Thomas R. Browne MD*
Barbara Leduc PhD**
* Professor of Neurology, Department of Neurology, Boston University School of Medicine, Boston, Massachusetts
** Massachusetts College of Pharmacy and Allied Sciences, Boston, Massachusetts
Phenytoin is the generic name for 5,5-diphenylhydantoin (acid form). The Chemical Abstracts name is 5,5-diphenyl-2,4-imidazolidinedione. It has the chemical structure shown inFigure 58.1. The free acid has a molecular weight of 252.26; the sodium salt has a molecular weight of 274.25, equivalent to acid content of 91.98%. The acid form is used in formulations of aqueous suspensions (Pediatric Dilantin-30 Suspension and Dilantin-125 Suspension; Pfizer, New York, NY) containing 30 mg or 125 mg of phenytoin acid per 5 mL. The free acid also is used in formulating chewable tablets (Dilantin Infatabs; Pfizer) containing 50 mg phenytoin acid per tablet. However, other products are formulated with the sodium salt of phenytoin (phenytoin sodium; acid equivalents = 91.98%). With these preparations, the drug content is expressed in terms of the sodium salt rather than the free acid. Thus, the gelatin capsules (Dilantin Sodium Kapseals; Pfizer) are formulated to contain either 30 mg or 100 mg of phenytoin sodium (= 27.6 mg or 92.0 mg of phenytoin acid equivalents) per capsule. The Mylan extended-release 100-mg capsules also are formulated with sodium phenytoin. This 8% difference in drug content should be taken into account when changing from one product to another. The sodium salt also is used in parenteral formulations [Parenteral Dilantin (Pfizer); phenytoin sodium injection (generic)]. The drug content is given in terms of the sodium salt.
Phenytoin is a weak organic acid that is poorly soluble in water. The apparent dissociation constant (pKNa), representing the pH at which half the drug is ionized, is in the range of 8.1 to 9.2. The acid essentially is nonionized at pH 5.4 (solubility approximately 19.4 µg/g at 25.4°C), whereas at pH 7.4 (approximately 80% nonionized), the acid has a water solubility of 20.5 µg/g (25.2°C). Higher concentrations of phenytoin required strongly alkaline solutions, with solubility measurements of 165 µg/mL at pH 9.1 and 1,520 µg/mL at pH 10. Parenteral phenytoin sodium is made up in an aqueous vehicle containing propylene glycol, ethanol, and sodium hydroxide. It contains 50 mg phenytoin sodium per mL (= 46 mg phenytoin acid per mL). The solubility of phenytoin in blood plasma is approximately 75 µg/mL (37°C), at least in part because of binding of the drug on the plasma proteins.
Phenytoin sodium is not recommended as an analytical standard because of its variable water content (hydrate formation) and partial conversion to the free acid on exposure to carbon dioxide.
This section was based on Glazko (61), which should be consulted for details and references.
Phenytoin (5,5-diphenylhydantoin; Dilantin) is eliminated almost entirely by metabolic transformation before excretion in the form of metabolites. Less than 5% of an administered dose is excreted unchanged in the urine (45,62, 114,121). The principal metabolic pathway of phenytoin in humans is the 5-(4-hydroxyphenyl)-5-phenylhydantoin (p-HPPH) and dihydrodiol pathway, accounting for 70% to 90% of administered phenytoin (Figure 58.2). The first
step of this pathway [involving the cytochrome P450 (CYP) enzymes CYP2C9 and CYP2C19] exhibits nonlinear enzyme kinetics, which has significant effects on phenytoin's clinical pharmacokinetics. A number of minor metabolic pathways for phenytoin also have been described, some of which involve the CYP enzymes CYP2C9, CYP2C19, and CYP3A4 (Figure 58.2). Species differences in the biotransformation of phenytoin have been reported.
FIGURE 58.1. Structure of phenytoin.
FIGURE 58.2. Pathways of phenytoin metabolism.
The Para-HPPH and Dihydrodiol Pathways
Para-HPPH accounts for 67% to 88%, and dihydrodiol accounts for 7% to 11% of human urinary metabolites of phenytoin (23,24,32,43,75,90). The first step in this pathway is the formation of an arene oxide intermediate through the enzymes CYP2C9 and CYP2C19 (Figure 58.2). The arene oxide is converted spontaneously to p-HPPH and is converted by the enzyme epoxide hydrolase to dihydrodiol (Figure 58.2).
Cytochrome P450 Enzymes Active in Phenytoin Metabolism and Their Genetic Variants
Enzymes belonging to the CYP2C subfamily and the CYP3A family have been implicated in the biotransformation of phenytoin and its metabolites (40,95). A substantial number of human allelic variants of these enzymes have been identified.
