Effects of anticancer drugs on the metabolism of the anticancer drug 5,6-dimethylxanthenone-4-acetic (DMXAA) by human liver microsomes

Abstract

Aims

To investigate the effects of various anticancer drugs on the major metabolic pathways (glucuronidation and 6-methylhydroxylation) of DMXAA in human liver microsomes.

Methods

The effects of various anticancer drugs at 100 and 500 µm on the formation of DMXAA acyl glucuronide (DMXAA-G) and 6-hydroxymethyl-5-methylxanthenone-4-acetic acid (6-OH-MXAA) in human liver microsomes were determined by high performance liquid chromatography (h.p.l.c.). For those anticancer drugs showing significant inhibition of DMXAA metabolism, the inhibition constants (K_i) were determined. The resulting in vitro data were extrapolated to predict in vivo changes in DMXAA pharmacokinetics.

Results

Vinblastine, vincristine and amsacrine at 500 µm significantly (P < 0.05) inhibited DMXAA glucuronidation (K_i = 319, 350 and 230 µm, respectively), but not 6-methylhydroxylation in human liver microsomes. Daunorubicin and N-[2-(dimethylamino)-ethyl]acridine-4-carboxamide (DACA) at 100 and 500 µm showed significant (P < 0.05) inhibition of DMXAA 6-methylhydroxylation (K_i = 131 and 0.59 µm, respectively), but not glucuronidation. Other drugs such as 5-fluoroucacil, paclitaxel, tirapazamine and methotrexate exhibited little or negligible inhibition of the metabolism of DMXAA. Pre-incubation of microsomes with the anticancer drugs (100 and 500 µm) did not enhance their inhibitory effects on DMXAA metabolism. Prediction of DMXAA–drug interactions in vivo based on these in vitro data indicated that all the anticancer drugs investigated except DACA appear unlikely to alter the pharmacokinetics of DMXAA, whereas DACA may increase the plasma AUC of DMXAA by 6%.

Conclusions

These results indicate that alteration of the pharmacokinetics of DMXAA appears unlikely when used in combination with other common anticancer drugs. However, this does not rule out the possibility of pharmacokinetic interactions with other drugs used concurrently with this combination of anticancer drugs.

Introduction

5,6-Dimethylxanthenone-4-acetic acid (DMXAA) (Figure 1) is an anticancer drug with an unusual mechanism of action compared with conventional cytotoxic anticancer drugs. DMXAA induces rapid vascular collapse and necrosis in transplantable murine tumours, thought to be due to immune modulation and the induction of cytokines, in particular tumour necrosis factor-α, interferons, serotonin and nitric oxide [1–4]. Co-administration of DMXAA with other drugs has been shown to result in enhanced antitumour activity and alterations in pharmacokinetics, as reported for the combination of DMXAA with melphalan, thalidomide, and the bioreductive agent tirapazamine, in mouse models [5–8]. These results suggest that coadministration of DMXAA with other anticancer drugs may be a useful strategy for enhancing their antitumour activity. DMXAA is extensively metabolized, mainly by glucuronidation of its acetic acid side chain and 6-methylhydroxylation [9–11], giving rise to DMXAA acyl glucuronide (DMXAA-G), and 6-hydroxymethyl-5-methylxanthenone-4-acetic acid (6-OH-MXAA), which are excreted into bile and urine. Studies have indicated that DMXAA glucuronidation is catalysed by uridine diphosphate glucuronosyltransferases (UGT1A9 and UGT2B7) [10], and 6-methylhydroxylation by cytochrome P450 (CYP1A2) [12].

Figure 1.

The structure of DMXAA.

The aim of this study was to investigate the effects of various anticancer drugs on the metabolism of DMXAA in human liver microsomes, and if significant inhibition is observed, to predict in vivo DMXAA–drug pharmacokinetic interactions.

Methods

Chemicals and reagents

DMXAA, 2,5-dimethylxanthenone-4-acetic acid (as internal standard), N-[2-(dimethylamino)-ethyl]acridine-4-carboxamide (DACA), and amsacrine were synthesized in the Auckland Cancer Society Research Centre (ACSRC) as described [13, 14]. DMXAA was protected from light exposure to avoid degradation [15]. Authentic DMXAA-G and 6-OH-MXAA were isolated and purified by a solid phase extraction method from the bile and urine of rats treated with DMXAA. Both metabolites had a purity of 99% as determined by h.p.l.c, and their structure was confirmed by mass spectrometry and [¹H]-nuclear magnetic resonance [11]. Daunorubicin, 5-fluorouracil, paclitaxel, cisplatin, tirapazamine, irinotecan, methotrexate, melphalan, 6-thioguanine, 6-mercaptopurine, cyclophosphamide, folic acid, vinblastine, vincristine and 6-methylguanine were purchased from Sigma-Aldrich Chemical Co. (Auckland, NZ). Uridine diphosphate glucuronic acid (UDPGA) and NADPH were purchased from Roche Diagnostics NZ Ltd. (Auckland, NZ). All other reagents were of analytical or h.p.l.c grade as appropriate.

