4β‐hydroxycholesterol correlates with dose but not steady‐state concentration of carbamazepine: indication of intestinal CYP3A in biomarker formation?

Abstract

Aim

4β‐hydroxycholesterol (4βOHC) is an endogenous CYP3A(4) biomarker, which is elevated by use of the CYP3A4 inducer carbamazepine. Our aim was to compare to what extent serum concentration of 4βOHC correlates with dose (presystemic exposure) and steady‐state concentration (systemic exposure) of carbamazepine.

Methods

The study was based on a therapeutic drug monitoring material, including information about daily doses and steady‐state concentrations (C _ss) of carbamazepine. 4βOHC concentrations were determined in residual serum samples of 55 randomly selected carbamazepine‐treated patients and 54 levetiracetam‐treated patients (negative controls) by UPLC‐APCI‐MS/MS after liquid–liquid extraction. Correlation analyses between 4βOHC concentration and daily dose and C _ss of carbamazepine, respectively, were performed by Spearman's tests. In addition, 4βOHC concentrations in females vs. males were compared in induced and non‐induced patients.

Results

Median 4βOHC concentration was ~10‐fold higher in carbamazepine‐ vs. levetiracetam‐treated patients (650 vs. 54 nmol l⁻¹, P < 0.0001). There was a significant, positive correlation between carbamazepine dose and 4βOHC concentration (Spearman r = 0.53, 95% confidence interval [CI] 0.27, 0.72, P < 0.001). No significant correlation between carbamazepine C _ss and 4βOHC concentration was found (Spearman r = 0.14; 95% CI −0.14, 0.40, P = 0.3). Enzyme‐induced females had significantly higher 4βOHC concentrations than males (P < 0.001), while no significant gender difference was found in non‐induced patients (P = 0.52).

Conclusion

Serum concentrations of 4βOHC correlate with presystemic, but not systemic exposure of the CYP3A4 inducer carbamazepine. This suggests a stronger inductive effect of carbamazepine on presystemic than systemic CYP3A4 phenotype and might indicate a role of the intestine in 4βOHC formation. Moreover, CYP3A4 inducibility seems to be higher in females than males.

What is already Known About This Subject

• 4β‐hydroxycholesterol (4βOHC) is a promising endogenous biomarker for CYP3A(4) activity.

• Since the liver is the primary site of cholesterol metabolism, it has been proposed that 4βOHC primarily reflects hepatic CYP3A activity and to a limited extent intestinal CYP3A phenotype.

What This Study Adds

• In CYP3A(4)‐induced patients, 4βOHC concentration significantly correlates with daily oral dose (presystemic exposure), but not steady‐state concentration (systemic exposure) of carbamazepine.

• This provides evidence that presystemic CYP3A4 phenotype affects 4βOHC concentration and might indicate involvement of the intestine in biomarker formation.

• Induction of CYP3A(4) by carbamazepine is apparently stronger in females than males.

Introduction

Cytochrome P450 3A4 (CYP3A4) is the most important enzyme in drug metabolism, and it has been estimated that more than 50% of all clinically used agents are CYP3A4 substrates 1, 2. The interindividual difference in CYP3A4 phenotype is extensive, but in contrast to, for example, CYP2D6, this variability is not tightly associated with genetic differences 3. Thus, genotyping is generally not an appropriate option for determination of an individual patient's metabolizing phenotype. Instead, endogenous CYP3A4 substrates might serve as phenotype biomarkers for dose individualization and treatment monitoring of drugs metabolized by this enzyme.

4β‐hydroxycholesterol (4βOHC) is an oxidation product of cholesterol, which has attracted great interest as a potential endogenous biomarker for CYP3A phenotype. Several studies by Diczfalusy and co‐workers indicate that the enzymatic oxidation of cholesterol to 4βOHC mainly involves CYP3A isoforms, i.e. CYP3A4 and CYP3A5 4, 5. In most ethnic groups, the majority are non‐expressers of CYP3A5 4, 6, so generally CYP3A4 is the most important CYP3A isoform in the overall enzyme‐catalyzed formation of 4βOHC, at least in Caucasians. Although CYP3A5 also catalyzes the formation of 4βOHC 4, 5, it is described as a CYP3A4 biomarker in the further text.

