Introduction

Tamoxifen (TAM) is a weak anti-estrogen that undergoes extensive biotransformation in humans to metabolites that have far greater anti-estrogenic potency. 4-Hydroxytamoxifen (4HT) represents less than 10 % of TAM primary oxidation [1, 2], but has 100-fold greater affinity for the estrogen receptor (ER) and 30-fold to 100-fold greater potency in suppressing estrogen-dependent cell proliferation compared with TAM [3–5]. The secondary TAM metabolite Z-N-desmethyl-4-hydroxytamoxifen (ENDX) has equivalent in vitro ER-binding potency [6] and suppression of ER-dependent human breast cancer cell proliferation compared to 4HT [6, 7], and is thought to substantially contribute to TAM activity [8].

In humans, ENDX plasma concentrations are typically sixfold–tenfold higher [9–11] than 4HT plasma concentrations (<10 nM), but exhibit substantial variability that is partially explained by genetic and drug-induced variation in CYP2D6 enzyme activity. TAM-treated women with reduced or absent CYP2D6 activity have low concentrations of ENDX [10–13]. While controversy continues to exist regarding the role of CYP2D6 and TAM in clinical outcomes [14, 15], recent secondary analyses from independent large prospective studies [16, 17], as well as data from the metastatic setting [18], have demonstrated that both CYP2D6 genotype [17] and ENDX concentrations [16, 17] are associated with disease-free survival. Based on these observations, studies are ongoing to develop endoxifen in patients with hormone refractory cancer (NCT01327781 and NCT01273168).

Because primary and secondary metabolism limits the concentrations of the active metabolites of TAM, we hypothesized that clinically relevant concentrations and exposures of ENDX could be achieved following direct oral administration of ENDX. Therefore, we characterized and compared the pharmacokinetics of intravenous (i.v.) and oral ENDX with oral and subcutaneous (s.c.) TAM in female mice.

Materials and methods

Z-Endoxifen hydrochloride (ENDX·HCl) was provided by the Developmental Therapeutics Program, NCI. TAM, 4HT, N-desmethyltamoxifen (NDMT), and toremifene (a selective ER modulator used as the internal standard) were purchased from Sigma Chemical Company (St. Louis, MO). The primary stock solutions of ENDX (2 mg/ml), 4HT (1 mg/ml), NDMT (1 mg/ml), TAM (1 mg/ml), and toremifene (1 mg/ml) were prepared in ethanol and stored at −20 °C. Working standard solutions were prepared in ethanol and stored at −20 °C. All animal studies were performed under protocols and guidelines approved by the Institutional Animal Care and Use Committee of the Mayo Clinic (Rochester, MN). Female CD1 mice were supplied by Charles River (Portage, MI). Female ovariectomized BALB/c athymic nude mice were obtained from Harlan Laboratories (Indianapolis, IN). The mice were housed in the Department of Comparative Medicine Animal Facility, which provided a pathogen-free environment under controlled conditions of light and humidity, and accredited by the International Association for Assessment and Accreditation of Laboratory Animal Care. Mice were acclimatized for 48 h after arrival and received food and water ad libitum.

HPLC assay

Plasma concentrations of TAM, ENDX, 4HT, NDMT were measured using a modification of a previously published HPLC assay [19] that utilized a concave gradient elution profile and a post-column photochemical reactor (PHRED-8, Aura Industries) with fluorescence detection (excitation wavelength—250 nm, emission wavelength—370 nm). This assay has been validated in our laboratory according to FDA guidance. The lower limit of quantification and linear range were 1 ng/ml and 1–500 ng/ml, respectively. In brief, TAM, ENDX, 4HT, NDMT, and toremifene were isolated from plasma samples using liquid–liquid extraction. Dried extracts were reconstituted in 100 μl of 55:45 ACN:20 mM KH2PO4, pH 3.0, transferred to an amber auto-sampler vial with glass insert and placed in an autosampler maintained at 4 °C. Compounds were separated on a Genesis C18 (250 mm × 3.0 mm, 3 µ particle size) analytical column protected by a Genesis C18 (10 mm × 3.0 mm, 3 µ particle size) pre-column using a concave gradient elution profile from 43:57 (v/v) ACN:20 mM KH2PO4, pH 3.0, to 75:25 (v/v) ACN:20 mM KH2PO4, pH 3.0 (flow rate—0.5 ml/min, injection volume—50 μl). All procedures were performed in the dark under minimum exposure to light.