Preparation of human liver microsomes

Human liver samples were obtained under strict ethical conditions from donors, and were stored at −80 °C prior to use. Histological examination of the resected livers ensured the use of healthy liver tissue. Relevant details of the donors have been described elsewhere [12]. Ethical approval from the Northern New Zealand Research Ethics Committee and written informed consent for liver tissue to be used for research was obtained. Liver microsomes were prepared by differential centrifugation as described [16] and microsomes were stored at −80 °C until used. Microsomal fractions used in this study were from HL6, HL7, HL8, HL12, HL13 and HL14 from our human liver bank. Microsomal protein concentration was determined by the bicinchoninic acid method [17]. The CYP content was determined as described [18].

In vitro metabolic inhibition studies

The effects of a number of anticancer drugs (100 and 500 µm) on DMXAA glucuronidation and 6-methylhydroxylation in human liver microsomes were investigated using optimized incubation conditions [12]. Typical incubations (total volume=200 µl, in triplicate) for DMXAA glucuronidation contained liver microsomal protein (0.1 mg ml⁻¹, pooled from HL6, HL7 and HL8), 10 mm UDPGA, 5 mm MgCl₂, 0.1 mg ml⁻¹ d-saccharic acid 1,4-lactone, Brij 58 (0.1–0.25 : 1, ratio of Brij 58 over microsome, w/w), inhibitor (100 and 500 µm), and DMXAA in 0.1 m phosphate buffer (pH 6.8). d-Saccharic acid 1,4-lactone was used to inhibit the activity of β-glucuronidase in microsomes. Typical incubations (total volume=200 µl, in triplicate) for 6-methylhydroxylation contained 1 mg ml⁻¹ liver microsomal protein (from three human livers, HL12, HL13, and HL14), 5 mm MgCl₂, 0.5 mm NADPH, inhibitor (100 and 500 µm), and DMXAA in 0.1 m phosphate buffer (pH 7.4). The concentrations of DMXAA were 100 µm for glucuronidation, and 25 µm for 6-methylhydroxylation (the corresponding apparent K_m values for each metabolic pathway). Pooled human liver microsomes were used for DMXAA glucuronidation as this pathway is catalysed by multiple UGT enzymes (UGT1A9/UGT2B7) [10], and the glucuronidation activity for DMXAA was similar between human livers; whereas CYP1A2 was responsible for DMXAA 6-methylhydroxylation, and significant interindividual variation in the activity was observed [12]. The reactions were initiated by the addition of NADPH or UDPGA as appropriate. Pre-incubations were also performed in duplicate in the presence of inhibitor and cofactor (NADPH or UDPGA) for 0 or 15 min prior to the addition of DMXAA at 37 °C in a shaking water-bath. Incubations were stopped by cooling on ice and adding 2 volumes of an ice-cold acetonitrile: methanol mixture (3 : 1, v/v) containing 2 µm internal standard, and vortexing vigorously. Mixtures were centrifuged (3000 g for 10 min) to remove the precipitated microsomal protein. The supernatant was removed, evaporated under nitrogen, and the residue reconstituted with mobile phase for injection onto the h.p.l.c. All anticancer drugs were dissolved in dimethyl sulphoxide (DMSO), which was used at a final concentration of 1% (v/v) in incubations. DMSO reduced the rate of DMXAA hydroxylation by 22%, but had no significant effect on DMXAA glucuronidation. Each drug was also incubated with microsomes and UDPGA or NADPH in the absence of DMXAA to identify any chromatographic peaks which might interfere with the measurement of DMXAA-G or 6-OH-MXAA.

For those drugs showing significant inhibitory effects, further inhibition kinetic studies were performed to determine the mechanism of inhibition and the apparent K_i values. To construct Dixon plots, DMXAA (25–100 µm for glucuronidation; 6.25–25 µm for 6-methylhydroxylation) was incubated at 37 °C with human liver microsomes in the presence of inhibitors. The inhibitor concentrations used were 50–400 µm for amsacrine, for 62.5–500 µm for vinblastine and vincristine, for 37.5–300 µm daunorubicin, and 0.625–5 µm for DACA.