In healthy, drug naive humans, the concentration of 4βOHC has been reported to range from 10 to 60 ng ml⁻¹ (~25–150 nmol l⁻¹) 7. Studies have shown that patients treated with CYP3A4 inducers, such as carbamazepine 8, 9 or rifampicin 10, 11, 12, 13, have up to 10‐fold increased concentrations of 4βOHC. Co‐administration of CYP3A4 inhibitors has, on the other hand, been reported to decrease concentrations of 4βOHC 10, 14, 15, but the responsiveness towards inhibitors seems to be less than for CYP3A4 inducers. However, the long elimination half‐life of 4βOHC, i.e. around 17 days 13, implies that several weeks of co‐administration would be required in order to detect the effect of enzyme inhibitors.

The ability to disclose CYP3A4‐related drug interactions shows that 4βOHC is a promising biomarker, but since the liver is the primary site of cholesterol metabolism, it has been proposed that 4βOHC concentration primarily reflects hepatic CYP3A4 phenotype 16, 17. However, intestinal cells also have access to cholesterol and CYP3A4 expression per tissue weight is higher in the small intestine than in the liver 18, 19. Oral bioavailability of the CYP3A4 probe drug midazolam is reported to be about 50% in healthy subjects 20. In patients where midazolam has been administered orally in the anhepatic phase during liver transplantation, the fraction of the absorbed midazolam dose that was metabolized on transit through the intestinal mucosa was approximately 0.4 21. Thus, there is little doubt that intestinal enzyme expression is a major determinant of the overall presystemic CYP3A4 phenotype in humans 21, 22.

In order to highlight the issue regarding a potential role of intestinal CYP3A4 in the formation of 4βOHC, the aim of the present study was to compare to what extent serum concentration of 4βOHC is correlated with dose (i.e. presystemic exposure) and steady‐state concentration (i.e. systemic exposure) of the CYP3A4 inducer carbamazepine.

Methods

Collection of patient samples and drug information

The study was approved by the Regional Committee for Medical and Health Research Ethics (case number 2014/1191) and the Hospital Investigational Review Board. Since the study was based on existing serum samples and retrospective data (anonymized) from a routine TDM service, ethical approval was given without the requirement of patient consent.

Serum samples from patients treated with carbamazepine originally analyzed for therapeutic drug monitoring (TDM) purposes were collected at the Center for Psychopharmacology, Diakonhjemmet Hospital, Oslo, Norway, over a random 6 week period in 2014/15. In the same period, serum samples from patients treated with levetiracetam, which does not induce or inhibit CYP3A4 metabolism, were included as negative (non‐induced) controls. All serum samples were re‐analyzed for determination of 4βOHC concentration. The samples were stored at −20 °C until analysis of 4βOHC.

For the carbamazepine‐treated patients, information about daily dose (mg day⁻¹) and comedication was obtained from requisition forms, while measured steady‐state concentration (μmol l⁻¹) was obtained from TDM files. Verification of steady‐state conditions of carbamazepine was based on assessment of information about treatment history available from requisition forms.

Anti‐epileptics assay

Serum concentrations of carbamazepine were measured by a routine LC‐MS/MS method for simultaneous determination of the most commonly used anti‐epileptic drugs, i.e. carbamazepine, levetiracetam, lamotrigine, valproic acid, phenytoin, phenobarbital, pregabalin, gabapentin, topiramat, oxcarbazepine and eslicarbazepine. Intra‐ and interday precision of the assay was <8%, while intra‐ and interday accuracy was <4% for carbamazepine. Calibration curves were prepared in the concentration range 10–50 μmol l⁻¹ and the lower limit of quantification of carbamazepine was 5 μmol l⁻¹.

