Patients with autoimmune encephalitis (AE) often present with symptoms that are broadly characterized as psychiatric or behavioral, yet little attention is given to the precise symptomatology observed. We sought to more fully define the psychiatric symptoms observed in patients with anti–N-methyl-D-aspartate receptor (NMDAR), anti–glutamic-acid-decarboxylase 65 (GAD65), and anti–voltage-gated-potassium-channel complex (VGKC) antibody-mediated AE using the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition nomenclature.
We present a case series (n = 25) using a retrospective chart review of 225 patients evaluated for AE in a tertiary care academic medical center between 2014 and 2018. The included patients were ≤18 years old with anti-NMDAR AE (n = 13), anti-GAD65 AE (n = 7), or anti-VGKC AE (n = 5). The frequency of neuropsychiatric symptoms present at the onset of illness and time to diagnosis were compared across groups.
Psychiatric symptoms were seen in 92% of patients in our cohort. Depressive features (72%), personality change (64%), psychosis (48%), and catatonia (32%) were the most common psychiatric symptoms exhibited. On average, patients experienced impairment in ≥4 of 7 symptom domains. No patients had isolated psychiatric symptoms. The average times to diagnosis were 1.7, 15.5, and 12.4 months for anti-NMDAR AE, anti-GAD65 AE, and anti-VGKC AE, respectively (P < .001).
The psychiatric phenotype of AE in children is highly heterogenous. Involving psychiatry consultation services can be helpful in differentiating features of psychosis and catatonia, which may otherwise be misidentified. Patients presenting with psychiatric symptoms along with impairments in other domains should prompt a workup for AE, including testing for all known antineuronal antibodies.
Autoimmune encephalitis (AE) describes a spectrum of illness characterized by an immune-mediated encephalopathy along with evidence of inflammation in the central nervous system. Frequently, antineuronal antibodies are detected in the serum and/or cerebrospinal fluid (CSF) of patients with AE. Within the AE diagnosis there may be a wide range of symptoms, including personality changes, psychosis, seizures, symptoms of movement disorders, cognitive decline, memory deficits, and autonomic instability.1,2 AE is often highly responsive to immunomodulatory therapy. In anti–N-methyl-D-aspartate receptor (NMDAR) AE (the most common antineuronal antibody type), 80% of patients make a full or substantial recovery after immunomodulatory therapy.3 The demonstration of early treatment as an independent predictor of positive outcomes adds to the importance of early recognition by clinicians.4
However, there are several barriers to early diagnosis of AE that should be addressed. AE has a relatively low incidence of disease: in 1 study in the pediatric population, the researchers estimate the incidence of NMDAR AE to be 0.07 per 100 000 person years, and the incidence of anti–glutamic-acid-decarboxylase 65 (GAD65) AE to be 0.055 per 100 000 person years.5 Cases of anti-NMDAR AE account for just 1% of PICU admissions, even at major academic centers.6 This leads to fewer opportunities for clinical exposure to a disease that can present with a wide phenotypic variation as well as significant overlap with other primary neurologic or psychiatric conditions.1 Although there have been many empirical studies in which researchers characterize the clinical presentation of anti-NMDAR AE, there are far fewer focusing on the less commonly detected antibodies, such as anti-GAD65 and anti–voltage-gated-potassium-channel complex (VGKC).7–9 In most of the existing studies in which researchers characterize the symptomatic presentation of seropositive AE, the researchers include an adult patient population.7,9,10–13 To our knowledge, the only published accounts of AE in children with detected anti-GAD65 antibodies have been in the form of case reports or case studies.14–16 In addition, in the majority of existing studies on clinical phenotypes within AE, researchers vaguely describe symptoms as “psychiatric” or “behavioral,” without further delineating the specific psychiatric symptoms exhibited.7–9,11,13,17–19 Having a more detailed understanding of the spectrum of psychiatric symptoms commonly seen in children with AE may improve providers’ clinical recognition of this disease and highlight the potential role of psychiatrists in assisting with both the diagnosis and symptomatic treatment. We aim to report and compare the neuropsychiatric clinical phenotypes at the onset of disease among a cohort of children and adolescents with anti-NMDAR, anti-VGKC, and anti-GAD65 AE.
