4β-hydroxycholesterol as an endogenous CYP3A marker in cancer patients treated with taxanes

Abstract

Aim

Taxanes are anti-cancer agents used to treat several types of solid tumours. They are metabolized by cytochrome P450 (CYP) 3A, displaying a large pharmacokinetic (PK) variability. In this study, we evaluated the endogenous CYP3A4 marker 4β-hydroxycholesterol (4β-OHC) as a potential individual taxane PK predictor.

Methods

Serum 4β-OHC and cholesterol concentrations were determined in 291 paclitaxel and 151 docetaxel-treated patients, and were subsequently correlated with taxane clearance.

Results

In the patients treated with paclitaxel, no clinically relevant correlations between the 4β-OHC or 4β-OHC : cholesterol ratio and paclitaxel clearance were found. In the patients treated with docetaxel, 4β-OHC concentration was weakly correlated with docetaxel clearance in males (r = 0.35 P = 0.01, 95% CI 0.08, 0.58). Of the 10% patients with taxane outlier clearance values, 4β-OHC did correlate with docetaxel clearance in males (r = 0.76, P = 0.03, 95% CI 0.12, 0.95).

Conclusion

There was no clinical correlation between paclitaxel clearance and the CYP3A4 activity markers 4β-OHC or the 4β-OHC : cholesterol ratio. A weak correlation was observed between 4β-OHC and docetaxel clearance, but only in males. This endogenous CYP3A4 marker has limited predictive value for taxane clearance in patients.

WHAT IS ALREADY KNOWN ABOUT THIS SUBJECT

• Patients with low or high taxane clearance have an increased risk of severe adverse events or decreased taxane efficacy, respectively.

• Predicting CYP3A metabolic capacity by using a phenotypic marker could help to individualize taxane therapy.

• In previous literature, 4β-hydroxycholesterol was hypothesized to be a reasonable marker for CYP3A activity.

WHAT THIS STUDY ADDS

• Studying the role of an endogenous marker for predicting taxane clearance instead of focusing on the more complicated and costly alternative, an exogenous marker, could improve clinical applicability of new markers.

• The correlation of taxane clearance with 4β-hydroxycholesterol and the correlation with the 4β-hydroxycholesterol : cholesterol ratio were not clinically relevant.

Introduction

The anti-microtubular agents paclitaxel and docetaxel are widely used for the treatment of breast, non-small cell lung, ovarian and prostate cancer [1]. These taxanes display a large inter-individual variability in pharmacokinetics, toxicity profiles and efficacy [2]. This large variability makes dosing within a small therapeutic window of these agents difficult. Therefore, patients with high taxane clearance are at risk for a suboptimal therapeutic effect due to low systemic drug concentrations. On the other hand, patients with low taxane clearance are at a higher risk of severe adverse events. For example, patients with low paclitaxel clearance are at risk of peripheral neuropathy and haematological toxicities [3], while docetaxel patients with low clearance develop febrile neutropenia, mucositis and skin toxicity more frequently [4,5]. Factors causing this large interindividual pharmacokinetic variation are still largely unknown, but include gender, testosterone concentrations, menopausal status, liver impairment and drug–drug interactions [6]. The cytochrome P450 (CYP) 3A family is responsible for the metabolism of a large number of drugs [7]. Paclitaxel and docetaxel are both metabolized by CYP3A4. For docetaxel, CYP3A5 has a minor contribution to this phase I metabolic route, whereas for paclitaxel CYP2C8 plays a prominent role (Figure1) [8,9]. Therefore, knowledge of CYP3A metabolic capacity of an individual patient could aid in the development of a personalized dosing strategy, especially for anti-cancer agents with a narrow therapeutic index. Predicting individual metabolic profiles by using a phenotypic marker could potentially allow for individual dose adjustments during successive courses [10]. CYP3A metabolic activity can be measured by determining the clearance of a marker specifically metabolized by CYP3A. The value and selection of a suitable CYP3A phenotyping method for clinical use has extensively been discussed [11,12]. Several CYP3A substrates such as midazolam, erythromycin, cortisol, alprazolam, alfentanil, dextromethorphan, nifedipine, lidocaine and dapsone have been suggested as probe drugs [10,13,14]. All these methods are laborious and require exogenous drug administration, which potentially limits their clinical use. Recently, the endogenous compound 4β-hydroxycholesterol (4β-OHC) has been proposed as a marker for CYP3A activity because in vivo concentrations of 4β-OHC are thought to reflect CYP3A4/5 activity [15–17]. The conversion of cholesterol to 4β-OHC is mainly catalyzed by CYP3A4 (Figure1) [15]. An advantage of this marker is the long plasma half-life of 4β-OHC (∼17 days). Therefore, changes over time in plasma concentrations within individuals will be relatively low. At the same time, this long half-life also limits its potential as a marker to predict a rapid CYP3A4 change [18,19]. It has already been shown that treatment with strong CYP3A4 inducers (e.g. carbamazepine, phenytoin and phenobarbital) increased plasma concentrations of 4β-OHC approximately 10-fold [15]. Also, treatment with the weak CYP3A4 inducer ursodeoxycholic acid resulted in a modest increase in 4β-OHC [15], indicating the ability of the endogenous marker to distinguish between weak and strong inducers. The ability of 4β-OHC and the 4β-OHC : cholesterol ratio to predict individual taxane pharmacokinetic profiles has not been investigated yet.

