Neural Antibodies in First-episode Psychosis Patients with Warning Signs for Autoimmune Encephalitis

Abstract

Objective

Autoimmune encephalitis (AE) remains an essential differential diagnosis in patients with first-episode psychosis (FEP). In this study, we aimed to assess to prevalence of AE in a cohort of FEP patients.

Methods

We used a phenotype-driven algorithm to detect AE in patients with FEP. Initially, we screened patients for warning signs with a low or high pre-test probability for AE, defined as “yellow” and “red flags”, respectively. In the next step, patients with red flags underwent cerebrospinal fluid analysis (including neural antibodies), while patients with yellow flags underwent tests for serum neural antibodies, electroencephalography, and brain magnetic resonance imaging.

Results

We screened 78 patients with FEP and found that eight (10.3%) had at least one warning sign for AE four (5.13%) patients had at least one red flag, while four (5.13%) had only yellow flags. Among these, two patients (2.56%) had anti-N-methyl-D-aspartate receptor encephalitis, while the remaining six (7.69%) received a primary psychiatric disorder diagnosis.

Conclusion

Our study highlights the importance of considering AE in the differential diagnosis of FEP.

INTRODUCTION

Autoimmune encephalitis (AE) remains an essential differential diagnosis in first-episode psychosis (FEP) [1]. While there is debate among researchers, some suggest that it may be beneficial to test all FEP patients for neural antibodies [2,3]. However, implementing this approach in real-world clinical settings presents several challenges. To differentiate AE from a primary FEP, one must confirm the presence of neural antibodies in cerebrospinal fluid (CSF) [4]. However, lumbar punctures are rarely performed in psychiatric patients despite calls for a change in practice [5]. Additionally, most psychiatric facilities lack the resources needed for large-scale antibody testing [6]. Furthermore, AE prevalence in FEP patients is likely less than 1% [7,8]. To date, only four studies have systematically searched for CSF neural antibodies in patients with FEP. Out of 383 subjects, none of them tested positive for AE [3,9-11].

After considering everything that was mentioned previously, global testing for AE remains unfeasible. Therefore, some researchers suggest selective testing for patients displaying warning signs [12-15]. However, many of these signs are neurological and are impractical in specific clinical settings. For example, during the initial phases of anti-N-methyl-D-aspartate receptor (anti-NMDAR) encephalitis, over 90% of patients have isolated or predominant psychiatric features [16]. About 1 to 2 weeks after the psychiatric onset, neurological features such as seizures, autonomic dysfunction, or abnormal movements emerge [16]. However, about 4% of patients exhibit only psychiatric symptoms and do not develop any neurological signs [17]. As a result, a team of researchers proposed alternative diagnostic criteria for a new category of so-called autoimmune psychosis (AP) [13]. These criteria are designed for cases where AE manifests with isolated or predominant psychiatric features, yet they rely heavily on neurological symptoms. In turn, this can impact their effectiveness in clinical practice [3].

Here, we describe how we tackled these practical challenges in order to assess the prevalence of AE in a cohort of FEP patients. First, we adapted a previously described algorithm [3] as per the capabilities and resources of most psychiatric facilities. Then, we screened FEP patients using a set of warning signs for AE that covered a broad range of psychiatric, neurological, paraclinical, and medical features. Based on the pre-test probability for AE, we further analyzed CSF or serum neural antibodies along with electroencephalography (EEG) and brain magnetic resonance imaging (MRI), respectively. Concurrently, we assessed the performance of previously described warning signs [12,13] and criteria for AP [13].

METHODS

This study was conducted in accordance with the principles of the Declaration of Helsinki, and all patients provided written informed consent prior to enrollment. Ethics Committee from the Iuliu Hațieganu University of Medicine and Pharmacy in Cluj-Napoca, Romania, approved the study (ref: AVZ116/17.05.2022).

