Printer Friendly

Evidence-based implementation of free phenytoin therapeutic drug monitoring.

There is increasing interest in measuring free drug concentrations for a variety of substances, based on the principle that the free drug is the pharmacologically active component (1). This is particularly true for phenytoin, which is highly bound to plasma protein (~90%), producing a very small free fraction. The common laboratory practice is to measure total phenytoin concentrations, which assumes that phenytoin protein binding is relatively constant, so that the concentration of the active unbound drug can be predicted from the total drug plasma concentration (2,3). Despite the high percentage of protein binding, there can be substantial interindividual variation in free phenytoin concentrations, ranging from 9% to 25% (4). The free drug fraction of phenytoin can be altered by various compounds and clinical situations, including hypoalbuminemia (5), uremia (6), pregnancy (7), critical illness (8), HIV infection (9), and administration of other drugs, including antiepileptic drugs (4,10,11).

The potential for clinically significant discordance between free and total phenytoin concentrations exists for the above reasons. Several studies have addressed the correlation between total and free phenytoin (12,13), with earlier studies supporting a strong correlation between total and free phenytoin (4,14) and more recent studies advocating free drug measurements (1,10,13 ). Kilpatrick et al. (13) examined the association between free phenytoin and clinical status and found that in all patients the unbound concentration of phenytoin reflected the clinical status equally or better than the total phenytoin concentration.

Recognizing that free phenytoin concentrations are the most appropriate information for clinical use, this report describes the process and data we used to implement monitoring of free phenytoin only, replacing monitoring of total phenytoin, at an urban medical center.

Materials and Methods

Over a 6-week period, both total and free phenytoin concentrations were measured on every clinical phenytoin request sent to the toxicology laboratory at Hennepin County Medical Center, a 400-bed primary care, urban medical center. Total and free phenytoin measurements are were performed on the Hitachi 911 using the Cedia Phenytoin II assay (Boehringer Mannheim Systems). Abbott TDx free phenytoin calibrators at 0.0, 0.5, 1.0, 2.0, 3.0, and 4.0 mg/L were used.

Serum or plasma was obtained from whole blood collected in sodium EDTA, sodium heparin, or lithium heparin tubes. Samples were separated into total phenytoin and free phenytoin aliquots. Separation of free phenytoin and protein-bound phenytoin was achieved by filtering the sample through an anisotropic hydrophilic YMT ultrafiltration membrane (Millipore Corporation), using centrifugation at 1000g at 25[degrees]C for 20 min as the driving force for the ultrafiltration. The temperature of 25[degrees]C was chosen for separating the free fraction of phenytoin because at 25[degrees]C,10% exists as the free fraction, compared with 15% at 37[degrees]C, when dissociation constants were studied in patient samples (15). Thus, our medical center practice of using 10% as the reference for free phenytoin agrees with the literature.

The therapeutic total and free phenytoin concentrations used in our laboratory are 10-20 mg/L and 1-2 mg/L, respectively (16). These ranges have been verified using in-house data (data not shown) and are reported on the laboratory information system patient reports. Total imprecisions (CVs) for total phenytoin and free phenytoin at 24, 2.3, and 1.1 mg/L were 4.7%, 11%, and 22%, respectively. For patients whose percentage of free phenytoin was >12% or <8%, the medical record was reviewed, looking specifically at current medical conditions, current and recent medications, blood urea nitrogen, albumin, whether the patient was pregnant, and the patient's age.


Over the 6-week period, 189 requests were received for phenytoin therapeutic drug monitoring on 139 patients (84 males, 55 females; age range, 7-77 years). Fifty-three patient data points were excluded from the study, leaving 136 patient data points (72%) for analysis. Reasons for data point exclusion included free phenytoin concentrations below the detection limit (0.5 mg/L), total phenytoin concentration <3 mg/L, and sample volume not sufficient for determination of both total and free phenytoin concentrations. The unbound phenytoin fractions were 6.8-35.3% of the total phenytoin concentrations. Sixty-eight of the patient data points (50%)had a free phenytoin fraction (FPF) of 8-12%, and 68 (50%) had a FPF outside of this range. Clinical history was available for 62 of the 68 patient data points for those patients whose FPF was >12% or <8%. Review of the medical records showed that 46 (74%) had an identifiable explanation or reason for this high or low FPF, leaving the remaining 16 (26%) with no identifiable cause for the abnormal FPF. Table 1 shows the incidence of clinical situations known to affect the FPF. Many patients had several reasons for increased or low FPF and were included in multiple categories. Hypoalbuminemia and valproic acid coadministration were the most common clinical situations associated with an abnormal FPF. Uremia, critical illness, and age extremes were also common clinical associations.