Briefly, serum samples were prepared by protein precipitation using acetonitrile. The chromatographic separation was performed on an Aquility UPLC BEH Shield RP18 column (1.7 μm, 1 × 100 mm; Waters, Milford, MA, USA) using gradient elution with a mix of ammonium acetate (pH = 4.8) and acetonitrile as mobile phase. The anti‐epileptic drugs were assayed on an Aquility ultra performance liquid chromatograph (UPLC) system with a Micromass Quattro Premier tandem MS detector from Waters. MS detection was achieved using the electrospray ionization (ESI) probe in positive ion mode (collision energy 15 eV, cone voltage 30 V, cone gas flow 90 l h⁻¹ and desolvation gas flow 500 l h⁻¹). Detection of carbamazepine was obtained with multiple reaction monitoring (MRM) at the transition m/z 237– > 194.1.

Determination of anti‐epileptics other than carbamazepine and levetiracetam in the assay was registered in both patient groups. The potential presence of enzyme‐inducing anti‐epileptics in samples from the levetiracetam patients was defined as an exclusion criterion from the control group. If multiple enzyme inducers were present in serum samples of the carbamazepine patients, these samples were included in the enzyme‐induced study group for re‐analysis of 4βOHC concentrations.

4βOHC analysis

The remaining volumes of the collected serum samples originally submitted for TDM analyses were re‐analyzed for determination of 4βOHC concentrations. The assay used for 4βOHC analysis was a modified version of a method previously published by van de Merbel et al. 23. Briefly, 500 μl serum sample was mixed with 50 μl internal standard, i.e. 5 μmol l⁻¹ deuterated 4βOHC (4βOHC‐D7; Toronto Research Chemicals Inc., Toronto, Ontario, Canada) dissolved in methanol and 1 ml of a 1 mol l⁻¹ sodium methoxide solution for hydrolysis of esterified 4βOHC. Then liquid phase extraction was carried out by adding 1 ml water and 4 ml hexane. For complete phase separation, the mix was extracted in ambient temperature for 2 min followed by centrifugation at 2500 rpm (1260 g) for 5 min at 20 °C, and finally 20 min of storage at −80 °C. By this time the aqueous layer was frozen and the upper organic layer was transferred to new tubes. The organic phase was evaporated to dryness at 37 °C by the use of nitrogen gas, reconstituted in 500 μl methanol, and transferred to UPLC vials.

The analysis of 4βOHC in 10 μl aliquots of the reconstituted samples was performed on the same type of UPLC‐MS/MS system as described for the anti‐epileptic assays. The UPLC column was also the same, while chromatographic separation was obtained at a temperature of 40 °C using gradient elution with a mix of water and methanol (85–95%) at a flow rate of 0.150 ml min⁻¹. The total run time per sample was 10 min and the retention time of 4βOHC was approximately 3 min. MS detection was achieved using an atmospheric pressure chemical ionization (APCI) probe in positive ion mode (collision energy 14 eV, cone voltage 30 V) with MRM at the transitions m/z 385.25– > 367.45 (4βOHC) and m/z 392.30– > 374.50 (4βOHC‐D7; IS). As confirmed by injection of reference standards in methanol, 4βOHC was separated from 4α‐hydroxycholesterol (4αOHC, Supplementary Figure 1).

The method was validated for accuracy and precision over six separate days. Calibration curves of 4βOHC prepared in methanol were linear throughout the concentration range 25–1600 nmol l⁻¹, as indicated by an average correlation coefficient of 0.998 obtained during validation. A large volume of pooled patient serum from the TDM service, containing a wide range of drugs, was used to prepare standard samples for the validation procedure. Precision and accuracy data of the assay were calculated from six samples spiked with 25 nmol l⁻¹ and 1600 nmol l⁻¹ of 4βOHC, respectively. The intra‐ and interday precision of the assay was <8% at 25 nmol l⁻¹ and <4% at 1600 nmol l⁻¹, while the intra‐ and interday accuracy was <15% at 25 nmol l⁻¹ and <2% at 1600 nmol l⁻¹. The signal : noise ratio was >20 at the lowest validated concentration (25 nmol l⁻¹). Extraction recovery of the internal standard (4βOHC‐D7) ranged from 70 to 90%.