Figure 1.

(A) Scheme explaining the metabolism of paclitaxel, primarily catalyzed by CYP2C8 and to a minor extent by CYP3A4. (B) Scheme explaining the metabolism of docetaxel, primarily catalyzed by CYP3A4/5. (C) Schematic conversion of cholesterol into 4β-hydroxycholesterol, primarily catalyzed by CYP3A4/5. OH = hydroxy

The aim of this study was to investigate correlations between the endogenous marker 4β-OHC and the clearance of the taxanes paclitaxel and docetaxel to assess the potential of 4β-OHC as a phenotyping method in taxane therapy. The association between taxane clearance and 4β-OHC parameters was also assessed in the 10% of patients with outlier clearance values, because in these patients availability of a strategy that allows a priori dose adjustments would be especially important. Because of previously published differences in CYP3A4/5 activity between males and females the correlations were analyzed separately for gender [20,21]. Also, paclitaxel metabolism has been described to be lower in females than in males [22].

Methods

Patients

Patients treated with paclitaxel or docetaxel who were enrolled in a pharmacokinetic study (Dutch trial registry: http://www.trialregister.nl, NTR2311) were included in this analysis. Inclusion criteria were as described previously [23,24]. In brief, patients included had (i) a histological or cytological confirmed diagnosis of cancer treated with paclitaxel or docetaxel, (ii) were aged ≥18 years, (iii) had a WHO performance score of 0–1 and (iv) had adequate haematopoietic, hepatic and renal functions. During the study, CYP3A4 and CYP2C8 inducers or inhibitors were not allowed (see reference [25] for enzyme nomenclature). The trial was approved by the medical ethics committee of the Erasmus University Medical Center (MEC2003-264) and all patients participating in this trial provided written informed consent.

Treatment

Patients treated with docetaxel were mainly administered weekly a 75 to 100 mg m^–2 dose intravenously, which depended on the tumour type and combination regimen (chemotherapy and/or radiotherapy) used. Patients treated with paclitaxel were mainly administered an intravenous dose of 50 mg m^–2, 90 mg m^–2 weekly or 175 mg m^–2 every 3 weeks. Patients did not receive other chemotherapy for 4–6 weeks before the start of docetaxel or paclitaxel treatment.

Pharmacokinetic analysis

Pharmacokinetic sampling was performed in any treatment cycle according to a limited sampling strategy and docetaxel and paclitaxel quantification in plasma was as described previously [23,24]. Samples were drawn pre-treatment, before the end of infusion and in the elimination phase of the drug. Docetaxel was quantitated in plasma by a validated high performance liquid chromatography (HPLC) method with u.v. detection [26] or by a validated LC MS/MS method [27,28]. The LLOQ for these methods was 2.00 ng ml^–1 (CV = 8.2%). Paclitaxel was quantitated by a validated u.v. detection HPLC method [29] or by a validated LC-MS/MS method based on the method described for docetaxel [27]. The LLOQ for these methods was 2.00 ng ml^–1 (CV = 7.6%). Individual pharmacokinetic parameters were based on a previously developed population pharmacokinetic model for docetaxel [30] or paclitaxel [2] with population Cremophor concentrations [31]. The pharmacokinetic parameters were estimated as Empirical Bayes estimates with the non-linear mixed effect modelling software nonmem version VI and 7 (Icon Development Solutions, Icon Development Solutions, Ellicott City, MD, USA). Unbound clearance of paclitaxel was used in the analysis instead of total paclitaxel, because the formulation vehicle of paclitaxel, Cremophor EL, causes the total fraction of paclitaxel to have non-linear pharmacokinetics [32].