High performance liquid chromatography (h.p.l.c)

The determination of DMXAA-G and 6-OH-MXAA has been described previously [19]. Briefly, the h.p.l.c system consisted of a solvent delivery system, a Model SF250 fluorescence detector (excitation and emission wavelength, 345 nm and 409 nm, respectively), a Model 460 autosampler, and a Model D450 data processing system (All from Kontron Instrument Co., Milan, Italy). A Luna C18 guard column and a 5-µm Spherex C18 analytical column (150 × 4.6 mm; Phenomenex) were used. The mobile phase was acetonitrile: 10 mm ammonium acetate buffer (24 : 76, v/v, pH 5.0) at a flow rate of 2.5 ml min⁻¹. The difference between the theoretical and measured concentration, and the coefficient variation, were less than 15% at the low quality control (QC) concentration (0.5 µm), and less than 10% at the medium (2.5 µm) and high (10 µm) QC concentrations. The limit for the determination of DMXAA-G and 6-OH-MXAA was 0.25 µm for a 75 µl injection volume. Assay specificity was indicated by the absence of interfering chromatographic peaks in microsomal samples and in incubations with anticancer drugs.

Prediction of drug interaction based on in vitro data

For the inhibition of a metabolic pathway by a drug, the degree of inhibition (R) can be calculated by the following equation 1 [20]:

1

where [I] is the unbound concentration of the inhibitor; K_i is the inhibition constant; and [S] is the unbound therapeutic substrate concentration.

As DMXAA has a low plasma clearance (2–5 ml min⁻¹ kg⁻¹) in cancer patients [21], the degree of change (R_c) in the plasma area under the concentration-time curve (AUC) by the inhibiting drug can be calculated as follows [22]:

2

where f_h is the fraction of hepatic clearance in total clearance, and f_m the fraction of metabolic pathway in hepatic clearance. Approximated f_h and f_m for DMXAA were estimated from the urine of a patient treated with DMXAA, where 2.4%, 35.9% and 5.5% was excreted as unchanged DMXAA, DMXAA-G and 6-OH-MXAA, respectively [23]. Thus the approximate f_h, f_m (glucuronidation) and f_m (6-methylhydroxylation) were 97.6%, 86.7% and 13.3%, respectively, assuming that biliary excretion of DMXAA in humans is similar to that in the urine, as was found in the rat and rabbit [9, 11]. The unbound fraction of DMXAA (2000 µm) in human plasma was 0.123 as determined by ultrafiltration followed by h.p.l.c [24].

Data analysis

Data were expressed as mean ± s.d. The initial estimate of the apparent K_i values and the nature of inhibition were obtained from Dixon plots, where the apparent K_i was given by the intersection point of the linear regression lines for data sets of 1/v against the concentration of inhibitor. Several inhibition models (competitive, uncompetitive, and mixed inhibition), represented by the following equations, were fitted to the data and compared using Prism 3.0 program (Graphpad Software Co., CA).

3

4

5

where v is the rate of metabolism, V_max, is the maximum velocity, K_m, the Michaelis-Menten constant, [S], the substrate concentration, [I], the inhibitor concentration, K_i; the apparent inhibition constant and subscripts c and u represent competitive and uncompetitive inhibition. The appropriate model was chosen by comparing and reviewing the relative residuals and the standard error of the parameter estimates. The significance of differences in the formation of DMXAA metabolites was assessed by Student's unpaired t-test. Differences were considered statistically significant when P values were < 0.05.

Results

Effects on anticancer drugs on DMXAA metabolism in vitro

The effects of various anticancer drugs on DMXAA glucuronidation and 6-methylhydroxylation by human liver microsomes are reported in Table 1. Vinblastine, vincristine and amsacrine at 500 µm significantly (P < 0.05) inhibited DMXAA glucuronidation (K_i = 319, 350 and 230 µm, respectively), but not 6-methylhydroxylation in human liver microsomes. Daunorubicin and DACA at 100 and 500 µm showed significant (P < 0.05) inhibition of DMXAA 6-methylhydroxylation (K_i = 131 and 0.59 µm, respectively), but not glucuronidation. Figure 2 illustrates the Dixon plots and the estimated apparent K_i values for inhibition of DMXAA metabolism by various anticancer drugs. The mechanism of inhibition of either DMXAA glucuronidation or 6-methyl-hydroxylation by these anticancer drugs was competitive, as indicated by the smallest mean sum of squares compared with noncompetitive and mixed models. Other drugs such as 5-fluorouracil, paclitaxel, cisplatin, irinotecan, tirapazamine and methotrexate exhibited little or negligible inhibition of the metabolism of DMXAA. Pre-incubation of microsomes with the anticancer drugs (100 and 500 µm) did not enhance their inhibitory effects on DMXAA metabolism.

