Eltrombopag, an oral thrombopoietin receptor agonist, has no impact on the pharmacokinetic profile of probe drugs for cytochrome P450 isoenzymes CYP3A4, CYP1A2, CYP2C9 and CYP2C19 in healthy men: a cocktail analysis
- 402 Downloads
It is likely that the thrombopoietin receptor agonist eltrombopag will be administered concomitantly with other medications in the treatment of thrombocytopenia. Therefore the potential for eltrombopag to interact with cytochrome P450 activity was investigated.
Twenty-four healthy men received eltrombopag 75 mg/day on days 3–9, midazolam 5 mg (a probe for CYP3A4) on days 1 and 8 and a probe cocktail on days 2 and 9 that included caffeine 100 mg (CYP1A2), flurbiprofen 50 mg (CYP2C9) and omeprazole 20 mg (CYP2C19).
Midazolam pharmacokinetic parameters were comparable before and after eltrombopag administration; geometric least squares (GLS) mean ratio (90% confidence intervals, CI) area under the curve from zero to infinity (AUC0-∞) was 1.03 (0.94,1.12) and maximum plasma concentration (Cmax) was 0.98 (0.86,1.07). Metabolic indices for other CYP isozymes were also equivalent before and after eltrombopag. GLS mean ratio (90% CI) for the paraxanthine:caffeine concentration ratio at 8 h postdose was 0.97 (0.92,1.03), for conjugated + unconjugated and unconjugated 4-hydroxy-flurbiprofen recovery in urine over 0–8 h was 0.95 (0.93,0.97) and 0.93 (0.88,0.98), respectively, and for the plasma omeprazole:5-hydroxyomeprazole concentration ratio at 2- and 3-h postdose was 1.00 (0.93,1.08) and 1.02 (0.88,1.18), respectively.
Once-daily administration of eltrombopag 75 mg for 7 days did not alter CYP3A4, CYP1A2, CYP2C9 or CYP2C19 activity in healthy volunteers.
KeywordsEltrombopag CYP1A2 CYP2C9 CYP2C19 CYP3A4
Thrombocytopenia, an abnormally low platelet count (<150,000 cells/μl) in the circulating blood, is a potentially disabling disorder associated with an increased risk of haemorrhage and prolonged bleeding. To date, the limited treatment options for thrombocytopenia have focussed on the main cytokine involved in megakaryopoiesis, thrombopoietin (TPO). Recombinant versions of TPO tested in patients undergoing chemotherapy [1, 2] were considered unsuccessful because of immunogenicity issues, and although peptide mimetics of TPO increased platelet counts in healthy volunteers and patients with idiopathic thrombocytopenic purpura (ITP) [3, 4], they are not orally bioavailable.
Eltrombopag (Promacta®, SB-497115) is a first-in-class, small-molecule, nonpeptide, orally bioavailable TPO-receptor agonist [5, 6, 7, 8, 9]. Preclinical studies have shown that eltrombopag interacts selectively with the transmembrane domain of the TPO receptor , thereby activating intracellular signal transduction pathways leading to increased proliferation and differentiation of human bone marrow progenitor cells . The ability of eltrombopag to increase platelet counts has been confirmed in healthy volunteers  and in studies in patients with ITP  and patients with thrombocytopenia in hepatitis-C-associated liver cirrhosis .
Once absorbed, eltrombopag is extensively metabolised, mainly through pathways including cleavage, oxidation and conjugation with glucuronic acid, glutathione or cysteine. The predominant route of eltrombopag excretion is via faeces (59%), with unchanged eltrombopag accounting for approximately 20% of the dose. Peak plasma eltrombopag concentrations occur between 2 and 6 h after dosing, and the plasma elimination half-life of eltrombopag is approximately 21–32 h in healthy volunteers. Specifically for the 75 mg once-daily dose used in this study, historical data for the same capsule formulation show that repeat doses of eltrombopag result in maximum plasma concentrations (Cmax) of approximately 7.3 μg/ml in healthy volunteers .