Measurement of 4β-hydroxycholesterol

Measurement of 4β-OHC was performed during the same paclitaxel cycle as the pharmacokinetic sampling. Blood samples were collected and centrifuged immediately after collection and then stored at –70°C until the day of analysis. After the addition of 50 µl of internal standard solution (1000 ng ml^–1 4β-hydroxycholesterol-d7 in water) and 500 µl of 1 m ethanolic potassium hydroxide to 50 µl plasma, oxyesterols and the internal standard were saponificated for 30 min at 37°C. After saponification, 300 µl of water was added to the solution and extracted twice with 1 ml n-hexane. The organic phase was evaporated at 45°C under reduced pressure. Hereafter, oxyesterols and the internal standard were derivated based on the mixed anhydride method previously described by Yamashita and colleagues with minor modifications [33]. Amounts of 10 mg 2-methyl-6-nitrobenzoic anhydride, 3 mg 4-dimethylaminopyridine and 8 mg picolinic acid were dissolved in 150 µl pyridine solution and added to the evaporated samples following addition of 20 µl triethylamine. Hereafter the samples were incubated for 45 min at 37°C. The oxysterols were extracted after the addition of 500 µl of water and 1 ml of n-hexane. The organic phase was evaporated and the residue was resuspended in 200 µl acetonitrile : methanol : water (3 : 6 : 1.8, v/v/v) and stored at 4°C until analysis. Oxysterols were separated by ultra performance liquid chromatography on an Acquity BEH Pheny™ 1.7 µm column Waters, Etten-Leur, the Netherlands) with a mobile phase composed of water acidified with 0.1% formic acid and acetonitrile : methanol (1 : 2, v/v) acidified with 0.1% formic acid (21 : 79, v/v) eluted at a flow rate of 0.300 ml min^–1. Baseline separation was achieved for the 4α-hydroxycholesterol and the 4β-OHC. Column effluents were analyzed by mass spectrometry with atmospheric pressure electropray ionization. The source temperature and the desolvation temperature were set at 130°C and 350°C respectively. The desolvation gas flow was set at 800 l h^–1 and the cone capillary voltage was kept at 1.5 kV. The multiple reaction monitoring (MRM) mode was applied for the quantitation of 4β-hydroxycholesterol and the internal standard with the following parameters: m/z 635 > 146, collision energy at 25 eV and m/z 642 > 146, collision energy at 15 eV, respectively. The cone voltage was 38V for all compounds and dwell times were set at 100 ms. The collision cell pirani was set at ∼7.42 e-3 mbar (argon). Calibration curves were linear over a range of 2–650 ng ml^–1. The lower limit of detection was 0.70 ng ml^–1 and the lower limit of quantification in water was set to 1.8 ng ml^–1 with a CV of 1.8%, which is well below the endogenous concentration (11.3–45.9 ng ml^–1, n = 112). Interday and intraday precision were, respectively, 4.3–2.4% CV at 14.5 ng ml^–1 and 6.8-2.6% at 50.4 ng ml^–1. Due to the absence of reference materials or reference methods to measure the 4β-OHC no accuracy could be assessed. The specificity of the assay was tested with six commercially available hydroxysterols (4α-hydroxycholesterol, 4β-OHC, 7β-hydroxycholesterol, 19-hydroxycholesterol, 22(R)-hydroxycholesterol and 25-hydroxycholesterol) to ensure complete separation of 4β-OHC from its isomers. To correct for cholesterol concentrations, total cholesterol was measured on a Roche Modular P800 analyzer (Roche Diagnostic Corp., Indianapolis, IN).

Statistics

Pharmacokinetic data are presented as median values with ranges unless stated otherwise. To test the associations between the pharmacokinetic parameters of paclitaxel or docetaxel and 4β-OHC, cholesterol and the 4β-OHC : cholesterol ratio, the Spearman correlation test was used. Patients with the 10% lowest and highest docetaxel and paclitaxel clearances were selected for separate analysis to test the correlation between the 4β-OHC marker and clearance in patients with pharmacokinetic outlier values. Because of previously seen differences in 4β-OHC concentrations between males and females, indicating a gender difference in CYP3A4/5 activity between males and females [19,20], the data were also analyzed separately for males and females. All P values are two-sided and a P value < 0.05 was considered statistically significant. Analysis was conducted with SPSS version 20.0 (SPSS Inc, Armonk, NY, USA) and Stata release 12 (StataCorp LP, College Station, TX, USA).

