they might be on multiple medications that contribute to cognitive dysfunction. They may have chronic pain syndrome beyond opioids, et cetera. And these patients were sometimes misdiagnosed if they had a low positive antibody.

And then there was a group of neurologic conditions. Many of these were neurodegenerative, that might be Alzheimer's disease, where it was more slowly in onset, but a young onset maybe might mimic an autoimmune encephalitis. Or patients with Lewy body disease, for example, may have fluctuations that mimic a subacute onset. Neoplastic disorders like lymphoma and glioma were others. And then there was a variety of other categories, for example, patients who had epilepsy, just regular epilepsy, were sometimes misdiagnosed. Or there was some mitochondrial disorders. There were certain infections including HIV infection involving the brain. And Wernicke's encephalopathy, associated with thiamine deficiency. Some of these have specific treatments and obviously we would like to get these diagnosed as soon as possible. 

Results: Imaging of misdiagnosis

Here's some examples of some of the cases. You can see a patient with a temporal lobe glioma mimicking limbic encephalitis. This second patient was a patient who had involvement to the corpus callosum of lymphoma. This next patient was a patient with a leukoencephalopathy associated with HIV infection. This next patient here had MELAS, which is a mitochondrial disorder that can have stroke-like episodes that are episodic and can resolve on their own but can mimic immunotherapy response if they improve coincidentally at the time immunotherapy is given. Here we see an example of a patient with Alzheimer's disease, another example of Alzheimer's disease here. And then this last patient was a patient with Creutzfeldt-Jakob disease with restricted effusion within the basal ganglia. 

Results: Most common reasons for misdiagnosis of AE

We looked at the most common reasons for misdiagnosis, and the major one was overinterpretation of a nonspecific serum antibody. That accounted for the reason in about 50%. Other times it was misinterpretation of the clinical findings, for example, mistaking functional symptoms as truly neurologic symptoms. And then in other cases, the imaging overlapped. For example, if there was MRI abnormalities, it overlapped with autoimmune encephalitis. And in some cases, there was reluctance to accept the psychiatric diagnosis if these patients were put on immunotherapy as sometimes a psychiatric diagnosis is less acceptable than an autoimmune encephalitis diagnosis, for example.

Results: Most common serum antibodies elevated

Some of the problematic antibodies that we found in our study were thyroid peroxidase antibodies, so, TPO antibodies are not a good marker of autoimmune encephalitis and often contributed to misdiagnosis. These patients were given what's termed a diagnosis of Hashimoto's encephalopathy, which is a very questionable syndrome and a little bit problematic, as the thyroid antibodies are quite problematic. And we'll come back to that. GAD65 antibodies at low titer. We know that high-titer GAD65 antibodies can sometimes be associated with an autoimmune encephalitis with epilepsy or other forms of encephalitis, but low titers are very common in the general population. Potassium channel antibodies that are negative for LGI1 and CASPR2 also caused a bit of a problem. We are moving more towards LGI1 CASPR2 testing alone in our lab, and many other labs, and moving away from the potassium channel complex antibody testing. Some patients had a positive NMDA antibody in serum, but negative in CSF, which is a red flag and should be recognized as that. And then there were some other antibodies with older generation techniques like immunoprecipitation, the ganglionic acetylcholine receptor antibody, the N-type calcium channel antibody, or some antibodies detected by western blot alone, or in noncertified laboratories, or even cell-based assays that were low positive that sometimes are recognized to be problematic at low levels, like CASPR2 antibodies.

Results: Treatments and adverse effects

There was both a diagnostic delay and some adverse effects to treatment. The median delay to the actual diagnosis was 16 months. And this can be problematic, as earlier treatment of the true diagnosis is very important. And also patients did develop adverse effects. Most of these were steroid related, including patients developing worsened psychosis, avascular necrosis of the hip, heart failure. Other patients may have developed some infections. And again, this was a retrospective study, this is likely underrepresented, and there may in fact have been more complications that weren't documented in the records. 

Discussion: Red flags

In terms of red flags, if the patient has an insidious onset, they won't be able to fulfill diagnostic criteria for autoimmune encephalitis, and that should be considered a red flag. If there's no evidence of inflammation, a normal white blood cell count and no oligoclonal bands in the CSF, or if the MRI is normal or shows no inflammation, that might be a red flag. If you have a nonspecific serum antibody positive, like thyroid peroxidase antibodies, TPO antibodies, or GAD65, be careful not to over-interpret those. And in general, if they're not fulfilling these criteria for possible autoimmune encephalitis, that should be considered a red flag. 