Methods
A total of 225 charts were reviewed from patients identified as having been evaluated for AE through the pediatric autoimmune brain disorders program and/or the pediatric psychiatry consultation service at a large academic institution from January 2014 to December 2018. Patients were included in this case series if they were ≤18 years of age at the onset of symptoms, had evidence of either anti-NMDAR or anti-GAD65 antibodies in their CSF or anti-VGKC antibodies in their serum, and met the Graus et al10 criteria for AE (Table 1) or limbic encephalitis (Table 2). The presence or absence of antibodies was determined before initiation of intravenous immunoglobulin treatment, either by report from an outside hospital after a patient’s transfer to our institution or by serum and CSF autoimmune antibody panels drawn at our institution and sent to a commercially available outside laboratory for analysis.20,21
Diagnostic Criteria for NMDAR AE, 2016
| Probable NMDAR Encephalitis Requires All of the Following |
|---|
| 1. Rapid onset of symptoms (<3 mo) in at least 4 of the 6 domains of altered consciousness, psychiatric, cognitive, language, movement, or autonomic dysfunction |
| 2. At least 1 of either abnormal EEG or CSF with pleocytosis or oligoclonal bands |
| 3. Reasonable exclusion of other disorders |
| Probable NMDAR Encephalitis Requires All of the Following |
|---|
| 1. Rapid onset of symptoms (<3 mo) in at least 4 of the 6 domains of altered consciousness, psychiatric, cognitive, language, movement, or autonomic dysfunction |
| 2. At least 1 of either abnormal EEG or CSF with pleocytosis or oligoclonal bands |
| 3. Reasonable exclusion of other disorders |
A definitive diagnosis requires NMDAR antibody to be present in CSF. Adapted from Graus F, Titulaer MJ, Balu R, et al. A clinical approach to diagnosis of autoimmune encephalitis. Lancet Neurol. 2016;15(4):391–404.
Diagnostic Criteria for Limbic Encephalitis, 2016
| Definite Limbic Encephalitis Requires All of the Following |
|---|
| 1. Subacute onset (<3 mo) of working memory deficits, seizures, or psychiatric symptoms |
| 2. Bilateral MRI or PET brain abnormalities of medial temporal lobes |
| 3. EEG with epileptic or slow wave activity in temporal lobes or CSF pleocytosis |
| 4. Reasonable exclusion of alternative causes |
| Definite Limbic Encephalitis Requires All of the Following |
|---|
| 1. Subacute onset (<3 mo) of working memory deficits, seizures, or psychiatric symptoms |
| 2. Bilateral MRI or PET brain abnormalities of medial temporal lobes |
| 3. EEG with epileptic or slow wave activity in temporal lobes or CSF pleocytosis |
| 4. Reasonable exclusion of alternative causes |
The detection of antineuronal antibodies allows for the diagnosis of definitive LE in the absence of any of the first 3 criteria. PET, positron emission tomography. Adapted from Graus F, Titulaer MJ, Balu R, et al. A clinical approach to diagnosis of autoimmune encephalitis. Lancet Neurol. 2016;15(4):391–404.
Data were retrospectively collected from patients’ electronic medical records. The demographic information collected included sex and age in years at onset of illness. Each patient’s chart was reviewed for a wide range of psychiatric and neurologic symptoms present at the onset of illness and before the initiation of immunomodulatory treatment (Table 3). The presence or absence of psychiatric and neurologic symptoms was determined by doing a retrospective review of electronic medical record documentation by child and adolescent psychiatry, pediatric neurology, and pediatric rheumatology providers who had conducted interdisciplinary interviews with patients and their families at the time of initial evaluation through the pediatric autoimmune brain disorders program. The presence or absence of symptoms was recorded in an all-or-none fashion to determine the relative prevalence of symptoms within each antibody subtype. Symptoms that were present before the onset of illness (on the basis of chart review and/or family report) were not included unless it was determined that these symptoms acutely worsened at the onset of illness and/or resolved with immunomodulatory treatment.