Results

Patients

Two hundred and ninety-one patients treated with paclitaxel and 151 patients treated with docetaxel were included in this study. In both cohorts patients were mostly of Caucasian origin (paclitaxel cohort 96%, docetaxel cohort 92%, Table 1). The median age in the paclitaxel cohort was 61 years (range, 18–82 years), 49% of these patients were female and oesophageal cancer was the most frequent primary tumour type (50%, Table 1). In the docetaxel treated group, the median age was 56 years (range 18–80 years), most patients were female (68%) and breast cancer was the most frequent primary cancer (64%), Table 1). Patients on paclitaxel received a median administered dose of 170 mg and patients on docetaxel a median dose of 160 mg (Table 1).

Table 1.

Patient characteristics^*

Characteristic	Paclitaxel cohort	Docetaxel cohort
Number of patients	291	151
Gender†
Male	149 (51)	49 (33)
Female	142 (49)	102 (68)
Ethnicity
Caucasian	278 (96)	139 (92)
Other	11 (4)	6 (4)
Unknown	2 (1)	6 (4)
Age (years)	61 (18–82)	56 (18–80)
BSA (m2)	1.89 (1.4–2.8)	1.89 (1.4–2.6)
Dose (mg)	170 (70–560)	160 (50–230)
Tumour type†
Oesophageal	144 (50)	-
Ovary	45 (16)	-
Breast	14 (5)	96 (64)
Cervix	21 (7)	-
Endometrial	15 (5)	-
Lung	11 (4)	4 (3)
Head/Neck	11 (4)	5 (3)
(A)CUP	8 (3)	-
Testis	6 (2)	-
Melanoma	2 (1)	6 (4)
Prostate	1 (0)	28 (19)
Other	13 (5)	12 (8)

^*All data are represented as median with range in parentheses, unless stated otherwise.

^†Number with percentages in parentheses.

BSA, body surface area; (A)CUP, (adeno)carcinoma of unknown primary.

Paclitaxel cohort

Paclitaxel and 4β-OHC pharmacokinetic parameters in paclitaxel treated patients are summarized in Table 2. Paclitaxel clearance differed approximately 8-fold between individual patients. The median 4β-OHC concentrations were 19.4 ng ml^–1 (range 2.9–155 ng ml^–1) and the median 4β-OHC : cholesterol ratio was 4.1 (range 1.0–26.0). The median 4β-OHC concentrations were higher in females than in males (21.0 ng ml^–1 vs. 17.5 ng ml^–1, P = 0.02). The median 4β-OHC : cholesterol ratio was similar in males and females (3.9, range 1.0–19.0 and 4.3, range 1.0–26.0, respectively).

Table 2.

Summary of docetaxel and paclitaxel pharmacokinetics and 4β-OH cholesterol pharmacokinetic parameters

	Paclitaxel cohort			Docetaxel cohort
Parameter	n	Median	Range	n	Median	Range
CL (l h–1)* taxane	291	477	138–1,037	151	44.1	16.2–95.9
4β-OH cholesterol (ng ml–1)	291	19.4	2.9–155	151	20.7	6.3–193
Cholesterol (mmol l–1)	291	4.7	2.1–20.0	151	5.4	2.7–10.7
Ratio†	291	4.1	1.0–26.0	151	4.0	1.0–27.0
Male
CL (l h–1)* taxane	149	540	142–1,037	49	45.0	26.6–95.9
4β-OH cholesterol (ng ml–1)	149	17.5	2.9–77.9	49	20.1	6.3–42.2
Cholesterol (mmol l–1)	149	4.3	2.1–20.0	49	4.7	2.7–7.7
Ratio†	149	3.9	1.0–19.0	49	4.4	2.0–9.0
Female
CL (l h–1)* taxane	142	425	138–906	102	42.1	16.2–84.9
4β-OH cholesterol (ng ml–1)	142	21.0	6.2–155	102	21.2	8.0–193
Cholesterol (mmol l–1)	142	5.1	2.4–8.6	102	5.7	3.1–10.7
Ratio†	142	4.3	1.0–26.0	102	3.9	1.0–27.0

^*For paclitaxel treated patients, unbound clearance was used;

^†Ratio 4β-OH cholesterol : cholesterol;

CL, clearance, that is, dose divided by area under the curve.

