Divyanshu Dubey, M.B.B.S., is a consultant in the Department of Laboratory Medicine and Pathology at Mayo Clinic in Rochester Minnesota. He holds the academic rank as an Assistant Professor of Laboratory Medicine and Pathology and Neurology.
Contact us: firstname.lastname@example.org.
Hi, I’m Matt Binnicker, the Director of Clinical Virology and Vice Chair of Practice in the Department of Laboratory Medicine and Pathology at Mayo Clinic. Based on recent studies a considerable proportion of patients with encephalopathy and/or epilepsy of unknown etiology may have an autoimmune or paraneoplastic cause. Many of these cases remain unrecognized and do not receive appropriate treatment in the form of immunotherapies. In this “Hot Topic,” my colleague, Divyanshu Dubey, will discuss two validated predictive models based on clinical features and initial neurological assessment. I hope you enjoy this month’s Hot Topic, and I want to personally thank you for allowing Mayo Clinic the opportunity to be a partner in your patient’s health care.
Thank you for this opportunity to present and discuss our predictive model for neural-specific antibody seropositivity among patients with epilepsy and encephalopathy.
These are my disclosures.
Autoimmune encephalitis and autoimmune epilepsy were once considered extremely rare, but lately this perception has changed. A considerable proportion of these cases have neural-specific antibody biomarkers. As we see in this figure, these antibodies are either directed towards cell-surface antigens or intracellular antigens. Some antibodies directed against extra-cellular or cell-surface antigens are considered pathogenic, whereas those against intra-cellular epitopes are biomarkers, and many of these diseases are considered T-cell mediated.
New serological biomarkers are being discovered nearly every year. Currently, we have more than 20 neural-specific biomarkers of autoimmune encephalitis. This has contributed to increased recognition of autoimmune encephalitis cases. In a population-based study, performed in Olmsted County, Minnesota, we found that the incidence of autoimmune encephalitis nearly tripled over two decades, increasing from 0.4/100,000 to 1.2/100,000. This was mostly due to increased recognition of antibody-positive cases. Furthermore, we found that the incidence and prevalence of autoimmune encephalitis was similar to infectious encephalitis with known pathogens.
We have noticed similar trends among patients with epilepsy of unknown etiology. Nearly 15‒20% of patients with cryptogenic epilepsies with or without other features of encephalitis have serological biomarkers suggestive of an autoimmune etiology. In a prospective study over a period of one year, we found about 23 of 112 cases with epilepsy of unknown etiology were seropositive for neural autoantibodies. The specificity of these antibodies varied considerably in patients with new-onset epilepsy compared to those with established or chronic epilepsy. New-onset epilepsy cases had a higher proportion of more specific serological biomarkers such as NMDA receptor antibody or LGI1 antibodies, which are likely to be pathogenic. Interestingly, patients found to be seropositive and subsequently treated with immunotherapy had a considerably better outcome compared to patients who continued to have epilepsy of unknown etiology and were treated symptomatically with anti-seizure medications. During this study we also analyzed various factors associated with autoantibody seropositivity. We applied a predictive model, the antibody prevalence in epilepsy (APE) score, to every patient who was enrolled in to the study. This is a composite model based on various predictors analyzed in our prior retrospective studies.
We subsequently evaluated this model among epilepsy patients at Mayo Clinic. We collected data from patients on whom autoantibody evaluations were sent from all three Mayo Clinic sites. This included 387 patients with the diagnosis of epilepsy.
On analysis we found factors which were predictors of seropositivity. The majority of these factors were already a part of the APE score.
We subsequently evaluated usefulness of this predictive model for autoimmune encephalopathy or dementia as well. In the process we included some revisions in the original scoring system, leading to the formulation and validation of the antibody prevalence in epilepsy and encephalopathy (APE2) score, which could be applied to both patients with epilepsy as well patients with cognitive dysfunction.
The variables in APE2 are based on clinical evaluation (such as new-onset epilepsy, neuropsychiatric manifestations, autonomic dysfunction, etc.), initial MRI, and cerebral spinal fluid (CSF) findings. MRI and CSF evaluations are usually a part of diagnostic evaluation of epilepsy or encephalopathy of unknown etiology patients. Therefore this score could be utilized by most clinicians even in a primary care setting or emergency departments where most esoteric evaluations are not available.
These are examples of medial temporal fluid-attenuated inversion recovery (FLAIR) or T2 hyperintensities which are included as part of the APE2 score. The first one is an example of a patient with the LGI1 antibody, the second a patient with Ma2 encephalitis, and the third a patient with ANNA1 or anti-Hu antibodies.