Patient Characteristics
| NMDAR (N = 13) | GAD65 (N = 7) | VGKC (N = 5) | Overall (N = 25) | |
|---|---|---|---|---|
| Characteristic | ||||
| Age at symptom onset, mean (minimum to maximum) | 9.5 (2.0 to 17.0) | 11.9 (4.0 to 18.0) | 7.4 (4.0 to 11.0) | 9.7 (2.0 to 18.0) |
| Female, n (%) | 4 (31) | 7 (100) | 4 (80) | 15 (60) |
| Previous psychiatric diagnosis, n (%) | 1 (8) | 0 (0) | 1 (20) | 2 (8) |
| Previous psychiatric medication, n (%) | 1 (8) | 0 (0) | 1 (20) | 2 (8) |
| Comorbid diabetes mellitus type 1, n (%) | 0 (0) | 5 (71) | 0 (0) | 5 (20) |
| Symptom | ||||
| Psychiatric disturbance, n (% [CIa]) | 13 (100 [77–100]) | 6 (86 [49–97]) | 4 (80 [38–96]) | 23 (92 [75–98]) |
| Depressive features | 9 (69 [42–87]) | 6 (86 [49–97]) | 3 (60 [23–88]) | 18 (72 [52–86]) |
| Anxiety | 1 (8 [1–33]) | 3 (43 [16–75]) | 2 (40 [12–77]) | 6 (24 [11–43]) |
| OCD | 2 (15 [4–42]) | 3 (43 [16–75]) | 1 (20 [4–62]) | 6 (24 [11–43]) |
| Personality change | 8 (62 [35–82]) | 5 (71 [36–92]) | 3 (60 [23–88]) | 16 (64 [45–80]) |
| ADHD | 0 (0 [0–23]) | 1 (14 [3–51]) | 3 (60 [23–88]) | 4 (16 [6–35]) |
| Psychosis | 8 (62 [35–82]) | 1 (14 [3–51]) | 3 (60 [23–88]) | 12 (48 [30–67]) |
| Mania | 0 (0 [0–23]) | 0 (0 [0–35]) | 1 (20 [4–62]) | 1 (4 [1–20]) |
| Catatonia | 7 (54 [29–77]) | 1 (14 [3–51]) | 0 (0 [0–43]) | 8 (32 [17–52]) |
| Cognitive decline, n (% [CIa]) | 13 (100 [77–100]) | 7 (100 [65–100]) | 5 (100 [56–100]) | 25 (100 [87–100]) |
| Concentration difficulty | 11 (85 [58–96]) | 6 (86 [49–97]) | 3 (60 [23–88]) | 20 (80 [61–91]) |
| Memory loss | 8 (62 [35–82]) | 5 (71 [36–92]) | 4 (80 [38–96]) | 17 (68 [48–83]) |
| Delirium | 6 (46 [23–71]) | 1 (14 [3–51]) | 0 (0 [0–43]) | 7 (28 [14–48]) |
| Language decline, n (% [CIa]) | 12 (92 [67–99]) | 5 (71 [36–92]) | 3 (60 [23–88]) | 20 (80 [61–91]) |
| Change in volume | 0 (0 [0–23]) | 1 (14 [3–51]) | 0 (0 [0–43]) | 1 (4 [1–20]) |
| Aphasia | 10 (77 [50–92]) | 3 (43 [16–75]) | 2 (40 [12–77]) | 15 (60 [41–77]) |
| Regressive speech | 1 (8 [1–33]) | 1 (14 [3–51]) | 1 (20 [4–62]) | 3 (12 [4–30]) |
| Dysarthria | 5 (38 [18–65]) | 1 (14 [3–51]) | 2 (40 [12–77]) | 8 (32 [17–52]) |
| Sleep disruption, n (% [CIa]) | 8 (62 [35–82]) | 3 (43 [16–75]) | 2 (40 [12–77]) | 13 (52 [34–70]) |
| Movement disorder, n (% [CIa]) | 9 (69 [42–87]) | 0 (0 [0–35]) | 0 (0 [0–43]) | 9 (36 [20–55]) |
| Seizures, n (% [CIa]) | 11 (85 [58–96]) | 6 (86 [49–97]) | 5 (100 [56–100]) | 22 (88 [70–96]) |
| Autonomic dysregulation, n (% [CIa]) | 7 (54 [29–77]) | 1 (14 [3–51]) | 3 (60 [23–88]) | 11 (44 [27–63]) |
| Average No. domains affected (out of 7), mean (CI) | 5.6 (5.0–6.2) | 4.0 (3.4–4.6) | 4.4 (3.1–5.7) | 4.9 (4.4–5.4) |
| Average time to diagnosis, mo, mean (CI) | 1.7b (0.01–3.4) | 15.5b (5.2–25.8) | 12.4b (10.2–14.6) | 7.7 (3.8–11.5) |
| Paraclinical parameters, n (% [CIa]) | ||||
| CSF + | 13 (100 [77–100]) | 7 (100 [65–100]) | 0 (0 [0–43]) | 20 (80 [61–91]) |
| Serum + | 8 of 9 (89 [56–98]) | 7 (100 [65–100]) | 5 (100 [56–100]) | 20 of 21 (95 [77–99]) |
| CSF pleocytosis | 9 (69 [42–87]) | 1 (14 [3–51]) | 1 (20 [4–62]) | 11 (44 [27–63]) |