Participants

We recruited adult participants from the 2nd Psychiatric Clinic of the Emergency County Hospital in Cluj-Napoca, Romania. To be included in the study, patients had to meet the following criteria: (1) a diagnosis of FEP, which is characterized by new-onset disorganized behavior accompanied by hallucinations or delusions, and (2) the onset of symptoms had to be recent (< 3 months). Patients were excluded from the study if they had (1) a secondary cause, such as acute drug intoxication or any medical condition that could explain their current psychotic symptoms, or (2) a previous diagnosis of intellectual disability, which is defined as having an intelligence quotient < 70 accompanied by functional impairment.

Study Design

The patients included in the study were screened for warning signs of AE. These warning signs were categorized as “yellow” or “red flags” based on their pre-test probability for AE (Table 1) [3,12-15]. Patients with at least one warning sign were assigned to one of two algorithm branches shown in Figure 1. While only patients with red flags underwent CSF analysis (including neural antibodies), all underwent general, psychiatric, and neurological evaluation, as well as blood (including serum neural antibodies) and urine tests, EEG, and brain MRI (Supplementary Table 1; available online).

Table 1.

Warning signs for autoimmune encephalitis in patients with first-episode psychosis were defined as “yellow” (low pre-test probability) and “red flags” (high pre-test probability)

Yellow flags (↓Pre-test probability for autoimmune encephalitis)	Red flags (↑Pre-test probability for autoimmune encephalitis)
∙Rapid progression of psychosis (despite antipsychotic treatment) ∙Insufficient response to antipsychotics ∙Persistent hyponatremia (not explained by side-effects of medication) ∙Other autoimmune disorders (such as systemic lupus erythematosus or autoimmune thyroiditis)	∙Altered level of consciousness ∙Severe or disproportionate cognitive dysfunction ∙Movement disorder (catatonia, dyskinesia, and/or dystonia) ∙Autonomic dysfunction ∙Focal neurological signs ∙Aphasia, decreased verbal output and/or unexplained dysarthria ∙Persistent headache ∙Seizures ∙Intolerance to antipsychotics or neuroleptic malignant syndrome
	Comorbidities: ∙Viral prodrome (mild fever, malaise, anorexia, and/or headache) ∙Recently diagnosed or active tumor (such as ovarian teratoma or thymoma) ∙Recent (< 3 months) history of viral encephalitis ∙History of autoimmune encephalitis

Fig. 1.

A phenotype-driven algorithmic approach for detecting autoimmune encephalitis in patients with first-episode psychosis. Adapted from the algorithm proposed by Guasp et al. [3].

AE, autoimmune encephalitis; CSF, cerebrospinal fluid; EEG, electroencephalography; MRI, magnetic resonance imaging.

^aEEG anomalies: slowing, epileptiform discharges, or extreme delta brush; ^bMRI anomalies: cortical or subcortical transient fluid-attenuated inversion recovery changes.

Antibody Detection

Blood and CSF samples were collected and analyzed on the same day. Antibody analyses were conducted in the Research Center for Functional Genomics, Biomedicine and Translational Medicine, Iuliu Hațieganu University of Medicine and Pharmacy Cluj-Napoca, Romania. Laboratory personnel were blinded to all clinical infor-mation.

To detect neural IgG antibodies against cell surface antigens, commercial fixed cell-based assays (CBA) transfected with NMDAR GluN1 subunit, LGI1, CASPR2, AMPAR1, AMPAR2, and GABA_BR B1/B2 antigens (Euroimmun) were used. Serum samples were analyzed in a 1:10 dilution, while CSF was undiluted. The samples were considered positive if they showed specific fluorescence at the cut-off dilutions specified by the manufacturer.

Neural immunoglobulin G (IgG) antibodies against intracellular antigens were detected by immunoblotting using a commercial assay with nine different antigens: amphiphysin, CV2/CRMP5, PNMA2 (Ma2/Ta), Ri, Yo, Hu, recoverin, SOX1, and titin (Euroimmun). The serum samples were analyzed in a 1:101 dilution. The incubated test strips were scanned using a flatbed scanner and evaluated with the EUROLineScan software provided by the manufacturer.

