To identify risk factors for hyperammonemia in pediatric patients with epilepsy.
To identify risk factors for hyperammonemia in pediatric patients with epilepsy.
A total of 2,944 pediatric patients (ages 0–15 years) were classified into the following three groups: a group without drug treatment (n = 445, group I), a group receiving antiepileptic drugs other than valproic acid (VPA) (n = 673, group II), and a VPA-treated group (n = 1,826, group III). Hyperammonemia was defined as a plasma ammonia level exceeding 100 μg/dl with reference to the standard range and previous reports.
The mean ammonia level of groups I, II, and III was 36.0, 56.0, and 86.8 μg/dl, respectively, and the incidence of hyperammonemia was 1.6%, 7.7%, and 31.7%, respectively. In each group, the mean ammonia level of patients aged 3 years or younger was significantly higher than that of patients aged 4–15 years. In group II, concomitant use of topiramate and zonisamide were risk factors for hyperammonemia (adjusted odds ratio [OR] 3.9, 95% confidence interval [CI] 1.7–9.2, and OR 3.5, 95% CI 1.9–6.5, respectively). In group III, the ammonia level increased in a VPA dose–dependent manner. At a VPA dose of 30 mg/kg, there was 4.3-fold increase in the incidence of hyperammonemia. The other significant risk factors identified were female gender (OR 1.3, 95% CI 1.0–1.6), symptomatic generalized epilepsy (OR 1.4, 95% CI 1.1–1.8), and the concomitant use of phenytoin (OR 4.7, 95% CI 3.3–6.9), phenobarbital (OR 2.2. 95% CI 1.6–3.2), acetazolamide (OR 6.6, 95% CI 2.5–17.2), topiramate, or zonisamide.
A young age and concomitant use of carbonic anhydrase inhibitors are associated with an increased risk of hyperammonemia regardless of whether the patient is taking VPA. In patients receiving VPA, concomitant use of phenytoin and/or phenobarbital enhances the risk of hyperammonemia. An increase in ammonia can be caused by multiple factors. Our results may help clinicians to avoid problems of hyperammonemia.
Hyperammonemia is a frequent problem associated with antiepileptic drugs (AEDs) that can lead to vomiting, aggression, ataxia, and exacerbation of seizures. Acute hyperammonemia can also cause cerebral edema and severe brain damage (encephalopathy), whereas chronic hyperammonemia due to metabolic disorders is associated with developmental delay and intellectual disability (Cagnon & Braissant, 2007; Lichter-Konecki, 2008). Among the AEDs, valproic acid (VPA) is recommended as a first-line treatment for generalized epilepsy. Although VPA can cause an increase of the blood ammonia level, VPA therapy is rarely associated with hyperammonemic encephalopathy.
The exact relationship between symptoms and the ammonia level remains unclear. Murphy and Marquardt (1982) reported that patients with hyperammonemia not exceeding 240 μg/dl were asymptomatic. In contrast, Coulter and Allen (1991) recommended reducing the VPA dose when the ammonia level exceeded 100 μg/dl. Our data (see later) indicate that hyperammonemia is common, but clinicians may be unaware of it and may attribute the clinical consequences to other causes.
Our previous study established that high-dose VPA therapy, concomitant use of hepatic enzyme inducers, and concomitant use of topiramate (TPM) were risk factors for an increase of the blood ammonia level in adult patients with epilepsy who were treated with VPA (Yamamoto et al., 2012). There have also been some reports regarding the risk of hyperammonemia in pediatric patients with epilepsy (Coulter & Allen, 1991; Batshaw & Brusilow, 1982; Murphy & Marquardt, 1982; Ohtani et al., 1982; Haidukewych et al., 1985; Laub, 1986; Iinuma et al., 1988; Thom et al., 1991; Kondo et al., 1992; Altunbaşak et al., 1997; Sharma et al., 2011). Verrotti et al. (1999) found that the VPA dose, the duration of treatment, and polytherapy were associated with hyperammonemia. In addition, Haidukewych et al. (1985) and Sharma et al. (2011) reported that the blood concentration of VPA was correlated positively with the ammonia level, but other studies have not detected a significant association. The number of subjects in most of these previous studies was fewer than 100, which may have contributed to the differing results. In addition, most of the previous studies were conducted before 2000, so it remains unknown whether new AEDs, such as zonisamide (ZNS), TPM, gabapentin, lamotrigine, and levetiracetam, can increase the ammonia level.