| CSF elevated protein | 0 of 12 (0 [0–24]) | 0 of 5 (0 [0–43]) | 1 (20 [4–62]) | 1 (4 [1–20]) |
| Oligoclonal banding | 4 of 9 (44 [19–73]) | 4 (57 [25–84]) | 1 (20 [4–62]) | 9 (36 [20–55]) |
| MRI abnormalities | 6 (46 [23–71]) | 4 (57 [25–84]) | 3 (60 [23–88]) | 13 (52 [34–70]) |
| EEG abnormalities | 12 (92 [67–99]) | 6 (86 [49–97]) | 5 (100 [56–100]) | 23 (92 [75–98]) |
| Inflammatory markers: ESR and CRP | 2 (15 [4–42]) | 2 (29 [8–64]) | 1 of 3 (33 [6–79]) | 5 of 23 (22 [10–42]) |
| Vital sign instability | 8 (62 [35–82]) | 0 of 5 (0 [0–43]) | 0 of 4 (0 [0–49]) | 8 of 22 (36 [20–57]) |
| NMDAR (N = 13) | GAD65 (N = 7) | VGKC (N = 5) | Overall (N = 25) | |
|---|---|---|---|---|
| Characteristic | ||||
| Age at symptom onset, mean (minimum to maximum) | 9.5 (2.0 to 17.0) | 11.9 (4.0 to 18.0) | 7.4 (4.0 to 11.0) | 9.7 (2.0 to 18.0) |
| Female, n (%) | 4 (31) | 7 (100) | 4 (80) | 15 (60) |
| Previous psychiatric diagnosis, n (%) | 1 (8) | 0 (0) | 1 (20) | 2 (8) |
| Previous psychiatric medication, n (%) | 1 (8) | 0 (0) | 1 (20) | 2 (8) |
| Comorbid diabetes mellitus type 1, n (%) | 0 (0) | 5 (71) | 0 (0) | 5 (20) |
| Symptom | ||||
| Psychiatric disturbance, n (% [CIa]) | 13 (100 [77–100]) | 6 (86 [49–97]) | 4 (80 [38–96]) | 23 (92 [75–98]) |
| Depressive features | 9 (69 [42–87]) | 6 (86 [49–97]) | 3 (60 [23–88]) | 18 (72 [52–86]) |
| Anxiety | 1 (8 [1–33]) | 3 (43 [16–75]) | 2 (40 [12–77]) | 6 (24 [11–43]) |
| OCD | 2 (15 [4–42]) | 3 (43 [16–75]) | 1 (20 [4–62]) | 6 (24 [11–43]) |
| Personality change | 8 (62 [35–82]) | 5 (71 [36–92]) | 3 (60 [23–88]) | 16 (64 [45–80]) |
| ADHD | 0 (0 [0–23]) | 1 (14 [3–51]) | 3 (60 [23–88]) | 4 (16 [6–35]) |
| Psychosis | 8 (62 [35–82]) | 1 (14 [3–51]) | 3 (60 [23–88]) | 12 (48 [30–67]) |
| Mania | 0 (0 [0–23]) | 0 (0 [0–35]) | 1 (20 [4–62]) | 1 (4 [1–20]) |
| Catatonia | 7 (54 [29–77]) | 1 (14 [3–51]) | 0 (0 [0–43]) | 8 (32 [17–52]) |
| Cognitive decline, n (% [CIa]) | 13 (100 [77–100]) | 7 (100 [65–100]) | 5 (100 [56–100]) | 25 (100 [87–100]) |
| Concentration difficulty | 11 (85 [58–96]) | 6 (86 [49–97]) | 3 (60 [23–88]) | 20 (80 [61–91]) |
| Memory loss | 8 (62 [35–82]) | 5 (71 [36–92]) | 4 (80 [38–96]) | 17 (68 [48–83]) |
| Delirium | 6 (46 [23–71]) | 1 (14 [3–51]) | 0 (0 [0–43]) | 7 (28 [14–48]) |
| Language decline, n (% [CIa]) | 12 (92 [67–99]) | 5 (71 [36–92]) | 3 (60 [23–88]) | 20 (80 [61–91]) |
| Change in volume | 0 (0 [0–23]) | 1 (14 [3–51]) | 0 (0 [0–43]) | 1 (4 [1–20]) |
| Aphasia | 10 (77 [50–92]) | 3 (43 [16–75]) | 2 (40 [12–77]) | 15 (60 [41–77]) |
| Regressive speech | 1 (8 [1–33]) | 1 (14 [3–51]) | 1 (20 [4–62]) | 3 (12 [4–30]) |
| Dysarthria | 5 (38 [18–65]) | 1 (14 [3–51]) | 2 (40 [12–77]) | 8 (32 [17–52]) |
| Sleep disruption, n (% [CIa]) | 8 (62 [35–82]) | 3 (43 [16–75]) | 2 (40 [12–77]) | 13 (52 [34–70]) |
| Movement disorder, n (% [CIa]) | 9 (69 [42–87]) | 0 (0 [0–35]) | 0 (0 [0–43]) | 9 (36 [20–55]) |
| Seizures, n (% [CIa]) | 11 (85 [58–96]) | 6 (86 [49–97]) | 5 (100 [56–100]) | 22 (88 [70–96]) |
| Autonomic dysregulation, n (% [CIa]) | 7 (54 [29–77]) | 1 (14 [3–51]) | 3 (60 [23–88]) | 11 (44 [27–63]) |