This is an example of a patient with multifocal MRI brain lesions which also been included in the APE2 scoring system. This is associated with GABAa receptor antibodies. Similar MRI features can also be found in patients with MOG antibodies, some of whom can present with epilepsy or encephalopathy.
An APE2 score of more than or equal to four performed well as a predictor of neural autoantibody positivity. The sensitivity for neural-specific antibody positivity among patients with epilepsy or encephalopathy for an APE2 score of more than or equal to four was 98%, whereas the specificity was approximately 84%. A higher APE2 score, such as one more than six, had a higher specificity for an autoimmune etiology.
This is a flow chart demonstrating the application of the APE2 score among patients with cognitive dysfunction. It shows how the APE2 score was able to identify patients consistent with the diagnosis of neural-specific antibody-positive autoimmune encephalitis. It also shows how most patients without underlying autoimmunity but who were seropositive for antibodies with lower specificities of autoimmune encephalitis or dementia scored lower on this predictive model.
Using our data we also utilized the APE2 score among patients with epilepsy of unknown etiology to propose diagnostic criteria for autoimmune epilepsy via this flow diagram.
Now we will discuss the application of the APE2 score using some patient examples.
A 58-year-old woman presented to an outside hospital with electrical shock-like sensations on the right side of her body and speech arrests lasting from 30 to 60 seconds, along with a few episodes of secondarily generalized tonic-clonic seizures. On initial workup no clear etiology was found. She was treated with multiple anti-seizure medications but continued to have refractory seizures. Three years later she presented to Mayo Clinic with orolingual dyskinesias and stridor. Her MRI demonstrated bilateral medial temporal FLAIR hyperintensities. Her CSF analysis demonstrated elevated CSF protein.
If you calculated this patient's APE2 score it comes out to be eight. She gets two points each for facial dyskinesia, two points for refractory seizures, two points for inflammatory CSF, and two points for medial temporal FLAIR hyperintensities. An autoimmune epilepsy panel was sent which came back to be positive for ANNA2 or anti-Ri antibody. This particular antibody is typically associated with brain-stem encephalitis but can present with involvement of the limbic system, as was the case with this patient. Among these cases it is important to screen for lung or breast cancer.
This is another example of a 60-year-old gentleman who had a long-time history of smoking. He was transferred to our hospital with three-week history of new-onset refractory seizures. He was also having significant behavior problems and short-term memory loss. His MRI brain was found to be normal. However, his CSF showed 10 white blood cells. CSF evaluation for infectious etiologies was unremarkable.
If we calculate the APE2 score for this patient, he gets one point for new-onset seizures, one point for neuropsychiatric changes, two points for refractory epilepsy, and two points for inflammatory CSF. His total score comes out to be six. In this case, the autoimmune encephalopathy panel ordered in both CSF and serum came back positive for GABA-B receptor antibodies. This particular antibody is associated with cognitive dysfunction and refractory epilepsy. As we are seeing in this case and about 50‒60% of cases, patients have small cell lung cancer.
In the third example, a 72-year-old man with progressive memory loss over six years was evaluated in the neurology clinic. His MRI showed global atrophy but his CSF showed mildly elevated protein (65 mg/dl). He only gets two points for elevated CSF proteins on an APE2 score. However, his clinician sent an autoimmune dementia panel which later came back to be negative. This highlights one of the main benefits of using the APE2 score: optimal utilization of resources and in some cases avoidance of unnecessary testing.
In the fourth example, we are applying the APE2 score to a previously reported case. A 71–year-old woman presented with a six-week history of frequent involuntary jerking movements involving the right face and arm, as shown in the video. Her MRI brain showed bilateral medial temporal FLAIR hyperintensities.
If we calculate an APE2 score for this patient, she gets one point for new-onset seizures, two points for medial temporal changes on brain MRI, and three points for faciobrachial dystonic seizure, leading to a total of six. Her diagnostic workup was found to be remarkable for LGI1 antibodies.
Faciobrachial dystonic seizures are pathognomic for LGI1 autoimmunity. Some of these cases can have normal MRI brain and CSF analyses; however, most of these cases would still score four or more based on differential weighting of the APE2 score.
To conclude, APE2 scoring is a useful predictive model based on clinical criteria, which can aid in diagnosis and management of autoimmune epilepsy and encephalopathy. Very low APE2 scores may be useful in identifying patients who are more likely to have any negative neural-specific autoantibody evaluation. Furthermore, in some cases with epilepsy of unknown etiology, after reasonable exclusion of alternative etiologies, this model can help in early diagnosis and early immunotherapy initiation.