Carbamazepine but Not Valproate Induces CYP2A6 Activity in Smokers with Mental Illness

This article is featured in Highlights of This Issue, p. 2419

More than 60 million Americans smoke cigarettes according to the 2007 National Survey on Drug Use and Health (1). Nicotine is essential to maintaining tobacco use and has importance as a pharmaceutical treatment for smoking cessation. When inhaled, nicotine is rapidly delivered to the systemic circulation and once in the bloodstream is distributed extensively to body tissues (2). The plasma half-life of nicotine averages about two hours with 70% to 80% metabolized in the liver by the cytochrome P450 enzyme CYP2A6. Nicotine is metabolized to cotinine, and cotinine is further metabolized to 3′-hydroxycotinine (3HC) primarily by the liver enzyme CYP2A6 (3). To a much lesser degree, nicotine is metabolized by other oxidative (CYP2B6 and CYP2D6) and nonoxidative pathways (2).

Because CYP2A6 enzyme activity is the major determinant of the rate of nicotine metabolism and smokers tend to regulate levels of nicotine in their bodies, it follows that the rate of CYP2A6 activity would influence smoking behavior. Genetic variants in CYP2A6 associated with slower than normal metabolism have been associated with reduced risk of becoming a smoker, smoking fewer cigarettes per day, and greater likelihood of quitting (4-6). The ratio of nicotine metabolites of 3′-hydroxycotinine to cotinine (3HC/cotinine, the nicotine metabolite ratio) is a marker of CYP2A6 metabolic activity and a noninvasive measure of the rate of nicotine metabolism that also has implications for smoking behavior (7). Studying phenotypes (i.e., metabolic ratios) is likely to be more informative than studying genotypes for nicotine metabolism because of the relatively low frequency of abnormal alleles in the Caucasian population and the large variability in CYP2A6 enzyme activity even among people with normal (wild-type) CYP2A6 genes (8).

Smoking is more prevalent in people with serious forms of mental illnesses, such as schizophrenia and bipolar disorder, than in the general population (9-11). The effect of medications used in the management of serious mental illness on nicotine metabolism and smoking behavior has not been well studied. Smokers with schizophrenia have no differences in 3HC/cotinine ratios compared with control smokers, indicating that schizophrenia per se does not affect the ability to metabolize nicotine at CYP2A6 (12). Although these smokers are usually taking antipsychotic medications, it is unlikely that drug interactions from antipsychotic medications impacted on these results given the lack of clinically significant interactions at CYP2A6 (13-15). Oral contraceptive medications are one of few medications known to increase nicotine metabolism at CYP2A6 (16). Antiepileptic drugs (AED) are of particular concern with respect to drug interactions because of their potential to induce many hepatic cytochrome P450 enzymes. AEDs are commonly used for conditions other than epilepsy, including bipolar disorder, neuropathic pain, and migraine prophylaxis (17). Adjunctive treatment with lithium or an AED is a commonly used strategy for managing aggression or mood in schizophrenia, and current estimates are that ≥50% of patients with schizophrenia receive treatment with these mood stabilizers (18, 19).

Carbamazepine, used in the treatment of bipolar disorder, has a very high potential for drug interactions. Carbamazepine is a potent inducer of CYP3A4 and other oxidative enzyme systems in the liver, and it increases glucuronyltransferase activity. This results in the acceleration of the metabolism of many drugs, including warfarin, tricyclic antidepressants, antipsychotics, oral contraceptives, glucocorticoids, cyclosporin, theophylline, chemotherapeutic agents, and cardiovascular drugs (20). Carbamazepine also induces the metabolism of other anticonvulsants, particularly valproic acid (21).

Oxcarbazepine, approved for use in the United States in 2000, is a ketoderivative of carbamazepine and has largely inactive metabolites. It was developed specifically to reduce the side effects and medication interactions associated with carbamazepine (22). Oxcarbazepine is metabolized through reduction and conjugation (glucuronidation), and unlike many other antiepileptic drugs its metabolism is not induced or inhibited via the cytochrome P450 system (23). Few clinically significant drug interactions with oxcarbazepine have been reported, although oxcarbazepine can induce CYP3A4 and CYP3A5, leading to reduction in levels and effectiveness of oral contraceptives (24, 25). Oxcarbazepine is generally used to treat the same conditions as is carbamazepine and is often preferred because it is better tolerated by patients and causes fewer rashes (26).