| Average No. domains affected (out of 7), mean (CI) | 5.6 (5.0–6.2) | 4.0 (3.4–4.6) | 4.4 (3.1–5.7) | 4.9 (4.4–5.4) |
| Average time to diagnosis, mo, mean (CI) | 1.7b (0.01–3.4) | 15.5b (5.2–25.8) | 12.4b (10.2–14.6) | 7.7 (3.8–11.5) |
| Paraclinical parameters, n (% [CIa]) | ||||
| CSF + | 13 (100 [77–100]) | 7 (100 [65–100]) | 0 (0 [0–43]) | 20 (80 [61–91]) |
| Serum + | 8 of 9 (89 [56–98]) | 7 (100 [65–100]) | 5 (100 [56–100]) | 20 of 21 (95 [77–99]) |
| CSF pleocytosis | 9 (69 [42–87]) | 1 (14 [3–51]) | 1 (20 [4–62]) | 11 (44 [27–63]) |
| CSF elevated protein | 0 of 12 (0 [0–24]) | 0 of 5 (0 [0–43]) | 1 (20 [4–62]) | 1 (4 [1–20]) |
| Oligoclonal banding | 4 of 9 (44 [19–73]) | 4 (57 [25–84]) | 1 (20 [4–62]) | 9 (36 [20–55]) |
| MRI abnormalities | 6 (46 [23–71]) | 4 (57 [25–84]) | 3 (60 [23–88]) | 13 (52 [34–70]) |
| EEG abnormalities | 12 (92 [67–99]) | 6 (86 [49–97]) | 5 (100 [56–100]) | 23 (92 [75–98]) |
| Inflammatory markers: ESR and CRP | 2 (15 [4–42]) | 2 (29 [8–64]) | 1 of 3 (33 [6–79]) | 5 of 23 (22 [10–42]) |
| Vital sign instability | 8 (62 [35–82]) | 0 of 5 (0 [0–43]) | 0 of 4 (0 [0–49]) | 8 of 22 (36 [20–57]) |
CI, confidence interval; CRP, C-reactive protein; ESR, erythrocyte sedimentation rate.
By using the Wilson score interval method.
P < .001 for Kruskal–Wallis 3-way comparison for the outcome of the average months to diagnosis. P = .001 for average months to diagnosis when comparing NMDAR with GAD65 by using 2-way Mann–Whitney U test. P = .003 for average months to diagnosis when comparing NMDAR with VGKC by using 2-way Mann–Whitney U test. P = 1.0 for average months to diagnosis when comparing GAD65 with VGKC by using 2-way Mann–Whitney U test.
Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition criteria were used for determination of the presence or absence of symptoms and to categorize the psychiatric disturbances (listed in Table 3) that coincide with a formal diagnosis, such as major depressive disorder, obsessive-compulsive disorder (OCD), attention-deficit/hyperactivity disorder (ADHD), catatonia, generalized anxiety disorder, psychosis, and mania. One category of psychiatric disturbance (personality change) did not coincide with a formal Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition diagnosis. We defined personality change as new onset aggression, disinhibition, social withdrawal, or irritability, if the symptom severity caused functional impairment(s).
Additionally, 7 domains were used to categorize areas of impairment for each patient. These domains were psychiatric symptoms, cognition, language, sleep, seizures, abnormal movements, and autonomic dysregulation. Cognitive impairment was determined by neuropsychological or psychoeducational testing, family report, and/or bedside neurocognitive assessments, such as the Montreal Cognitive Assessment test, when available. The presence of seizures was determined on the basis of familial report or the presence of seizure activity on EEG, when available. Autonomic dysregulation was defined as persistent vital sign changes, syncope, or diaphoresis.
