Therapeutic Drug Monitoring of Phenytoin in Critically Ill Patients

Sandrina L. von Winckelmann, Pharm.D.; Isabel Spriet, Pharm.D.; Ludo Willems, Ph.D., Pharm.D.


Pharmacotherapy. 2008;28(11):1391-1400. 

In This Article

Methods to Account for Altered Pharmacokinetics of Phenytoin in Critically Ill Patients

Measurement of Free Drug

Phenytoin meets the theoretical criteria that justify monitoring of free drug levels. The criteria are extensive (> 80%) binding to serum proteins, a narrow therapeutic range, a volume of distribution less than 2 L/kg, a free fraction likely to vary in the therapeutic range, and a correlation between free serum concentrations and pharmacologic effects.[31]

The free fraction varies widely.[12,48] Approximately 20% of patients with epilepsy who received phenytoin had elevated free fractions due to impaired serum protein binding.[49]

Only the free fraction is available to cross the blood-brain barrier, and the free serum concentration reflects the drug concentration in the central nervous system, where phenytoin exerts its action.[13,50,51] However, phenytoin is a substrate for efflux carriers of the blood-brain barrier.[45,52] Persistent subtherapeutic levels were reported in a patient with refractory epilepsy associated with overexpression of MDR1, which encodes the efflux carrier P-glycoprotein.[52]

Free phenytoin concentrations are more strongly correlated with clinical toxicity than are total serum concentrations.[4,49,53–55] Case reports of critically ill patients with phenytoin toxicity showed that free phenytoin levels were disproportionally higher than values expected on the basis of total serum concentrations ( Table 4 ).[4–7,13] This was secondary to increased free phenytoin fractions. High free fractions in the serum raise concentrations in the brain.[13] However, in most cases, total levels exceed the therapeutic range. Increasing the maintenance dosage in critically ill patients with normal or slightly elevated total levels without knowledge of the free fraction must be discouraged.

In addition, free phenytoin serum concentrations reflect a patient's clinical status as well as or better than total concentrations.[40,48,49] This situation can be attributed to highly variable interindividual ranges in the free fraction.

Techniques for monitoring serum concentrations of free drug in clinical laboratories are based on the separation of free drug and subsequent chromatography or immunoassay.[56] Equilibrium dialysis is regarded as the reference method, whereas ultrafiltration is the most common technique for monitoring free drug concentrations in clinical laboratories. Comprehensive reviews of these methods have been published.[10,57–59] Free drug concentration measurements are not widely implemented in clinical practice. An extra step in drug analysis adds time and cost and requires validation of the method under standardized conditions. The pH, temperature, and time for centrifuging are crucial for measuring free drug concentrations.[10,11] Factors that make this approach impractical for routine use are technical difficulties, a lack of established reference ranges for serum concentrations of free drug, and a need for sensitive assays to accurately measure free concentrations of highly bound drugs.[1,28,60] Although measuring both total and free levels is not cost-effective, determination of a free level alone costs about 10% more than obtaining a total level.[55] The availability of relatively inexpensive, commercial ultrafiltration devices has greatly enhanced the ability of clinical laboratories to determine free drug concentrations in patients' samples.[10]

Saliva has been suggested as an alternative matrix for monitoring phenytoin under well-controlled and standardized conditions.[1,11,31,61–63] Excellent correlations between saliva levels and free phenytoin concentrations in the serum are reported.[48,61] In addition, phenytoin concentrations in saliva have been positively correlated with toxicity but not efficacy.[61] In critically ill patients, collection of blood samples and ultrafiltration are simpler than collecting and processing saliva.[48] Further research is necessary to determine the clinical value of this technique in intensive care.

Use of Theoretical Equations

In patients with low serum albumin levels, theoretical equations can be used to determine the phenytoin serum concentration that would have been observed if the albumin levels were normal.[3] Because measurement of free phenytoin concentrations may add costs or may be unavailable at some institutions, use of these theoretical equations has been widely adopted.[64] Adjusted, or normalized, total serum concentrations are obtained from measured total serum concentrations, free fractions in people with normal albumin levels, and albumin serum concentrations in healthy individuals and patients with hypoalbuminemia.

