1 Introduction
Sodium valproate (VPA) is a widely utilized antiepileptic drug(AEDs)employed in the management of epilepsy and various psychiatric disorders.
1 Although generally well tolerated, VPA has been linked to specific adverse effects involving the hematopoietic, hepatic, and gastrointestinal systems.
2 Notably, hyperammonemia represents a significant idiosyncratic adverse reaction associated with VPA therapy.
3 This condition may precipitate valproate-induced hyperammonemic encephalopathy (VHE), a serious drug-related adverse event characterized by clinical manifestations including somnolence, vomiting, cognitive deficits, neurological impairments, and decreased consciousness in patients receiving VPA treatment.
4VPA is extensively utilized across multiple disease categories; however, a notable disparity exists in the incidence of VHE among these populations. This phenomenon is well-documented in published case reports. For example, Xianyun Liu described a case involving a 29-year-old female patient with epilepsy who developed impaired consciousness and significantly elevated blood ammonia levels following VPA administration, culminating in a diagnosis of VHE.
5 Such cases are commonly reported within the epilepsy cohort. Conversely, occurrences of VHE in psychiatric patients are exceedingly rare, with only isolated instances documented in the literature. One such example is provided by Baumgartner et al (2019), who reported a 26-year-old male with bipolar disorder who developed VHE subsequent to VPA treatment.
6 This patient presented with encephalopathic symptoms, including profound sedation and bradykinesia, alongside markedly elevated serum ammonia concentrations; notably, these symptoms resolved rapidly upon cessation of VPA. Collectively, these findings imply that the risk of VHE may be contingent upon the underlying disease. This observation prompts a critical scientific inquiry that remains insufficiently explored:whether systematic differences exist in the risk factors, clinical manifestations, and prognostic outcomes of VHE among populations with distinct primary diagnoses. To date, no comprehensive studies have rigorously examined this hypothesis beyond isolated case reports, thereby highlighting a significant gap in the current evidence base.
To address this research gap, the present study performed signal detection and quantitative analysis of VHE utilizing the U. S. FDA Adverse Event Reporting System (FAERS) database, complemented by a systematic review of published case reports. By conducting a comprehensive analysis of 134 clinical cases derived from both FAERS real-world data and the literature, the study specifically compared risk profiles, clinical presentations, and prognostic variations of VHE between patients with epilepsy and those with psychiatric disorders. The FAERS analysis offers robust validation for clinical case identification, while detailed case documentation further elucidates the clinical spectrum associated with these risks. Through the integration of multi-source data, this research aims to systematically characterize differential risk patterns of VHE across distinct patient populations, thereby providing high-quality evidence to inform precision clinical monitoring and medication safety decision-making.
2 Materials and methods
2.1 Data sources and study desig
The FAERS, one of the largest publicly available databases, facilitates the monitoring of adverse drug reactions.
7 It collects reports of adverse reactions reported from healthcare professionals, pharmaceutical man-ufacturers, and consumers worldwide. This database enables the detection and quantification of associations between drugs and reported adverse events. The present study aims to investigate potential risk factors for VHE. First, we performed a comprehensive exploratory analysis of VHE cases from the FAERS database. Subsequently, we conducted a review to thoroughly describe the clinical features of patients with epilepsy and psychiatric disorders at the onset of VHE. The flow chart for this investigation is displayed in supporting Fig. 1A and B.
2.2 Pharmacovigilance analysis
Adverse event reports related to VPA were retrieved from the FAERS database covering the period from January 2013 to September 2024. Each report encompassed clinical variables such as patient age, sex, indication for VPA use, medication details, therapy initiation date, as well as the timing and date of adverse event occurrence and outcomes. Adverse events were coded according to the Medical Dictionary for Regulatory Activities (MedDRA), with cases of VHE specifically identified using the Preferred Term (PT) code 10067327, based on MedDRA version 24.1.
Reports documenting VHE were subsequently selected for detailed analysis. Patients were stratified into two cohorts according to the indication for VPA therapy: epilepsy and psychiatric disorders. Descriptive statistics are presented using significance level and range. Differences in proportions between groups were assessed using the chi-square test with the N-1 correction, as implemented in MedCalc® Statistical Software. elevant demographic variables, including age and sex, were examined through binary logistic regression to determine risk factors associated with the development of VHE. Furthermore, the latency period from VPA initiation to VHE onset was compared between the epilepsy and psychiatric groups using violin plots generated with GraphPad Prism version 10.1.2.