Mouse pharmacokinetic studies

Pharmacokinetics were characterized in female CD1 and ovariectomized BALB/c athymic nude mice. Blood samples were collected from mice anesthetized under isoflurane by cardiac puncture into a syringe containing 0.15 ml of 10 % heparin/CPDA anticoagulant and transferred to silanized amber microcentrifuge tubes. Plasma and red blood cells were separated immediately by centrifugation in a refrigerated centrifuge set at 4 °C. Plasma was transferred to a separate tube, and both tubes were immediately frozen at −20 °C until analysis.

Single-dose pharmacokinetics in CD1 mice

For i.v. dosing, ENDX·HCl was dissolved in 5 % ethanol, 5 % PEG 400, and 90 % normal saline (final concentration—0.2 mg/ml ENDX) for the 1 mg/kg i.v. dose. Blood samples (2–6 mice per time-point) were collected 5, 15 min, 0.5, 1, 2, 4, 8, 16, and 24 h after administering the dose. For oral dosing, TAM (4, 10, and 20 mg/kg) and ENDX·HCl (10, 25, 50, 75, and 200 mg/kg) were suspended in 0.5 % carboxymethyl cellulose (CMC). Blood samples (three mice per time-point) were collected 0.5, 1, 2, 4, 8, 16, and 24 h after administering the dose. For subcutaneous dosing, TAM (4, 10, and 20 mg/kg) was suspended in 0.5 % CMC and ENDX·HCl (2.5 and 25 mg/kg) was prepared in 1 mM ascorbic acid:PEG400, 1:1 v/v. Blood samples (three mice per time-point) were collected 0.5, 1, 2, 4, 8, 16, and 24 h after administering the dose.

Multiple-dose pharmacokinetics in CD1 mice

ENDX·HCl (25 or 100 mg/kg) was suspended in 0.5 % CMC and administered orally for five consecutive days. Blood samples (five mice per time-point) were collected 4 h after administering the dose on day 1 and day 5.

Single-dose pharmacokinetics in ovariectomized female nude mice

Ovariectomized female BALB/c athymic nude mice 4–6 weeks of age supplemented with 1.7-mg 90-day-release estrogen pellets (17β-estradiol, NE-121, Innovative Research, Novi, MI). Mice were randomly assigned to receive oral TAM (75 mg, suspended in 0.3 % hydroxypropyl cellulose) or oral ENDX (75 mg/kg dissolved in 3 mM PEG400:ascorbic acid, 50:50 v/v). A third group of female CD1 mice received oral ENDX (75 mg/kg dissolved in PEG400:ascorbic acid, 50:50 v/v). Blood samples (three mice per time-point) were collected before treatment and 1, 4, 8, and 24 h after administering the dose.

Data analysis

Pharmacokinetics were estimated by standard non-compartmental analysis methods using WinNonlin (Professional Version 3.0; Pharsight Corp; Mountain View, CA). The apparent terminal elimination rate constants (λz) were determined by linear least-squares regression through the linear terminal portion of the graph of the log plasma concentration versus time. The apparent elimination half-life (t 1/2) was calculated as 0.693/λz. Area under the plasma concentration–time curve (AUC0–T ) was determined using the linear trapezoidal rule from time zero to the time of the last detectable sample (T). Area under the plasma concentration–time curves through infinite time (AUC0–∞) was calculated by adding C T /λz to AUC0–T . The CLp was calculated as dose/AUC0–∞. Bioavailability was calculated as follows: (AUCpo/AUCiv)·(Doseiv/Dosepo)·100 %.

Differences in concentration levels between groups were assessed using Kruskal–Wallis test. A p value <0.05 was considered significant.

Results

The pharmacokinetics of TAM and its major circulating metabolites were characterized in female CD1 mice administered s.c. and oral doses of 4, 10, and 20 mg/kg, which are the most commonly utilized doses for in vivo studies [20]. The pharmacokinetic data are summarized in Table 1. Plasma profiles for oral and s.c. administration of 20 mg/kg TAM are shown in Fig. 1a, b, respectively.

Table 1 Single-dose pharmacokinetics tamoxifen in mice
Full size table
Fig. 1
figure 1

Plasma profiles for average (n = 3) TAM (closed circle), 4HT (open circle), NDMT (closed inverted triangle) and ENDX (open triangle) following a s.c. and b oral administration of 500 μg (20 mg/kg) TAM

Full size image

For both the oral and s.c. TAM doses, plasma concentrations increased significantly across dose levels (oral: p values ≤0.044 for 0.5, 1, 2, 4, and 16 h; s.c.: p values ≤0.039 for 0.5, 2, 4, 16, and 24 h). At the 20 mg/kg dose, TAM peak plasma concentrations (C max) were achieved 0.5–1 h after oral administration; however, only one of the seven mice achieved C max > 30.0 ng/ml. TAM C max values were achieved 0.5–4 h after s.c. administration, and 11 of 12 mice achieved C max > 30.0 ng/ml. TAM AUC increased with dose; however, comparing equivalent oral and s.c. doses, TAM AUC values following oral administration were 2.5–14 % of the AUC values achieved following s.c. administration (Table 1).