All the analyses were carried out following the manufacturer’s recommendations.

Statistical Analyses

Data analysis was performed using IBM SPSS Statistics 29 (IBM Co.). We calculated descriptive statistics and conducted a t test to compare the age means between patients with and without warning signs for AE. In contrast, Fisher’s exact test was used to compare sex ratios between the same two groups. However, due to the low number of patients with warning signs, we did not attempt to statistically compare the patient subgroups.

RESULTS

We identified 90 eligible FEP patients, out of whom we excluded 12 due to secondary causes (n = 9) or intellectual disability (n = 3). We included and screened 78 FEP patients, of whom 70 (89.7%) had no warning signs for AE. None of these patients developed any warning signs afterward and were discharged with a primary psychiatric diagnosis. Meanwhile, eight patients (10.3%) had at least one warning sign (Fig. 2). Demographic information of the study cohort is given in Table 2. The two groups (with and without warning signs) were comparable for age (t = −1.872, p = 0.065) and sex (χ² = 0.116, p = 0.734).

Fig. 2.

Selection and inclusion of patients with first-episode psychosis (FEP) and warning signs for autoimmune encephalitis (AE).

Table 2.

Demographic information of the study cohort

	W−	W+	p value
Patients	70 (89.7)	8 (10.3)
Age (yr)	38.86 ± 12.73 (18, 81)	29.75 ± 15.7 (18, 59)	0.065
Sex (female/male)	69/1	8/0	0.734

Values are presented as number (%) or mean ± standard deviation (minimum, maximum).

W−, first-episode psychotic patients with no warning signs for autoimmune encephalitis; W+, first-episode psychotic patients with at least one warning sign for autoimmune encephalitis.

Warning Signs and Criteria for Autoimmune Psychosis

A detailed overview of the warning signs displayed by the eight patients is given in Table 3. Warning signs indicated by other studies [12,13] and AP criteria [13] were also assessed. Four patients (5.13%) had only yellow flags, three patients (3.85%) had both yellow and red flags, and one patient (1.28%) had only red flags (Fig. 2). Only the patients with at least one red flag (n = 4, 5.13%) fulfilled the possible AP criteria.

Table 3.

Warning signs for autoimmune encephalitis and criteria for autoimmune psychosis in the eight selected patients of our study

Patient no.	Sex	Age (yr)	Warning signs for autoimmune encephalitis	Warning signs for autoimmune encephalitis [12]	Warning signs for autoimmune encephalitis [13]	Criteria for autoimmune psychosis [13]
P1	F	21	2 Y: rapid progression of psychosis + insufficient response to antipsychotics	1 Y: rapid progression of psychosis (despite therapy)	2 WS: rapid progression + insufficient response to antipsychotics	None
P2	F	24	2 Y: rapid progression of psychosis + insufficient response to antipsychotics 2 R: disproportionate cognitive dysfunction + movement disorder (catatonia)	2 Y: rapid progression of psychosis (despite therapy) + catatonia	4 WS: rapid progression + insufficient response to antipsychotics + movement disorder (catatonia) + mutism	Possible autoimmune psychosis (2 criteria: movement disorder [catatonia] + severe or disproportionate cognitive dysfunction)
P3	F	50	3 Y: rapid progression of psychosis + insufficient response to antipsychotics + other autoimmune disorder (Hashimoto thyroiditis)	2 Y: rapid progression of psychosis (despite therapy) + other autoimmune disease	3 WS: rapid progression + insufficient response to antipsychotics + other autoimmune disorders	None
P4	F	22	5 R: altered levels of consciousness + movement disorder (catatonia + orofacial dyskinesia) + autonomic dysfunction + focal neurological signs + aphasia	6 Y: decreased levels of consciousness + abnormal postures or movements (orofacial dyskinesia) + autonomic instability + focal neurological deficits + aphasia + catatonia	5 WS: movement disorder (catatonia + dyskinesia) + focal neurological disease + decreased consciousness + autonomic disturbance + aphasia	Possible autoimmune psychosis (3 criteria: movement disorder [catatonia & dyskinesia] + a decreased level of consciousness + a clinically significant autonomic dysfunction)
P5	F	18	2 Y: rapid progression of psychosis + insufficient response to antipsychotics	1 Y: rapid progression of psychosis (despite therapy)	2 WS: rapid progression + insufficient response to antipsychotics	None
P6	F	18	2 Y: rapid progression of psychosis + insufficient response to antipsychotics	1 Y: rapid progression of psychosis (despite therapy)	2 WS: rapid progression + insufficient response to antipsychotics	None
P7	F	59	1 Y: other autoimmune disorder (cutaneous lupus erythematosus) 3 R: altered levels of consciousness + unexplained dysarthria + persistent headache	4 Y: decreased levels of consciousness + dysarthria + headache + other autoimmune disease 1 R: EEG abnormalities (epileptic activity)	4 WS: new-onset severe headache + decreased consciousness + dysarthria + other autoimmune disorders	Possible autoimmune psychosis (1 criterion: a decreased level of consciousness)
P8	F	26	2 Y: rapid progression of psychosis + insufficient response to antipsychotics 1 R: movement disorder (catatonia)	2 Y: rapid progression of psychosis (despite therapy) + catatonia 1 R: EEG abnormalities (slowing)	3 WS: rapid progression + insufficient response to antipsychotics + movement disorder (catatonia)	Possible autoimmune psychosis (1 criterion: movement disorder [catatonia])