VPA is metabolized mainly by uridine diphosphate glucuronosyltransferases (UGTs) and partially via β-oxidation and various cytochrome P450 (CYP) enzymes. Metabolites of VPA, such as valpronyl–coenzyme A (CoA), propionate, and 2-n-propyl-4-pentenoic acid (4-en-VPA), reduce the activity of enzymes involved in the urea cycle, resulting in the accumulation of ammonia (Kondo et al., 1992; Verbiest et al., 1992; Aires et al., 2011). Growth and development of pediatric patients may influence the metabolism of VPA, leading to age-related differences in the risk of hyperammonemia. Altunbaşak et al. (1997) reported that children younger than 2 years old had a higher risk of hyperammonemia related to VPA therapy, but other authors have not found a relationship between age and the blood level of ammonia. This may be because patient background factors varied widely among the studies. To identify risk factors for hyperammonemia, a large-scale study of pediatric patients covering a wide age range is needed.
Although hyperammonemia is generally associated with VPA, monotherapy with phenytoin (PHT), carbamazepine (CBZ), primidone, or acetazolamide can occasionally cause hyperammonemia (Ambrosetto et al., 1984; Katano et al., 2002; Kim et al., 2007; Adams et al., 2009; Labib et al., 2011). In addition, ZNS, TPM, acetazolamide, and sulthiame inhibit carbonic anhydrase and block bicarbonate re-uptake, so drug-induced metabolic acidosis could lead to an increase of the blood ammonia level in patients taking these AEDs. However, there is limited information regarding the risk of elevation of the ammonia level due to AED therapy other than VPA.
In recent years, several reports have indicated that generalized tonic–clonic (GTC) seizures can cause hyperammonemia (Liu & Su, 2008; Yanagawa et al., 2008; Hung et al., 2011; Liu et al., 2011). Therefore, it seems that an increase of ammonia could be caused by multiple factors, including those related to the patient's profile, seizure control, AED regimen, and drug doses. Accordingly, the aim of the present study was to evaluate the risk factors and prevalence of hyperammonemia among pediatric patients in relation to their AED therapy.
This was a retrospective study of 2,944 pediatric patients aged 0–15 years who consulted the National Epilepsy Center (Shizuoka, Japan) from January 2006 to December 2011. These patients were classified into three groups. Group I was 445 patients with newly diagnosed or suspected epilepsy or intellectual disability who had not received any prior AED therapy. The other 2,499 patients were taking AEDs including 673 patients treated with AEDs other than VPA (group II) and 1,826 patients treated with VPA (group III). Patients with serious hepatic dysfunction (aspartate aminotransferase or alanine aminotransferase >200 U/L), metabolic disorders, or severe infections were excluded. The study protocol was approved by the ethics committee of our hospital (protocol No. 2012–11).
Venous blood samples were obtained from the patients at 2–6 h after a meal. Measurement of the plasma ammonia level was done by the method described previously (Yamamoto et al., 2012). If multiple measurements were performed during the study period, the highest blood ammonia level was used. If VPA was added or discontinued during the study, we obtained two blood samples from the relevant patient. At our hospital, the normal range of plasma ammonia is 16–76 μg/dl. In this study, hyperammonemia was defined as a plasma ammonia level exceeding 100 μg/dl with reference to the above standard range and previous reports.
PHT, phenobarbital (PB), and CBZ are enzyme-inducing drugs; in this study, they will be referred to as inducers. Because primidone is converted to PB, we considered it to be equivalent to PB. ZNS, TPM, acetazolamide, and sulthiame were classified as carbonic anhydrase inhibitors (CAIs).
Quantitative variables were analyzed by the unpaired t-test or analysis of variance (ANOVA) with a post hoc Scheffe's multiple comparison test. Nominal variables were analyzed by the chi-square test.