There were no significant correlations between the clearance of paclitaxel and 4β-OHC concentrations in both males and females (Table 3, Figure2), except for a weak correlation between cholesterol concentrations and paclitaxel clearance (r = −0.13, P = 0.03, 95% CI −0.24, −0.01). There were also no correlations between the 4β-OHC : cholesterol ratio and paclitaxel clearance (P > 0.4, Table 3). And, when comparing patients with the 10% highest or lowest paclitaxel clearances (n = 58), there were no correlations between the clearance of paclitaxel and 4β-OHC parameters (P > 0.05, Table 4). This outcome did not change when the data were analyzed separately for gender (males P ≥ 0.5 and females P > 0.2).

Table 3.

Correlations between 4β-OH cholesterol parameters and paclitaxel and docetaxel clearance (l h^–1)

	r†	P value	95% CI
Paclitaxel(n = 291)			Lower bound	Upper bound
4β-OH cholesterol (ng ml–1)	–0.06	0.29	–0.18	0.05
Cholesterol (mmol l–1)	–0.13	0.03‡	–0.24	–0.01
Ratio*	0	0.98	–0.11	0.12
Male (n = 149)
4β-OH cholesterol (ng ml–1)	0.03	0.75	–0.14	0.19
Cholesterol (mmol l–1)	–0.01	0.88	–0.17	0.15
Ratio*	0	0.98	–0.16	0.16
Female (n = 142)
4β-OH cholesterol (ng ml–1)	0	1.0	–0.16	0.17
Cholesterol (mmol l–1)	–0.05	0.57	–0.21	0.12
Ratio*	0.06	0.47	–0.11	0.22
Docetaxel (n = 151)
4β-OH cholesterol (ng ml–1)	0.07	0.40	–0.09	0.23
Cholesterol (mmol l–1)	–0.01	0.90	–0.17	0.15
Ratio*	0.04	0.67	–0.13	0.19
Male (n = 49)
4β-OH cholesterol (ng ml–1)	0.35	0.01‡	0.08	0.58
Cholesterol (mmol l–1)	0.11	0.45	–0.18	0.38
Ratio*	0.18	0.22	–0.11	0.44
Female (n = 102)
4β-OH cholesterol (ng ml–1)	–0.04	0.70	–0.23	0.16
Cholesterol (mmol l–1)	–0.02	0.81	–0.22	0.17
Ratio*	–0.04	0.69	–0.23	0.16

CI, confidence interval;

^*Ratio 4β-OH cholesterol : cholesterol;

^†Spearman rank correlation coefficient was used to evaluate associations between 4β-OH cholesterol and docetaxel and paclitaxel pharmacokinetics. All statistical tests were two sided.

^‡P < 0.05.

Figure 2.

(A) Scatter plot of 4β-hydroxycholesterol concentration (ng ml^–1) vs. paclitaxel unbound clearance (l h^–1). (B) Scatter plot of 4β-hydroxycholesterol : cholesterol ratio vs. paclitaxel unbound clearance (l h^–1). male; female

Table 4.

Correlations between 4β-OH cholesterol parameters and docetaxel and paclitaxel clearance (l h^–1) in patients with 10% outlier clearance values

	nd	r‡	P value	95% CI
Paclitaxel				Lower bound	Upper bound
4β-OH cholesterol	58	–0.15	0.27	–0.39	0.11
Cholesterol	58	–0.25	0.06	–0.48	0.01
Ratio*	58	–0.09	0.48	–0.35	0.17
Male
4β-OH cholesterol	35	–0.06	0.71	–0.39	0.28
Cholesterol	35	–0.08	0.63	–0.41	0.26
Ratio*	35	–0.12	0.50	–0.43	0.22
Female
4β-OH cholesterol	23	0.05	0.83	–0.37	0.45
Cholesterol	23	–0.24	0.26	–0.60	0.19
Ratio*	23	0.02	0.92	–0.39	0.43
Docetaxel
4β-OH cholesterol	30	0.09	0.64	–0.28	0.44
Cholesterol	30	0.05	0.79	–0.32	0.40
Ratio*	30	0.03	0.86	–0.33	0.39
Male
4β-OH cholesterol	8	0.76	0.03§	0.12	0.95
Cholesterol	8	–0.22	0.60	–0.80	0.57
Ratio*	8	0.52	0.19	–0.30	0.90
Female
4β-OH cholesterol	22	–0.08	0.71	–0.49	0.35
Cholesterol	22	0.10	0.66	–0.34	0.50
Ratio*	22	–0.17	0.45	–0.55	0.27

