This study included 157 patients with a diagnosis of symptomatic epilepsy based on the etiologic classification of epilepsy (Fisher et al., 2014). All patients were recruited at Yuhuangding Hospital between December 2021 to February 2023. For each patient, demographic characteristics (age, BMI, VPA dose, concomitant drug therapy and liver function) were recorded. In addition, patients were divided into a normal liver function [NLF, the levels of all liver function indicators did not exceed the upper limit of normal (ULN), 112 subjects] group and an abnormal liver function (ABLF, at least one indicators exceed two-times of the ULN, 45 subjects) group according to liver function indicators as described previously (Chen et al., 2019). Meanwhile, patients with the following conditions were excluded from this study: 1) pre-existing of ABLF; 2) hyperhomocysteinemia; 3) other potential causes of liver disease including hepatitis A/B/C, HIV- positive or metabolic disease 4) smoking, alcohol or drug abuse; 5) less than 3-month of VPA-based treatment.

This study was approved by the Ethics Committee of the Affiliated Yuhuangding Hospital of Qingdao University (No. QDU-HEC-2023108). Informed consents were signed by patients or their legal guardians. Sample handling and data analysis protocols followed the clinical principles of Yuhuangding Hospital.

After fasting overnight for approximately 10 h, venous blood was collected from each patient using sterile clot activator tubes. Blood samples were centrifuged at 3500 rpm for 5 min and then analyzed within 30 min or stored at −80 °C for later analysis.

Liver function indicators, such as alanine aminotransferase (ALT), aspartate aminotransferase (AST), alkaline phosphatase (ALP), gamma-glutamyltransferase (γ-GT), total bilirubin (TBiL), total protein (TP), albumin (ALB) and Hcy were determined by an automatic biochemistry analyzer (Au5800, Beckman Coulter, United States). The FA and VB12 concentrations were determined by chemiluminescent immunoassay using automated techniques (e601, COBAS, Roche, Germany) based on the protocols given by the manufacturer.

In this study, patients underwent more than 3 months of VPA-based therapy to ensure a steady-state VPA concentration. Ethylenediaminetetraacetic acid (EDTA)-2Na tubes were used to obtain venous blood samples from each patient. To measure the steady-state VPA concentration, blood samples were collected just before the last VPA administration. Subsequently, the VPA concentration was quantitatively measured using an automatic fluorescence immunoassay system (Abbott, Chicago, United States) as described previously (Ma et al., 2019).

A total of 5 mL of peripheral blood was obtained from each participant using an EDTA tube and stored at −20 °C. Genomic DNA was extracted by DNA Extraction Kit (OMEGA, Norcross, United States) according to the manufacturer’s protocols. The MTHFR A1298C, MTHFR C677T, MTR A2756G and MTRR A66G polymorphisms were detected directly by DNA sequencing after PCR amplification using an automatic genetic analyzer (Applied Biosystems, United States). Details of genetic polymorphisms and sequences of primers used for genotyping procedures are presented in Supplementary Table S1.

The concentration of GSH was quantitatively measured according to the protocol given by Chien’s study (Chien et al., 2016). Briefly, 10 μL of serum was mixed in 990 μL cold phosphate (10 μM)/EDTA (5 μM) buffer. Then, add 1 volume ice trichloroacetic acid (TCA) into 5 volumes of sample, and centrifuge samples at 12,000 g for 5 min. Next, the supernatant was neutralized by NaHCO₃ and measured spectrophotometrically at 405 nm. Serum SOD activity and TBARS concentrations were determined by Beckman Coulter ACCESS ^® (Brea, CA, United States) and suitable kits according to the manufacturer’s instructions.

The statistical analysis in this study was performed using SPSS (version 20.0; IBM, United States). For demographic characteristic analysis, Student’s t-test was used to evaluate significant differences between the NLF and ABLF groups. The statistical significance of the differences in VPA concentration and concomitant drug were determined by Student’s t-test or Fisher’s exact test. Comparisons of genotype frequencies were performed by χ2 (chi-square) tests. Logistic regression was used to evaluate the risk factors for VPA-induced liver dysfunction. Data are presented as the mean ± standard deviation. p-value less than 0.05 indicated a statistically significant difference.

This study included 157 pediatric patients (aged: 6.93 ± 6.02 years in the ABLF group and 6.16 ± 4.80 years in the NLF group) with VPA-based therapy. Patients were divided into ABLF and NLF groups based on the results of liver function tests. The demographic characteristics of ABLF and NLF patients are summarized and listed in Table 1. Patients in ABLF group exhibited significantly higher ALT, AST, γ-GT and TBiL levels than those in NLF group (P < 0.05). Moreover, no differences in age, BMI, TP, ALB or globulin (GLO) levels were observed between ABLF and NLF groups (Table 1).

In this study, all patients underwent VPA-based therapy (monotherapy or polytherapy). We recorded the VPA dose and concomitant drug used for each patient in ABLF and NLF groups. Moreover, we analyzed the VPA concentration by an automatic fluorescence immunoassay system. As shown in Table 2, there were no significant differences in VPA dose, VPA concentration or adjusted VPA concentration between two groups, suggesting that neither the VPA dose nor the concentration may be directly associated with VPA-induced liver dysfunction.

In addition, carbamazepine (ABLF: 2, NLF: 4), clonazepam (NLF: 2), lamotrigine (ABLF: 2, NLF: 5), levetiracetam (ABLF: 1, NLF: 3), topiramate (ABLF: 2, NLF: 4) and oxcarbazepine (NLF: 1) were combined with VPA therapy. However, no significant association was found between concomitant drugs and liver dysfunction in the ABLF or NLF groups (Table 2).