The clinical importance of the FPF became obvious when total phenytoin concentrations were compared with free phenytoin concentrations. For all patient data points, 56 (41%) had therapeutic total phenytoin concentrations, 32 (24%) had total phenytoin concentrations greater than therapeutic, and 48 (35%) had subtherapeutic total phenytoin concentrations. For the 56 patient data points with therapeutic total phenytoin concentrations, 5 had free phenytoin concentrations >2 mg/L and 3 had free phenytoin concentrations <1 mg/L. For the 32 patient data points with supratherapeutic total phenytoin concentrations, there were no subtherapeutic free phenytoin concentrations and 5 with therapeutic free phenytoin concentrations. The majority of the subtherapeutic total phenytoin patient data points had therapeutic free phenytoin concentrations (28 of 48).

The last two columns in Table 2 divide the patient data points into those with an FPF between 8% and 12% (group 1) and those with a FPF <8% or >12% (group 2). In group 1, 51% had both therapeutic total and free phenytoin concentrations (35 of 68). Two of 37 patient data points had subtherapeutic free phenytoin concentrations with therapeutic total concentrations, and 3 of 19 patient data points had therapeutic free concentrations with supratherapeutic total concentrations. Patient data points in group 2 had fewer (19%) therapeutic total and free concentrations (13 of 68) when compared with group 1. Of the 19 patient data points with therapeutic total phenytoin concentrations, 5 had supratherapeutic and 1 had subtherapeutic free concentrations. Group 2 demonstrated a relatively high number of patient data points (28 of 68; 41%) with therapeutic free phenytoin concentrations and subtherapeutic total phenytoin concentrations. This finding was not observed in group 1.

When the concordance for all patient data points was compared, 95 of 136 (70%) of the free and total phenytoin concentrations were in agreement; i.e., both the free and the total phenytoin concentrations were supra-, sub-, or therapeutic. For those patient data points in group 1, 93% were concordant, whereas those in group 2 had only 47% concordant, leaving a majority of 53% discordant. These data were distributed to the hospital staff along with notification of plans to drop routine measurement of total phenytoin. The change was well accepted. Only two total phenytoin concentrations have been requested in the 5 months since the change was made.


There is growing literature demonstrating that the relationship between free and total phenytoin concentrations is more unpredictable than previously thought. Kilpatrick et al. (13) reported a range of 6.7-33.3% free fraction, and Peterson et al. (4) reported a range of 9.7-24.7%. Our free fraction range of 6.8-35.3% compares well to these previously published reports. However, 50% of our patient data points fell outside the 8-12% range expected from review of the literature.

Several clinical conditions and medications are known to cause alterations in FPF. The maj ority of the patient data points in this study (74%)had an identifiable clinical diagnosis or concurrent drug therapy to account for an FPF outside the 8-12% range. This leaves 26% of the patient data points for which the free, pharmacologically active fraction cannot be accurately predicted based on the total phenytoin concentration. Coadministration of valproic acid is a well-documented cause of altered phenytoin protein binding, but the FPF is not predictable (17). Valproic acid coadministration was commonly associated with altered phenytoin binding in this study. In our patients, hypoalbuminemia accounted for the largest clinical diagnosis associated with an abnormal FPF. Even when albumin was within the reference range, there was an inverse relationship between the plasma albumin concentration and the unbound phenytoin fraction (13), such that those patients with borderline albumin concentrations might not have predictable FPF.

