Autoimmune encephalitis patients identified among individuals attending Swedish memory clinics

A retrospective study of Swedish memory clinic patients revealed that 0.8% suffer from undiagnosed autoimmune encephalitis characterized by subtle symptoms and a lack of standard inflammatory markers, highlighting the need for more sensitive diagnostic tools to identify this treatable condition.

The Gist

Imagine your brain is a bustling, high-tech city. Usually, when people visit a "Memory Clinic," they are looking for signs that the city's infrastructure is slowly crumbling due to age or wear and tear. This is what we call neurodegenerative dementia (like Alzheimer's). The roads are getting potholes, the lights are flickering, and the city is just fading away.

But there is another problem that looks very similar on the surface: Autoimmune Encephalitis (AE). In this scenario, the city isn't just fading; it's under attack by its own security forces (the immune system). The security guards have gotten confused and are mistakenly shooting at the city's power plants (neurons), causing blackouts and chaos.

The big problem? The security guards are very quiet. In older patients, this attack doesn't always look like a war zone with explosions (seizures) or smoke (fever). Instead, it just looks like the city is running a bit slower. Because of this, doctors often mistake this "friendly fire" for simple aging, leaving the real problem untreated.

The Study: A Detective Search in Stockholm

Researchers in Sweden decided to play detective. They knew that in the Netherlands, about 1 out of every 100 people visiting a memory clinic actually had this "friendly fire" (autoimmune encephalitis) instead of just aging. They wondered: Is this happening in Sweden too?

They went into the archives of the Karolinska University Hospital in Stockholm, looking at the blood samples of 1,021 people who had visited the memory clinic between 2019 and 2023. These people all had memory problems, but most had been diagnosed with standard dementia or mild cognitive issues.

The Search Method:
Think of the researchers as using a super-sensitive metal detector. They didn't just look for rust (aging); they looked for specific "warrants" (antibodies) that prove the immune system is attacking specific parts of the brain. They tested for five different types of "warrants" (antibodies against proteins like CASPR2, LGI1, GABABR, NMDA-R, and IgLON5).

The Findings: Hidden Culprits Found

Out of the 1,021 people, the researchers found 8 individuals who had these "warrants."

• 0.8% of the group had autoimmune encephalitis.
• 7 of these 8 people had never been diagnosed with it before. They were walking around with an active immune attack on their brains, thinking they just had "old age memory loss."

The Twist:
These 7 people didn't look like the "classic" autoimmune patients you see in movies (who might be having seizures or acting wildly).

• They mostly just complained about forgetting things and mood swings.
• Their brain scans (MRIs) looked normal—no smoke, no fire.
• Their spinal fluid (CSF) often looked normal too—no signs of inflammation.

It was like finding a house that is on fire, but the smoke detectors haven't gone off, and the walls look fine. The fire was small and smoldering, but it was still there.

Why This Matters: The "Silent" Fire

The study highlights a major gap in our medical toolkit.

1. The Current Tools are Blunt: The standard tests doctors use to check for brain inflammation (like looking for white blood cells in spinal fluid or checking MRI scans) often miss these "silent" cases. In fact, 2 out of the 7 newly found patients had zero signs of inflammation in their standard tests.
2. The Cost of Missing It: Autoimmune encephalitis is treatable. If you catch it, you can give medicine to calm down the confused security guards, and the patient can often get better. If you miss it, the patient is treated for dementia (which doesn't work) and the brain continues to be damaged.

The Takeaway

The researchers are saying: "We need better smoke detectors."

Currently, if a patient walks into a memory clinic with just memory loss and no seizures, doctors often assume it's dementia. This study suggests that about 1 in 100 of these patients might actually have a treatable immune disorder that is hiding in plain sight.

The authors are calling for new, more sensitive ways to screen for this "silent fire" so that these patients can get the right treatment and potentially save their memories. It's a reminder that sometimes, the problem isn't that the city is old; it's that the security system is broken, and it needs a tune-up, not a demolition.

Technical Summary

1. Problem Statement

• Clinical Challenge: Neurodegenerative dementias are increasing in prevalence, but a treatable differential diagnosis, Autoimmune Encephalitis (AE), is frequently misdiagnosed as dementia, particularly in older adults where symptoms are subtle and progressive.
• Diagnostic Gap: Previous studies (e.g., in the Netherlands) suggest ~0.8% of patients with suspected neurodegenerative dementia actually have AE. However, in Sweden, AE prevalence data is inconsistent, suggesting potential under-diagnosis.
• Limitations of Current Markers: Standard diagnostic criteria (Graus criteria) often rely on overt signs of inflammation (seizures, MRI abnormalities, CSF pleocytosis). Late-onset AE often lacks these overt signs, making it difficult to distinguish from neurodegenerative dementia using current biomarkers (e.g., CSF cell count, albumin quotient, neurofilament light chain).
• Hypothesis: The authors hypothesized that a subset of Swedish patients seeking care for memory impairment harbor undetected AE and that these patients might exhibit aberrations in neuroinflammation markers, even if they lack classic clinical signs.

