Fulminant Onset of Valproate-Associated Hyperammonemic Encephalopathy

Valproate in its various forms (valproic acid [sodium valproate] and valproate semisodium), originally developed as an antiepileptic drug, is a first-line therapy for the treatment of acute mania, with recommendation grade 1 and clinical evidence A, except in younger women (recommendation grade “2”), according to current guidelines (1). Valproate acts by increasing gamma-aminobutyric acid (GABA) levels via the inhibition of GABA-metabolizing enzymes and GABA synthesis catalysis, and further by inhibiting N-methyl-d-aspartate (NMDA) receptors and neuronal sodium and calcium channels (2). Interferences in cell metabolism and structure of lipid membranes have been described. Some of the most common side effects of valproate treatment include tremor, confusion, anemia, thrombocytopenia, nausea, and weight gain. Severe adverse events such as pancreatitis, hepatotoxicity, coagulatory defects, and potential teratogenicity (associated with exposure during pregnancy) also may occur with valproate treatment (3). Notably, chronic therapy with valproate can occasionally produce a syndrome of cognitive deterioration and/or parkinsonism, sometimes accompanied by ataxia and/or hearing loss, which is reversible after valproate is stopped but advances if valproate is continued. Cortical pseudoatrophy and enlargement of the lateral ventricles are typical radiologic signs of this side effect, which can occur weeks to years into treatment with valproate (4). An incidence between 1% and 2% was reported in a cohort of epilepsy patients under valproate treatment (5).

Valproate-associated encephalopathy is described as a rare but serious adverse event, predominantly reported in young children with epilepsy who have innate metabolic disorders (6). Combinations of anticonvulsant drugs seem to increase the risk of encephalopathy (6). Valproate-associated encephalopathy is an acute and critical condition characterized by reduced consciousness, focal neurological deficits, vomiting, vertigo, lethargy, and cognitive slowing, often with hyperammonemia and sometimes with hepatopathy (3). However, hyperammonemia and hepatopathy do not occur in all patients (3, 7). Indeed, in a study of patients with epilepsy, different phenotypes of valproate-associated encephalopathy could be identified, and different pathogenic pathways were suspected as the explanation for this heterogeneity (3).

Data on incidence rates of valproate encephalopathy in epilepsy patients are inconclusive, and some cases may be overlooked (3). In psychiatric patients, 25 cases of valproate-associated encephalopathy were reported between 1980 and 2017 (7–15). However, there is evidence for broader relevance, as a recent cross-sectional study (16) found a rate of 2.5% for valproate-induced hyperammonemic encephalopathy in a general hospital population with at least one psychiatric disorder. Possibly, in some cases, signs and symptoms of encephalopathy may go unnoticed because of heterogeneity in clinical presentation and laboratory findings.

Our patient developed a fulminant onset of encephalopathy and hyperammonemia associated with antimanic treatment with valproic acid. Interestingly, he showed clinical improvement and tolerated the medication well for 2 weeks with plasma levels within normal ranges until a sudden onset of clinical signs. The acute clinical onset and the 10-fold elevation of ammonia, while valproate had initially been within a therapeutic range, could suggest an underlying genetic vulnerability in the patient. In fact, direct inhibition of the urea cycle by valproate has been suggested as the primary mechanism in patients with isolated hyperammonemic encephalopathy without hepatopathy (3). Valproate inhibition of ammonia elimination in the urea cycle is mediated by the inhibition of N-acetylglutamate synthase (NAGS) and carbamylphosphate synthase 1 (CPS1) activity, and by the interference in carnitine transport of acyl-coenzyme A via the mitochondrial matrix (Figure 3) (17). Polymorphisms in CPS1 and NAGS genes have been associated with hyperammonemia in valproate treatment (18).

A screening for innate defects of urea cycle was negative in our patient; plasma glutamic acid was only slightly increased to 87.8 μmol/L (normal range, 1–57 μmol/L), and there were no signs of orotic acid in the urine. Furthermore, no anamnestic hints were found in the patient’s own and in his family history (e.g., no unexplained deaths of young male children), and his ammonia levels were normal in the absence of valproate. Hence, severe innate urea cycle defects (Figure 4) can be excluded as the mechanism of action in our patient. However, in the plasma amino acid screening, months after the event, the patient’s serum showed hyperprolinemia and elevated leucine and valine levels as markers of ketosis. Alanine and glycine plasma levels were also elevated to a medium extent, indicating a secondary rather than a primary mitochondrial disease, that is, with an environmental etiology rather than a genetic etiology (nuclear DNA/mitochondrial DNA). As valproate depletes mitochondrial acetyl-CoA (Figure 3), an inborn mitochondriopathy, demasked by valproate, could also be a relevant mechanism for hyperammonemia in our patient. However, with the diagnostic tests described above, a definite diagnosis cannot be made yet. Further genetic testing at a specialized center for inborn errors of metabolism was recommended. Figure 4 lists inborn errors of metabolism that favor the occurrence of hyperammonemic encephalopathy under valproate treatment. A review by Segura-Bruna et al. (19) offers an extensive discussion on pathophysiology and a broad differential diagnostic list of factors predisposing to hyperammonemia.

In our patient, signs of encephalopathy were obvious and swiftly led to the right diagnostic and therapeutic approach. In many cases, however, clinical signs may not present in such a clear manner. New onset of neurologic symptoms during valproate treatment should lead to diagnostic measures such as examination of ammonia levels and EEG, even when valproate levels and liver enzymes are within normal ranges. Nonetheless, we do not recommend routine measurements of ammonia in all patients under valproate treatment; in a 5-year single-center observational study (N=158) (20), hyperammonemia was observed in 20%−30% of epilepsy patients on valproate and was dose dependent. The risk of ammonia elevation was further increased when valproate was combined with liver enzyme–inducing antiepileptic or antipsychotic drugs. In an acutely ill psychiatric intensive care unit population, hyperammonemia was observed in every second patient on valproate, and also in every fifth patient on another mood stabilizer, without the occurrence of valproate-associated encephalopathy (21).

In conclusion, valproate-associated encephalopathy remains a possibly underestimated and heterogeneous entity with a variety of mechanisms compatible with the broad therapeutic spectrum of valproic acid. Based on the reported case, we recommend close monitoring of newly developed neurologic symptoms such as cognitive deterioration, ataxia, tremor, confusion, and changes in consciousness during valproate treatment, even if the medication was not started recently. If valproate-associated encephalopathy is considered, tests of serum ammonia levels and EEG should be performed immediately. Mild cases of encephalopathy, characterized mostly by confusion, may be easily overlooked.