Detection of active TAM metabolites in plasma was dependent on the TAM dose and route of administration. Metabolite concentrations of 4HT and ENDX were consistently higher following oral administration compared to s.c. administration. 4HT was detected after each dose and both routes, but concentrations were low (C max, 2.0–7.3 ng/ml) after s.c. dosing. ENDX was not detected until the 20 mg/kg s.c. dose and 10 and 20 mg/kg oral doses. At the 20 mg/kg dose, plasma concentrations of 4HT and ENDX were lower for the s.c. compared to the oral route of administration, and at 1 h, this difference was statistically significant (p = 0.0339 and p = 0.0323, respectively).

Since low ENDX concentrations of were produced following s.c. and oral TAM, we investigated the pharmacokinetics and bioavailability of ENDX after direct dosing of this active TAM metabolite (Table 2). Following an i.v. dose of 1 mg/kg ENDX·HCl, a graph of the plasma ENDX concentration versus time showed rapid distribution and slower elimination (Fig. 2). The ENDX C max, terminal half-life and plasma clearance values were 96 ng/ml, 6.5 h, and 11.8 l/h/kg, respectively.

Table 2 Single-dose pharmacokinetics endoxifen in mice
Full size table
Fig. 2
figure 2

ENDX plasma profile after a single i.v. dose of 1 mg/kg ENDX·HCl

Full size image

After oral administration, ENDX C max was achieved within 0.25–2 h. Plasma concentrations increased significantly across dose levels (10, 25, 50, 75, and 200 mg/kg) at 0.5, 1, 2, and 4 h after oral administration (p value ≤0.001). As the dose was increased from 10 to 200 mg/kg, the C max and AUC values increased 88-fold and 400-fold, respectively. Apparent oral bioavailability increased from 12 to >200 % over this same dose range. Moreover, plasma concentrations exceeding 1 μM could be achieved and maintained for 24 h following a single oral dose of 75 mg/kg ENDX·HCl. Thus, high plasma concentrations, associated with previously observed in vitro ENDX cytotoxicity [8], were achieved and maintained after a single oral ENDX·HCl dose.

The pharmacokinetic data for ENDX following equivalent s.c. and oral doses of 20 mg/kg TAM and 25 mg/kg ENDX·HCl are shown in Tables 1 and 2, respectively. ENDX C max and AUC following oral ENDX·HCl (103 ng/ml and 660 ng/ml h, respectively; Table 2) were eightfold higher than those same values following the equivalent oral dose of TAM (13.4 ng/ml and 83.1 ng/ml h; Table 1). ENDX C max and AUC following s.c. ENDX·HCl (935 ng/ml and 4,920 ng/ml h, respectively; Table 2) were >140-fold higher than those same values following the equivalent oral dose of TAM (1.9 ng/ml and 34.6 ng/ml h; Table 1). Finally, when comparing oral with s.c. ENDX·HCl dosing, C max and AUC values achieved with a 2.5 mg/kg s.c. dose were similar to those achieved with 10–25 mg/kg oral doses (Table 2).

Based on the single-dose pharmacokinetic data, we selected ENDX·HCl doses of 25 and 100 mg/kg for multiple oral dose regimens to yield ENDX plasma concentrations of 100 nM (37 ng/ml) and 1–2 μM (370–740 ng/ml), respectively. The ENDX concentrations were measured 4 h after the daily dose on day 1 and day 5. As shown in Table 3, ENDX accumulated in plasma after 5-day dosing and this difference was significant at the 100 mg/kg dose (Wilcoxon rank sum test p = 0.036).

Table 3 Endoxifen plasma concentration (ng/ml) 4 h after a single dose or daily doses for five consecutive days
Full size table

To confirm the similarity of ENDX and TAM pharmacokinetics between CD1 mice and nude mice used for tumor xenograft studies, we determined ENDX C max and AUC0–24h values in CD1 mice and tumor-bearing nude mice treated with an oral dose of 75 mg/kg ENDX·HCl or TAM. The ENDX C max and AUC0–24h values for tumor-bearing nude mice (1,030 ng/ml and 15,500 ng/ml h) were similar to the C max and AUC0–24h values for CD1 mice (888 ng/ml and 12,700 ng/l h) and were 27-fold higher than the ENDX C max and AUC0–24h values (33 ng/ml and 471 ng/ml h) following TAM administration.