Y, yellow flag; R, red flag; WS, warning signs; EEG, electroencephalography.

Relevant History and Main Clinical Features in the Eight Patients with Warning Signs

Three patients (3.85%) had a relevant medical history, while none had a psychiatric history. The prodrome was psychiatric in all patients with warning signs, with a median duration of 13 (interquartile range [IQR] 18.25) days. The median rating on the Positive and Negative Syndrome Scale was 103 (IQR 24.5), while the median score on the Montreal Cognitive Assessment (available for 5/8 patients with warning signs) was 18 (IQR 8). Besides psychosis, six patients (7.69%) with warning signs had other psychiatric symptoms as well: sleep disturbance (n = 4, 5.13%), catatonia (n = 3, 3.85%), mood symptoms (n = 2, 2.56%), other (short-term memory deficits; n = 2, 2.56%). Two patients (2.56%) developed neurological symptoms during the hospital stay. In one patient (1.28%), a tumor (ovarian teratoma) was found (Table 4).

Table 4.

Relevant history and main clinical features in the eight patients with warning signs

Patient no.	Sex	Age (yr)	Medical/ psychiatric historya	Prodromeb	Prodrome duration (days)	PANSS total (positive, negative, general)	MoCA	Psychiatric symptoms (other than psychosis)c	Neurological symptoms	Tumor
P1	F	21	None	Psychiatric	6	104 (36, 14, 54)	N/A	None	None	None
P2	F	24	Other (non-autoimmune hypothyroid-ism)	Psychiatric	1	136 (29, 42, 65)	18/30	Catatonia, other (short-term memory deficits)	None	Ovarian teratoma
P3	F	50	Autoimmune (Hashimoto thyroiditis)	Psychiatric	30	59 (20, 11, 28)	N/A	None	None	None
P4	F	22	None	Psychiatric	20	102 (32, 15, 55)	13/30	Mood, sleep, catatonia	Altered level of consciousness, ocular flutter, deglutition impairment, autonomic dysfunction, orofacial dyskinesia	None
P5	F	18	None	Psychiatric	50	101 (27, 23, 51)	13/30	Sleep	None	None
P6	F	18	None	Psychiatric	5	159 (35, 42, 82)	N/A	Mood	None	None
P7	F	59	Autoimmune (cutaneous lupus erythematosus)	Psychiatric	5	68 (28, 9, 31)	25/30	Sleep	Altered level of con-sciousness, persistent headache, dysarthria, gait disturbance	None
P8	F	26	None	Psychiatric	21	111 (26, 38, 47)	21/30	Sleep, catatonia, other (short-term memory deficits)	None	None

PANSS, Positive and Negative Syndrome Scale; MoCA, Montreal Cognitive Assessment; N/A, not available.