There are some caveats and some rare exceptions. For example, in the elderly, some patients with LGI1 antibodies or AMPA receptor antibodies won't have much evidence of inflammation on their MRI or their spinal fluid and may not fulfill the full criteria. IgLON5 would be another example. There are some rare exceptions. But overall, the main idea is that, watch out for these red flags and if you have them at least consider a broad differential diagnosis in your antibody testing, make sure you interpret that in the correct context. 

Discussion: Antibody positivity in healthy/disease controls

Some of the antibodies that we found in this study are very common in the general population, like TPO antibodies are found in 20% of elderly adults. That makes it not a useful biomarker for autoimmune encephalitis because we know 20% of elderly adults do not have autoimmune encephalitis. Other antibodies we've removed from some of our encephalopathy panels because they do have a positivity rate of 3% to 5% in the general population, and that includes the ones mentioned here. Sometimes the titers can be helpful. For example, with GAD65, very high titer is more predictive. Or with MOG antibodies, high titer is more predictive of a true MOG antibody disease.

The dilutions can help. We now offer dilutions in CASPR2, and we're looking into the possibility of dilution for GABA-B, because sometimes you can have some low, nonspecific binding with some of these cell-based assays. We're always trying to work to improve our assays and to try and give the best, most important result.

Discussion: Isolated western blot results problematic

Dr. Dalmau and colleagues and others have shown that if you use a line blot or western blot in isolation — we don't really offer that here at Mayo Clinic, but some centers do offer line blot in isolation — and when those results come back positive and everything else is negative on the tissue immunofluorescence, then that should be a red flag, because line blot and isolation has a high risk of false positives.

Published study

I'd like to just mention that this study that we did and a discussion around it was published in JAMA Neurology at the end of November in 2022. And it was also an accompanying editorial by Dr. Dalmau and Graus, who are known experts in the field, and I think highlighting some of the issues around misdiagnosis and how to prevent it.


I'd like to just conclude by summarizing some of what we mentioned in today's "Hot Topic." Firstly, as I mentioned, about 25% of patients who are referred with autoimmune encephalitis actually had an alternative condition. This is a frequent and important clinical problem, autoimmune encephalitis misdiagnosis. 

Remember that was two major categories: the non-neurologic conditions, for example, primary psychiatric, that was also functional neurologic disorder, or the other category was a brain fog type, non-encephalopathy. And then there were neurologic categories, which included neurodegenerative disorders or gliomas, particularly involving the temporal lobe. And then a wide variety of other neurologic conditions. 

Remember to use the diagnostic criteria for autoimmune encephalitis, and if patients don't fulfill that, consider that to be a red flag. Many of the patients, almost three-quarters of the patients in our study, did not fulfill the criteria for even possible autoimmune encephalitis. 

Remember some of the red flags that we mentioned, like completely normal MRI. That can happen, for example, with NMDA receptor antibodies. That occurs in the majority. But if there's no inflammation on the spinal fluid and the clinical syndrome does not fit, then you want to question the diagnosis. If the CSF does not show much inflammation too, that can be a red flag. 

Remember, some of the nonspecific antibodies like TPO antibodies, GAD65 antibodies at low titer, and some of the other ones we mentioned today can be a bit more problematic.

Remember that if a patient receives immunotherapy but continues to develop expansile brain lesions, you might want to think about a neoplastic disorder, like in this case was a glioma. Or if a patient has prominent atrophy on MRI and on PET scan has hypometabolism, you might want to consider testing for Alzheimer's disease. Biomarkers are looking into other neurodegenerative etiologies. Misdiagnosis of autoimmune encephalitis is not trivial because the treatments that we use like immunotherapies can have harm, and 20% in our study had adverse effects of treatment. And there was a median delay to diagnosis of 16 months. While it's okay to use a short course of steroids in patients where you're not sure, it's important to also consider a broad differential diagnosis, particularly if there are red flags that were outlined in this study.


I'd like to just acknowledge our team here at Mayo Clinic. This is all of our neuroimmunology team in the laboratory, and we have a great team also of coordinators and then of clinical service representatives out in the field who help us a lot to educate our clients. And I'd also like to thank the many contributors across the multiple centers with whom this study would not have been possible without them. I'd like to thank all of those and thank you all for listening.


Contact us:

Image credit: Getty Images

MCL Education

This post was developed by our Education and Technical Publications Team.