Quantification of 4βOHC was based on the ratio between the top height of 4βOHC and the top height of 4βOHC‐D7 (IS). Due to the natural presence of 4βOHC in human serum, calibration curves were prepared from standard samples directly dissolved in methanol. To evaluate potential suppressive effects of matrix (serum) components on the MS detection response, a direct infusion method was used. No matrix suppression was observed in signal response of 4βOHC in the testing.

When serum concentrations of 4βOHC were determined in the present study, the prepared patient samples were analyzed twice and mean values applied in the statistical calculations.

Statistics

Median serum concentrations of 4βOHC were compared in patients treated with carbamazepine and levetiracetam by Mann–Whitney analysis. The same test was also applied for the comparison of 4βOHC concentrations in females vs. males in both subgroups. Fisher's exact test was applied for comparisons of gender distribution and comedication of anti‐epileptics between the subgroups. In the CYP3A4‐induced subgroup, correlation analyses between 4βOHC level and dose/steady‐state serum concentration (C _ss) of carbamazepine were performed by Spearman's signed rank tests.

GraphPad Prism version 6 (GraphPad Software, Inc., San Diego, CA, USA) was used for statistical analyses and graphical illustrations. P < 0.05 was considered statistically significant.

Results

A total of 55 patients treated with carbamazepine and 54 patients treated with levetiracetam were included in the study. Patient characteristics of the study subgroups are summarized in Table 1. The observed median age of the carbamazepine‐treated patients was higher than the levetiracetam‐treated patients, i.e. 60 vs. 44.5 years, but the difference was not significant (P = 0.07). The gender distribution was balanced and quite similar in the two groups.

Table 1.

Characteristics of subgroups treated with carbamazepine and levetiracetam

Variable	Carbamazepine n = 55	Levetiracetam n = 54	P value
Age (years), median (range)	60 (18–93)	44.5 (13–88)	0.07
Females, n (%)	29 (53)	25 (46)	0.57
Use of other anti‐epileptics, n (%)	38 (69)	16 (30)	<0.0001
One	25	12	‐
Two	10	4	‐
Three	2	‐	‐
Four	1	‐
Non‐inducers *, n
Valproic acid	13	7	0.22
Lamotrigine	4	10	0.09
Pregabalin	3	‐	‐
Topiramat	3	1	0.62
Eslicarbazepine	‐	1	‐
Oxcarbazepine	‐	1	‐
Inducers, n
Phenobarbital	4	‐	‐
Phenytoin	4	‐	‐
Dose† (mg day−1), median (range)	600 (100–1600)	n.r.	‐
<500 mg, n	11
500–1000 mg, n	26
>1000 mg, n	7
Time since last dose (h), median (range)‡	13 (3–25)	n.r.	‐
C ss (μmol l−1), median (range)	33 (11–51)	n.r.	‐

^* 22 patients in the carbamazepine group were also comedicated with levetiracetam.

^† Information available for 44 of the carbamazepine‐treated patients.

^‡ Information available for 32 of the carbamazepine‐treated patients.

C _ss, steady‐state concentration; 4βOHC, 4β‐hydroxycholesterol.

not relevant.

In the carbamazepine‐treated patients, median time between last dose intake and blood sample withdrawal was 13 h (range 3–25 h), while median steady‐state serum concentration was 33 μmol l⁻¹ (range 11–51 μmol l⁻¹). Information about daily dosing of carbamazepine was written on the requisition forms for 44 of 55 patients, and among these the median daily dose was 600 mg day⁻¹ (range 100–1600 mg day⁻¹). In patients with available dosing information, carbamazepine was administered twice daily in 34 patients, once daily in seven patients and three times daily in three patients.