^aNeurological/autoimmune/psychiatric/other/none.

^bNeurological/psychiatric/viral/other.

^cMood/sleep/catatonia/other/none.

Antibody Findings

Per our protocol, we investigated serum antibodies on all patients who showed warning signs (n = 8, 10.3%). However, only the patients with at least one red flag (n = 4, 5.13%) had paired serum/CSF antibody investigation. Two patients (2.56%) had serum and CSF anti-NMDAR antibodies, leading to a diagnosis of definite anti-NMDAR encephalitis [4]. No discrepancies were observed between the paired serum/CSF samples. No other types of neural antibodies were found in either serum or CSF (Table 5).

Table 5.

Antibody and other paraclinical findings in the eight patients with warning signs

Patient no.	Cerebrospinal fluid								Serum		EEG	Brain MRI
	Neural IgG antibodies against cell surface antigensa	WBC count (/mm3)b	Total protein (mg/dl)c	AQd	IgG indexe	OCBs	Infectious and other markersf		Neural IgG antibodies against cell surface antigensa	Neural IgG antibodies against intracellular antigensg
P1	N/A								Negative	Negative	Normal	Normal
P2	Anti-NMDAR (1:32)	2	35.3	4.9	0.49	Negative	Negative		Anti-NMDAR (1:2,560)	Negative	Normal	Normal
P3	N/A								Negative	Negative	Normal	Normal
P4	Anti-NMDAR (1:64)	3	17.7	2.8	0.54	Negative	Negative		Anti-NMDAR (1:40)	Negative	Normal	Normal
P5	N/A								Negative	Negative	Normal	Normal
P6	N/A								Negative	Negative	Normal	Normal
P7	Negative	0	33.1	3.8	0.46	Negative	Negative		Negative	Negative	Abnormal (epileptiform discharges)	Normal
P8	Negative	0	22.9	2.7	0.42	Negative	Negative		Negative	Negative	Abnormal (slowing)	Normal

IgG, immunoglobulin G; WBC, white blood cell; AQ, albumin quotient; OCBs, oligoclonal bands; EEG, electroencephalography; MRI, magnetic resonance imaging; Anti-NMDAR, anti-N-methyl-D-aspartate receptor.

^aNMDAR, N-methyl-D-aspartate receptor; LGI1, leucine-rich glioma-inactivated 1; CASPR2, contactin-associated protein-2; AMPAR1, α-amino-3-hydroxy-5-methyl-4-isoxazole propionic acid receptor 1; AMPAR2, α-amino-3-hydroxy-5-methyl-4-isoxazole propionic acid receptor 2; GABA_BR B1/B2, gamma-aminobutyric acid B receptor B1/B2.

^bReference range: < 5 cells/mm³.

^cReference range: 15−45 mg/dl.

^dCerebrospinal fluid (CSF) albumin/serum albumin ×100, a marker of blood-brain barrier permeability. An AQ > 2.4 may indicate increased blood-brain barrier permeability [23].

^e(CSF IgG/serum IgG) × (serum albumin/CSF albumin), a measure of intrathecal IgG synthesis. An IgG index value > 1 may indicate intrathecal IgG production [23].

^fBacterial, fungal and BioFire multiplex PCR for microbial detection and cytopathology.

^gAmphiphysin; CV2/CRMP5, collapsin response mediator protein 5; PNMA2 (Ma2/Ta), paraneoplastic antigen Ma2; Ri, Yo, Hu, recoverin, SOX1, and titin.

Other Paraclinical Findings

By our algorithm, four patients (5.13%) had CSF analysis. None of the four patients had pleocytosis, elevated proteinorachy, increased IgG index, oligoclonal bands, infectious or other markers. However, all four patients had albumin quotients (AQ) > 2.4. Two patients (2.56%) with warning signs had abnormal EEG, while none had abnormal MRI (Table 5). The other paraclinical findings (Supplementary Table 1; available online) were unremarkable.