To extract the factors influencing the plasma ammonia level, stepwise multiple regression analysis was performed using the age, gender, type of epilepsy (classified as focal epilepsy, idiopathic generalized epilepsy, symptomatic generalized epilepsy, or Dravet syndrome), and 14 AEDs (PHT, CBZ, PB, TPM, ZNS, clobazam, clonazepam, nitrazepam, acetazolamide, sulthiame, gabapentin, lamotrigine, levetiracetam, and ethosuximide) as the factors. In addition, the dose of VPA was added to the multiple regression model in group III. Finally, multiple logistic regression analysis was performed to calculate adjusted odds ratios for hyperammonemia, which was defined as a maximum ammonia level exceeding 100 or 150 μg/dl.
Results are expressed as the mean ± standard error. Statistical analyses were conducted with SPSS software Ver 19.0 (IBM Japan, Tokyo, Japan).
Table 1 shows the 2,944 pediatric patients classified into three groups based on their treatment. There was no difference in age among the three groups, but group I had the highest percentage of males (60.4%). The mean ammonia level of group I was 36.0 μg/dl, which was significantly lower than that of the patients treated with AEDs (groups II and III). Among the 1,826 patients treated with VPA, 578 patients (31.7%) had ammonia levels >100 μg/dl and 110 of these 578 patients (19.0%) required l-carnitine therapy. In this study, symptoms of hyperammonemia were not investigated in all of the patients, but 37 of 166 patients with ammonia levels exceeding 150 μg/dl had symptoms such as somnolence, lethargy, nausea, vomiting, and anorexia.
|Group I (non-AED group)||Group II (non-VPA group)||Group III (VPA group)||p-Value|
|No. of patients||445||673||1,826|
|Age (years)||7.2 ± 0.20||6.9 ± 0.17||7.1 ± 0.11||NS|
|Ammonia level (μg/dl)||36.0 ± 1.0*†||56.0 ± 1.1†||85.8 ± 1.0||<0.001|
|VPA dose (mg/kg)||–||–||21.4 ± 0.23|
|Hepatic enzyme inducers||–||519||566||<0.005|
|Carbonic anhydrase inhibitors||–||256||534||NS|
Epilepsy was classified as focal epilepsy in 1,492 (50.7%) patients, idiopathic generalized epilepsy in 168 (5.7%) patients, symptomatic generalized epilepsy in 523 (17.8%) patients, and Dravet syndrome in 108 (3.7%) patients. Other categories were unclassified epilepsy, situation-related syndrome, and undiagnosed epilepsy type/syndrome.
Figure 1 displays the plasma ammonia levels of the patients classified into four age groups. The mean ammonia level varied across the age groups (ANOVA: p < 0.001). In all three treatment groups, patients aged from 0 to 3 years had the highest ammonia levels of all age groups (Scheffe's test; p < 0.001). There was also a statistical difference between patients aged 4–7 and 12–15 years in groups I and III (p < 0.005).
Table 2 shows the effect of AED monotherapy on the blood ammonia level. The mean ammonia levels in patients receiving any type of monotherapy were significantly higher than those of the untreated group. Among the inducers, PHT monotherapy was associated with higher blood ammonia levels despite the patients having the greatest mean age, but there were no significant differences. The mean ammonia level of patients taking CAIs was significantly higher than that of patients taking CBZ (p < 0.005), and hyperammonemia was significantly more frequent (chi-square test: p < 0.05).
|Regimen||No AEDs (reference)||PHT||PB||CBZ||CAIs||BZs||p-Valuea|
|Age (years)||7.2 ± 0.2||10.8 ± 0.6||4.2 ± 0.8c||7.7 ± 0.3e||6.2 ± 0.6d||7.0 ± 1.2||<0.001|
|Ammonia level (μg/dl)b||36.0 ± 1.0||59.9 ± 4.9||52.5 ± 5.1||44.9 ± 1.6||60.0 ± 3.3f||48.5 ± 8.5||<0.001|
|>100, n (%)||7 (1.6)||0 (0)||1 (3.4)||3 (1.6)||6 (10.3)||1 (6.7)||<0.05|
Table 3 shows the blood ammonia levels of patients receiving VPA monotherapy or VPA plus other AEDs. The mean ammonia level of the patients on VPA plus AEDs was significantly higher than that of those on VPA monotherapy, but the combined effects of VPA and each AED were different. Concomitant use of PHT, PB, or CAIs was associated with a significantly higher blood ammonia level compared with concomitant CBZ. The combination of VPA plus PHT, VPA plus PB, or VPA plus CAIs was associated with a higher incidence of hyperammonemia than the other AED regimens (p < 0.001). In particular, concomitant use of PHT was most likely to cause hyperammonemia exceeding 150 μg/dl.