Comparisons of measured serum concentrations of 4βOHC in females and males in both study groups are illustrated in Figure 1. There was a significantly higher median serum concentration of 4βOHC in females compared with males in carbamazepine‐treated patients (780 vs. 467 nmol l⁻¹, P = 0.0046), while no significant gender difference was found in levetiracetam‐treated patients (median 57.6 [females] vs. 48.2 nmol l⁻¹, P = 0.52). In the carbamazepine‐treated patients, four females and four males were treated with an additional inducer.

Figure 1.

Individual serum concentrations of 4β‐hydroxycholesterol (4βOHC) for females and males in patients treated with carbamazepine (A) and levetiracetam (B). Lines show median values in the subgroups and open circles (○) indicate use of multiple enzyme inducers

A higher percentage of the patients in the carbamazepine group was comedicated with one or more additional anti‐epileptic drugs (AEDs), compared with the levetiracetam group (P < 0.0001). Of the 55 patients treated with carbamazepine, eight subjects (14.5%) were comedicated with an additional CYP3A4 inducer (phenytoin or phenobarbital). Of the total number of comedicated AEDs in the whole study population (n = 54), 7 cases (13%) were identified from the UPLC‐MS/MS analyses only (i.e. comedicated AED not written on the requisition forms).

The distribution of the individual serum 4βOHC concentrations in patient subgroups is shown in Figure 2. In the levetiracetam‐treated patients, four measurements of 4βOHC were below the validated concentration range. Two patients in the carbamazepine group, both treated with an additional inducer, had 4βOHC serum concentrations above the validated concentration range. A more than 10‐fold higher median concentration of 4βOHC was observed in the group of patients treated with carbamazepine compared with the control group treated with levetiracetam, i.e. 650 nmol l⁻¹ (range 151–1945) vs. 54 nmol l⁻¹ (range 15–228), P < 0.0001. An approximately 70% higher median concentration of 4βOHC was observed in the group of patients treated with multiple CYP3A4 inducers (n = 8) compared with patients treated with carbamazepine as the only enzyme inducer (1008 vs. 598 nmol l⁻¹, P = 0.042).

Figure 2.

Individual serum concentrations of 4β‐hydroxycholesterol (4βOHC) in patients treated with levetiracetam (n = 54) and carbamazepine (n = 55), where the latter are divided into patients using carbamazepine as the only enzyme inducer (n = 47; middle) and patients using carbamazepine + phenobarbital or phenytoin (multiple inducers, n = 8; right). Lines indicate median values in the subgroups

A significant correlation (r = 0.53, 95% CI 0.27, 0.72, P < 0.001) was observed between serum concentration of 4βOHC and carbamazepine dose (Figure 3A). In contrast, there was no significant correlation between C _ss of carbamazepine and 4βOHC level (r = 0.14, 95% CI −0.14, 0.40, P = 0.3; Figure 3B).

Figure 3.

Correlation between 4β‐hydroxycholesterol (4βOHC) serum concentration and daily dose (A) and steady‐state concentration of carbamazepine (B). P values are estimated from Spearman's tests, while linear trend lines are added only for visual purposes. Open circles (○) indicate use of multiple enzyme inducers

Discussion

Previous studies have shown that 4βOHC concentrations are increased in patients treated with carbamazepine 8, 9, but as far as we know, this is the first study to investigate to what extent 4βOHC is correlated with administered dose (i.e. presystemic exposure) and steady‐state concentration (i.e. systemic exposure) of a potent CYP3A4 inducer. We observed a highly significant correlation between carbamazepine dose and serum concentration of 4βOHC, whereas no association between C _ss of carbamazepine and 4βOHC concentrations was found. These findings suggest that carbamazepine has a stronger inductive effect on presystemic than systemic CYP3A4 phenotype, and might indicate the role of intestinal CYP3A4 in the formation of 4βOHC.