2. Methodology

• Study Cohort: A retrospective screening of 1,021 patients (median age 63, 58% female) attending memory clinics at Karolinska University Hospital (Stockholm) between 2019 and 2023. Patients presented with rapid or early-onset memory impairment.
• Screening Strategy:
  • Target Antigens: Sera were screened for autoantibodies against CASPR2, LGI1, GABABR, NMDAR (GRIN1), and IgLON5.
  • Primary Assay: Live Cell-Based Assays (CBAs) using HEK-293-MSR cells transfected with specific plasmids. Imaging was performed via automated microscopy (Nikon Ti2) with visual scoring.
  • Confirmation Protocol: Samples showing consistent positive signals underwent rigorous confirmation:
    1. Titrations of serum and CSF via live CBA.
    2. Commercial Fixed CBAs (Euroimmun).
    3. Tissue-Based Assays (TBAs) using rodent brain sections (performed by the Odense Autoimmune Neurology Group).
  • Definition of Positivity: A sample was confirmed positive if validated by at least two other tests beyond the initial serum live CBA. For NMDAR, CSF confirmation in two independent tests was required due to the risk of false positives in serum for affective disorders.
• Biomarker Analysis: The study compared antibody-positive patients against the antibody-negative cohort using:
  • Neuroinflammation markers: CSF leukocyte count, albumin quotient, kappa-free light chain intrathecal fraction (KFLC-IF), and oligoclonal bands (OCB).
  • Neuronal damage markers: CSF Neurofilament light (NFL) and Total Tau.
  • Neurodegeneration markers: Amyloid-beta, p-tau, and MRI findings (atrophy, white matter lesions).
• Statistical Analysis: Non-parametric tests (Mann-Whitney U, Fisher's exact test) were used to compare demographic and clinical features.

3. Key Results

• Prevalence: 8 out of 1,021 patients (0.8%) were confirmed to have autoantibodies.
  • Breakdown: 3 CASPR2, 2 GABABR, 1 LGI1, 2 NMDAR.
  • IgLON5: No positive cases found.
• Clinical Presentation:
  • 7 of the 8 positive patients had not been previously diagnosed with AE.
  • Symptoms: The primary presentation was memory impairment and affect lability. Only 3 patients showed clear psychiatric symptoms (hallucinations, OCD), and 2 showed neurological signs (seizures, neuropathic pain, neuromyotonia).
  • Cognitive Scores: MoCA and MMSE scores were not significantly different from the antibody-negative cohort.
  • Diagnoses: Most were diagnosed with Mild Cognitive Impairment (MCI); one was diagnosed with Alzheimer's disease.
• Paraclinical Findings (The "Silent" Nature of Late-Onset AE):
  • MRI: 0% of patients showed signs of active encephalitis (e.g., medial temporal lobe hyperintensity). Some showed age-appropriate atrophy or vascular changes.
  • CSF Inflammation: 2 out of 7 (28.6%) of the undiagnosed AE patients had completely normal CSF inflammatory markers (normal cell count, albumin quotient, KFLC-IF, and OCB).
  • Commercial Assays: 3 of the 8 confirmed patients were negative on commercial fixed CBAs, highlighting the superior sensitivity of the live CBA and tissue-based assays used in this study.
• Graus Criteria: Only 2 of the 7 undiagnosed patients fulfilled the "possible AE" paraclinical criteria (requiring abnormal MRI or CSF), yet all were considered clinically significant AE cases based on the antibody confirmation and clinical context.

4. Key Contributions

• Validation of Screening in Memory Clinics: Confirmed that AE is a present, albeit rare (0.8%), differential diagnosis in Swedish memory clinic cohorts, mirroring findings from the Netherlands.
• Demonstration of "Silent" AE: Provided strong evidence that late-onset AE can present with normal MRI and normal CSF inflammatory markers, challenging the reliance on these standard biomarkers for screening.
• Methodological Superiority: Demonstrated that live CBAs combined with tissue-based assays are more sensitive than commercial fixed CBAs for detecting specific neuronal surface antibodies in this population.
• Identification of Diagnostic Gaps: Highlighted that current standard biomarkers (CSF cell count, NFL, Tau) lack the sensitivity to identify the subset of AE patients who do not exhibit overt neuroinflammation.

5. Significance and Conclusion

• Clinical Implication: Autoimmune encephalitis should be considered in patients with memory impairment even in the absence of seizures, MRI abnormalities, or CSF pleocytosis. The "treatable" nature of AE makes missing these cases clinically significant.
• Need for New Biomarkers: The study concludes that current neuroinflammation markers are insufficient for screening memory clinic populations. The authors suggest investigating more sensitive markers, such as CSF IL-6, CXCL13, or flow cytometry for antigen-specific plasmablasts, to improve risk stratification.
• Public Health Impact: If even a fraction of the estimated 75,000 annual memory clinic investigations in Sweden harbor undiagnosed AE, the number of missed treatable cases could exceed the total number of conventionally diagnosed AE cases in the country.
• Future Directions: The authors advocate for the development of cost-effective, high-sensitivity screening protocols to identify these patients early, before irreversible neurodegeneration occurs.

Note: This is a preprint (medRxiv) and has not yet undergone peer review. The authors explicitly state the findings should not yet guide clinical practice without further validation.