These methods are highly dependent on the accuracy and precision of the population pharmacokinetic parameters used to construct them and on the assumptions of similarity between the study population and the patients for whom the equation is to be used.[41]

Widely adopted is the Sheiner-Tozer equation (equation 1). However, it is reliable only in patients without clinically significant renal impairment or in those not receiving other drugs that highly bind to albumin.[3,8,18]

where C is the concentration of phenytoin (normalized or measured) in micrograms/ milliliter, α is the normal free fraction, and S is the serum concentration of albumin (patient's and normal values) in grams/deciliter.

If α is 0.1 and Salbumin-normal is 4.4 g/dl, the term (1 – α )/Salbumin-normal = (1 – 0.1)/4.4 = 0.2. Therefore, equation 1 becomes equation 2:

Investigators have compared measured and normalized free serum concentrations of phenytoin by applying the Sheiner-Tozer equation in critically ill adults ( Table 5 ).[64–66]

One group found that clinical estimates of free phenytoin serum concentrations based on the Sheiner-Tozer equation were unbiased and precise in a uniform population of critically ill patients with neurosurgical conditions.[65] In contrast, previous researchers[66,67] had found the Sheiner-Tozer equation unreliable and, therefore, unsuitable for application in clinical practice. Temperature effects might account for these contrary results. The free fraction of phenytoin is considerably higher in vivo than in vitro at an ultrafiltration temperature of 25°C.[64,68]

Other investigators discourage the use of equations to estimate free phenytoin serum concentrations in critically ill patients because they found considerable variation despite serum albumin levels that were essentially unchanged.[4,69] A notable bias of more than 10% in over-prediction and underprediction is reported. For most patients, this bias appears to be small enough to warrant an initial empiric appraisal of hypoalbuminemia. However, the Sheiner-Tozer method is probably not accurate enough to guide dosing based on the normalized total phenytoin concentration. Besides albumin levels, many other factors can influence the pharmacokinetics of phenytoin, especially in critically ill patients.

In patients with end-stage renal disease requiring hemodialysis, equation 3 should be applied, as it accounts for alterations in serum albumin concentrations and in binding affinity.[3]

If 0.48 x 0.2 = 0.1, equation 3 is simplified to equation 4:

A reliable method for predicting the free fraction of phenytoin in patients with a wide range of renal function and hypoalbuminemia has yet to be developed. Accumulation of endogenous ligands continue to interfere with protein binding.[34]

Concurrent use of phenytoin and valproic acid is common and usually uneventful. However, given the complexity of their interaction, serum concentrations of phenytoin are difficult to predict. Valproic acid both displaces phenytoin from serum protein binding sites and weakly inhibits CYP2C9.[2] Serum concentrations of free phenytoin initially rise because of the displacement interaction. Total serum levels of phenytoin may fall by 20–50% as free drug is exposed to hepatic metabolism. In some patients, the combination of the nonlinear kinetics of phenytoin and the enzymatic inhibition of CYP2C9 by valproic acid can increase total phenytoin in serum, as well as sustain and significantly increase free phenytoin in serum. These effects may be associated with symptoms of toxicity for which dosage reduction is needed.[21,23,70]

As a CYP2C9 inducer, phenytoin increases the metabolism of valproic acid, reducing its serum concentrations by a mean of 50%.[22] Phenytoin possibly increases the formation of a minor but hepatotoxic metabolite of valproic acid (2 propyl-4-pentenoic acid).[23,25] Close monitoring for seizure control and toxicity is necessary. Equations such as the following equation 5[71] have been designed to estimate serum concentrations of free phenytoin when valproic acid and phenytoin are administered concurrently:

where C is the serum concentration of each drug.

Other researchers[72] compared the predictive performance of equation 5[71] with another equation.[73] Equation 5 leads to predicted free phenytoin concentrations with the least bias and most precision.


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.