It has been estimated that about 90% of patients with dementia will develop behavioral problems at some point during their illness.¹ Although behavioral disturbances may be caused by environmental stresses, medical problems, or psychiatric comorbidities such as depression, in about 50% of cases, the behavioral disturbance is due to effects of the dementia itself.¹ A pharmacological intervention may be recommended in this situation; however, no medications have been FDA-approved for treatment of dementia with behavioral disturbance. Given the 2005 FDA warnings for atypical antipsychotics in this population, physicians may turn increasingly to other classes of drugs, such as anticonvulsants, to treat behavioral issues in patients with dementia.

Valproic Acid (VPA) has been reported in the literature as a possible treatment for agitation in dementia since 1986.² The literature is conflicting on the efficacy and safety of VPA in this population, potential side effects being somnolence, thrombocyotopenia, weakness, and respiratory tract infections,³ in addition to weight gain, tremor, pancreatitis, and liver toxicity.⁴ Less commonly known is the potential for elevated ammonia levels in the absence of hepatic damage, known as valproate-induced hyperammonemic encephalopathy (VHE).⁴

Although there have been case reports of VHE since the 1980s, these have been mainly in the neurology literature, with less attention in the psychiatric literature.⁵ In the geriatric psychiatry literature, there has been only one case report.⁶ Mechanisms have been proposed as to how VPA may elevate ammonia. In the kidneys, VPA is thought to increase ammonia production by enhancing uptake of glutamine, increasing glutaminase activity, or both.^4,7 In the liver, VPA reduces ammonia metabolism by disruption of the urea cycle. Proposed mechanisms of this disruption include decreasing carnitine availability,⁸ depleting N-acetyl-glutamate,⁵ and inhibiting the enzyme carbamoylphosphate synthetase I.⁷ Carnitine’s role in the urea cycle disruption has led to the suggestions that its supplementation may aid in VHE’s treatment and that carnitine deficiency may be a risk factor.^7,8

Although ammonia levels are elevated in 20%–50% of patients prescribed VPA,⁴ the majority are felt to be subclinical or asymptomatic.⁹ However, detailed neuropsychological testing has revealed deficits in motor speed, visual perception, construction, concentration, and attention in so-called subclinical stages of VHE.⁴ Ammonia levels greater than 60 µmol/L are commonly associated with clinical symptoms including anorexia, irritability, lethargy, vomiting, somnolence, disorientation, asterixis, coma, and death.¹⁰ Because elevated ammonia occurs without functional disruptions of organs detectable by routine lab testing (such as elevated liver-function tests), elevated ammonia may not be considered.¹⁰

To our knowledge, there has been no study of elevated ammonia in an elderly psychiatric population treated with VPA. In this preliminary study, we report on the prevalence of hyperammonemia in elderly psychiatric patients treated with valproic acid.

In this retrospective, chart-review study, all patients age 65 and over being treated with valproic acid from two nursing homes in a university psychiatric nursing home service over a 1-year period were included, resulting in a total of 12 patients for the study. Although there are different pharmacologic formulations of valproic acid, such as divalproex, all 12 patients were on valproic acid. Patients’ charts were reviewed by one author (JTO), and the following data were obtained, including demographics (age, race, gender), psychiatric diagnoses, Folstein Mini-Mental State Exam score, number and type of medical conditions, medications, liver-function tests, valproic acid dosages and levels, ammonia levels, and any possible side effects, including lethargy or change in mental status as noted in the patient chart. The ammonia lab values were measured by two different hospital laboratories in differing units, µmol/L and µg/dL. Measurements taken in µg/dL were converted to µmol/L by multiplying them by 0.714. Variables were entered into SPSS, and the data were analyzed.

The study was approved by the University Institutional Review Board.

Twelve patients were included in the study. Mean age was 80 (standard deviation [SD]: 10.4]) years. Psychiatric diagnoses were Any Dementia in 75% (N=9), with Alzheimer’s disease in 41.7% (N=5), and vascular dementia in 41.7% (N=5); bipolar disorder in 25% (N=3), and schizoaffective disorder 8.3% (N=1). Some patients had more than one diagnosis. Folstein MMSE scores ranged between 0/30 and 28/30, with a mean of 14/30 (SD: 11.4). The number of other medications prescribed ranged between 2 and 12, with a mean of 7.3 (SD: 7.4) medications.

Coexisting medical problems included the following: 58.3% (N=7) had hypertension; 33.3% (N=4) had diabetes mellitus; 2 of 12 had hyperlipidemia; 2 of 12 had coronary artery disease; 1 of 12 had chronic renal insufficiency; and 2 of 12 had congestive heart failure. No patients had liver disease or a history of alcohol abuse or dependence. One patient was also prescribed phenytoin, and one was prescribed risperidone, but no other anticonvulsants or antipsychotics were prescribed. No patients were taking amiodarone or chemotherapy treatments known to cause hyperammonemia.

Table 1 shows the results of the valproic acid dosing information and ammonia levels. Of note, 83.3% of the sample had elevated ammonia levels (N=10). All liver-function studies (AST, ALT, alkaline phosphatase, bilirubin) were normal.

Three of the patients with elevated ammonia were noted to have possible effects from elevated ammonia. One had lethargy; one, delirium; and one had a nonspecific cognitive decline noted. The three with documented side effects had a mean ammonia level of 74.7 µmol/L.

Despite the small sample size, an exploratory t-test analysis was done to compare patients with elevated ammonia levels (83.3%; N=10) with those in the normal range (16.7%; N=2). No significant differences were noted for age, MMSE score, total dose of VPA, or VPA blood level between groups.

It is notable that, in this preliminary study, 83.3% of geriatric nursing home patients prescribed VPA had ammonia levels above the normal range. Of these, only three patients were noted to have apparent symptoms associated with the elevated ammonia, although it is possible this is an underestimate, since this was a retrospective, chart-review study and only included such patients where the change of mental status was clinically obvious. It is also possible that lower levels of ammonia may affect cognition and physical symptoms in elderly perons, just as they are more vulnerable to the effects of urinary tract infections and anticholinergic side effects. As noted previously, subtle cognitive changes are seen in adults with elevated ammonia who appear to be asymptomatic.⁴

Also, many of the symptoms of hyperammonemia overlap with the neurological side effects of VPA itself, and, in the geriatric population, where dementia and delirium are prevalent, it may be difficult to detect subtle symptoms due to elevated ammonia. This is also difficult because of the lack of liver enzyme elevation seen in such patients. If lethargy, somnolence, or any other mental status change is observed in an elderly patient prescribed VPA, an ammonia level should be checked as part of the differential diagnosis.

There are many limitations to this study. This was a preliminary, retrospective chart-review of a small sample living in a nursing-home setting, and there was no control group. The patients had different neuropsychiatric diagnoses of differing severities and differing comorbidities. We cannot definitely conclude that it was the VPA that caused the elevated ammonia levels, as no ammonia levels were drawn before beginning VPA treatment in these patients. Other causes could be polypharmacy or malnutrition. No tests for genetic risk factors such as urea cycle disorders or carnitine deficiency were done. Prospective studies are needed to check ammonia levels before and during treatment with VPA to determine the effect of VPA on ammonia levels in the geriatric psychiatric population. Also, cognitive testing should be performed to determine what, if any, effect such levels have in elderly patients. Given the elevated ammonia levels seen in this study, further research is clearly warranted.