Four members of the CYP2C subfamily, CYP2C8, CYP2C9, CYP2C18, and CYP2C19, have been identified in humans (64,110). The CYP2C enzymes, which constitute approximately 20% of total hepatic CYP content, are encoded by a cluster of four genes at chromosomal location 10q24 (85,138). Both the CYP2C9 and CYP2C19 enzymes are active in phenytoin hydroxylation, with CYP2C9 being the more vigorous catalyst (10,51,82,154, 167). Although earlier in vitro studies suggested a role for CYP2C8 and CYP2C18, their contribution to phenytoin
metabolism in vivo has been questioned, and studies have indicated that only negligible amounts of CYP2C18 are expressed in human liver (130). CYP2C9 catalyzes the formation of both (R)- and (S)-HPPH [5-(4′-hydroxyphenyl)-,5-phenylhydantoin] but is highly stereoselective for (S)-HPPH, whereas CYP2C19 exhibits little stereoselectivity and thus contributes more to the formation of the (R)-enantiomer (82,110,167). In addition, studies have identified CYP2C19, CYP2C9, and several CYP3A forms as the catalysts for the further hydroxylation of 3′- and 4′-HPPH to the catechol 3′,4′-diHPPH [5-(3′,4′-dihydroxyphenyl)-,5-phenylhydantoin] (40,95).
Four alleles of CYP2C9 have been identified. The wild-type enzyme, CYP2C9*1, contains arginine at position 144 and isoleucine at position 359 (132). The variant CYP2C9*2 contains a single amino acid substitution at position 144 (Arg144Cys), whereas in CYP2C9*3, leucine is substituted at position 359 (Ile359Leu) (39,69,122, 128,141). Another mutation generating an amino acid substitution at position 359 has been identified in CYP2C9*4 (Ile359Thr) (84). (For a current listing of human CYP450 alleles, see www.imm.ki.sec/CYPalleles). Although the activity of CYP2C9*2 for HPPH formation is reported to be only moderately diminished compared with that of CYP2C9*1, CYP2C9*3 shows markedly decreased catalytic activity both in vitro and in vivo (9,40,52,72,110,122, 128,143,154). Interestingly, a white patient with epilepsy displaying a very low clearance of phenytoin was found to be homozygous for CYP2C9*3 (91). The CYP2C9*4 mutation (Ile359Thr) is extremely rare, but was associated also with a diminished rate of phenytoin hydroxylation in vivo (83). It has been suggested that amino acid 359 is located in the enzyme's substrate recognition site; this would explain the large detrimental effect of substitutions at this position (67). To date, poor metabolizers of phenytoin have been shown to have mutations of either CYP2C9 or CYP2C19. Although CYP2C9*3 is the primary determinant of slow phenytoin metabolism, defective CYP2C19 alleles also contribute, especially at high doses (120).
The incidence of allelic variants of CYP2C9 varies significantly between different racial groups. The gene frequency of CYP2C9*2 has been estimated as 0.08 to 0.125 in whites, 0.01 in Africans, and zero in East Asians (9,122,140,141). The predicted frequency of CYP2C9*3 in whites is 0.06 to 0.10, in East Asians, 0.017 to 0.026, and in Africans, 0.005 (9,92,122,140,141,157).
In contrast to CYP2C9, most variant alleles of CYP2C19 identified thus far appear to result in nonfunctional or absent enzymes. The CYP2C19 polymorphism is responsible for the variation in human 4′-hydroxylation of (S)-mephenytoin (41,42). Poor metabolizers of (S)-mephenytoin constitute approximately 2% to 6% of white or African, and 18% to 23% of East Asian populations (92,119,164, 165, 166). Two wild-type and seven defective alleles of CYP2C19 have been identified thus far (87). As with CYP2C9, the incidence of specific alleles varies significantly among different ethnic groups. CYP2C19*1A and CYP2C19*1B are the wild-type forms (129,132). The allelic variants CYP2C19*2 (CYP2C19m1), CYP2C19*3 (CYP2C19 m2), and CYP2C19*5 account for approximately 99% of East Asian, but only 75% to 85% of white poor metabolizers of (S)-mephenytoin (54,87). CYP2C19*2B, CYP2C19*4, CYP2C19*5B, and CYP2C19*6 account for the remainder of slow metabolism in whites (54). The mutations found in CYP2C19*2A and CYP2C19*2B are due to G6A point mutations in exon 5, leading to aberrant splice sites (41,42,80). CYP2C19*3 is due to a G6A point mutation in exon 4, causing premature termination (41,42). In CYP2C19*4, an A→G mutation disrupts the initiation codon (50). CYP2C19*5A contains an amino acid substitution (Arg433Trp), and CYP2C19*5B contains this, plus an additional substitution (Arg433Trp; Ile331Val); both enzymes appear to lack activity (79,80,163). CYP2C19*6 (Argl32Gln; Ile331Val), CYP2C19*7 (splicing defect), and CYP2C19*8 (Trp120Arg) apparently are absent or inactive as well (80,81).