2.3 Systematic literature review
The literature search encompassed two online databases, PubMed and Embase. A comprehensive search was conducted from the inception of these databases through February 10, 2025. The inclusion criteria were as follows: (i) human case reports, (ii) cases predominantly involving VHE, and (iii) cases providing relatively detailed patient information. Each article was categorized as included, excluded, or uncertain based on these criteria. The selection procedure was conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines. Two reviewers worked independently screened the literature and extracted the data. Extracted data included basic patient information(including age, gender, height, weight, etiology, and types of underlying diseases), administration details, clinical presentations, intervention strategies, and outcomes. In instances of incomplete or ambiguous data, the original publications were reviewed for clarification, and when necessary, corresponding authors were contacted to obtain additional details. Following data extraction, two researchers performed cross-validation, resolved discrepancies through discussion, and compiled a structured dataset for subsequent analysis. The World Health Organization's defined daily dose(DDD) metric was employed to quantify the daily dosage of each anti-seizure medication (ASM) and to calculate the total ASM burden.
8 Dosage standardization was achieved by computing the ratio of the prescribed daily dose (PDD) to the DDD (PDD/DDD).
9This study employed statistical analyses with recovery time following intervention as the primary outcome variable. Initially, univariate analyses were conducted using Kaplan–Meier (KM) survival curves to examine the effects of patient indications, age, and serum VPA concentrations. Subsequently, univariate Cox proportional hazards models were applied to evaluate the influence of age, clinical indications, blood ammonia levels, serum VPA concentrations, and VPA dosage on recovery time. Finally, a multivariate Cox proportional hazards analysis, adjusted for age and gender, was performed to assess their combined impact on recovery time and patient outcomes among individuals diagnosed with epilepsy and psychiatric disorders.
3 Results
3.1 Risk of VHE occurrence in epilepsy and psychiatric disorders from the FAERS database
A total of 549 cases of VHE were identified from the FAERS database, with VPA being the primary suspected agent. To systematically compare the clinical characteristics across different indications, this study included 312 cases involving patients with epilepsy and 237 cases involving patients with psychiatric disorders for subsequent analysis. Both cohorts predominantly comprised adult patients, with fewer than 50 pediatric and elderly patients in each group. Analysis of clinical outcomes revealed a significantly higher mortality rate in the epilepsy group compared to the psychiatric disorder group (7.1% vs. 3.0%)(Fig. 2A). Multivariate logistic regression, adjusted for age and sex, demonstrated that clinical indication independently influenced the risk of VHE occurrence. Specifically, patients with psychiatric disorders exhibited a significantly lower risk of developing VHE compared to those with epilepsy ([OR] = 0.60, 95% [CI]: 0.51–0.70), suggesting a potential protective effect of psychiatric disorders against VHE. Further stratification by age indicated that, relative to adult patients, both minors and elderly patients had a significantly reduced risk of VHE. Regarding pharmacotherapy, VPA monotherapy served as the control. Both VPA combined with one or two antiepileptic drugs were found to significantly increase the risk of HE, with the risk being particularly pronounced when combined with two drugs (OR = 1.56, 95% CI[1.21–2.00], P < 0.001) (Fig. 2B). Analysis of onset timing, based on 29 patients with complete temporal data (19 from the epilepsy group and 10 from the psychiatric disorder group), revealed a median onset time of 493 days ([IQR]: 245–1005 days) for the epilepsy group and 730.5 days(IQR: 351–1641 days) for the psychiatric disorder group (Fig. 2C).
3.2 VHE clinical characteristics in patients with epilepsy and psychiatric disorders from published
This study conducted a systematic review of the existing literature, encompassing a total of 134 cases of VHE. Analysis of case distribution indicated that 79.1% of patients (n = 106) were diagnosed with epilepsy, whereas 20.9% (n = 28) presented with psychiatric disorders. Demographically, adults constituted 59.0% of the cohort, with a slight predominance of female patients (53.7%) over males. Regarding clinical outcomes, the prognosis was generally favorable: 55.2% of patients achieved full recovery, and 38.8% exhibited symptomatic improvement. The overall mortality rate was 3.7%, with all fatalities occurring exclusively within the epilepsy subgroup (Table 1). Comparative analyses between the psychiatric disorder and epilepsy groups revealed statistically significant differences in several clinical parameters. Notably, all patients in the psychiatric disorder group demonstrated elevated blood ammonia levels, whereas 19.8% of the epilepsy group did not exhibit hyperammonemia (p < 0.05). Additionally, the proportion of patients with serum VPA concentrations exceeding 100 μg/mL was significantly greater in the epilepsy group (39.3%) compared to the psychiatric disorder group (17.0%) (p < 0.05). No other clinical parameters showed significant intergroup differences (Table 1).