The time in months from the onset of symptoms to diagnosis and initiation of definitive treatment was recorded for each patient. Lastly, the presence or absence of abnormalities among vital signs, various serum and CSF laboratories, and imaging parameters before initiation of treatment was also recorded for each patient.
Categorical variables (ie, frequencies of the presence of symptoms or frequencies of paraclinical parameters) were reported as percentages ± 95% confidence intervals. Continuous variables (ie, months to diagnosis and number of domains affected) were reported as means ± 95% confidence intervals. Further statistical analyses, such as 3-way χ2 comparisons of the categorical variables, were considered to evaluate for statistically significant differences in symptom frequency among the antibody subtypes. However, these analyses were not used because of the small sample size of this case series and the subsequent implications on power. Given that such a large clinical difference was present in the average time to diagnosis, we performed a Kruskal–Wallis test to compare this measure across all 3 antibody subtypes. We then performed 2-way Mann–Whitney U test comparisons for the variable of average time to diagnosis across the possible 2-way permutations of antibody subtypes (anti-NMDAR versus anti-GAD65, anti-NMDAR versus anti-VGKC, and anti-GAD65 versus anti-VGKC).
Results
Demographic information, clinical symptoms, and other pertinent laboratory results are organized by antibody status in Table 3. A total of 25 pediatric patients diagnosed with AE were identified (13 anti-NMDAR, 7 anti-GAD65, and 5 anti-VGKC). All of the patients included in this study met the Graus et al10 criteria for AE on the basis of their clinical course, including the presence of psychiatric symptoms, seizures, cognitive and/or memory impairments, CSF changes, and/or MRI changes. The average age at the onset of symptoms for the total cohort was 9.7 years, with a mean time to diagnosis of 1.7 months for anti-NMDAR, 15.5 months for anti-GAD65, and 12.4 months for anti-VGKC AE (P < .001). When the 3 groups by antibody type are compared in pairwise fashion, the average time to diagnosis for patients with anti-NMDAR AE differed significantly from those with anti-GAD65 AE (P = .001) as well as those with anti-VGKC AE (P = .003). The majority of anti-NMDAR patients were male (9 of 13), whereas both the anti-GAD65 and anti-VGKC cohorts were majority female (11 of 12).
The most frequent symptoms were cognitive impairment (100%), psychiatric disturbances (92%), seizures (88%), language impairment (80%), and sleep disturbance (52%). Although psychiatric symptoms were present at disease onset in nearly all patients, it is important to note that none of the patients had psychiatric symptoms in isolation. Of the 23 patients with psychiatric symptoms, 100% had impairment in at least 2 other domains. On average, patients across the cohort had impairment in ≥4 domains (Fig 1). Table 3 and Figure 1 reveal that the type and severity of psychiatric symptoms were heterogenous across antibody subtype. Depressive features (72%) and personality changes (64%) were the most common psychiatric symptoms in our total cohort. Nearly one-half of our patients presented with psychosis (48%), and almost one-third of our patients had catatonia (32%). The most common symptoms exhibited by our patients with anti-NMDAR antibodies were personality changes, depressive features, psychosis, and catatonia. Additionally, the most common symptoms seen in our patients with anti-GAD65 receptor antibodies were personality changes, depressive features, anxiety, and OCD. Personality changes, depressive features, ADHD, and psychosis were the most common symptoms exhibited by patients with anti-VGKC antibodies.
Constellation of domains affected for each patient, by antibody type. Shaded boxes indicate the presence of the symptom. Individual psychiatric symptoms are presented in color for emphasis.
Constellation of domains affected for each patient, by antibody type. Shaded boxes indicate the presence of the symptom. Individual psychiatric symptoms are presented in color for emphasis.
In Table 3, we also summarize the prevalence of sleep disruption, symptoms of movement disorders, seizures, and autonomic dysregulation by antibody subtype. Half (52%) of our total cohort experienced disruption in sleep, and a large majority (88%) of our total cohort exhibited seizures. Abnormal movements indicative of a movement disorder were present in 69% of patients with NMDAR antibodies but absent in all other patients.
Diagnostic workup data, including laboratory and imaging findings, are also summarized in Table 3. All patients diagnosed with anti-NMDAR AE or anti-GAD65 AE had antibodies detected in their CSF (Table 3). However, none of the patients with anti-VGKC antibodies in their serum had anti-VGKC; leucine-rich, glioma inactivated 1 (LGI1); or contactin-associated protein-like 2 (CASPR2) subunit antibodies detected in their CSF.