Although CYP2C9 is the primary determinant, a number of studies have confirmed that the activity of CYP2C19 is clinically significant to overall phenytoin metabolism. Diminished renal elimination of (R)-HPPH and total HPPH, a substantially decreased urinary (R)-HPPH/phenytoin ratio, elevated serum phenytoin concentrations, and an elevated Michaelis constant (Km) for phenytoin have been demonstrated in Japanese subjects who were either heterozygous or homozygous for mutant CYP2C19 alleles (82,110,116,120,122,157,158). Furthermore, ticlopidine, a potent inhibitor of CYP2C19, has been implicated in drug interactions resulting in phenytoin toxicity (47).
The gene frequency of CYP2C19*2 has been estimated as 0.11 to 0.13 in whites and in Africans, but 0.27 to 0.37 in East Asians (16,30,92,112,133,142,163). In contrast, the CYP2C19*3 allele is rare or absent in whites, whereas its frequency is approximately 0.5 to 0.11 in East Asians (92,133,163). Interestingly, in Japanese but not whites, mutations of CYP2C18 and CYP2C19 appear to be completely linked, that is, the occurrence of CYP2C18m2 coincides with that of CYP2C19*32, and CYP2C18ml occurs with CYP2C19*3 (96,109,116). The CYP2C9 mutations were found to be independent of either CYP2C18 and CYP2C19 alleles (86).
Collectively, the CYP3A enzymes constitute most of the total hepatic CYP. Four genes have been described: CYP3A4, CYP3A5, CYP3A7, and a newly identified form, CYP3A43 (56,147). cDNA analysis of CYP3A4, CYP3A5, and CYP3A7 sequences confirms at least 90% homology between them (71,160). CYP3A43 is predicted to be at least 71% homologous with other CYP3A forms (56). The CYP3A forms display broad and overlapping substrate specificities,
and are thought to participate in the metabolism of at least 50% of all drugs (49,103). Although CYP3A4 exhibited the greatest activity, recombinant CYP3A4, CYP3A5, and CYP3A7 isoforms were shown to be capable of catalyzing the conversion of 3′- and 4′-HPPH to the catechol diHPPH in vitro (40,93). In humans, total CYP3A activity displays large interindividual and interethnic variations that might be due to differences in induction, inhibition, or regulation of expression (97,138). To facilitate study of the control of CYP3A gene expression, the human CYP3A gene locus at 7q21.1 has been sequenced (56).
CYP3A4 is the most common form of CYP450 in liver and intestine. CYP3A4*1A is the wild-type form (65). Although CYP3A4*1B contains an A→G point mutation in the 5′-flanking region, there appears to be no difference in the level of hepatic expression of the enzyme compared with the wild-type form (89,161). The frequency of this mutation is 4.2% to 9% in whites, 53% to 66% in African Americans, and 0% in Taiwanese subjects (134,156). CYP3A4*2 contains a Ser222Pro substitution, resulting in lower in vitro nifedipine metabolism but, curiously, no difference in testosterone β-hydroxylation (134). This variant was found in 2.7% of white subjects, but was absent in both East Asians and African Americans (134). A CYP3A4*3 variant, containing a Met445Thr substitution, was found in only one Chinese subject (132). Recently, Hsieh et al. identified three rare variants in Chinese subjects, CYP3A4*4 (Ilel18Val), CYP3A4*4 (Prol02Arg), and CYP3A4*6 (frameshift), which were associated with decreased β-hydroxycortisol formation (77).
CYP3A5 appears to be expressed considerably more frequently than previously thought (13,97). The isoform constitutes up to 50% of total CYP3A in subjects who express the gene, and may be an important contributor to interindividual and interracial differences in CYP3A-dependent drug disposition (97). Only subjects having at least one copy of CYP3A5*1 express large amounts of the enzyme (95). The CYP3A5*1B and CYP3A5*1C variants differ from the wild-type gene (CYP3A5*1A) by having nucleotide substitutions in the promoter region, although these mutations are not associated with differences in enzyme activity in vivo (1,97). The CYP3A5*2, CYP3A5*3, and CYP3A5*6 variants appear to result in absence of the hepatic enzyme (89,97). Hepatic CYP3A5 is expressed much more frequently in African Americans (66%) than in whites (33%) (97).