Regarding pharmacological treatment characteristics, further analysis identified intergroup disparities. The median serum VPA concentration in the epilepsy group (237.3 μg/mL; n = 18) was significantly elevated relative to that in the psychiatric disorder group (142.0 μg/mL;n = 11) (p < 0.05). Both cohorts exhibited a drug load exceeding 2, indicative of prevalent polypharmacy. The epilepsy group most frequently received concomitant medications such as topiramate (TPM;10 cases), phenobarbital (PB; 7 cases), and phenytoin (PHT; 7 cases). Conversely, the psychiatric disorder group primarily utilized risperidone(RIS), lamotrigine (LTG), and lorazepam (LZP), each reported in a single case (Table 2). Concerning the temporal onset of VHE, the median interval from VPA initiation to symptom manifestation was 12 days in the psychiatric disorder group, marginally shorter than the 18 days observed in the epilepsy group; however, this difference did not reach statistical significance (p > 0.05). No significant differences were detected in the onset times of other suspected drug-related adverse events between the groups. Clinical management strategies were largely consistent across cohorts, predominantly involving cessation of VPA, substitution with alternative agents, and supportive symptomatic treatments (e.g., lactulose, levocarnitine), with detailed intervention protocols provided in Table 2.
3.3 Recovery time and influencing factors after VHE intervention from published
The statistical analysis of recovery times following intervention revealed that the median recovery duration for the mental illness cohort was significantly shorter than that observed in the epilepsy cohort (2 vs.5 days, p < 0.05) (Fig. 3A). This observation was corroborated by univariate Cox proportional hazards modeling, which demonstrated a faster recovery rate among patients with mental illnesses (Fig. 3D). Further subgroup analyses indicated that this disparity was especially marked in adult patients ([HR] = 2.55) and female patients (HR = 3.34) (Fig. 3E). Beyond clinical diagnosis, the study also examined additional variables potentially influencing recovery time, including age category (adult vs. children), VPA serum concentration (using a threshold of 100 μg/mL), gender, and blood ammonia levels. The findings indicated that none of these factors exhibited a statistically significant association with recovery time (Fig. 3B and C).
4 Discussion
VPA is employed across a wide range of patient populations, with its safety profile potentially varying according to specific clinical indications. This study primarily examines cases of VHE in individuals diagnosed with epilepsy and psychiatric disorders, aiming to elucidate the incidence risk, clinical characteristics, and outcomes of VHE within these two groups. Utilizing 549 encephalopathy reports entries from the FAERS database and 134 cases documented in the published literature, our analysis reveals that patients with epilepsy are at a comparatively higher risk of developing VHE than those with psychiatric conditions. Clinically, epilepsy patients exhibited elevated VPA concentrations and predominantly increased blood ammonia levels, although some cases presented with normal ammonia values. The prognosis of VHE was generally favorable, with an observed mortality rate of 3.7%. Notably, the duration of recovery following intervention differed between the two cohorts, with epilepsy patients experiencing longer recovery periods. This variation was especially pronounced among female and adults.
Due to their broad range of indications, AEDs are used to manage both epilepsy and various psychiatric disorders. The safety profiles of these medications across different patient populations is a key focus of research. Although previous studies have not specifically addressed VHE within these two cohorts, existing literature has examined differential toxicities associated with AEDs between them. For example, a pharmacovigilance study involving 2917 cases of lamotrigine use reported that patients with psychiatric conditions exhibited a higher incidence of arrhythmia and a sixfold increased risk of cardiac arrest compared to patients with epilepsy.
10 Our findings indicate that individuals with epilepsy are at an elevated risk of developing VHE. Clinically, VPA concentrations tend to be higher in epilepsy patients; notably, while some patients maintain normal blood ammonia levels, all psychiatric patients in our sample presented with elevated ammonia concentrations. We propose that these disparities may arise from two primary factors:first, the potential differences between epilepsy and psychiatric disorders in blood ammonia metabolism and hyperammonemia tolerance may arise from distinct core pathophysiological mechanisms, including variations in hepatic enzyme activity, blood-brain barrier integrity, and the efficiency of astrocytic ammonia detoxification. second, variations in combination pharmacotherapy regimens may contribute to differential risks of VHE. Preliminary data from our study suggest that combination therapy confers an additional risk for VHE. Compared to VPA monotherapy, VPA combined with one or two antiepileptic drugs was found to significantly increase the risk of HE, with the risk being particularly pronounced when combined with two drugs. Previous studies have shown that in epilepsy patients, concomitant administration of VPA with topiramate or levetiracetam was associated with a 2.012-fold increase in encephalopathy incidence compared to VPA monotherapy. Among psychiatric patients, co-treatment with VPA and antipsychotics such as olanzapine, quetiapine, or risperidone similarly heightened the risk of encephalopathy.