|Regimen||VPA mono (reference)||VPA + PHT||VPA + PB||VPA + CBZ||VPA + CAIs||VPA + BZs||p- Valuea|
|Age (years)||7.8 ± 0.2||8.0 ± 0.6||5.2 ± 0.7||7.9 ± 0.4e||6.1 ± 0.3h||6.8 ± 0.4||<0.001|
|Ammonia level (μg/dl)||67.0 ± 1.5||115.0 ± 8.0||107.1 ± 7.9||76.6 ± 3.0b,d||93.7 ± 2.5c,g||74.2 ± 3.1b,d,i||<0.001|
|>100, n (%)||87 (15.5)||29 (54.7)||21 (53.8)||38 (24.4)||84 (41.8)||21 (16.7)||<0.001|
|>150, n (%)||18 (3.2)||15 (28.3)||6 (15.4)||6 (3.8)||15 (7.5)||4 (3.2)||<0.001|
|VPA dose (mg/kg)||17.5 ± 0.4||24.2 ± 1.7||26.5 ± 2.2||19.6 ± 0.8f||23.2 ± 0.7h||21.5 ± 0.7||<0.001|
There was a significant positive correlation between the plasma ammonia level and the dose of VPA in the patients receiving VPA monotherapy (Pearson's correlation coefficient analysis, r = 0.44, p < 0.001). When patients were classified into four VPA dose groups: <10 mg/kg, ≥10 and <20 mg/kg, ≥20 and <30 mg/kg, and ≥30 mg/kg, the mean plasma ammonia levels increased in a VPA dose–dependent manner (46.8, 60.4, 85.7, and 92.6 μg/dl, respectively, ANOVA: p < 0.001). In particular, there was significant difference between the dose ranges of <10 mg/kg and ≥20 mg/kg (p < 0.001, See Fig. S1.).
By stepwise multiple regression analysis, the factors influencing the ammonia level were extracted (see Table S1). These factors were then used as independent variables for multiple logistic regression analysis.
In group II, age and the concomitant use of PHT, PB, ZNS, TPM, or levetiracetam had a significant influence on the ammonia level, and these factors were entered as independent variables in multiple logistic regression analysis. Age and concomitant use of ZNS and TPM were found to be significant risk factors for hyperammonemia (Table 4). PHT and PB were also risk factors, but were not statistically significant.
|Risk factor||Plasma ammonia level (>100)|
|OR (95% CI)||p value|
|Age (per year)||0.89 (0.82–0.96)||<0.005|
In group III, the age, symptomatic generalized epilepsy, VPA dose, and concomitant use of PHT, PB, TPM, ZNS, or acetazolamide were all found to be significant risk factors (Table 5). In contrast, the VPA dose and concomitant use of PHT, PB, or TPM were significantly associated with an increased risk of an ammonia level exceeding 150 μg/dl), with high-dose VPA and concomitant PHT being the most important factors.
|Risk factor||Plasma ammonia level|
|OR (95% CI)||p-Value||OR (95% CI)||p-Value|
|Age (per year)||0.97 (0.94–0.99)||<0.05||1.00 (0.96–1.04)||0.91|
|Gender (female = 1)||1.30 (1.04–1.61)||<0.05||1.55 (1.10–2.20)||<0.05|
|Symptomatic generalized epilepsy||1.40 (1.09–1.80)||<0.01||1.17 (0.79–1.73)||0.44|
|VPA dose (per mg/kg)||1.05 (1.03–1.06)||<0.001||1.05 (1.03–1.07)||<0.001|
|Phenytoin||4.73 (3.25–6.88)||<0.001||5.45 (3.54–8.40)||<0.001|
|Phenobarbital||2.22 (1.59–3.12)||<0.001||2.44 (1.57–3.79)||<0.001|
|Zonisamide||1.72 (1.33–2.23)||<0.001||1.42 (0.95–2.11)||0.089|
|Topiramate||2.69 (1.78–4.05)||<0.001||2.13 (1.20–3.76)||<0.05|
|Acetazolamide||6.55 (2.50–17.2)||<0.001||2.51 (0.91–6.95)||0.076|
The present cross-sectional study of 2,944 pediatric patients demonstrated that the mean ammonia level was 2.4-fold higher in patients treated with VPA than in patients without AED therapy (Table 1). A literature review performed by Chicharro and Kanner (2007) identified a significant difference between the VPA group and the control group in 11 of 17 cross-sectional studies, whereas the other six studies showed no difference. The large sample size of the present study allowed us to detect significant differences of the ammonia level among groups I, II, and III.
