Inaccuracies With Valproic Acid Free Concentration Identified in Status Epilepticus

January 19, 2021
Marco Meglio
Marco Meglio

Marco Meglio, Associate Editor for NeurologyLive, has been with the team since October 2019. Follow him on Twitter @marcomeglio1 or email him at

The calculation of fVPA based on this equation resulted in a misinterpretation in 40% of measurements with around 75% of them overestimating the actual measured fVPA.

Newly published data from a retrospective study that included adults with status epilepticus (SE) demonstrated considerable discordance between calculated and actual measured free concentration of serum valproic acid (fVPA) within this patient population.1

Senior author Raoul Sutter, MD, neurologist, University Hospital Basel, and colleagues aimed to test the accuracy of an equation in adults with SE that calculates fVPA from the total concentration of serum valproic acid (tVPA) and serum albumin. Valproic acid (VPA) has been shown to be as equally effective as levetiracetam or fosphentyoin to treat SE, but supratherapeutic VPA levels have been associated with dose-dependent toxicities.2

Despite a linear association between measured and calculated fVPA (regression coefficient, 1.1; 95% CI, 0.9–1.2; P <.0001), the agreement on effective drug levels did not match in 39.8% of measurements regardless of serum albumin levels, with calculated fVPA overestimating measured fVPA in 30.4% of measurements.

"Because of the potential critical impact of overestimation and the considerable variability of fVPA, we advocate therapeutic drug monitoring in critically ill patients with SE receiving VPA as a second-line ASD to be based on measured free fractions,” Fisch et al concluded.

READ MORE: Epileptic Peripartum and Postpartum Seizure Frequency Unchanged by Pregnancy

For VPA measurements with patients with serum albumin levels between 18 g/l and 42 g/l, linear regression models were performed to assess the associated between calculated and measured fVPA and serum albumin levels. If only measurements with corresponding serum albumin within this range were considered, the proportion of agreement between the categories remained essentially unchanged, according to the study authors.

Of 676 patients with SE, 106 had measurements of serum albumin, tVPA, and fVPA available. Patients were further stratified by the clinical characteristics of SE (FIGURE). Two patients without parallel measurements were removed, resulting in 104 patients with a total of 506 concurrent measurements of serum albumin tVPA, and fVPA.

When categorizing the fVPA measurements into subtherapeutic (<5 mg/l), therapeutic (5–10 mg/l), and subtherapeutic (>10 mg/l) levels, 305 (60.3%) of 506 measurements showed an agreement between calculated and measured fVPA, while calculated underestimated measured fVPA in 44 (8.7%) cases, and the aforementioned overestimated in 157 (31%) cases, respectively.

Agreements were defined as the proportion of measured and calculated fVPA falling within the same category.

The median tVPA was 43.5 mg/l (interquartile range [IQR], 27.4–63.6), with a median measured fVPA of 9.1 mg/l (IQR, 4.5–14.7) and calculated fVPA of 10.1 mg/l (IQR, 7.0–13.0), respectively. Furthermore, the median deviation of calculated from measured fVPA was –0.8 mg/l (IQR, –3.2 to 2.5) with 336 measurements >1 mg/l.

In 12 patients, 21 measurements of fVPA were less than 3 mg/l despite tVPA of more than 20 mg/l, most likely due to preanalytical errors, Fisch and colleagues noted.

The researchers found a significant association between measured fVPA and tVPA (regression coefficient, 1.7; 95% CI, 1.5–1.9; P <.0001), with less fitting with increasing fVPA serum concentrations.

Patients with (regression coefficient, 1.69; 95% CI, 1.48–1.90; P >.0001) or without (regression coefficient, 2.10; 95% CI, 1.87–2.33; P >.0001) renal- or liver-insufficiency saw similar results in terms of the association between measured fVPA and tVPA. However, the difference between calculated and measured fVPA significantly increased with higher albumin levels (regression coefficient, 0.3; 95% CI, 0.2–0.4; P <.0001) in patients with and without liver- or renal-insufficiency, and likewise with higher tVPA.

1. Fisch U, Baumann SM, Semmlack S, Marsch S, Ruegg S, Sutter R. Accuracy of calculated free valproate levels in adult patients with status epilepticus. Neurology. Published online October 14, 2020. doi: 10.1212/WNL.0000000000011000
2. Kapur J, Elm J, Chamberlain JM, et al. Randomized Trial of Three Anticonvulsant Medications for Status Epilepticus. N Engl J Med. 2019;381:2103- 2113.