Anti-NMDAR encephalitis is a recently described disorder that appears to be mediated by antibodies to the NR1 subunit of the receptor.9
In the current study, the diagnosis of eight pediatric patients over 8 months in a single institution suggests the disorder in children is more frequent than previously thought. A similar increase in the recognition of the disorder in children has been noted in referrals from other institutions. In a previous series of 100 patients, 22 were 18 years old or younger and of these 55% had an underlying tumor. In the current study of 81 patients, 40% were 18 years old or younger and of these only 25% (31% if only females were considered) had a tumor, all ovarian teratomas.9
The younger the patient the less likely a tumor was to be identified. Our findings also indicate the disorder is increasingly recognized in young boys.
In most respects the clinical picture of pediatric patients with anti-NMDAR encephalitis is similar to that of adults although there are differences in the frequency and manifestation of some symptoms. In the first series, which mostly included adults, 85% initially presented to psychiatrists for anxiety, agitation, paranoia, and visual or auditory hallucinations.9
In the current series, most adolescents had similar symptoms, but the recognition of psychosis in younger children posed more of a challenge. Parents described temper tantrums, behavioral change, agitation, aggression, and progressive speech deterioration as initial symptoms; however, these behaviors in children can initially be overlooked. Three patients had dystonia or dyskinesias as the presenting symptom with no behavioral change. Yet, as occurred in adults, most children eventually developed abnormal movements, including complex stereotyped motions involving the face, limbs, trunk, or abdomen, among other movements (see descriptions in ).2,10
Some of these movements were suspicious for seizures. Although electrographic seizures were verified in several children, the ictal changes were often subclinical and some movements causing concern for seizures had no EEG correlate. These findings caution against overtreating these patients with antiepileptic medications before a thorough video EEG evaluation.
The autonomic manifestations in children appear to be less severe than in adults. While 66% of patients in the original series exhibited central hypoventilation usually requiring weeks of mechanical support,9
only 23% of patients of the current series had hypoventilation. It is unclear whether these differences are the result of earlier diagnosis, or if they reflect developmental differences in the brainstem control of breathing or expression of NMDAR. Other signs of autonomic lability were more common and, in fact, all patients from CHOP had urinary incontinence and sleep dysfunction, including erratic sleep patterns, insomnia, and less frequently hypersomnia. Episodes of hypertension, tachycardia, or hyperthermia were frequent and often correlated with agitation, reminiscent of autonomic storming. However, different from adults, severe cardiac dysrhythmia (eg, requiring a pacemaker) did not occur in any of the children.4
Given the neuropsychiatric manifestations of the disorder, the differential diagnosis can be broad as evidenced by the extensive testing performed on the patients. Viral encephalitis is often the first presumptive diagnosis, suggested by the acute neurological change, CSF pleocytosis, and occasional hyperthermia.11
In a review of encephalitis by investigators of the California Encephalitis Project, 63% of cases ultimately had no identified etiology, suggesting that some subtypes could be immune-mediated.12
This hypothesis gains support with the identification of NMDAR antibodies in 10 of 19 (53%) of their cases that after extensive viral studies had been classified as “idiopathic encephalitis with psychiatric manifestations or dyskinesias” (C.A.G. and J.D., unpublished results). Another disorder frequently considered is neuroleptic malignant syndrome.4
However, symptoms of muscle rigidity, hyperthermia, elevated serum levels of creatine kinase (eg, case #8) and rhabdomyolysis may occur in patients with anti-NMDAR encephalitis that have not been treated with neuroleptics or antipsychotics.
The mechanisms that initiate this disorder are unknown. In a subgroup of patients, the presence of a tumor, usually a teratoma of the ovary that expresses NMDAR, likely contributes in triggering the immune response.1
The paradigm is similar to other autoimmune disorders such as myasthenia gravis or the Lambert-Eaton syndrome, which may occur with or without tumor association.13,14
Four patients of the current study had serum antibodies to antinuclear antibody (ANA), thyroid peroxidase, or both, which may suggest a propensity to autoimmunity. Extensive studies for a common infectious agent were negative, making an immune response by molecular mimicry unlikely (Supporting Information
). It should be noted that five patients had positive mycoplasma serologies but negative CSF polymerase chain reaction (PCR); however, the significance of this finding is unclear given the high prevalence of positive mycoplasma serologies in most series of pediatric encephalitis.11
Although in young girls with anti-NMDAR encephalitis the frequency of a teratoma is low compared with that of the adults, it does raise the question of what screening to perform and at what interval. Considering the age-related limitations of some tests (eg, transvaginal ultrasound; exposure to radiation with repeat CT scans), we recommend ultrasound and MRI of abdomen and pelvis. Based on the experience with some adults whose initial teratoma or second teratoma was identified many months after the first episode of encephalitis or at a relapse,2
periodic MRI or ultrasound studies for at least 2 years appears prudent.15
At this time, the number of male patients is too small for recommendations on tumor surveillance. Our initial series showed that two of nine male patients had tumors (one bilateral testicular teratoma and seminoma,16
and one small-cell lung cancer),9
but none of 12 subsequently diagnosed patients (six children) had tumors.
At CHOP, the experience with immunotherapy has been mixed. Some children appeared to respond rapidly to IVIg or methylprednisolone; however, this was not universal and, at times, even those who responded to an initial course did not seem to benefit after repeat treatment; some of these patients continued to slowly improve independent of the treatments used. A similar experience regarding the varied response to corticosteroids, IVIg, or plasma exchange was noted in patients from other institutions. This could relate to the fact that most children did not have tumors. Indeed, as described in adults,9
children who had a teratoma and were treated with tumor removal and immunotherapy had more frequent full recovery (p
= 0.03) and less relapses than those without a tumor. Nevertheless, 23 of 31 (74%) assessable patients had either full recovery (29%) or substantial improvement (45%). Although these results are notable, one should keep in mind the short follow-up of the current study. It is likely that many patients with mild deficits will attain full recovery, as seen in adults, but some may have relapses with or without residual deficits.
Based on this and previous studies,9
anti-NMDAR encephalitis should be suspected in children with acute behavioral change, seizures, dystonia, or dyskinesias, usually accompanied by (1) CSF lymphocytic pleocytosis or oligoclonal bands; (2) EEG with infrequent epileptic activity, but frequent slow, disorganized activity that does not correlate with most abnormal movements; and (3) brain MRI that is often normal or shows transient fluid attenuation inversion recovery (FLAIR) or contrast-enhancing abnormalities. Detection of antibodies to NR1 subunits of the NMDAR confirms the diagnosis. Patients should be examined for a tumor, usually an ovarian teratoma, and treated with immunotherapy. The recovery is slow, frequently over months, and requires a multidisciplinary team including physical rehabilitation and psychiatric management of the protracted behavioral symptoms.
Despite substantial progress in the recognition of this disorder, many questions remain. The trigger of the immune response is unknown, particularly in patients without teratoma; human leukocyte antigen (HLA) profiling to determine a genetic predisposition to autoimmunity should be considered in future studies. After tumor removal or immunotherapy, patients who do not improve usually have persistent high CSF antibody titers even though serum titers may be substantially decreased (J.D., personal observation). Prospective analysis of paired serum and CSF antibody titers in patients who do and do not improve is needed to determine the optimal clinical-immunological follow-up. It is unclear whether the early use of other immunotherapy (eg, cyclophosphamide, rituximab)6
to control the immune response within the central nervous system (CNS) would shorten the duration of symptoms. Given that early relapses appear to occur while tapering the immunotherapy and in patients without teratoma, the effectiveness of long-term immunosuppression should be assessed.