Our data also suggested that inducers and CAIs had the potential to increase the ammonia level in patients who were not taking VPA (Tables 2 and 4). Several previous studies have found a decrease of carnitine in patients treated with PHT, PB, or CBZ rather than VPA. Hug et al. (1991) reported that the mean total carnitine level was 57.8 nmol/ml in healthy children, whereas the carnitine level of pediatric patients with epilepsy who were receiving VPA, PB, PHT, or CBZ was 35.6, 32.7, 39.7, and 41.5 nmol/ml, respectively. Reduction of the blood carnitine level by these AEDs may cause mitochondrial dysfunction, resulting in an increase of ammonia.
Tables 2 and 3 show that VPA monotherapy was more closely associated with hyperammonemia in comparison with monotherapy taking PHT, PB, or CBZ (15.5% vs. 0%, 3.4%, and 1.6%). Figure S1 shows that the increase of ammonia in our pediatric patients with epilepsy was dependent on the daily dose of VPA. Several authors have reported that the blood carnitine level decreases with an increase in the VPA dose or concentration (Hamed & Abdella, 2009; Nakajima et al., 2011). Another possible mechanism of hyperammonemia related to VPA therapy is inhibition of carbamoyl phosphate synthase 1 and N-acetylglutamate synthase by metabolites of VPA such as valpronyl-CoA, propionate, and 4-en-VPA. That is, high-dose VPA monotherapy could increase harmful metabolites and reduce the carnitine level, resulting in an increased risk of hyperammonemia.
In patients receiving VPA with PHT or PB, a high incidence of hyperammonemia was observed (Table 3). According to our multivariate model, combined therapy with VPA and PHT was the strongest risk factor for hyperammonemia in the pediatric patients with epilepsy (Table 5). These findings agree with the results of a previous study in adult patients with epilepsy (Yamamoto et al., 2012). Although hyperammonemia associated with VPA + PHT and/or PB therapy has long been reported, the mechanism remains obscure. PHT and PB increase the activity of CYP enzymes and UGTs, which would facilitate metabolism of VPA. Among the CYP enzymes, CYP2C9, CYP2A6, and CYP2B6 are involved in the synthesis of 4-en-VPA and propionate, which inhibits urea cycle enzymes. In contrast, a number of studies have shown that the blood carnitine concentration is lower in patients receiving VPA combined with inducers than in patients receiving VPA monotherapy (Ohtani et al., 1982; Coulter 1991). Hyperammonemia in patients receiving VPA + PHT and/or PB therapy could be caused by a combination of these mechanisms. Among the inducers, CBZ had the smallest impact on the plasma ammonia level. Measurement of VPA metabolites and the carnitine concentration could be valuable for clarifying the mechanisms of hyperammonemia associated with concomitant use of VPA and inducers.
The influence of new AEDs such as ZNS, TPM, lamotrigine, and levetiracetam on the blood ammonia level remains poorly understood. Several case reports have been published about hyperammonemia in patients on TPM therapy (Hamer et al., 2000; Longin et al., 2002; Knudsen et al., 2008; Deutsch et al., 2009), but there have been none about the other new AEDs. The present study demonstrated that concomitant use of ZNS and TPM was an important risk factor for an increase of the ammonia level, irrespective of whether the patient was being treated with VPA. Coppola et al. (2006) reported that lamotrigine and TPM did not have a significant effect on the serum carnitine level. In contrast, ZNS and TPM inhibit carbonic anhydrase and block bicarbonate reuptake, resulting in metabolic acidosis that may lead to an increase of the plasma ammonia level regardless of concomitant VPA therapy. We also found that treatment with acetazolamide had the potential to increase the ammonia level, presumably via the same mechanism as that for TPM and ZNS. However, there were only five patients receiving acetazolamide in group II, contributing to the lack of a significant result. In this study, bicarbonate was not measured, so further studies will be needed to determine the association between bicarbonate and ammonia levels.
