Valproate Monitoring in Patients With Hypoalbuminemia

Valproic acid, or 2-propylpentanoic acid, is a branched-chain fatty acid derived from valerian acid, with Food and Drug Administration approval for treatment of seizures, migraine prophylaxis, and manic or mixed episodes associated with bipolar disorder (1, 2). It is available in oral and intravenous formulations as immediate-release and enteric-coated delayed-release as well as extended-release valproate sodium and divalproex sodium.

The mechanisms of action are not fully understood, but broadly, the acute effects appear to be mediated by enhancement of gamma-aminobutyric acid (GABA)-mediated neurotransmission via interference with GABA metabolism and effects on signaling pathways (1). The long-term effects are a result of alteration in multiple gene expression, which is at least partially mediated through direct inhibition of histone deacetylase (which leads to increased acetylation of lysine residues and, consequently, enhanced transcriptional activity) (1).

Valproic acid is highly protein bound, metabolized extensively by the liver, and eliminated by first-order kinetics (3). It also has unusual and variable pharmacodynamic and pharmacokinetic characteristics, including a nonlinear level-dose relationship, a variable degree of plasma protein binding, wide interpatient differences, and pharmacological effect that outlasts presence in plasma (4, 5).

Many studies have demonstrated high rates of medical comorbidities in patients with psychiatric disorders, including sequelae of the disorder (e.g., cirrhosis secondary to alcohol abuse), side effects of psychotropic medications, poorer self-care as a result of mental illness, and less efficient utilization of the health care system (6). Poor management of chronic illnesses such as diabetes and hypertension can result in serious sequelae, mainly renal failure. The kidneys serve as primary sites for excretion of many drugs, and thus renal disease can significantly affect drug clearance and steady-state levels (7). It also affects pharmacokinetics, because renal failure, characterized by an excessive loss of protein in the urine, leads to a hypoalbuminemic state (8).

Albumin is the principal plasma protein responsible for binding to acidic drugs (9). Most psychotropic drugs are highly protein bound, and thus only a small fraction of the total serum drug concentration—the free fraction—is available for pharmacological action (10). In a hypoalbuminemic state such as renal disease, there is less protein available for a drug to bind, and therefore a greater free fraction of drug is found in the plasma (9). Additionally, renal failure leads to accumulation of endogenous binding inhibitors such as uremic toxins and organic acids, which compete with drugs for protein-binding sites and displace them into the plasma (9). Albumin itself also undergoes a conformational change that is hypothesized to alter its binding properties (9).

Sometimes more than one psychotropic agent is used for treatment, and in a setting in which even a single drug can be displaced from protein-binding sites by endogenous factors, the presence of multiple drugs results in greater competition for those limited binding sites. To complicate matters further, many common nonpsychotropic medications are highly protein bound (e.g., aspirin, statins, and antihypertensives) (11). The competition for binding sites can result in many-fold increases in free plasma levels of the displaced drug (11).

There is considerable debate in the literature concerning the relationship between valproic acid plasma levels and therapeutic effect, but the generally accepted therapeutic range (for all indications) is 50–100 µg/mL (12, 13). At higher levels, there is potential for toxicity characterized by neurologic, cardiac, respiratory, hematologic, gastrointestinal, or metabolic sequelae (5, 14). Although the effects of valproate overdose are usually mild, more severe and life-threatening events may occur, including coma, arrhythmia, shock, and bone marrow failure (14). Because of the potential for adverse outcomes and the narrow therapeutic range of valproic acid, therapeutic drug monitoring is oftentimes performed. The serum total valproic acid level is usually measured on the assumption that it is reflective of the free fraction of the drug. However, as discussed above, the free fraction usually increases in renal failure and other hypoalbuminemic states (alcoholic cirrhosis, acute hepatitis, burns, etc.). The total concentration of a drug (a sum of free and bound drug) often does not change appreciably with these shifts in equilibrium between free and bound fractions (11). Many studies have shown that monitoring the total valproic acid concentration can be misleading, because it can be normal or even low despite high concentrations of free valproic acid (11). If dosages are adjusted on the basis of total valproic acid levels, patients who are hypoalbuminemic can be overdosed, leading to clinically significant neurologic symptoms or toxicity (10, 15). Thus, there is greater clinical utility and importance in monitoring the free valproic acid concentration, rather than the total, in patients with hypoalbuminemia (3, 4, 11). The largest barrier to regularly monitoring free levels is the absence of widely available free valproic acid assays. In one survey conducted by the College of American Pathologists, only 2% of the laboratories that routinely performed total valproic acid determinations also offered free assays (10). Different theoretical models have been proposed to indirectly estimate the free fraction of valproic acid or to calculate a corrected total concentration for varying albumin levels, but the models are not validated and are not applicable to patients with renal failure, jaundice, or a high total drug concentration, criteria which essentially exclude many of the patients for whom the model is needed (10).