The highly significant correlation observed between daily dose of carbamazepine and individual 4βOHC concentration clearly shows that the inductive effect of carbamazepine on CYP3A4 metabolism is dose‐dependent. We have been unable to find former studies showing increased CYP3A4‐mediated metabolism by increasing doses of carbamazepine, but a dose‐dependent increase in CYP3A4 metabolism has been described for rifampicin 11, which is another potent enzyme inducer. In a study where rifampicin was administered at doses of 20, 100 or 500 mg daily over 2 weeks, the 4βOHC concentrations increased by 1.5‐, 2.5‐ and 4‐fold, respectively 11. However, the correlation between 4βOHC concentration with rifampicin dose and C _ss was not compared in this study.

It is well‐known that CYP3A4 expression is high in the intestinal wall 18, and that presystemic metabolism of CYP3A4 substrates during intestinal absorption could be important for the bioavailability following oral administration, e.g. of midazolam 21, 22. Backman et al. found that intestinal CYP3A4 phenotype is also increased by the use of enzyme inducers 24. In a study with midazolam, they reported that AUC and half‐life of this CYP3A4 probe agent were reduced by 95% and 58%, respectively, during co‐administration with either carbamazepine or phenytoin 24. The fact that the relative decrease in midazolam half‐life, reflecting hepatic CYP3A4 phenotype, was much smaller than the decrease in AUC, provides firm evidence that carbamazepine and phenytoin induce intestinal CYP3A4 phenotype.

The hepatic selectivity of 4βOHC formation has been proposed as a limitation of this endogenous agent as a CYP3A(4) phenotype biomarker 16, 17. However, there has not been published evidence showing that intestinal CYP3A4 expression does not influence 4βOHC concentration. In our study, 4βOHC was correlated with oral dose rather than serum concentration of carbamazepine. This finding indicates that the 4βOHC concentration is more closely associated with presystemic than systemic CYP3A4 induction, which might suggest an involvement of intestinal CYP3A4 in formation of 4βOHC. In accordance with this, Tomalik‐Scharte et al. found a better correlation of 4βOHC concentrations with oral compared with intravenous midazolam clearance 25. The potential importance of intestinal CYP3A4 in 4βOHC formation could also be a reason why de Graan et al. in a recent study reported no association between 4βOHC concentrations and clearance of the CYP3A4 substrates paclitaxel and docetaxel following intravenous administration 26.

Although our findings provide indications that intestinal CYP3A4 might catalyze the formation of 4βOHC, further evidence is necessary to confirm this. To provide additional evidence that 4βOHC also reflects intestinal CYP3A4 phenotype, a key study will be to investigate the effect of grapefruit juice co‐administration, which selectively inhibits CYP3A4 in enterocytes 27, 28, on 4βOHC concentrations. Moreover, it will be of interest to study the association of 4βOHC with directly measured CYP3A4 expression in enterocytes, as well as the correlation of 4βOHC concentrations with serum concentrations/AUC of drugs whose oral bioavailability is known to be restricted by intestinal CYP3A4 metabolism.

We found a 10 times higher median serum concentration of 4βOHC in patients treated with carbamazepine compared with patients treated with levetiracetam. This is in line with previously published studies where patients have been treated with carbamazepine or other potent CYP3A inducing drugs, e.g. phenytoin, phenobarbital and rifampicin 8, 9, 10, 11, 12, 13. These studies have been performed with co‐administration of a single CYP3A inducer only, but in the present study eight patients were using multiple inducers. In the subgroup with multiple inducers, the median 4βOHC concentration was approximately 70% higher compared with patients treated with carbamazepine. This rather modest additive enzyme‐inducing effect probably reflects that carbamazepine, phenytoin and phenobarbital share the same mechanism of CYP3A4 upregulation.

The individual variability in serum concentrations of 4βOHC in the study population was extensive, both in the levetiracetam and carbamazepine group, with 10–15‐ fold ranges. There are many potential sources to this variability, but in this naturalistic setting we had limited access to information that might be of relevance. However, gender was available for all patients and previous studies have reported that females have higher CYP3A4 activity than males 4, 26, 29. In our patient material, the median observed 4βOHC concentration was higher in females in both levetiracetam and carbamazepine treated patients, but the difference was most pronounced, and only significant, in the carbamazepine group. This suggests that the CYP3A4 inducability is higher in females than males, which actually could be a reason for the generally higher CYP3A4 activity among females.