Valproic acid was approved for use as anticonvulsant in the United States in 1978 and treats a variety of seizure disorders. Since the 1990s use of valproate has increased rapidly for psychiatric indications and was approved by the Food and Drug Administration for treatment of mania in 1996. Various forms of the chemical (divalproex sodium or valproic acid) dissociate to the active valproate ion in the gastrointestinal tract, and divalproex is the most commonly prescribed formulation (27). Valproic acid is highly protein bound and metabolized primarily via mitochondrial oxidation with <20% of the dose eliminated by other oxidative mechanisms (28).

The aim of this study was to investigate the association of three commonly used AEDs, namely, carbamazepine, oxcarbazepine, and valproic acid, with nicotine and nicotine metabolite levels in smokers with schizophrenia and bipolar disorder.

This study consisted of an analysis of serum nicotine and nicotine metabolite levels in smokers with mental illness taking various psychotropic medications.

The sample included 65 smokers with schizophrenia who participated in quit-smoking studies or studies of nicotine intake that required them to have blood sampling for nicotine and are described elsewhere in greater detail (12). The sample also included consecutively enrolled smokers from an ongoing study of nicotine intake and smoking topography in smokers with schizophrenia or bipolar disorder (36 schizophrenia; 48 bipolar) that is currently under way.

Subjects had to be currently enrolled in mental health treatment and stable on psychiatric medications to participate. All subjects had their diagnosis of schizophrenia or bipolar disorder confirmed with the Structured Clinical Interview for DSM (29) and smoked more than 8 cigarettes per day (cpd). Individuals with a diagnosis of schizoaffective disorder were excluded. Seriously cognitively impaired patients were also excluded, and subjects were required to score 24 or higher on the Folstein Mini Mental Status Examination (30) to be eligible.

Any subject using nicotine replacement therapy (NRT) (nicotine gum, patch, inhaler, nasal spray, lozenge), clonidine, bupropion, nortriptyline, or varenicline was excluded. We excluded subjects using tobacco products other than cigarettes as well as anyone who was pregnant, because this is associated with accelerated metabolism of nicotine and cotinine (31). All subjects were required to bring their own cigarettes in for testing procedures. Analyses were carried out using data from these 149 smokers (101 schizophrenia, 48 bipolar). The study was approved by the UMDNJ-Robert Wood Johnson Medical School Institutional Review Board (IRB).

After signing the consent forms, subjects completed an assessment battery including a smoking history and assessment of their current tobacco use (measured as cpd), demographic and medication questionnaire, and the Fagerstrom Test for Nicotine Dependence (FTND; ref. 32). Baseline expired carbon monoxide (CO) was measured using an EC-50 Smokerlyzer (Bedfont Scientific). All medications were recorded and antipsychotic medication dose was converted to chlorpromazine equivalents (CPZ; ref. 33). Use and dose (mg) of AED drugs was recorded. All subjects participated in an afternoon blood draw for nicotine, cotinine, and 3HC on a usual smoking day, taken approximately two minutes after smoking one of their own cigarettes. For subjects enrolled in the quit-smoking study this was done prior to setting the quit date. Serum was frozen at −20°C for later analysis. Specimens were sent to the Clinical Pharmacology Laboratory at the University of California, San Francisco for analysis of nicotine, cotinine, and 3-hydroxycotinine, which were quantified by liquid chromatography-mass spectrometry (7).

Independent sample t-tests or Mann-Whitney test (for continuous variables) and χ² tests (for categorical variables) were used to compare baseline differences between smokers with schizophrenia and smokers with bipolar disorder on their sociodemographic, smoking history, and clinical variables. For all other analyses, groups with schizophrenia and bipolar disorder were combined to evaluate the effects of medications. Subjects taking all forms of valproic acid were consolidated into one group. ANOVA was used to compare nicotine, exhaled CO, and cpd in smokers taking and not taking AED. Ratios of 3HC/cotinine were calculated for all subjects. Because cotinine values and 3HC/cotinine ratios are not normally distributed, we used the Kruskal-Wallis test for comparisons of these means. Analyses were adjusted for gender and race because these have been shown to affect nicotine metabolism (2).

We conducted backward stepwise linear regression models to identify predictors of the log transformed 3HC/cotinine ratios. The variables entered into the model included age, gender, race, education, body mass index (BMI), cpd, expired CO, time of blood draw, FTND score, and smoking mentholated cigarettes. Antipsychotic medication type (typical versus atypical), antipsychotic dose (chlorpromazine equivalents), taking carbamazepine (versus not taking), taking valproic acid (versus not taking), taking oxcarbazepine (versus not taking), and diagnosis group (schizophrenia versus bipolar) were also included in the regression model. The criterion for eliminating variables from the model was set at P ≥ 0.10. We also examined the correlation between the 3HC/cotinine ratio and dose in mg of AED (carbamazepine-oxcarbazepine, valproic acid) using Spearman's correlation coefficient. All analyses were carried out using SPSS 16.0 and SAS 9.1.