Diagnosis and Treatment at Last Follow-up

Performance of our algorithm

All patients with either “highly unlikely AE” (n = 4, 5.13%) or “AE ruled out” (n = 2, 2.56%) had primary psychiatric diagnoses and antipsychotic treatment at the last follow-up. None of them developed any warning signs in the meantime. The two patients (2.56%) with “confirmed AE,” according to our algorithm, had a diagnosis of definite anti-NMDAR encephalitis and tumor removal/immunomodulatory treatment at the last follow-up (Table 6).

Table 6.

Diagnosis and treatment at the last follow-up for the eight patients with warning signs

Patient no.	Diagnosis according to our algorithm	Diagnosis according to autoimmune psychosis criteria [13]	Cerebrospinal fluid study (neural antibodies included)	Diagnosis at 1-month follow-up	Treatment at 1-month follow-up
P1	Highly unlikely AE	-	N	Schizophrenia	Antipsychotic (clozapine)
P2	Confirmed AE	Possible APa	Y (antibody positive)	Anti-NMDAR encephalitis	Tumor removal, prednisone & azathioprine
P3	Highly unlikely AE	-	N	Delusional disorder	Antipsychotic (risperidone)
P4	Confirmed AE	Possible APa	Y (antibody positive)	Anti-NMDAR encephalitis	Methylprednisolone, IVIg & azathioprine
P5	Highly unlikely AE	-	N	Schizophrenia	Clozapine
P6	Highly unlikely AE	-	N	Schizophrenia	Antipsychotic (flupentixol, risperidone, quetiapine)
P7	AE ruled out	Non-autoimmune psychosis	Y (antibody negative)	Brief psychotic disorder	Antipsychotic (olanzapine)
P8	AE ruled out	Non-autoimmune psychosis	Y (antibody negative)	Schizophrenia	Antipsychotic (olanzapine)

AE, autoimmune encephalitis; AP, autoimmune psychosis; Anti-NMDAR, anti-N-methyl-D-aspartate receptor.

^aNo probable or definite autoimmune psychosis.

Performance of autoimmune psychosis criteria

Based on the authors’ recommendations [13], we performed CSF analysis in all (n = 4, 5.13%) patients who met the criteria for possible AP. Two patients (2.56%) were diagnosed with definite anti-NMDAR encephalitis [4] but did not meet the criteria for probable or definite AP due to a lack of paraclinical evidence. In the other two patients (2.56%), we established a diagnosis of non-AP due to a lack of CSF neural antibodies (Table 6).

Case Vignettes for Patients with Autoimmune Encephalitis

Patient 2 (P2)

A 24-year-old woman presented with interoceptive hallucinations, delusions, and disorganized behavior that started a few hours before admission. She was admitted to the psychiatric ward and received antipsychotics with minimal effect. Due to the acute onset of psychosis, she was referred to a neurologist. The neurological exam, screening work-up, EEG, and MRI were unremarkable. After 4 weeks, she developed catatonic signs (mutism, mannerisms, and echolalia), which prompted a lumbar puncture. Serum and CSF were both positive for anti-NMDAR antibodies. She was started on corticosteroids and underwent a pelvic MRI, which revealed an ovarian teratoma. After tumor removal, her psychotic and cognitive symptoms improved rapidly. During the hospital stay, she was started on azathioprine, which was discontinued after a few months. At the 1-year follow-up, the patient reported being virtually symptom-free without medication. She returned to work and did not have any relapse in the meantime.