CYP3A7 is the fetal form of the enzyme, although expression has been known to persist into adulthood in some people (70,135). A partial explanation for this continued expression may be the discovery of the CYP3A7*1C variant, in which a segment of the CYP3A4 promoter is substituted into the equivalent CYP3A7 promoter region. Interestingly, this CYP3A7*1C allele is three times more common in African Americans than in whites (97).
The arene oxide of phenytoin has never been isolated from plasma or urine, presumably because it is rapidly converted to further metabolic products. The existence of an arene oxide intermediate in the formation of p-HPPH was suggested by early kinetic experiments (148) and was established using the “NIH shift” technique (38,117).
The NIH shift technique depends on the observation that arene oxides spontaneously break down to form a hydroxylated metabolite, and that when this happens the hydrogen molecule at the site of the hydroxyl group is lost 50% of the time and is shifted (NIH shift) to the adjacent carbon, where the oxygen molecule had been attached 50% of the time. The NIH shift experiment of Claesen et al. (38) is shown in Figure 58.3. Racemic 5-(4-deuteriophenyl)-5-phenylhydantoin (p-[2H]-DPH) was administered to volunteers, and urine was collected. After enzymatic hydrolysis of the urine, deuterium retention by p-HPPH metabolites was determined. The expected percentage of deuterium retention by p-HPPH in the presence of an arene oxide/NIH shift pathway was 75%, and the measured values were 65% to 75% in four volunteers.
Phenytoin is known to exhibit nonlinear pharmacokinetics in humans (see later), implying that substrate saturation must be occurring in one or more of the enzymes metabolizing phenytoin. The CYP2C9 and CYP2C19 enzymes appear to be the site of substrate saturation resulting in nonlinear pharmacokinetics in humans based on the following observations: (a) Browne et al. (18) demonstrated that the rate of urinary excretion of p-HPPH varies inversely with phenytoin plasma concentration in humans (r = -0.640, p < 0.005) (9); (b) Tsuru et al. (149) demonstrated that the rate of formation of p-HPPH from rat hepatocytes and rat hepatic microsomes varies inversely with phenytoin concentration in the medium. Phenytoin is the probable substance causing competitive inhibition of the arene oxidase enzyme (25). There is some evidence that high concentrations of p-HPPH competitively inhibit hydroxylation of phenytoin in animals (76), but this phenomenon has not been demonstrated in vivo in humans (25,32). Arene oxides have attracted considerable attention because of their reactivity and possible role in toxicity and teratogenicity mechanisms.
Para-HPPH and Meta-HPPH
The principal urinary metabolite of phenytoin is p-HPPH in humans (27,28,146). Most p-HPPH is excreted as a glucuronide, and only small amounts of free p-HPPH are found in human urine (31,114). Very small amounts of m-HPPH also are found in human urine (5,29,63). Species differences in the patterns of p-HPPH and m-HPPH exist (5,28,73,121). The details of these differences have been reviewed previously (17).
FIGURE 58.3. Arene oxide-NIH shift pathway for para-hydroxylation of phenytoin. (From Claesen M, Moustafa MAA, Adline J, et al. Evidence for an arene oxide-NIH shift pathway in the metabolic conversion of phenytoin to 5-(4-hydroxyphenyl)-5-phenylhydantoin in the rat and in man. Drug Metab Dispos 1982;10:667-671, with permission.)
Mechanism of Formation
NIH shift experiments indicate that most p-HPPH in humans is produced through an arene oxide intermediate (see earlier). CYP2C9 and CYP2C19 are the enzymes responsible for the formation of arene oxide (see earlier).
A minority of patients are “slow metabolizers” of phenytoin (32,63,98,99,152,153). In these individuals, toxic phenytoin plasma concentrations develop at average dosing rates of phenytoin. Mutations of CYP2C9 or CYP2C19 are the causes of slowed metabolism (see earlier).
We (17) have reviewed the key papers on phenytoin dihydrodiol.
Mechanism of Formation
Phenytoin dihydrodiol is believed to be formed from phenytoin arene oxide through the enzyme epoxide hydrolase (Figures 58.2 and 58.4). This belief is supported by several lines of evidence. First, the usual metabolic pathway leading to formation of dihydrodiols is conversion of an epoxide intermediate to a dihydrodiol by the enzyme epoxide hydrolase (32,68,88,123). Second, administration of an epoxide hydrolase inhibitor (1,2-epoxy-3,3,3-trichlorophropane) to pregnant Swiss mice reduced the incidence of phenytoin-related teratogenesis and the covalent binding of 14C radioactivity in
the fetus after administration of 14C-labeled phenytoin, while maternal phenytoin serum concentrations remained unchanged (111). This implies that the initial step of phenytoin metabolism (presumably the arene oxide formation) is unaffected by epoxide hydrolase inhibition and that a metabolic intermediate with teratogenic and covalent binding properties (presumably arene oxide) accumulates in greaterthan-usual amounts when epoxide hydrolase is inhibited. Third, incubation of phenytoin with human liver fractions known to contain epoxide hydrolase results in the formation of phenytoin dihydrodiol (127). There is evidence that chronic administration of phenytoin may induce epoxide hydrolase activity in humans (18,127).