Smokers with schizophrenia (n = 101) were compared with smokers with bipolar disorder (n = 48) on smoking and demographic variables, including age, ethnicity, gender, cpd smoked, expired CO at baseline, years smoked, total FTND score, and age of first smoking (Table 1). Smokers with schizophrenia were older (44 versus 37 years, P < 0.001), smoked more cpd (24.7 versus 20.1, P < 0.05), and had a higher FTND score (6.8 versus 5.8, P < 0.01) compared with smokers with bipolar disorder.

Although there were no significant differences in serum nicotine concentrations between the schizophrenia and bipolar disorder groups, respectively [mean nicotine 29.2 versus 26.8 ng/mL; P not significant (NS)], there were differences for serum cotinine (mean cotinine 377.8 versus 322.4 ng/mL; P = 0.046) and 3HC/cotinine ratios (0.49 versus 0.68, P < 0.01; Table 1). Seventy-one percent of smokers with bipolar disorder were taking antipsychotic medications compared with all of the schizophrenia smokers, and most were taking the newer, second-generation atypical antipsychotics (90% schizophrenia versus 94% bipolar, NS). Mean chlorpromazine equivalents were higher for schizophrenia compared with bipolar disorder (593.5 versus 281.9 mg, P < 0.001), indicating higher doses of antipsychotic medication in the schizophrenia group. Forty-four percent of bipolar disorder smokers were taking AEDs compared with 31% of schizophrenia smokers (NS) and the frequency of use of carbamazepine, oxcarbazepine, and valproic acid was not different between groups (Table 1). Two subjects (one bipolar disorder, one schizophrenia) were taking both valproic acid and carbamazepine simultaneously.

For these analyses, smokers with schizophrenia and bipolar disorder were combined to evaluate the effects of medications, and analyses were adjusted for diagnosis group. For most analyses except where noted, subjects taking carbamazepine and oxcarbazepine were consolidated into one group (carbamazepine-oxcarbazepine). A total of 14 smokers in the sample were taking carbamazepine or oxcarbazepine (7 schizophrenia, 7 bipolar disorder). The clinical characteristics of these smokers are detailed in Table 2. There were no significant differences in serum nicotine (mean nicotine 26.4 versus 28.6 ng/mL; P = NS) in individuals taking carbamazepine-oxcarbazepine versus those not taking (Table 3). Differences in cotinine levels in individuals taking carbamazepine-oxcarbazepine, however, approached significance (mean cotinine 268.2 versus 369.5 ng/mL; P = 0.067), and the cotinine/cpd ratio was significantly lower in individuals taking carbamazepine-oxcarbazepine (11.1 versus 16.1; P < 0.05). The 3HC/cotinine ratios were significantly higher in individuals taking carbamazepine-oxcarbazepine versus those not taking (mean 0.993 versus 0.503; P < 0.001). The 3HC/cotinine ratios were also significantly higher in individuals taking carbamazepine-oxcarbazepine in the subgroups of smokers with schizophrenia but not bipolar disorder (Fig. 1). These results were also not changed when we repeated these analyses controlling for gender and race. We ran a 2 × 2 ANOVA to test for the possible interaction effect between diagnostic group and taking carbamazepine-oxcarbazepine. There was a main effect for taking carbamazepine, but no effect for the group and no interaction effect.

We also repeated these analyses to examine the same effects in subgroups of smokers taking carbamazepine (n = 8) or taking oxcarbazepine (n = 6). Carbamazepine use was still associated with significant differences in the 3HC/cotinine and cotinine/cpd ratios, but oxcarbazepine was not.

Forty smokers in the sample were taking valproic acid (24 schizophrenia, 14 bipolar disorder). There were no significant differences in nicotine (mean nicotine 28.5 versus 28.3 ng/mL; P = NS) or cotinine levels (mean cotinine 389.8 versus 349.1 ng/mL; P = NS) in individuals taking valproic acid versus those not taking it. This was true for the total group as well as within the diagnostic subgroups (Fig. 2). There were also no differences in 3HC/cotinine ratios (mean 0.62 versus 0.51; F-stat = 3.368, P = 0.069), although values of the 3HC metabolite were higher in the valproic acid group (mean 160.9 versus 128.7; P = 0.026). All of the valproic acid analyses were adjusted for group and concurrent use of carbamazepine. Results were not changed when we repeated these analyses removing the two individuals using both valproic acid and carbamazepine and not adjusting for carbamazepine (Table 3). Results were also not changed when we repeated these analyses adjusting for gender and race.