CI confidence interval;

^*Ratio 4β-OH cholesterol : cholesterol;

†Spearman rank correlation coefficient was used to evaluate associations between 4β-OH cholesterol and docetaxel and paclitaxel pharmacokinetics. All statistical tests were two sided.

^‡number of patients evaluable for correlation analysis.

^§P < 0.05.

Docetaxel cohort

The docetaxel pharmacokinetic parameters and the 4β-OHC parameters in the docetaxel treated cohort are shown in Table 2. There was almost a 6-fold difference in docetaxel clearance between patients. The median 4β-OHC concentrations were 20.7 ng ml^–1 (range 6.3–193 ng ml^–1). The median 4β-OHC : cholesterol ratio was 4.0 (range 1.0–27.0). The median 4β-OHC concentrations did not differ between males and females (20.1 ng ml^–1, range 6.3–42.2 ng ml^–1 and 21.2 ng ml^–1, range 8.0–193 ng ml^–1, respectively). Also, the median 4β-OHC : cholesterol ratio was similar in males and females (4.4, range 2.0–9.0 and 3.9, range 1.0–27.0, respectively).

There was no correlation between 4β-OHC parameters and docetaxel in the total cohort (P ≥ 0.4, Figure3). In males treated with docetaxel (n = 49), there was a significant but weak correlation between docetaxel clearance and 4β-OHC concentrations (r = 0.35, P = 0.01, 95% CI 0.08, 0.58, Table 3). This correlation was not found in females. There were no correlations between the 4β-OHC : cholesterol ratio and docetaxel clearance (P > 0.2, Table 3). Of the patients with the 10% lowest and highest docetaxel clearances, only males showed a significant correlation between docetaxel clearance and 4β-OHC (r = 0.76, P = 0.03, 95% CI = 0.12, 0.95, Table 4).

Figure 3.

(A) Scatter plot of 4β-hydroxycholesterol (4β-OHC) concentration (ng ml^–1) vs. docetaxel clearance (l h^–1). (B) Scatter plot of 4β-hydroxycholesterol (4β-OHC) : cholesterol ratio vs. docetaxel clearance (l h^–1). male; female

Discussion

Recently, the metabolic conversion of cholesterol into 4β-OHC has been described as a useful tool to predict CYP3A4 activity after treatment with strong CYP3A inducers such as carbamazepine, phenytoin or phenobarbital [15]. Treatment with these inducers resulted in highly elevated plasma concentrations of 4β-OHC, as a result of an intensified conversion of cholesterol into 4β-OHC [15]. On the other hand, treatment with CYP3A4 inhibitors, such as ritonavir or itraconazole, has led to decreased plasma concentrations of 4β-OHC. Taken together, these results suggest a potential use of this marker to assess CYP3A activity after enzyme induction or inhibition [17]. Our current study is the first to test the ability of the endogenous marker 4β-OHC and the 4β-OHC : cholesterol ratio to predict individual clearance profiles in cancer patients treated with the taxanes, docetaxel and paclitaxel.

The 4β-OHC values measured in our study were comparable with those reported in another study in healthy Caucasians (mean 20.5 ng ml^–1) [34]. Also, gender differences in 4β-OHC concentrations in the paclitaxel treated patients were similar to those previously reported, as were the higher 4β-OHC concentrations we found in females [20,21]. The variability observed in 4β-OHC measurements was large in both the docetaxel and the paclitaxel-treated patients. This could potentially be due to the use of CYP3A4 inhibiting or inducing co-medication before the start of docetaxel or paclitaxel treatment. Because of the long half-life of the effects of CYP3A4 inhibition or induction could affect 4β-OHC concentrations during therapy.