FIGURE 58.4. Scheme depicting stereoselective metabolic pathways involved in the production of 5-(4-hydroxyphenyl)-5-phenylhydantoin (p-HPPH) and dihydrodiol (DHD). Possible stereoselective direct hydroxylation pathways are indicated with an “X.” [From Maguire JH, Butler TC, Dudley KH. Absolute configuration of (+)-5-(3-hydroxyphenyl)-5-phenylhydantoin, the major metabolite of 5,5-diphenylhydantoin in the dog. J Med Chem 1978;21:1194-1297, with permission.]
Stereoselective Formation of HPPH and Dihydrodiol
Phenytoin is a prochiral compound. Introduction of a hydroxyl group in one of the phenyl rings leads to the creation of a chiral center and results in the formation of enantiomeric phenolic metabolites (Figure 58.4). CYP2C19 catalyzes the formation of both (R)-HPPH and (S)-HPPH, whereas CYP2C9 is more stereoselective in the formation of (S)-HPPH (82,110,167). The p-HPPH from human urine consists of a 10:1 mixture of levorotatory and dextrorotatory isomers (29,51,108). The amount of m-HPPH in human urine is too low to permit isolation and measurement of optical rotation. Most dihydrodiol in human urine is in the (S) configuration [3:1 ratio of (S):(R)] (105,107,108,124). Although an arene oxide of phenytoin has not been directly isolated and characterized, a large body of evidence has accumulated in support of the existence of (S) and (R) arene oxide intermediates. The proposed pathway is depicted in Figure 58.4. There are species differences in the chirality of p-HPPH, m-HPPH, and dihydrodiol excreted in the urine (29,51,105,107,108).
Effects of Head Trauma and Pregnancy
After head trauma, there is decrease in plasma albumin concentration that begins immediately, reaches a minimum at 5 to 7 days, and does not return to normal for 3 to 4 weeks (2). After head trauma, there also is an increase in phenytoin metabolism (3,15). These changes are accompanied by a decrease in free and total phenytoin plasma concentration and an increase in free fraction of phenytoin (2,4,15). Frequent monitoring of free phenytoin plasma concentration and frequent adjustment of phenytoin dosage often are necessary after head injury.
Phenytoin does not appear to be metabolized to p-HPPH or derivatives by the human placenta (94).
Minor Metabolic Pathways
Catechol and 3-O-Methylcatechol Metabolites
Relatively small amounts of a 3,4-catechol metabolite, -(3,4-dihydroxyphenyl)-5-phenylhydantoin, and a 3-O-methylcatechol metabolite, 5-(4-hydroxy-3-methoxyphenyl)-5- phenylhydantoin, have been identified in the urine of humans and other species (12,14,33,34,115). The same metabolites also have been identified as glucuronides in the bile from isolated, perfused rat liver (60).
The catechol could be formed through dehydrogenation of the dihydrodiol metabolite by a mechanism similar to those described by Ayengar et al. (8). In a preliminary experiment,14C-labeled dihydrodiol metabolite isolated from rat urine was administered perorally to rats. It did not result in the appearance of the catechol metabolite in urine (33). The catechol metabolite could be formed by further hydroxylation of m-HPPH or p-HPPH. Gerber and Thompson (57) identified small amounts of the catechol and 3-O-methylcatechol in rat bile by gas chromatography/mass spectrometry (GC/MS) techniques after administration of m-HPPH or p-HPPH to rats or by addition of these compounds to an isolated, perfused rat liver preparation. More recently, Billings and Fischer (12) and Chow and Fischer (36) provided evidence that the catechol metabolite of phenytoin was formed from both dihydrodiol and p-HPPH in rats and mice. However, the predominant route was through p-HPPH (36).
In rat urine, the concentration of the 3-O-methylcatechol metabolite was approximately fivefold greater than that of the catechol, indicating extensive O-methylation in this species (33). Evidence obtained by Chang et al. (34) indicates that the 3-O-methylcatechol metabolite of phenytoin is formed from the catechol. Administration of the synthetic 3,4-dihydroxycatechol to rats resulted in the prompt appearance of 3-O-methylcatechol in the urine (33). It is reasonable to assume that the enzyme catechol-O-methyltransferase is involved in the formation of this metabolite because it is known to methylate other catechols, including catecholamines (7).