Backward stepwise linear regression analyses were conducted to identify the predictors of the log transformed 3HC/cotinine ratio. Taking carbamazepine (B = 0.596; SE B = 0.228, P = 0.011), and number of cigarettes smoked per day (B = 0.025; SE B = 0.009; P < 0.01), were found to be significant determinants of log 3HC/cotinine (Table 4).

Figure 3 shows the correlation between 3HC/cotinine ratio and total oral daily dose of carbamazepine or valproic acid (in mg). There was a moderate correlation between 3HC/cotinine ratio and carbamazepine dose that approached significance (R² linear = 0.485, P = 0.08) but no correlation for 3HC/cotinine ratio and valproic acid dose [R² linear = 0.048, NS (not significant)].

The study found significantly higher 3HC/cotinine and cotinine/cpd ratios in smokers with schizophrenia and bipolar disorder taking carbamazepine or oxcarbazepine. The likely mechanism is through induction of CYP2A6 by carbamazepine and oxcarbazepine. It is unlikely that this effect was due to mental illness alone because we had previously found nicotine metabolite ratios that are no different from control smokers without mental illness (12). In addition, values for 3HC/cotinine ratios in those not taking carbamazepine in this study were similar to those previously reported in other samples of non–mentally ill smokers (34, 35). Although these smokers were often taking several psychiatric medications simultaneously it is unlikely that the effect was due to these other medications because none are known to have significant effects on CYP2A6 activity and antipsychotic medications were not predictors of nicotine metabolite ratio in the regression analysis. This is the first report of either carbamazepine or oxcarbazepine causing a clinically significant interaction with nicotine metabolism. Although we lacked power in this small sample to detect a significant effect on metabolism from oxcarbazepine, 3HC/cotinine ratios were still elevated (mean 0.81), suggesting induction of CYP2A6 similar to carbamazepine and warranting further study.

Increased nicotine metabolite ratios have several implications for smokers taking carbamazepine or oxcarbazepine. Oral contraceptives are known to significantly accelerate nicotine metabolism. A still unanswered question is whether medications such as sex hormones or AEDs that increase the rate of nicotine metabolism influence either how much a person smokes or how much smoke a person takes in from a cigarette (or both). Although our study did not show evidence of higher CO values in subjects taking carbamazepine-oxcarbazepine, more detailed measurements of CO or topography would support the hypothesis that rapid metabolizers are smoking cigarettes more intensively.

CYP2A6 gene variants associated with higher rates of nicotine metabolism and higher clearance of nicotine (4, 36) are associated with increased smoking behaviors. Fast metabolizers are more prone to develop nicotine dependence and also smoke more compared with slow metabolizers (37). Fast metabolizers may also have greater difficulty quitting smoking and increased severity of abstinence symptoms. CYP2A6 is also the enzyme responsible for the metabolic activation of procarcinogenic compounds (like nitrosamines) that cause lung cancer (38). Thus, persons with higher levels of CYP2A6 activity (i.e., fast metabolizers) may also be at higher risk of tobacco-caused cancer.

Greater CYP2A6 activity, as indicated by a higher nicotine metabolite ratio and higher cotinine/cpd ratio, would be expected to result in a lower serum nicotine level for a given nicotine dose. If the serum nicotine level is the same in faster and slower metabolizers, the faster metabolizers must be taking in more nicotine (and more tobacco smoke) than slower metabolizers. Because smoking-caused disease is related to intake of tobacco smoke, a person with carbamazepine-induced rapid metabolism of nicotine who is compensating by inhaling more smoke is likely to be at greater risk of disease.

A limitation of this study is that we examined only one pathway for nicotine and cotinine metabolism. Although the majority of the clearance of nicotine proceeds via oxidative metabolism, primarily via CYP2A6, to cotinine, there are other metabolic pathways that must be considered that might not be reflected by the 3HC/cotinine ratio. For example, some nicotine may be metabolized by CYP2B6 (2). Both nicotine and cotinine are metabolized to glucuronides, mediated by UDP-glucuronosyltransferase (UGT) enzymes.

Further research is needed to examine the question of whether carbamazepine or oxcarbazepine influences either smoking behavior or intake of tobacco smoke from cigarettes in groups of smokers who may or may not have mental illness. As this is also the first report of nicotine metabolites in bipolar disorder it is possible that differences seen in this group (compared with schizophrenia) are also due to medication effects because diagnostic group was not predictive of faster metabolism although this also warrants further evaluation.

No potential conflicts of interest were disclosed.

Grant Support: Grants from the National Institute of Mental Health (MH076672 to J.M. Williams) and the National Institute on Drug Abuse (K23-DA140090 to J.M. Williams and DA12393 to N.L. Benowitz).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.