Learn the nuances in diagnosing pediatric autoimmune encephalitis and distinguishing it from other causes of acute changes in behavior such as PANS/PANDAS. A Norton Children’s neuroimmunologist breaks down symptoms, diagnostic limits and referral guidance.

Published: October 14, 2026

A child shows up with a sudden, dramatic change in behavior — new obsessive-compulsive symptoms, food refusal or strange movements — and a worried family who’s already read online about pediatric autoimmune neuropsychiatric disorder associated with streptococcal infections (PANDAS), a subclass of pediatric acute-onset neuropsychiatric disorder (PANS). Could this be an autoimmune or inflammatory disease of the brain that requires urgent evaluation?

Acute behavioral changes in children can be caused by a wide spectrum of different factors, including autoimmune encephalitis and PANS. While there are overlapping clinical features that can make them appear similar at first glance, they differ in several aspects: On one end, we have well-documented brain inflammation; on the other, PANS-suspected but unproven immune process. That distinction should drive how you evaluate and refer, according to Michael L. Sweeney, M.D., pediatric neurologist and neuroimmunologist with Norton Children’s Neuroscience Institute, affiliated with the UofL School of Medicine.

PANDAS was first described in the early 1990s by National Institute of Mental Health researchers who identified a group of children with abrupt-onset tics or other movement disorders temporally linked to Group A streptococcal infection, with a hypothesized autoimmune mechanism. In 2008, the criteria were broadened into PANS — an umbrella term that keeps PANDAS as a subset but drops the requirement for a movement disorder and the requirement for a close temporal link to strep.

“PANS and PANDAS are not settled science — and treating it that way is what protects our patients. As we get better at finding real evidence of inflammation, we’ll get better at targeting treatment precisely to the kids who actually need it,” Dr. Sweeney said.

2 ends of 1 spectrum

Neuroinflammatory disorders exist on a spectrum. On one end are conditions where the inflammation is well-characterized — the antibody known and where it acts, and how patients respond to treatment. On the other end are conditions that are theoretically inflammatory, where a role for immune dysfunction is suspected but far from proven. Autoimmune encephalitis sits at the well-characterized end. PANS and PANDAS sit at the other.

Autoimmune encephalitis: What it looks like

Autoimmune encephalitis is inflammation of the cortex and deep gray structures of the brain. It can occur at any age and typically presents subacutely, building over weeks to months rather than appearing overnight. Watch for a symptom cluster that includes:

• New-onset seizures
• Movement disorders
• Encephalopathy — though this can wax and wane and isn’t always obvious on a single visit
• Autonomic instability

There’s no single gold-standard test. Diagnosis is a matter of building confidence — possible, probable or definite — using specific biomarkers (antibody testing in blood and spinal fluid) and nonspecific biomarkers (EEG, MRI, inflammatory markers in the blood and spinal fluid), alongside the clinical picture.

• What triggers it? In some cases, a preceding infection is the spark — anti-NMDA receptor encephalitis, for example, can follow herpes simplex virus infection, sometimes mimicking a relapse of the original infection when it’s actually antibody-driven inflammation. Tumors also can trigger it (an ovarian teratoma expressing neural tissue as the classic example), prompting the immune system to develop antibodies that then cross-react with the nervous system.
• Testing is a mixed bag. Spinal fluid findings vary by subtype. In anti-NMDA receptor encephalitis — the most common pediatric autoimmune encephalitis alongside myelin oligodendrocyte glycoprotein (MOG) antibody disease — cerebrospinal fluid (CSF) is usually abnormal: elevated, typically lymphocytic white cell count, elevated protein, and sometimes oligoclonal bands. In leucine-rich glioma-inactivated 1 (LGI1) antibody encephalitis, by contrast, CSF is often normal, and the antibody is more reliably found in serum than spinal fluid — suggesting a more systemic driver.
• Imaging is not reliable enough to rule anything out. MRI can show cortical or deep gray swelling certain sequences in some cases of autoimmune encephalitis, but in the majority of cases, imaging is unrevealing. Neither a normal MRI nor a normal spinal fluid study rules out autoimmune encephalitis.
• Treatment often starts with steroids, intravenous immunoglobulin (IVIG) and/or plasma exchange, escalating to other forms of immunosuppressing medications such as rituximab or cyclophosphamide, and in some centers to newer agents like tocilizumab, bortezomib, or inebilizumab, for refractory cases. Two clinical trials for pediatric anti-NMDA receptor encephalitis are currently underway.

PANS and PANDAS: An evolving hypothesis

PANS diagnostic criteria require all of the following:

• Abrupt, dramatic onset of obsessive-compulsive disorder (OCD) or severe food restriction, with a short (not precisely defined — hours to days, or days to weeks) time to peak severity
• Enough supportive symptoms that were not prominent features of the child before onset
• Symptoms that cannot be better explained by a preexisting condition (e.g., Tourette syndrome, autism or another known medical disorder)

That third criterion is easy to overlook clinically but is central to the diagnosis — PANS is a diagnosis of a genuinely new, abrupt change, not a label for a chronic condition that happens to flare.