4,4N-Dihydroxy Metabolite and N-Glucuronide of Phenytoin
Two minor metabolites of phenytoin—namely, 5,5-bis(4-hydroxyphenyl) hydantoin (145) and an N-glucuronide of phenytoin (139)—were identified in rats and in humans (Figure 58.2). The 4,4N-dihydroxy metabolite was excreted as a glucuronide and accounted for approximately 1% of the total hydroxylated metabolites. The pathway that leads to the formation of this metabolite is not known. When p-HPPH was added to the perfusate of an isolated rat liver preparation, there was no evidence of the formation of the 4,4N-dihydroxy metabolite (145). However, addition of the synthetic 4,4N-dihydroxy compound to the same in vitro preparation (145) resulted in the formation of a monoglucuronide, a trihydroxyphenytoin glucuronide, and a dihydroxymethoxyphenytoin glucuronide, indicating further hydroxylation of the dihydroxy metabolite. Whether the trihydroxylated product represents normal phenytoin metabolites in intact animals is not known.
The glucuronide of phenytoin was isolated from a patient receiving phenytoin and was characterized as the N-3 glucuronide by GC/MS. The same metabolite also was present in the bile of an isolated, perfused rat liver preparation. The structure assignment was based on the mass spectra of various permethylated derivatives and a comparison of the reaction of the metabolite and 5,5-diphenyl-3-methylhydantoin with diazomethane (139). Subsequently, Hassell et al. (73) reported that phenytoin N-glucuronide was the major metabolite in cat urine.
A number of other minor metabolites of phenytoin, some not yet identified, are known to exist (31,32).
The metabolite distribution patterns in human urine were reported by Chang and coworkers (32,35): 68% to 81% p-HPPH, 7% to 11% dihydrodiol, 2.5% 3-O-methylcatechol, 1% unchanged phenytoin, and approximately 1% catechol. Metabolites in human feces (1% of fecal radioactivity) were identified as p-HPPH (50% to 80%), dihydrodiol (15% to 35%), and catechol (5% to 10%). Later studies have confirmed these values (43).
SPECIES DIFFERENCES IN METABOLISM
With the exception of the dog and cat, the major metabolic product of phenytoin in all species, including humans, is a glucuronide conjugate of p-HPPH (17,66,106). Significant amounts of free p-HPPH were observed in the urine of mice and rats (32). The dihydrodiol metabolite was found in higher concentrations in urine of rats and monkeys than in that of mice, dogs, or cats. The catechol and 3-O-methylcatechol metabolites were detected in rat urine after repeated administration of phenytoin in the diet and accounted for 2% and 20%, respectively, of the total metabolite present (34). In dog urine, the major metabolite was the glucuronide conjugate of m-HPPH, whereas phenytoin N-glucuronide was the major urinary product in cat.
Species differences in the formation of glucuronide conjugates of m-HPPH or p-HPPH in rat or dog liver supernatant were reported by Gabler (53). Conjugation of m-HPPH was greater than that of p-HPPH in the dog liver, whereas the opposite effect was found with rat liver supernatant.
NONLINEAR PHARMACOKINETIC PROPERTIES OF PHENYTOIN
The rate of change of plasma concentration (C) of a drug by an enzyme system can be expressed by the Michaelis-Menten equation:
where t is time, Vmax is the maximum velocity of the enzyme system, and Km is the Michaelis constant of the enzyme system (plasma concentration at which half of the maximum velocity of the enzyme system is attained). Mean steady-state plasma drug concentration (Css) can be expressed as
where R is dosing rate (102). When C is similar to or greater than Km, dC/dt varies in a nonlinear fashion with C; when R is equal to or greater than 0.1 Vmax, Css will vary in a nonlinear fashion with R. These observations are the basis of nonlinear pharmacokinetics.
The mean apparent value for phenytoin Km in adults is 6.2 µg/mL, with a range of 1.5 to 30.7 µg/mL based on 55 reported determinations; the mean apparent value for phenytoin Vmax in adults is 0.45 µg/mL/hr, with a range of 0.14 to 1.36 µg/mL/hr based on 54 reported determinations (6,19,48,55,59,74,113,126). These values appear to be determined principally by CYP2C9 and CYP2C19 Km and Vmax values. The other pathways shown in Figure 58.2 potentially could have modifying effects on the apparent values of Km and Vmax for phenytoin in humans (104). However, attempts to demonstrate an effect of these other pathways on phenytoin pharmacokinetic parameters in humans have been negative (23).