One of the most striking findings of this study was the 30% overall rate of discordance between the total and free phenytoin concentrations (47% for group 2). Put another way, in 30% of patients on phenytoin therapy, the total phenytoin concentration was not accurately portraying the pharmacologically active component, free phenytoin concentration. This has serious implications for the management of epilepsy with phenytoin and for therapeutic drug monitoring of phenytoin. Phenytoin has a narrow therapeutic range, with potentially significant adverse events associated with under- or overmedicating a particular patient (18). Additionally, our data show that even in patients without known causes of altered FPF, the free fraction cannot be predicted reliably based on the total phenytoin concentration.

On the basis of the evidence presented in this study, our clinical laboratory replaced total phenytoin measurements with free phenytoin measurements after this 6-week period of parallel free and total phenytoin concentration measurements. In the 5 months subsequent to this, only two total phenytoin measurements have been performed, one because of technician error and one in response to an unusual clinical situation. We postulate that in a given patient, the FPF fluctuates in comparison with the total phenytoin over time because of changes in the clinical status of the patient. Because we and the practitioners taking care of these patients cannot predict how a change in management or clinical status will affect the free and total phenytoin concentrations, we recommend measuring only free phenytoin in all patients. This has been widely accepted at our medical center, without reports of any adverse outcomes. In contrast, for example, before implementation of our free phenytoin only policy, four of our patients who had therapeutic total concentrations were found to have toxicity with free phenytoin concentrations >3.0 mg/L. Additionally, to identify those patients with normal vs abnormal binding fractions, both total and free phenytoin concentration measurements would have to be done on every new patient and at every point when the clinical status of the patient changed, to verify if the FPF had changed. In comparison, measuring only free phenytoin on all patients is much simpler, less costly than performing both the total and free measurements, and is a true measurement of the pharmacologically active drug. Therefore, we recommend free phenytoin concentration determinations for all therapeutic drug monitoring laboratories, with the elimination of total phenytoin determinations for routine practice.

Finally, we point out that the parameter "functional sensitivity", defined as an imprecision (CV) of 20%, has been proposed to serve as a clinically relevant estimate of the lowest reporting limit for an assay (19). This parameter accounts for the imprecision associated with interassay variables, such as different calibration and reagent lots, and biological variables. A potential limitation of our findings could be attributed to our low-end CV of 22% at 1.1 mg/L, which falls above the 20% functional sensitivity required. Our patient groupings of data could possibly have been influenced, thus affecting our conclusions.


(1.) Soldin S. Free drug measurements: when and why? An overview. Arch Pathol Lab Med 1999;123:822-3.

(2.) Kutt H, Winters W, Kokenge R, McDowell F. Diphenylhydantoin metabolism, blood levels and toxicity. Arch Neurol 1964;11: 642-8.

(3.) Lund L. Anticonvulsant effect of diphenylhydantoin relative to plasma levels. Arch Neurol 1974;31:289-94.

(4.) Peterson GM, McLean S, Aldous S, von Witt RJ, Millingen KS. Plasma protein binding of phenytoin in 100 epileptic patients. Br J Clin Pharmacol 1982;14:298-300.

(5.) Fedler C, Stewart MJ. Plasma total phenytoin: a possibly misleading test in developing countries. Ther Drug Monit 1999;21:155-60.

(6.) Reidenberg MM, Odar-Cederlof I, von Bahr C, Borga 0, Sjoqvist F. Protein binding of diphenylhydantoin and desmethylimipramine in plasma from patients with poor renal function. N Engl J Med 1971;285:264-7.

(7.) Perucca E, Richens A, Ruprah M. Serum protein binding of phenytoin in pregnant women. Br J Clin Pharmacol 1981;11:409-10.

(8.) Boucher BA, Hanes SD. Pharmacokinetic alterations after severe head injury. Clinical relevance. Clin Pharmacokinet 1998;35:20921.

(9.) Dasgupta A, McLemore JL. Elevated free phenytoin and free valproic acid concentrations in sera of patients infected with human immunodeficiency virus. Ther Drug Monit 1998;20:63-7.

(10.) Lenn NJ, Robertson M. Clinical utility of unbound antiepileptic drug blood levels in the management of epilepsy. Neurology 1992;42: 988-90.

(11.) Kwong TC. Free drug measurements: methodology and clinical significance. Clin Chim Acta 1985;151:193-216.