To evaluate the association between OCM and VPA-induced liver dysfunction, we first analyzed the levels of OCM-related nutrients in the ABLF and NLF groups. As shown in Table 2, there was a clear separation between the two groups. Patients in ABLF group had significantly higher Hcy level than that in NLF group (P < 0.05). In contrast, the FA and VB₁₂ concentrations were significantly lower in ABLF group than those in NLF group (P < 0.05).

Then, we explored the effect of VPA treatment on OCM, correlation analysis was performed and exhibited in Figure 2. As shown in Figure 2, there was a significant negative correlation between the plasma VPA concentration and FA concentration (r = −0.2962, P < 0.05), and a significant positive correlation between the plasma VPA concentration and Hcy concentration (r = 0.4387, P < 0.05). Interestingly, although the VB₁₂ concentration displayed a separation between ABLF and NLF groups, there was no significant correlation between the plasma VPA concentration and VB₁₂ concentration (p > 0.05). These results suggest that OCM pathway contributes to VPA-induced liver dysfunction.

To further elucidate the association between OCM and VPA-induced liver dysfunction, we analyzed the OCM pathway-related genetic variants (MTHFR A1298C, MTHFR C677T, MTR A2756G and MTRR A66G) in the ABLF and NLF groups. The Hardy-Weinberg equilibrium test and allele frequencies of the four genes are listed in Supplementary Table S2. The associations between the four gene polymorphisms and VPA-induced liver dysfunction are listed in Table 3. In detail, the frequencies of AC and CC genotypes in MTHFR A1298C were significantly higher in ABLF group than those in NLF group (48.9% versus 31.2%, p = 0.004; 15.5% versus 1.8%, p = 0.001, respectively). Meanwhile, there was a significantly increased frequency of TT genotype of MTHFR C677T in ABLF group when compared to NLF group (15.6% versus 6.3%, p = 0.023). However, there were no statistically significant correlations between the MTR A2756G or MTRR A66G genotype and VPA-associated liver dysfunction (p > 0.05, Table 3). These results suggested that MTHFR A1298C and MTHFR C677T may be risk factors for VPA-induced liver dysfunction.

The associations between OCM-related genetic polymorphisms and nutrient levels (FA, Hcy and vitamin B₁₂) in ABLF and NLF patients were analyzed. As shown in Table 4, we observed that the MTHFR A1298C polymorphism was associated with serum Hcy (AA: 7.25 ± 1.70; AC: 8.16 ± 1.93; CC: 9.63 ± 3.81, p = 0.001) and FA (AA: 19.17 ± 2.89; AC: 17.92 ± 3.70; CC: 17.97 ± 3.56, p = 0.001) levels. In addition, the MTHFR CC genotype resulted in higher Hcy levels (P < 0.01) and lower FA levels (P < 0.01) than the AA genotype (Figure 3). Interestingly, although MTHFR C677T showed a different genotype frequency between the ABLF and NLF groups, the differences in the MTHFR C677T polymorphism among OCM-related nutrient levels were not statistically significant (p > 0.05, Table 4; Figure 3).

We further analyzed the associations between MTR A2756G and MTRR A66G polymorphisms and OCM-related nutrition levels. However, there were no statistically significant differences between MTR A2756G and MTRR A66G polymorphisms in OCM-related nutrition levels (p > 0.05, Supplementary Table S3).

Oxidative stress is well known to be involved in the progression of VPA-associated liver dysfunction (Chang and Abbott, 2006; Gai et al., 2020). Hence, we examined oxidative stress-related parameters. As shown in Table 5, a significantly higher TBARS concentration and a lower GSH concentration were observed in ABLF patients than in NLF patients (P < 0.05). These results demonstrated that oxidative stress is associated with VPA-induced liver dysfunction.

Considering the potential link between VPA administration and oxidative stress, we next performed correlation analysis to explore the influences of VPA concentration and oxidative stress. In this study, the VPA concentrations was significantly positively correlated with TBARS concentration (r = 0.2694, p = 0.006) and a significantly negative correlation with GSH concentration (r = −0.2233, p = 0.0049, Figures 4A,B), suggesting that the VPA concentration is associated with oxidative stress.

Importantly, Hcy is involved in redox signaling via regulating ROS generation and GSH levels (Gomez et al., 2011). For this reason, we analyzed the correlation between Hcy levels and oxidative stress. As shown in Figures 4C,D, we found that Hcy concentration exhibited highly correlated with GSH and TBARS concentrations (r = −0.6065, P < 0.0001; r = 0.6564, P < 0.0001, respectively). Based on these findings, Hcy-related oxidative stress might be involved in the pathogenesis of VPA-induced liver dysfunction.

To further clarify the associations of OCM-related nutrients and genetic factors with VPA-induced liver dysfunction, logistic regression analysis was performed. As shown in Table 6, the MTHFR A1298C polymorphism (r = 1.646, p = 0.001), MTHFR C677T polymorphism (r = 1.162, p = 0.007) and Hcy level (r = 1.334, p = 0.030) exhibited a significant positive correlation with VPA-associated liver dysfunction. Moreover, there were significant negative correlations between FA, GSH concentrations and VPA-induced liver dysfunction (r = −0.602, p = 0.025; r = −1.258, p = 0.006, respectively). Interestingly, although TBARS concentrations displayed only a marginally significant association with VPA-caused liver dysfunction (p = 0.067), there was a 2.6-fold (95% CI: 0.693–6.436) increased risk for liver dysfunction progression. These results indicated that MTHFR A1298C/C677T polymorphisms and GSH/Hcy/FA concentrations may be risk factors for VPA-induced liver dysfunction.