Patient 4 (P4)

A 22-year-old female developed visual hallucinations, delusions, and disorganized behavior, which led her to be involved in a car accident. Three weeks earlier, she began exhibiting manic symptoms, including euphoria, increased energy, decreased need for sleep, and impulsive and risky behavior. The patient was admitted to the psychiatric ward, and 2 days later, she became lethargic. Over several days, the patient developed catatonic symptoms such as posturing, catalepsy, mutism, and staring. In addition, she also showed neurological signs such as ocular flutter, deglutition impairment, autonomic dysfunction, and orofacial dyskinesia. While the EEG and MRI were normal, she was found to have anti-NMDAR antibodies in both serum and CSF. She had no tumor on computed tomography (thorax, abdomen, and pelvis) or pelvic MRI. She received intravenous corticosteroids, intravenous immunoglobulins, and azathioprine and showed progressive improvement to baseline over six months. At 1-year follow-up, the patient was symptom-free without medication. She resumed her studies and did not have any relapse in the meantime.

DISCUSSION

In this prospective cohort of 78 patients with FEP, we found eight patients with at least one warning sign for AE, either yellow or red flags. Four patients had at least one red flag and underwent CSF analysis, which included neural antibodies. The remaining four patients had only yellow flags and underwent serum neural antibody testing, EEG, and brain MRI. Four of the patients met the criteria for possible AP. Two patients with both serum and CSF anti-NMDAR antibodies were diagnosed with anti-NMDAR encephalitis, indicating a 2.56% AE prevalence in this FEP cohort. The other six patients had no neural antibodies and received a primary psychiatric diagnosis. The AP criteria failed to identify patients with definite anti-NMDAR encephalitis as definite AP due to a lack of paraclinical criteria.

We designed this study by considering the challenges of real-world clinical practice. Since most psychiatric facilities cannot employ large-scale antibody testing [6], we opted for selective testing. In the first step, we screened our patients for a set of warning signs we compiled from previous studies [3,12-15]. We decided to exclude paraclinical warning signs such as CSF, EEG, or MRI abnormalities since these are not routinely performed in all FEP patients. We followed the footsteps of Herken and Prüss [12] by designing our warning signs as “yellow” and “red flags” to indicate low or high pre-test probability for AE, respectively. What sets us apart is that we designated non-neurological features as yellow flags while considering neurological symptoms and other severe features as red flags. We did this to fit the algorithm we used in the next step. The phenotype-driven algorithm we implemented was adapted from the one proposed by Guasp et al. [3]. Although serum neural antibodies are not detected in approximately 15% of patients with anti-NMDAR encephalitis [18], we believe this approach is still sound. However, whether adding EEG and brain MRI would increase sensitivity to acceptable levels is still unknown.

After reviewing the warning signs in the eight patients of our study, we observed that there were only minor differences compared to previously proposed warning signs [12,13]. The discrepancies were mainly related to paraclinical features (EEG changes in P7 and P8), which we did not include in our list. However, we observed a significant difference when assessing for AP criteria [13], as our warning signs were less stringent. The prevalence of possible AP in our cohort (5.13%) was much lower than in a recent study (20%) [3].

In addition to psychotic symptoms, most patients who displayed warning signs had other psychiatric symptoms. Among these, three patients showed signs of catatonia, two of whom were diagnosed with anti-NMDAR encephalitis. In one case (P2), the presence of catatonia was crucial in raising the suspicion of AE. Our findings confirm previous reports that catatonia is a crucial indicator of anti-NMDAR encephalitis [19,20].

In our study, the prevalence of AE was found to be 2.56%, which is about three times higher compared to previous studies conducted by Kelleher et al. [7] and Scott et al. [8], where it was 0.89% and 0.88%, respectively. We observed that in all cases where anti-NMDAR antibodies were present, they were found in both serum and CSF. The seropositivity rate was consistent with previous reports ranging from 0% to 12% [3]. Unsurprisingly, we found only anti-NMDAR antibodies. Up to 40% of patients with anti-NMDAR encephalitis are initially admitted to psychiatric wards due to isolated or predominant psychiatric symptoms [21]. In contrast, patients with other types of AE generally exhibit more pronounced neurological symptoms already from the initial phase of the disease [22], which reduces the likelihood of being admitted to a psychiatric ward.