The apparent Km values computed for humans are based on total (protein-bound and non-protein-bound) phenytoin plasma concentration. Because only non-protein-bound phenytoin can be acted on by the metabolizing enzyme system and the non-protein-bound fraction for phenytoin is approximately 10% in humans (162), the Km of the enzyme responsible for parahydroxylation of phenytoin actually should be approximately 0.6 µg/mL. This prediction has been verified in rat liver microsomes (149).
Eadie et al. (48) performed the largest comprehensive comparison of phenytoin Km and Vmax values in children and adults. The Km values from 21 adults (mean = 5.8 µg/mL) and 15 children (mean = 5.3 µg/mL) were not significantly different. The Vmax values from 21 adults (mean = 0.48 µg/mL/hr, assuming phenytoin volume of distribution = 0.7 L/kg) were significantly (p < .025) less than the Vmax values from 15 children (mean = 0.74 µg/mL/hr, assuming phenytoin volume of distribution = 0.7 L/kg). These observations predict that the clearance (Vmax/[Km + C]) of phenytoin should be greater in children than in adults. This prediction is confirmed by the observations that the elimination half-life of phenytoin is shorter in children than in adults and that the average dosing rate of phenytoin (in milligrams per kilogram per day) required to achieve a given plasma concentration is greater in children than in adults (46,98).
The Vmax for phenytoin increases significantly during pregnancy (44). This explains part or all of the observed decreases in phenytoin steady-state plasma concentration during pregnancy (equation ).
Phenytoin exhibits nonlinear pharmacokinetic properties in most patients because the usual therapeutic plasma concentration values (10 to 20 µg/mL; Chapter 60) exceed the usual Km (6.2 µg/mL), and the usual dosing rate (0.15 to 0.45 µg/mL/hr) is greater than 0.1 times the usual value of Vmax (0.45 µg/mL/hr). Note that the metabolism of phenytoin is linear at low plasma concentrations and nonlinear at therapeutic and higher plasma concentrations (151). The consequences of nonlinear pharmacokinetics are discussed later.
Steady-State Plasma Concentration Varies in a Nonlinear Fashion with Dosing Rate
For drugs with nonlinear pharmacokinetics, steady-state plasma concentration increases faster than dosing rate when dosing rate is increased, and plasma concentration decreases faster than dosing rate when dosing rate is decreased (equation 2) (27,44,131) (Figure 58.5). Thus, the steady-state plasma concentration of a drug with nonlinear pharmacokinetic properties at one dosing rate does not directly predict the steady-state plasma concentration of the drug at another dosing rate.
FIGURE 58.5. Relationship between serum phenytoin concentration and daily dose in five patients. Each point represents the mean (± standard deviation) of three to eight measurements of serum phenytoin concentration at steady state. The curves were fitted by computer using the Michaelis-Menten equation. (From Richens A, Dunlop A. Serum phenytoin levels in the management of epilepsy. Lancet 1975;2:247-248, with permission.)
If the clinician attempts to increase or decrease the phenytoin steady-state plasma concentration by simple linear extrapolation from a known plasma concentration versus dosing rate point, the result often is an unexpectedly high or low plasma concentration when the new steady-state value is attained (Figure 58.5). Numerous mathematical and tabular methods have been published that claim to be able to predict the phenytoin dosing rate necessary to produce a given steady-state plasma concentration from a single steady-state plasma concentration versus dose point, and these methods have been critically reviewed elsewhere (7,19,27,37,78,118,126,136,137). A useful rule of thumb in titrating phenytoin dosage upward in adults is to increase dosing rate in increments of 100 mg/day at monthly intervals (see later) until a steady-state phenytoin plasma concentration of 5 to 10 µg/mL (a value approximately equal to Km) is attained; later increases should not exceed 50 mg/day at monthly intervals.
Nonlinear pharmacokinetics complicate the issue of generic equivalence of phenytoin products. The weighted mean value for absolute bioavailability of brand-name (Dilantin Kapseals, 100 mg) phenytoin was 86% in three studies (18). Less-than-complete absorption of brand-name phenytoin is, at least in part, a consequence of the use of a sustained-release preparation (Chapter 13). Different generic preparations of phenytoin thus have the potential to differ in absolute bioavailability from the brand-name preparation by +14% and -14% or more. Because of phenytoin's nonlinear pharmacokinetics, a 14% difference in bioavailability would result in a >14% increase or decrease in steady-state plasma concentration. A national epidemic of phenytoin intoxication occurred in Australia when a more bioavailable formulation was substituted for an older formulation (125,150). Ludden et al. (104) and Browne et al. (26) have reviewed the effect of nonlinear pharmacokinetics on bioavailability studies in more detail.