(12.) Ohshima T, Hasegawa T, Johno I, Kitazawa S. Variations in protein binding of drugs in plasma and serum. Clin Chem 1989;35: 1722-5.

(13.) Kilpatrick CJ, Wanwimolruk S, Wing LMH. Plasma concentrations of unbound phenytoin in the management of epilepsy. Br J Clin Pharmacol 1984;17:539-46.

(14.) Barth N, Alvan A, Borga 0, Sjoqvist. Two-fold interindividual variation in plasma protein binding of phenytoin in patients with epilepsy. Clin Pharmacokinet 1976;1:444-52.

(15.) Ratnaraj N, Goldberg VD, Hjelm M. Temperature effects on the estimation of free levels of phenytoin, carbamazepine, and phenobarbitone. Ther Drug Monit 1990;12:465-72.

(16.) Winkler S, Luer MS. Antiepileptic drug review: part I. Surg Neurol 1998;49:449-52.

(17.) Anderson GD. A mechanistic approach to antiepileptic drug mechanisms. Ann Pharmacother 1998;32:554-63.

(18.) Smith DB, ed. Antiepileptic drug selection in adults in epilepsy: current approaches to diagnosis and treatment. New York: Raven Press, 1990:111-39.

(19.) Spencer CA, Takeuchi M, Kazarosyan M. Current status and performance goals for serum thyrotropin (TSH) assays. Clin Chem 1996;42:140-5.


Departments of Laboratory Medicine and Pathology and Neurology, Hennepin County Medical Center, University of Minnesota School of Medicine, Minneapolis, MN 55415.

* Address correspondence to this author at: Hennepin County Medical Center, Clinical Laboratories 812,701 Park Ave., Minneapolis, MN 55415. Fax 612-904-4229; e-mail

Received January 27, 2000; accepted May 11, 2000.
Table 1. Clinical findings likely responsible for abnormal free
phenytoin percentages.

Indication Number

Hypoalbuminemia 18
Valproic acid coadministration 16
Age 6
Uremia 6
Critical illness 5
Phenobarbital coadministration 3
Pregnancy 1
Carbamazepine coadministration 1

Table 2. Comparison of total and free phenytoin concentrations in 136
study patients.

 Group 1:
Phenytoin, mg/L Number with
 [greater than or equal to]8% and
 [less than or equal to]12%
Total Free free

Therapeutic (a) Therapeutic 35
Therapeutic >2.0
Therapeutic <1.0 2
>20.0 >2.0 16
>20.0 Therapeutic 3
>20.0 <1.0
<10.0 <1.0 12
<10.0 Therapeutic
<10.0 >2.0

 Group 2:
Phenytoin, mg/L Number with
 <8% or >12%
Total free

Therapeutic (a) 13
Therapeutic 5
Therapeutic 1
>20.0 11
>20.0 2
<10.0 8
<10.0 28

(a) Therapeutic represents 10-20 mg/L for total and 1-2 mg/L for free.
COPYRIGHT 2000 American Association for Clinical Chemistry, Inc.
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2000 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Title Annotation:Drug Monitoring and Toxicology
Author:Burt, Martha; Anderson, David C.; Kloss, Julie; Apple, Fred S.
Publication:Clinical Chemistry
Article Type:Clinical report
Date:Aug 1, 2000
Previous Article:Analytical and clinical evaluation of two homogeneous assays for LDL-cholesterol in hyperlipidemic patients.
Next Article:Space flight is associated with rapid decreases of undercarboxylated osteocalcin and increases of markers of bone resorption without changes in their...

Related Articles
Serum phenytoin levels of patients on gastrostomy tube feeding.
Falsely increased immunoassay measurements of total and unbound phenytoin in critically ill uremic patients receiving fosphenytoin.
Prodrug metabolites: implications for therapeutic drug monitoring.
Glucuronidation of prodrug reactive site: isolation and characterization of oxymethylglucuronide metabolite of fosphenytoin.
Cross-reactivity of fosphenytoin in four phenytoin immunoassays.
Cross-reactivity of fosphenytoin in two human plasma phenytoin immunoassays.
Standards of laboratory practice: antiepileptic drug monitoring.

Terms of use | Privacy policy | Copyright © 2019 Farlex, Inc. | Feedback | For webmasters