Unfortunately, paclitaxel and docetaxel clearance could not be predicted accurately by the 4β-OHC marker, which thus precludes clinical use. This could theoretically be due to the fact that CYP3A4 is not the major determinant in the metabolism of these taxanes, but CYP3A4 and CYP3A5 have been implicated earlier as crucial enzymes in the metabolism in these drugs. Although a correlation was seen in a small subgroup of men with the lowest and highest docetaxel clearances, this correlation may still result from chance, and has limited predictive value for individual patients. As the conversion of cholesterol to 4β-OHC is mainly catalyzed by CYP3A, it is understandable that paclitaxel clearance, which is mediated by both CYP3A and CYP2C8, is not predicted by this endogenous marker. For docetaxel clearance this is more surprising, as docetaxel is largely metabolized by CYP3A, although drug transporters (i.e. ABCB1 and OATP1B) may also affect the exposure to this drug [6,24]. The use of CYP3A4 and CYP2C8 inducers and inhibitors was strictly prohibited during the study, making this potential explanation for a lack of correlation less plausible. An explanation for the lack of correlation in the docetaxel-treated patients with 4β-OHC parameters could be that docetaxel metabolism is not solely dependent on CYP3A4, but may include other (currently unknown) enzymes and transporters. For example, it was recently reported that administration of the CYP3A inhibitor, imatinib, did not affect docetaxel clearance, while it was shown to inhibit CYP3A4 [35]. In contrast, it is also possible that 4β-OHC cholesterol is being metabolized by other enzymes than just CYP3A4, making the predictive power of this marker for CYP3A4 activity less strong.

Although we were not able to measure 3'-p-OH paclitaxel (Figure1), analysis of this metabolite could have been of interest to assess the ability of our endogenous marker to detect CYP3A4 clearance in our group of cancer patients.

A lot of research in the field of ‘phenotyping’ has already been done but a perfect phenotyping method for CYP3A4 still has not been found. Multiple exogenous CYP3A phenotyping methods have been proposed, including the midazolam clearance test, the erythromycin breath test and the administration of cortisol [10]. Clear correlations have been observed between midazolam clearance or erythromycin clearance and hepatic CYP3A concentrations and the amount of CYP3A4 protein levels, respectively, making these drugs potentially usable probe drugs [36,37]. Midazolam metabolism has also been associated with the induction and inhibition of CYP3A4 activity in patients receiving rifampicin and erythromycin, respectively [38]. Also, midazolam metabolism has been shown to be highly correlated with the clearance of ciclosporin and irinotecan, while erythromycin and cortisol were correlated with docetaxel metabolism [14,36,39–41]. However, in another recent study, the utility of the erythromycin test to predict docetaxel pharmacokinetics could not be confirmed [42].

The use of phenotyping methods is still not widely adopted. There are several explanations for this. First of all, the administration of an exogenous compound and additional blood sampling is a burden to the patient and is time consuming. Secondly, the administration of radioactively labelled material, as is the case for erythromycin, makes this method even less attractive. Finally, all these methods are relatively expensive and complex and may not be suitable for every individual patient [10]. We may therefore wonder if an exogenous drug will ever be adopted as a clinically implementable probe drug, or whether we should focus more on endogenous options. Although this first study on correlations between an endogenous marker and two anti-cancer drugs may be disappointing, we should further explore other options including other endogenous markers and other drugs. Whether the 4β-OHC marker could be of potential clinical use for other CYP3A4 substrates remains to be elucidated.

Competing Interests

There are no competing interests to declare.

We thank Inge Ghobadi Moghaddam-Helmantel, Mei-Ho Lam and Walter Loos for the pharmacokinetic analyses, Bertrand van Zelst for the 4β-OHC measurements and Lena Friberg for the pharmacokinetic nonmem modelling. We thank our physicians for including patients in the clinical trials.

Financial support

This study was sponsored by the Dutch Cancer Society (EMCR 2010-4664).

Authorship contributions

Anne-Joy M. de Graan: Wrote manuscript, designed research, performed research and analyzed data.

Alex Sparreboom: Wrote manuscript, designed research and performed research.

Peter de Bruijn: Wrote manuscript, performed research and contributed analytical tools.

Evert de Jonge: Performed research and contributed analytical tools.

Bronno van der Holt: Wrote manuscript, performed research and analyzed data.

Erik A.C. Wiemer: Wrote manuscript and designed research.

Jaap Verweij: Wrote manuscript and designed research.

Ron H.J. Mathijssen: Wrote manuscript, designed research, performed research and analyzed data.

Ron H.N. van Schaik: Wrote manuscript, designed research, performed research and analyzed data.