Clearance Varies Inversely, and Elimination Half-Life Directly, with Plasma Concentration
Drug clearance is equal to Vmax/(Km + C). Drug elimination half-life is equal to 0.693 × volume of distribution/clearance. Thus, phenytoin clearance varies inversely with plasma concentration, and phenytoin elimination half-life varies directly with plasma concentration (18,20,27) (Figure 58.6). Browne et al. (20,27) described and validated a method for calculating phenytoin elimination half-life at any given phenytoin plasma concentration if the patient's Km and Vmax values for phenytoin are known. The results were as follows for a group of six adult men on phenytoin monotherapy (plasma concentration - mean calculated elimination half-life): 1 µg/mL—12.8 hours; 10 µg/mL—25.8 hours; 20 µg/mL—40.2 hours; 40 µg/mL—69.1 hours. Note that the often-quoted elimination half-life of 24 hours for phenytoin applies principally to plasma concentration values in the low therapeutic range (10 µg/mL) and that the elimination half-life often is longer at higher plasma concentrations (Figure 58.6). The range of elimination half-life values at phenytoin plasma concentration 40 µg/mL was 37.1 to 96.8 hours. Because of phenytoin's long and variable elimination half-life values at toxic plasma concentration values, the time required for phenytoin plasma concentration to fall from a toxic value to a therapeutic value cannot be predicted in a given individual. In such circumstances, the clinician must withhold phenytoin and obtain daily plasma concentration determination values until the plasma concentration has fallen back into the therapeutic range.
FIGURE 58.6. Phenytoin clearance and elimination half-life values determined by stable isotope tracer techniques at 30 different serum concentration values in 18 adult men on phenytoin monotherapy. (Based in part on data from references 18, 21, and 23, with permission.)
Time to Reach Steady-State Plasma Concentration Varies Nonlinearly with Dosing Rate and Linearly with Plasma Concentration
As phenytoin plasma concentration rises, phenytoin clearance decreases (see earlier). This results in a further rise in phenytoin plasma concentration and a further decrease in phenytoin clearance. This self-propagating cycle can require a long period to go to completion. The time (t) required to attain a given plasma concentration can be computed by the equation
FIGURE 58.7. Plot of minimum serum concentration after the Nth dose, Cmin, versus dose number n. Symbols and dosing rates (g/day) are: ♦, 0.50; ●, 0.40; ◊, 0.30; ▪, 0.25; and▲, 0.20. (From Wagner JG. Time to reach steady state and prediction of steady state concentration for drugs obeying Michaelis-Menten elimination kinetics. J Pharmacokinet Biopharm 1978;6:209-225, with permission.)
where Vd is the volume of distribution, Cs,t is the plasma concentration at time t, and Cs,0 is the plasma concentration at time 0 (89). Assuming average values for Km and Vmax, it is possible to compute an accumulation 1/half-life (t½A) for phenytoin as follows:
where t½,A has units of days, and Css has units of µg/mL (100). Equations  and  predict, and empirical data confirm, the following: (a) the time to reach steady-state plasma concentration varies nonlinearly with dosing rate; (b)
the time to reach steady-state plasma concentration varies linearly with plasma concentration; and (c) the time required to attain new steady-state plasma concentration values after starting phenytoin therapy or increasing or decreasing phenytoin dosing rate may be as long as 28 days (19,27, 101,144,155) (Figures 58.7 and 58.8). Therefore, a plasma phenytoin concentration value measured less than 28 days after a change in phenytoin dosing rate may not be an accurate indication of the ultimate new steady-state plasma concentration that will result from the change in dosing rate.
FIGURE 58.8. Changes in serum phenytoin concentration after reduction in phenytoin dosing rate from 250 to 200 mg/day. Serum phenytoin concentration stabilized after day 23. (From Theodore WH, Qu P, Tsay JY, et al. Phenytoin: the pseudosteady-state phenomenon. Clin Pharmacol Ther 1984;25:822-825, with permission.)
TABLE 58.1. PHENYTOIN BIOTRANSFORMATION AND PHARMACOKINETIC VALUES FOR SIX PATIENTS AT THREE TIMES DURING MONOTHERAPY DETERMINED WITH 150-MG TRACER DOSES OF STABLE ISOTOPE-LABELED PHENYTOINa
Typical Changes in Phenytoin Pharmacokinetics during Chronic Administration
When a patient starts phenytoin monotherapy, his or her plasma phenytoin concentration progressively rises. This has the following effects on phenytoin biotransformation and pharmacokinetic values (18): Rate of formation of p-HPPH decreases, clearance decreases, and elimination half-life increases (Figure 58.6). Table 58.1 illustrates typical values for these changes in a group of six patients started on phenytoin monotherapy.
This work was supported in part by the Department of Veterans Affairs.
(µg/m) × 3.96 = µmol/L ÷ 3.96 = µg/mL
(µmol/L) ÷ 3.96 = µg/mL