Discussion
Although a distinct clinical phenotype for anti-NMDAR AE is well described and becoming more widely recognized, there are fewer descriptions of the clinical phenotype of anti-GAD65 and anti-VGKC AE in children in the existing literature. To our knowledge, this is the first case series in which the psychiatric phenotypes of a cohort of children with seropositive AE beyond anti-NMDAR AE are described. Children and adolescents are being admitted to pediatric medical wards at increasing rates for “psychiatric boarding” while awaiting inpatient psychiatric admission.22 It is critical for pediatric hospitalists to be aware that patients with anti-NMDAR, anti-VGKC, and anti-GAD65 AE can present with a wide variety of psychiatric symptoms, including those that can be more subtle, such as features of depression, negative symptoms of psychosis, or personality changes. Although catatonia and psychosis were present in many of the anti-NMDAR AE patients in our cohort, the patients with anti-GAD65 AE and anti-VGKC AE often presented with more subtle psychiatric symptoms. Patients presenting with psychiatric chief complaints can also be proactively evaluated for impairment in other core domains that may be associated with AE. Our cohort of 25 patients had impairment in ≥4 neuropsychiatric domains, on average. None of these patients, regardless of antibody type, presented with psychiatric symptoms alone, emphasizing the importance of diagnosing AE on the basis of the clinical syndrome of impairment in several neuropsychiatric domains. Establishing a premorbid baseline is another important component of case conceptualization.
Consultation and liaison psychiatrists can assist with diagnostic clarity, symptom management, and helping families navigate the illness course inherent to a diagnosis of AE.23 At the academic institution specific to this chart review, the psychiatry consultation team was involved early in the evaluation of any patient suspected to have AE. Psychiatrists can offer diagnostic clarity, particularly when distinguishing between symptoms of psychosis, agitated catatonia, and delirium. This distinction is essential, given the prevalence of these symptoms in patients with anti-NMDAR AE and potentially life-threatening complications of underdiagnosing and undertreating catatonia. Unrecognized catatonia can become a life-threatening condition (malignant catatonia), and unrecognized psychosis can lead to behavioral code events or other secondary harms. When patients are admitted with symptoms of catatonia and psychosis, medical decision-making will be necessary to determine which patients warrant a workup for organic etiologies (including a lumbar puncture). Additionally, symptom management with psychotropic medications (mood stabilizers, typical or atypical antipsychotics, benzodiazepines, antidepressants, stimulants, etc) is also frequently required to maintain patient safety and improve functionality.
In the setting of extreme agitation, children may be prescribed atypical antipsychotics and/or benzodiazepines. However, first- and second-generation antipsychotics may exacerbate symptoms of catatonia, whereas benzodiazepines have been observed to exacerbate symptoms of delirium.1 Given this diagnostic dilemma, psychiatric providers familiar with the use of clinical tools, such as the Bush Francis Catatonia Rating Scale and the Cornell Assessment of Pediatric Delirium, may be called on to aid in discerning the underlying etiology of altered mental status.24 As another concrete example, patients with anti-NMDAR AE may respond poorly to antipsychotic medications before initiation of immunomodulatory therapy. This phenomenon is thought to be a consequence of the intracellular internalization of the NMDAR after binding of the antibody and subsequent extravasation of the receptor with immunomodulatory treatment.16
Our sample size is not sufficient enough to elucidate differences in psychiatric presentations among the antibody subtypes. However, our data do illustrate the presence of overlap among the wide variety of psychiatric symptoms exhibited by patients with each of the 3 antibody subtypes. Because of the limited power of this study, it would not be possible to reliably identify the disease-causing antibody for the patients in this cohort solely on the basis of their clinical features. Therefore, diagnostic confirmation may be dependent on sending the full autoimmune encephalopathy serum and CSF antibody panels. Screening for a single antibody subtype may lead to a missed or unnecessarily delayed diagnosis. Early initiation of immunotherapy has been identified as an important predictor of improved outcomes for AE.25
In this cohort, patients with anti-NMDAR AE required an ICU level of care more often than patients with the other antibody subtypes. This difference in symptom severity may be one factor contributing to the clinically and statistically significant delay in diagnosis for each of the groups with anti-GAD65 and anti-VGKC AE, as compared with patients with anti-NMDAR AE. In theory, the symptom burden with anti-NMDAR AE is so high that providers may be more inclined to test patients earlier in their clinical course. Additionally, AE mediated by the NMDAR antibody has been described more frequently in the literature than cases mediated by the GAD65 and VGKC autoantibodies, likely contributing to higher provider awareness of anti-NMDAR AE.