Discussion

Tamoxifen is an effective and widely used drug for treatment of ER+ breast cancer. While TAM itself has both estrogenic and antiestrogenic activity, it is a prodrug with complex metabolism [21]. Cytochrome P450-catalyzed oxidation yields several hydroxylated metabolites including 4HT and ENDX, which have much greater anti-estrogenic activity than the parent drug [8, 21]. In TAM-treated humans, plasma concentrations of 4HT are low (<5 ng/ml) [21]. In marked contrast, plasma concentrations of ENDX are up to tenfold higher than 4HT, but vary widely (3–26 ng/ml) based in part on CYP2D6 polymorphisms and concomitant administration of drugs that inhibit CYP2D6 enzyme activity [21].

Given the importance of ENDX as it relates to disease-free survival in human TAM-treated breast cancer [16], recent data on the pharmacologic activity of TAM metabolites [8, 21], as well as differences in P450 enzyme isoform expression in mice and humans [22, 23], it is essential to reexamine the murine pharmacokinetics of tamoxifen. This point is critical since the current standard is to use 4HT to model the in vitro antitumor activity of TAM, but to use s.c. administration of the parent drug, TAM for in vivo studies. This is based in part on a head-to-head comparison of s.c. TAM with s.c. mono-hydroxytamoxifen where TAM was more effective than mono-hydroxytamoxifen in an in vivo DMBA-induced rat mammary carcinoma model [24].

Therefore, our first goal was to characterize the murine pharmacokinetics of the major TAM metabolites following oral and s.c. administration. An early study of TAM pharmacokinetics in mice using high oral doses (50 and 200 mg/kg) demonstrated dose-dependent formation of NDMT and 4HT, exposure to higRh concentrations of 4HT that did not reflect pharmacokinetics in humans, and trace concentrations of ENDX that could not be quantified due to the lack of pure compound [25]. Here, we demonstrate that the pharmacokinetics of TAM and 4HT are consistent with earlier studies, with very low 4HT concentrations detected after s.c. dosing and much higher 4HT concentrations observed after oral dosing. In fact, s.c. administration of 20 mg/kg TAM to female mice yields low exposure to 4HT (7.3 ng/ml) and ENDX (1.9 ng/ml), and mirrors the concentrations of active metabolites observed in TAM-treated humans with deficient CYP2D6 metabolism. The 20 mg/kg oral dose yielded a tenfold higher 4HT (26.4 ng/ml) than observed in humans receiving an oral TAM dose of 20 mg/day. We observed much lower NDMT concentrations than either TAM or 4HT. The reason for the low NDMT is unclear, but may be due to differences in dose or variability in cytochrome P4503A enzyme expression across mouse strains including CD1 mice [23]. Importantly, with the availability of a pure standard for ENDX, we were able to confirm the presence of very low ENDX levels.

Our second goal was to select the route and dose of TAM that most closely resembled the concentrations of the active metabolites in TAM-treated humans, and to compare concentrations of 4HT and ENDX based upon the route of TAM administration (oral vs. s.c.). Our observation that 4HT and ENDX concentrations were substantially lower following s.c. compared with oral administration is consistent with the fact that the s.c. route circumvents first-pass metabolism and results in lower systemic availability of the active metabolites. Specifically, peak concentrations of 4HT and ENDX were 7.3 and 1.9 ng/ml, respectively, following a single s.c. dose of 20 mg/kg TAM, and very similar to the concentrations observed in humans with deficient CYP2D6 metabolism. In contrast, oral administration of TAM resulted in a 4HT concentration that was twice that of ENDX (26.4 and 13.4 ng/ml, respectively) and tenfold higher than known to be achievable in TAM-treated humans. This difference is likely due to the lower expression of CYP2D and CYP3A enzymes in mice compared to humans [22] and differences in P450-catalyzed metabolism of TAM between humans and mice. CYP2D6 was originally characterized as debrisoquine hydroxylase in rat and human liver, but does not appear to be homologous to the P450 enzyme that catalyzes debrisoquine hydroxylation in mice since it is not inhibited by an antibody against purified rat liver CYP2D that inhibits debrisoquine hydroxylase activity in rat liver microsomes [23, 26]. Consequently, mouse debrisoquine hydroxylase either is a mouse CYP2D enzyme that is not recognized by anti-rat CYP2D or belongs to another P450 subfamily. Therefore, oral TAM disposition in mice is not representative of human oral TAM disposition, and this route should not be used to model human TAM antitumor activity.