Apart from the presence of anti-NMDAR antibodies, the CSF analysis in AE patients was unremarkable. While not reaching the pleocytosis threshold, patients with AE had detectable white blood cells in their CSF compared to those without AE. Nonetheless, all patients showed an increase in AQ, which suggests an increase in blood-brain barrier permeability [23]. Brain MRI was normal in all patients, including those with anti-NMDAR encephalitis, despite changes occurring in about 40% of cases [24]. In our two anti-NMDAR encephalitis patients, EEG abnormalities were absent, which is challenging as they usually occur in more than 95% of cases [25]. This peculiarity is probably due to the brief duration of the recording (20 minutes) along with the lack of cooperation from patients. On the other hand, two patients who showed suggestive EEG changes (focal epileptiform discharges in P7 and focal slowing in P8) did not have AE.

To be considered for possible AP, a patient must demonstrate at least one of seven features [13], all listed as red flags in our protocol. By considering other features listed as yellow flags, our algorithm identified twice as many patients as the AP criteria. Indeed, patients with yellow flags and “highly unlikely AE” had no paraclinical anomaly and did not develop warning signs for AE afterward. However, we believe that incorporating yellow flags in the assessment process will enhance the identification of AE with isolated or predominant psychiatric symptoms [17]. Due to their strong reliance on neurological features, the AP criteria may overlook these cases [3]. Our algorithm identified patients with definite anti-NMDAR encephalitis as “confirmed AE.” However, because they lacked paraclinical evidence, the AP criteria labeled them only as possible AP, raising further concerns about their performance in clinical practice.

There are several strengths to this study. Firstly, we were able to determine the prevalence of autoimmune cases of FEP. Data on this topic is limited due to recent studies being unable to provide prevalence figures [3,9-11]. Secondly, we provided a phenotype-driven algorithm that matches most psychiatric facilities’ capabilities and resources. This is important because large-scale neural antibody testing is still out of reach for most centers, even in developed countries [6]. Finally, this is the first study of its kind in Romania and, to our knowledge, in Eastern Europe. It remains to be determined whether the higher AE prevalence is due to population or patient sample peculiarities.

Nonetheless, there are significant limitations that must be taken into consideration. Firstly, our algorithm is provisional and needs to be validated for accuracy. Moreover, some warning signs, such as insufficient response to antipsychotics or disproportionate cognitive dysfunction, lack a clear definition and thus require further clarification. Secondly, the prevalence rate of AE in our cohort may have been underestimated since CSF analysis was not performed in all recruited patients. Thirdly, the participants were almost exclusively women due to the profile of the psychiatric ward we recruited from. The 2nd Psychiatric Clinic of the Emergency County Hospital in Cluj-Napoca provides care for men as well, but the closed ward is for women only. However, anti-NMDAR encephalitis has a more significant impact on females than males (8:2 distribution) [24], so the impact on the prevalence is likely to be small. Although our study was limited to an almost exclusively female population, including males would have likely decreased the prevalence. Finally, we did not perform confirmatory tests such as brain immunohistochemistry or live CBA, as they were unavailable in our center. Commercial CBAs have significant limitations, which are nevertheless more apparent in diagnosing anti-LG1, GABA_BR, or AMPAR encephalitis [26]. Since isolated psychiatric symptoms are almost exclusively found in anti-NMDAR encephalitis [16], commercial CBAs might prove useful in psychiatric settings.

To sum up, it is crucial to consider AE as a possible cause of FEP. While we did not perform large-scale neural antibody testing, we developed a phenotype-driven algorithm that could be applied in most psychiatric facilities. Selective testing, based on the pre-test probability, has yielded valuable results, although we may have underestimated the prevalence of AE in our cohort. To obtain more precise data, it is necessary to conduct international multicenter studies.

Funding Statement

Funding This work was supported by the Iuliu Hațieganu University of Medicine and Pharmacy, Cluj-Napoca, Romania (grants no. 881/41/12.01.2022 and 772/32/11.01.2023).