The naturalistic nature is a methodological limitation of the present study. For example, doses filled in on the requisition forms may not be the same as those actually administered by the patients. Moreover, we have no control with nutritional factors and use of herbal drugs that potentially may affect CYP3A(4) activity. In addition, steady‐state conditions of carbamazepine were confirmed from information available on the requisition forms, which is an approach implying some uncertainty. Together with variable or lacking information about time between last dose intake and blood sampling, this limits the validity of the correlation analysis between C _ss of carbamazepine and serum concentration of 4βOHC. Dose information was also lacking in 20% the carbamazepine‐treated patients. However, the tight and highly significant correlation between daily dose of carbamazepine and 4βOHC concentration, despite some uncertainties associated with the naturalistic methodology, underscores the strength of this main study finding.

In this study, we used absolute serum concentration as a measure of CYP3A(4) activity instead of the cholesterol‐adjusted ratio. Use of 4βOHC: total cholesterol ratio (4βOHC : C) would eliminate potential bias caused by variation in plasma cholesterol, but Diczfalusy et al. reported that differences in cholesterol concentrations only explain about 9% of the variation in 4βOHC concentrations 4. In line with this, Tomalik‐Scharte et al. found marginal differences in correlations between the 4βOHC : C ratio and unadjusted 4βOHC concentration with midazolam clearance 25. Thus, monitoring of the unadjusted 4βOHC concentration will generally be sufficient for determination of individual differences in CYP3A(4) phenotype, as measured in the present study.

Although the cross‐sectional 4βOHC concentrations in the carbamazepine‐treated patients probably reflect their CYP3A4 inducibility, an exact measure of individual CYP3A4 induction would have been the relative change in 4βOHC concentration following carbamazepine treatment. Pre‐treatment samples were not available in this study, so it was not possible to calculate relative 4βOHC values. However, the respective correlations between absolute, cross‐sectional 4βOHC concentrations and oral dose/C _ss of carbamazepine are suitable for comparing the impact of presystemic and systemic carbamazepine exposure on induction in CYP3A4‐mediated 4βOHC synthesis.

In conclusion, our study shows that the serum concentration of 4βOHC is significantly associated with presystemic, but not systemic exposure, of carbamazepine. This suggests a stronger inductive effect of carbamazepine on presystemic than systemic CYP3A4 phenotype and might indicate a role for intestinal CYP3A4 in formation of 4βOHC. However, additional studies are necessary to evaluate further the possible intestinal formation of 4βOHC. The secondary finding that CYP3A4 inducibility appears to be higher in females than males should also be further investigated.

Competing Interests

All authors have completed the Unified Competing Interest form at www.icmje.org/coi_disclosure.pdf (available on request from the corresponding author) and declare i) no support from any organization for the submitted work, ii) no financial relationships with any organizations that might have an interest in the submitted work in the previous 3 years and iii) no other relationships or activities that could appear to have influenced the submitted work.

The authors thank Niclas Lunder at Diakonhjemmet Hospital for excellent support with the establishment of the analytical method for quantification of 4β‐hydroxycholesterol in serum. The South‐Eastern Norway Regional Health Authority is acknowledged for PhD funding to author CG.

Supporting information

Supplementary Figure 1 Example chromatograms showing the detection of 4β‐hydroxycholesterol (4βOHC) in patients treated with carbamazepine (A) and levetiracetam (B). The chromatographic separation of 4α‐hydroxycholesterol (4αOHC) from 4βOHC was confirmed by analysis of reference standards

Gjestad, C. , Huynh, D. K. , Haslemo, T. , and Molden, E. (2016) 4β‐hydroxycholesterol correlates with dose but not steady‐state concentration of carbamazepine: indication of intestinal CYP3A in biomarker formation?. Br J Clin Pharmacol, 81: 269–276. doi: 10.1111/bcp.12833.