PANDAS-specific criteria additionally require:

• Onset in the prepubertal range (most cases are 5 to 7 years old)
• An episodic, not chronic-progressive, course
• Episodes temporally correlated with Group A strep
• Some neurological abnormality on exam

Telling them apart at the bedside

	Autoimmune encephalitis	PANS/PANDAS
Driven by	Confirmed inflammation	Inflammation theorized, not established
Seizures	Can occur	Not seen
True encephalopathy	Can occur (may wax/wane)	Not seen
Typical age	Skews older, though can occur at any age	Typically younger, early school age
Spinal fluid	Variable — can be normal or abnormal depending on subtype	Normal by definition
MRI EEG	Variable — can be normal even with active disease	Normal by definition

Major professional bodies do not group PANS/PANDAS with autoimmune encephalitis, according to Dr. Sweeney, who noted that these are treated as distinct diagnostic categories, not points on the same continuum, despite how families sometimes conflate them.

Is there really inflammation in PANS/PANDAS?

The honest answer, according to Dr. Sweeney, is: still under investigation, and the target keeps moving. Early hypotheses focused on antibodies against basal ganglia antigens; more recent animal work has looked at antibodies targeting striatal cholinergic interneurons. Infusing serum from PANDAS patients into mice has shown antibody binding in the brain, but this hasn’t translated into consistent, reproducible human findings.

A commercially available panel tests for five antibodies (including those targeting dopamine receptors and CaM kinase II). A 2018 Swedish study found this panel has very poor specificity — cited by the American Academy of Pediatrics as roughly 6% to 8% — meaning nearly every child tested comes back with “abnormal” results regardless of whether they actually have PANS/PANDAS. This test is not covered by insurance and is not clinically useful; ordering it (or acting on an outside result) is more likely to mislead than to help, according to Dr. Sweeney.

MRI is normal in PANS/PANDAS by definition. Research-level studies have found subtle diffusion changes in deep gray structures (basal ganglia, thalami) in small cohorts, but these findings look like chronic, baseline structural differences rather than acute inflammatory changes, and haven’t translated into a clinically useful test. According to Dr Sweeney, that distinction argues against these being markers of active inflammation.

Dr. Sweeney points to an emerging theory worth knowing about (but not yet actionable): The driver of PANS/PANDAS may originate in the periphery rather than the brain itself. The thinking is that changes in the gut microbiome or innate immune regulation secondarily affect the blood-brain barrier or neurotransmitter metabolism. This is an active area of research, not established science.

Treatment reality check for PANS/PANDAS

Despite frequent parent reports that “something worked,” such as nonsteroidal anti-inflammatory drugs, various antibiotics, N-acetylcysteine or supplements, controlled studies have not shown benefit. Trials of amoxicillin, Augmentin, and azithromycin have not changed outcomes, reduced flare frequency or reduced Group A strep recurrence. A 2018 review of the PANS/PANDAS treatment literature found significant bias across most existing studies, meaning the evidence base is genuinely weak.

There is currently no recommendation to treat PANS/PANDAS with anti-inflammatory or immunosuppressive drugs as standard practice. The PANS/PANDAS consortium guidelines do describe using IVIG, steroids, plasmapheresis or rituximab in more severe or treatment-refractory cases, but these recommendations rest on low levels of evidence. Before committing a patient to a high-risk immunosuppressive therapy, it’s worth pausing to consider what else could explain the presentation.

Dr. Sweeney doesn’t rule these treatments out, but cautions that clinicians should think carefully about what else could explain the presentation before committing a patient to a high-risk immunosuppressive therapy.

When to refer

Refer urgently (neurology, and consider emergency evaluation) for suspected autoimmune encephalitis when you see:

• New-onset seizures
• Encephalopathy — altered awareness, cognition or behavior that may fluctuate
• Autonomic instability
• Movement disorder plus a subacute (weeks-to-months) decline
• Any of the above regardless of a normal MRI or normal initial labs — normal imaging and normal spinal fluid do not rule out autoimmune encephalitis, so a reassuring scan is not a reason to hold off on referral if the clinical picture fits

Refer for possible PANS/PANDAS workup when you see:

• Abrupt, dramatic new-onset OCD or food restriction in a young, prepubertal child, without seizures or encephalopathy
• A clear break from the child’s prior baseline — new symptoms, not a preexisting condition flaring
• Absence of a better explanation (rule out Tourette syndrome, autism or another known diagnosis first)
• For PANDAS specifically: An episodic course temporally tied to documented strep infection, plus some neurological exam finding

In either case:

Don’t order (or lean on) the commercial basal-ganglia antibody panel — its poor specificity makes it more likely to confuse the picture than clarify it.

A normal MRI/EEG/CSF is expected and doesn’t argue against PANS/PANDAS, but doesn’t support a diagnosis of exclusion for autoimmune encephalitis either — the two require different diagnostic logic.

According to Dr. Sweeney, a multidisciplinary approach is genuinely valuable here — neurology, psychiatry, psychology, rheumatology, immunology and infectious diseases specialties each can contribute a piece of the picture. Looping these specialties in early, rather than sequentially after each workup dead ends, tends to serve these patients better.

Bottom line

In Dr. Sweeney’s assessment, autoimmune encephalitis — at least the well-described subtypes like anti-NMDA receptor encephalitis and MOG antibody disease — has solid evidence behind its biomarkers and treatment response. PANS is best understood as a working, evolving clinical hypothesis rather than settled science. Some children who meet full PANS criteria with severe symptoms do respond to treatments like IVIG, but that decision involves considerably more nuance than a positive antibody panel or a parent-reported response to antibiotics. For both conditions, normal labs and normal imaging are not reassurance enough on their own — the clinical picture must drive the referral.