11 In case-report cohort, combination therapy accounted for 71.6% of cases, whereas monotherapy was comparatively infrequent, indicating that the augmented risk associated with combination regimens may be a contributing factor. Due to limitations in sample size, adjustments for potential confounders were not performed, and only the five most commonly used medications were enumerated. Consequently, the possibility of confounding effects arising from drug-induced factors within the treatment protocols remains a consideration.
Subsequent investigations have focused on established risk factors associated with VHE, including VPA dosage and blood ammonia levels. However, the precise relationship between these variables and the onset of VHE remains unclear. For example, a case series involving 30 patients diagnosed with VHE reported an average VPA dosage of 1336 ± 575 mg, with symptom onset occurring within 1 to 30 days. Notably, the occurrence and severity of hyperammonemic encephalopathy did not appear to correlate with VPA dosage, serum VPA levels, or treatment duration.
12 Most studies suggest that VHE is independent of both dosage and serum drug concentrations, with VHE manifesting even at therapeutic doses and serum levels.
13–17 Consistent with these findings, our study observed a mean VPA dose of 1225.2 ± 625.4 mg, with serum concentrations predominantly exceeding 100 μg/mL. However, existing literature predominantly documents early onset of VHE, typically within 30 days following initiation of VPA therapy. For instance, one investigation reported that among 20 VHE patients, nine achieved full recovery solely after discontinuation of VPA, whereas the remaining eleven required adjunctive treatments. The latency period for VHE onset ranged from 2 days to 2 months post-VPA initiation, with symptom resolution occurring between 1 and 14 days.
18 In contrast, data derived from the FAERS database indicate a predominance of delayed onset cases, with symptom emergence approximately 510 days after commencing VPA treatment, thereby revealing a notable discrepancy relative to published reports. To address these inconsistencies, we integrated data from both sources to reconcile differences attributable to heterogeneous data origins. Moreover, the association between blood ammonia levels and encephalopathy remains equivocal. Clinical evidence suggests that the correlation between blood ammonia levels and encephalopathic manifestations is not absolute. Some patients with elevated ammonia levels remain asymptomatic, whereas others exhibiting severe encephalopathy may present with only mildly elevated or even normal ammonia concentrations.
19 Our findings further revealed that 10.4% of patients with epilepsy maintained normal blood ammonia levels, underscoring the need for vigilance. Consequently, it is advisable that clinicians closely monitor for emergent neurological symptoms during VPA therapy regardless of blood ammonia measurements.
20Although the precise mechanisms underlying VHE are not yet fully understood, there is a general consensus regarding several critical components of its pathophysiology. VPA causes carnitine depletion of carnitine and undergoes extensive hepatic metabolism via glucur-onidation, mitochondrial β-oxidation, and ω-oxidation—metabolic pathways that rely on carnitine to the transport of VPA into mitochondria. Prolonged administration of high-dose VPA or acute overdose has been shown todeplete carnitine, which may enhance ω-oxidation pathways and increase levels of the toxic metabolite 4-en-VPA. These toxic metabolites inhibit carbamoyl phosphate synthetase, the enzyme responsible for catalyzing the conversion of ammonia to carbamoyl phosphate in the initial step of the urea cycle. Consequently, this inhibition contributes to elevated blood ammonia concentrations, exacer-bating the clinical condition.
21–25.
This study aimed to validate the differences in VHE between patients with epilepsy and those with psychiatric disorders by analyzing data from the FAERS database and case reports. However, the study has several limitations. First, the FAERS database functions as a voluntary reporting system, which inherently introduces selection bias, incomplete case information, and lacks a defined denominator for the drug-exposed population.
26 Consequently, it was not possible to determine the true incidence rate of VHE. Second, as a retrospective analysis based on previously reported cases, the study was limited by a small sample size, heterogeneity among patient populations, and insufficient data, resulting in non-significant statistical outcomes. Finally, potential confounding variables, including know drug interactions, were neither controlled for nor excluded in this investigation.
5 Conclusion
When VPA is used to treat epilepsy and psychiatric disorders, significant differences exist in the risk of VHE, clinical manifestations, and prognosis between these patient groups. Compared to patients with epilepsy, those with psychiatric disorders have a lower risk of developing VHE, often presenting with elevated blood ammonia levels and lower serum concentrations of VPA. Overall, patients with psychiatric disorders tend to recover more quickly, whereas the epilepsy group experiences some fatalities and a longer recovery period. These findings have important clinical implications for the safe use of VPA in these two populations. It is recommended to enhance monitoring of blood ammonia levels and VPA serum concentrations in high-risk patients and to develop individualized treatment strategies to reduce the risk of VHE. Further research is needed to elucidate the underlying mechanisms and optimize treatment management strategies for both patient groups.
2026 The Authors. Publishing services by Elsevier B. V. on behalf of Higher Education Press and KeAi Communications Co. Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).