To overcome the limitations of primary and secondary metabolism necessary for the metabolic activation of TAM in humans, we investigated the ENDX pharmacokinetics. A single i.v. dose of 1 mg/kg ENDX·HCl achieved a peak concentration that was more than twofold higher than the concentration associated with activity against breast cancer in women treated with TAM [16]. A single oral dose of 10 mg/kg ENDX·HCl achieved a peak concentration associated with maximal in vitro activity [8]. Clinically active concentrations of ENDX were sustained for 8 h with a single dose of 50 mg/kg ENDX·HCl, and concentrations greater than 1 μM (370 ng/ml) were sustained for 24 h with a single dose of 75 mg/kg ENDX·HCl.

Two prior studies have evaluated the in vivo antitumor activity of ENDX [27, 28]. In the first study, Ahmed et al. evaluated low doses of oral ENDX (2–8 mg) compared with oral TAM (20 mg) using MCF7 xenografts and observed that both TAM and ENDX suppressed antitumor activity, but demonstrated no difference comparing the 8 mg oral ENDX dose with the 20 mg oral TAM dose. Pharmacokinetics were not evaluated in either the TAM or ENDX treated mice, and given our observations that oral TAM results in high concentrations of 4HT, the antitumor activity in oral TAM-treated mice is likely due to 4HT with a small contribution from ENDX. Recently, Gong et al. evaluated the relationship between ENDX exposure and tumor growth inhibition and demonstrated 100 % growth inhibition in MCF7 xenografts using a 30–60 mg/kg dose of ENDX. While Gong et al. did not compare ENDX with TAM, it should be noted that the doses, concentrations, and antitumor activity of ENDX were similar to our findings and suggest that the optimal concentrations of ENDX needed to inhibit antitumor activity appear to be higher than those achievable in TAM-treated humans.

Of note, the data from our studies with mice show evidence of dose-dependent pharmacokinetics as illustrated by greater than dose-proportional increases in AUC and C max over the dose range studied (Table 2). There are two possible explanations for dose-dependent pharmacokinetics of ENDX. First, TAM and its primary metabolites, NDMT and 4HT, are effective inhibitors of CYP3A activity [29]. The low K i values for reversible CYP3A inhibition by NDMT and 4HT, as well as the low K i values for mechanism-based CYP3A inhibition by TAM and NDMT, are similar to the steady-state plasma of those found when TAM is administered at doses of 20–40 mg/day. By analogy, it can be hypothesized that CYP3A activity may be inhibited by ENDX at concentrations achievable following oral doses. Thus, it is possible that inhibition of CYP3A in small intestine epithelium would increase drug absorption, while inhibition of CYP3A activity in the liver would decrease hepatic clearance. Second, higher doses may inhibit drug transporters involved in ENDX absorption and clearance. Several investigators have examined the role of p-glycoprotein in the disposition of TAM. TAM, NDMT, and 4HT were inhibitors, but not substrates, for p-glycoprotein [30]. 4HT was the most potent inhibitor with an IC50 value that was threefold lower than the values observed for TAM and NDMT. In an in vivo study quercetin, an inhibitor of p-glycoprotein and CYP3A activity increased the rate of absorption and AUC of TAM, increased the 4HT AUC, and decreased the 4HT/TAM ratio [31]. These observations were consistent with an effect of quercetin on both CYP3A and p-glycoprotein. Similarly, ENDX has been shown to be a substrate for p-glycoprotein in both in vitro studies of p-glycoprotein-dependent flux across LLCPK cells that overexpressed p-glycoprotein and in vivo studies of plasma and brain disposition of ENDX in Mdr1a-deficient mice [32]. The result of both mechanisms would be to increase systematic exposure and apparent bioavailability of ENDX.

Conclusions

In conclusion, our studies provide important insight into the murine pharmacokinetics of both TAM and ENDX, and demonstrate the importance of dose and route of administration as it relates to the concentrations of the active metabolites. These data further suggest that the administration of ENDX may result in greater antitumor activity by means of increased ENDX exposure and provide support for the ongoing development of ENDX as a new hormonal therapy for ER+ breast cancer. Based on these data and in collaboration with the National Cancer Institute, phase I and II clinical trials of ENDX are ongoing/planned in women with ER-positive breast cancer and in patients with hormonally positive tumors (NCT01327781 and NCT01273168).