Table 1.

Effects of various anticancer drugs on DMXAA glucuronidation and 6-methylhydroxylation by human liver microsomes¹

	Glucuronidation (% enzyme activity remaining)		6-Methylhydroxylation (% enzyme activity remaining)
Drugs	at 100 µm inhibitor	at 500 µm inhibitor	100 µm inhibitor	500 µm inhibitor
Folic acid	73 ± 8	70 ± 2	102 ± 5	111 ± 12
Vincristine	74 ± 14	46 ± 8	101 ± 0	67 ± 14
Amsacrine	82 ± 12	27 ± 1	102 ± 19	91 ± 5
Daunorubicin	84 ± 19	63 ± 7	59 ± 2	15 ± 2
Paclitaxel	88 ± 7	78 ± 5	98 ± 8	95 ± 6
Cisplatin	91 ± 5	67 ± 2	100 ± 10	86 ± 7
DACA	96 ± 1	71 ± 16	17 ± 2	6 ± 1
Irinotecan	97 ± 7	71 ± 4	95 ± 9	92 ± 5
Tirapazamine	97 ± 7	95 ± 7	93 ± 3	86 ± 4
Vinblastine	99 ± 4	40 ± 0	104 ± 9	113 ± 1
6-Thioguanine	100 ± 2	98 ± 7	89 ± 2	64 ± 1
Cyclophosphamide	101 ± 29	74 ± 12	95 ± 2	85 ± 4
6-Mercaptopurine	102 ± 5	96 ± 10	97 ± 8	94 ± 6
5-Fluorouracil	104 ± 11	103 ± 7	97 ± 7	96 ± 10
Methotrexate	108 ± 8	88 ± 2	109 ± 8	89 ± 14
Melphalan	113 ± 15	104 ± 15	99 ± 7	101 ± 5

¹For glucuronidation, DMXAA (100 µm, ≈ K_m) was incubated for 20 min at 37 °C with 0.1 mg ml⁻¹ Brij 58-activated liver microsomes from pooled HL6, HL7 and HL8 in the presence of various anticancer drugs. For 6-methylhydroxylation, DMXAA (25 µm, ≈ K_m) was incubated for 40 min at 37 °C with 1.0 mg ml⁻¹ liver microsomes from HL12, HL13 and HL14 in the presence of various anticancer drugs. Results are the mean ± s.d. of at least three determinations from pooled human liver microsomes for glucuronidation and mean ± s.d. of data from three individual livers for 6-methylhydroxylation.

Figure 2.

Dixon plots for the inhibition of DMXAA glucuronidation by amsacrine (a), vinblastine (b) and vincristine (c), and 6-methylhydroxylation by daunorubicin (d) and DACA (e) in human liver microsomes. Concentrations of DMXAA are shown alongside individual plots. Each data point is the mean of at least three determinations.

Quantitative prediction of DMXAA–drug interaction in vivo

The predicted percentage inhibition (R) and percentage increase in AUC of DMXAA (R_c) that might be caused by coadministration of various anticancer drugs based on our in vitro studies are shown in Table 2. An increase of 6% in the plasma AUC was predicted with DACA, which is of insignificant magnitude to be of clinical relevance. With amsacrine, daunorubicin, vinblastine and vincristine, there was no predicted in vivo inhibition of DMXAA metabolism and change in plasma AUC of DMXAA. Predictions of in vivo DMXAA–drug interactions in patients based on these in vitro data indicated that all the anticancer drugs tested except DACA appear unlikely to alter the pharmacokinetics of DMXAA.

Table 2.

Predicted DMXAA–anticancer drug interactions in vivo.

Anti-cancer drugs	Apparent Ki value (µm)	Metabolic pathway of inhibition	[I] (µm)	Predicted R (%)	Rc (%)
Daunorubicin	131	Hydroxylation	0.01–1.001	0	0
DACA	0.59	Hydroxylation	0.21–18.52	0.1–6.4	0.1–6.6
Vinblastine	319	Glucuronidation	0.01–0.173	0	0
Vincristine	350	Glucuronidation	0.01–0.083	0	0
Amsacrine	230	Glucuronidation	0.71–3.704	0	0

¹Paul et al. [28]; Galettis et al. [29]. Approximately 63% of daunorubicin binds to human plasma [30].