Our study demonstrated that pediatric patients with epilepsy aged 3 years or younger who were treated with VPA had an increased risk of hyperammonemia, confirming a previous report by Altunbaşak et al. (1997). The same trend was also seen in the untreated and non-VPA groups, suggesting that ammonia metabolism differs among infants, children, and adolescents. Galal et al. (2010) reported that there was no relationship between age and the ammonia level in patients attending a pediatric emergency department. In contrast, Colombo et al. (1984) reported that the plasma ammonia level was higher in neonates than in children aged 2–6 years (range 30–144 vs. 24–48 μmol/l). Because of the large number of patients in our study it was possible to establish a more reliable indication of the impact of age on the plasma ammonia level.
Among patients receiving VPA, female gender was associated with a risk of hyperammonemia. In group III, female patients had a significantly lower mean body weight than males (24.5 vs. 26.7 kg, p < 0.001), suggesting that female patients would have greater hepatic blood flow and VPA clearance. Furthermore, a review by Pleym et al. (2003) suggested that UGT activity is higher in male patients than female, indicating that CYP-derived metabolites may be more important in female patients. Therefore, gender-based differences of body weight and UGT activity presumably resulted in differences of VPA metabolism and the ammonia level.
Continuous muscle contraction due to GTC seizures can elevate the blood ammonia level by deamination of adenosine monophosphate in the purine nucleotide cycle (Mutch & Banister, 1983; Wilkinson et al., 2010). Several reports have indicated that the risk of hyperammonemia in adult patients with epilepsy is significantly related to GTC seizures (Liu & Su, 2008; Yanagawa et al., 2008; Hung et al., 2011; Liu et al., 2011). These studies included patients without AEDs and excluded patients taking VPA. In contrast, our study showed that symptomatic generalized epilepsy was one of the risk factors for hyperammonemia in group III (with VPA), but not in group II (without VPA). Because VPA is a first-line treatment for GTC seizures, the prevalence of generalized epilepsy was significantly higher in group III than in group II (23.2% vs. 11.9%), so this lower prevalence may have contributed to the lack of any association with GTC seizures in group II.
Our previous study showed that 40.7% of adult patients on VPA therapy had symptomatic hyperammonemia (Yamamoto et al., 2012). In the present study, 37 (22.3%) of 166 patients with an ammonia level exceeding 150 μg/dl had symptoms of hyperammonemia. Therefore, careful attention should be paid to whether symptoms are related to hyperammonemia in patients with an ammonia level >100 μg/dl.
There were several limitations of the present study. Although the dose of VPA showed a high partial regression coefficient, blood samples were not collected at a specific time and VPA trough concentrations were not obtained. In addition, if multiple measurements were obtained from same patients, the highest ammonia level was used, which may have contributed to selection bias. Long-term VPA therapy reduces the blood carnitine level, but we did not investigate the duration of AED therapy. Furthermore, the retrospective design of this study meant that we could not confirm seizure frequency, nitrogen intake, and symptoms at the point of examination. We were also unable to investigate the symptoms of hyperammonemia in all 2,944 patients and could not set clinical criteria for symptomatic hyperammonemia. In particular, the relationship between symptoms and the ammonia level remains unclear, although it is known that chronic hyperammonemia can have harmful effects on brain development and seizure control. Therefore, further studies will be required to evaluate the ammonia level below which there will be no influence on seizure control and brain development.
In conclusion, our study established a number of risk factors for hyperammonemia in pediatric patients with epilepsy who are on AED therapy. We emphasize the need for measurement of ammonia when patients have multiple risk factors, especially those receiving high-dose VPA and regimens such as VPA plus PHT and/or PB. In addition, use of CAIs and a younger age can be associated with an increase of ammonia regardless of whether a patient is receiving VPA. These results have several implications for minimizing the risk of hyperammonemia in pediatric patients with epilepsy treated with VPA-based AED therapy.
None of the authors has any conflict of interest to disclose. We confirm that we have read the Journal's position on issues involved in ethical publication and affirm that this report is consistent with those guidelines.