The presence of serum antibodies alone should not be used to diagnose AE because there is a high prevalence of some antibody subtypes in the general population even in the absence of autoimmune disease. Serum anti-GAD65 antibodies in particular have a high prevalence within the general population. Testing for serum or CSF antibodies after immunotherapy has started can lead to false-negatives if the disease is being treated, whereas testing for serum antibodies after giving intravenous immunoglobulin can lead to false-positives. Testing for all antibody subtypes as part of the initial diagnostic workup can be helpful in preventing delays in diagnosis and minimizing the risk of inaccurate testing once immunotherapy has begun.
It should also be noted that none of the patients in our cohort with positive serum anti-VGKC screens had antibodies against the LGI1 or CASPR2 subunits of the VGKC antigen. This is consistent with descriptions of VGKC autoimmunity in children and adolescents in the existing literature in which pediatric patients with VGKC antibodies detected in the serum only rarely are positive for either the LGI1 or CASPR2 subunits.11,12,26–28 The absence of the anti-VGKC antibody in the CSF and “double negative” nature of the VGKC antibody (negative for LGI1 and CASPR2 subunits) in children differs from the antibody profile seen in adults.29 The significance of this difference between adults and children is unknown. Specifically, it is unclear whether these children have a subunit target of the VGKC antibody other than the LGI1 or CASPR2 subunits (yet to be identified) or if the VGKC antibody detected in the serum is simply a marker of autoimmunity.28,29 All 5 of our patients with anti-VGKC AE met the Graus criteria10 for AE,5 in addition to having the antibody in their serum; the lack of detection of LGI1 and/or CASPR2 subunits did not ultimately determine their clinical diagnosis.
Although, in our results, we detail the heterogenous neuropsychiatric phenotypes among the 3 most common antibody subtypes of AE in our patient panel, there are several limitations to our study. Because of the retrospective nature of our chart review, we relied on interpretation of the inherently unstandardized report of premorbid and new onset symptoms by patients and caregivers. Similarly, our methods relied heavily on the documentation by providers for gathering data regarding the presence or absence of symptoms, as opposed to consistent use of validated scales or measures. Many patients were seen as referrals (ie, to a tertiary care institution) several weeks after the onset of symptoms, precluding a standardized workup at the onset of disease for all patients. The severity of some patients’ symptoms, such as patients with anti-NMDAR AE who required an ICU level of care, may have inhibited the team’s ability to assess that patient for other specific symptoms, such as memory deficits, features of ADHD, mood symptoms, or anxiety. In particular, the language deficits exhibited by the majority of patients in our cohort may have prevented the detection of symptoms requiring self-report. Finally, our ability to perform a high-powered statistical analysis on the frequency of specific neuropsychiatric symptoms was limited by the small size of our cohort. The 95% confidence intervals for the frequency of many of our neuropsychiatric symptom variables are large. Despite these limitations, our descriptive statistics may help emphasize the wide heterogeneity with which psychiatric symptoms can present within seropositive AE in children. Multicenter prospective studies powered to find statistical differences in the prevalence of symptoms among a variety of antibody subtypes may be used to aid in future efforts to identify individual biomarkers and/or clinical symptoms as predictors of treatment response and outcomes.
Conclusions
In the existing literature, researchers frequently describe the psychiatric symptoms observed in children and adolescents with AE nonspecifically using inconsistent and vague terminology (ie, “behavioral” or “psychiatric”). With this case series we provide a more detailed description of the psychiatric symptoms observed in a cohort of pediatric patients with 3 antibody subtypes of seropositive AE. Successful diagnosis and treatment of patients with AE is dependent on a consistent framework for evaluation, timely testing for all known antibodies, and intentionally pursuing collaborative multidisciplinary partnerships between pediatric hospitalists, pediatric subspecialists, and consultation psychiatry teams.
Acknowledgments
We thank Marie Sarratt for her organization and coordination of the Autoimmune Brain Diseases clinic, and Daniel Childers, PhD, Julie Weber, MD, and Tracy Spears, MS, for their guidance on our statistical analysis.
FUNDING: No external funding.
Dr Mooneyham completed all work related to this article while employed at the Duke University School of Medicine. Dr Mooneyham is now an employee of the National Institute of Mental Health in Bethesda, Maryland. The views expressed in this article do not necessarily represent the views of the National Institutes of Health, the Department of Health and Human Services, or the United States Government.
Dr Adams conceptualized and designed the study, performed the chart review, completed the statistical analysis, and drafted the initial manuscript; Dr Mooneyham conceptualized and designed the study and guided the manuscript production process; and all authors approved the final manuscript as submitted.
Comments