²Kestell et al. [14]. Approximately 97% of DACA binds to human plasma [14].

³Nelson et al. [31]. Approximately 75% of vinblastine and vincristine binds to human plasma proteins [32].

⁴Petros et al. [33]. Approximately 97% amsacrine binds to human plasma [34].

Discussion

There is an increasing interest in the drug–drug interaction studies using human liver microsomes or recombinant human drug-metabolizing isozymes (mainly CYPs) in the early stages of drug development, as this may help identify possible drug interactions and avoid drug toxicity in vivo [22]. Our results indicate that of the anticancer drugs investigated, only vinblastine, vincristine, amsacrine, DACA and daunorubicin at supratherapeutic concentrations (≥ 100 µm) caused a significant inhibition of either glucuronidation or 6-methylhydroxylation of DMXAA with an apparent K_i of 0.59–350 µm. The inhibition by these anticancer agents appears to be competitive, with no involvement of mechanism-based inhibition. When interpreting the clinical relevance of these inhibition studies, the unbound concentration of both the substrate and inhibitors utilized in vitro, and those that exist in vivo, must be taken into consideration. The concentrations for DMXAA (25–140 µm) used in this study have been observed as unbound plasma concentrations in patients in a Phase I trial [21].

Numerous models have been used to correlate in vitro with in vivo drug interactions, with some success [22]. A model for low hepatic clearance drugs administered by intravenous injection was used to extrapolate our in vitro drug interaction data to the in vivo state. In this model, it is assumed that both inhibitor and DMXAA are metabolized only in the liver, with possible competitive inhibition, and that there are only two metabolic pathways for DMXAA. Our results indicate that none of the anticancer drugs investigated would cause a clinically significant reduction in DMXAA plasma clearance. However, these conclusions must be regarded with caution due to the inherent assumptions within the model. In addition, nonspecific microsomal binding may be another important factor influencing the accuracy of the extrapolation from the in vitro to in vivo state [25–27]. Non-specific microsomal binding of substrates will result in a higher apparent K_m value determined from the total (added) concentration, rather than the unbound concentration. However, DMXAA as a weak acid is predicted to have insignificant nonspecific microsomal binding, since the microsomal membrane has a net negative charge, and acidic drugs such as caffeine, tolbutamide and naproxen do not bind significantly to it [27]. Furthermore, the model used in this study does not take into account inhibition of metabolism by indirect mechanisms such as alterations in cytokine and/or nitric oxide concentrations, or changes in other drug disposition processes, such as absorption, renal metabolism, or active transport mechanisms.

According to equation 2, the contribution to overall clearance of a particular metabolic pathway subject to inhibition, is an important determining factor in the prediction of in vivo drug interactions [22]. Our in vitro and in vivo studies have indicated that glucuronidation is the major metabolic pathway for DMXAA, whereas 6-methylhydroxylation is a minor metabolic pathway [11, 23]. The intrinsic clearance ratio of glucuronidation: 6-methylhydroxylation was 2.6 from our in vitro human studies [Paxton et al. personal communication]. Previously Miners et al. [10] screened various drugs, mainly substrates of UGTs, for their potential to interact with DMXAA using human liver microsomes and cDNA-expressed UGTs. Significant inhibition of DMXAA glucuronidation was observed with diclofenac, epirubicin, indomethacin, R,S-ketoprofen, lorazepam, S-naproxen, oxazepam, and temazepam with apparent K_i values ranging from 10 to 318 µm. Recently, our mouse studies indicated that diclofenac at 100 mg kg⁻¹ administered by intraperitoneal injection increased the plasma AUC of DMXAA by 24% [8]. This was considered to be mainly due to inhibition of DMXAA glucuronidation. DMXAA is 6-methylhydroxylated by CYP1A2 [12]. However, alterations in this relatively minor pathway of DMXAA metabolism would not be expected to have a clinically significant effect on DMXAA disposition. As none of the anticancer drugs investigated are known substrates of the UGTs, it is perhaps of no surprise that there was little inhibition of DMXAA glucuronidation by these drugs, with the exception of amsacrine, vinblastine and vincristine at high concentration.

In conclusion, despite the occurrence in vitro of inhibition of DMXAA metabolism by several anticancer drugs, such as daunorubicin, DACA and amsacrine, it appears unlikely that combinations of these drugs will result in clinically significant drug interactions. However, this does not rule out the possibility of pharmacokinetic interactions with other drugs used concurrently with this combination of anticancer drugs.