As it is likely that eltrombopag will be administered concomitantly with other medications in the treatment of thrombocytopenia, we investigated eltrombopag’s potential for interaction. An initial in vitro screen showed that eltrombopag was not an inhibitor (at concentrations up to 100 μM) of CYP1A2, CYP2A6, CYP2C19, CYP2D6, CYP2E1, or CYP3A4/5. These studies demonstrated that eltrombopag was an inhibitor of CYP2C8 and CYP2C9, with 50% inhibitory concentration (IC50) values of 24.8 μM (11 μg/ml) and 20.2 μM (8.9 μg/ml), respectively (data on file). In addition, eltrombopag did not show potential for CYP induction based on weak activation of human pregnane X receptor (PXR) at concentrations up to 10 μM (4.4 μg/ml) and no increase in the messenger RNA (mRNA) level or catalytic activity of CYP1A2, CYP2B6 or CYP3A4 in a human hepatocyte study at concentrations up to 30 μM (13.2 μg/ml). For clinical study, we selected CYP2C9, for which eltrombopag showed moderate inhibition potential in vitro, and CYPs, for which eltrombopag showed no inhibition potential in vitro, including CYP1A2, CYP2C19 and CYP3A4. Therefore, this study was conducted to assess the potential for repeat oral dosing of eltrombopag 75 mg/day for 7 days to inhibit or induce the activity of CYP1A2, CYP2C9, CYP2C19 and CYP3A4, which have all been reported as being frequently involved in drug metabolism . Although in vitro data have shown that eltrombopag inhibits CYP2C8 with a similar IC50 to that of CYP2C9, CYP2C8 was not tested in this study because there was no validated and specific CYP2C8 probe available.
Materials and methods
Brent Medical Ethics Committee, London, UK, approved this study. The study was conducted at a single centre (Hammersmith Medicines Research Unit at the Central Middlesex Hospital, London, UK) and in accordance with the Declaration of Helsinki (South Africa Revision, 1996) and the International Conference on Harmonization Guidelines for Good Clinical Practice. All subjects provided written informed consent.
Healthy nonsmoking male subjects, aged 18–45 years with a body mass index (BMI) of 19–30 kg/m2, were enrolled at a single site. The study was also open to women of nonchildbearing potential, but none was recruited before the required sample size was attained. All participants were in good health on the basis of medical history, physical examination, routine laboratory screening and electrocardiogram examination. Subjects with the following conditions were excluded: significant cardiovascular disease, history of deep vein thrombosis, thrombocytopenia or bleeding due to abnormal platelet counts or function, atrial fibrillation, mitral valve prolapse, significant heart murmur or vascular bruit and family history of myocardial infarction or stroke. Subjects were also excluded if the following were observed during screening: signs of clinically relevant abnormality, abnormal laboratory results (e.g. elevated liver function tests or C-reactive protein), elevated lipids (especially triglycerides) or elevated blood pressure (systolic >140 mmHg, diastolic >85 mmHg). Use of concomitant medications or vitamins was not permitted within 7 days of the first dose of study medication, with the exception of acetaminophen at doses of ≤2 g/day, which was permitted for up to 24 h before pharmacokinetic sampling days. Consumption of red wine, Seville oranges, grapefruit or grapefruit juice was not permitted within 14 days before the first dose of study medication and throughout the study. Caffeine-containing food and beverages were not allowed from day 1 until the last blood sample was drawn.
This open-label, repeat-dose study investigated the potential of eltrombopag 75 mg/day to inhibit or induce CYP450 activity. Probe compounds (and target CYP450 isoenzymes) included midazolam (Hypnovel®, Roche Products Limited, Welwyn Garden City, UK) 5 mg (CYP3A4) and a cocktail of caffeine (ProPlus®, Bayer HealthCare, Newbury, UK) 100 mg (CYP1A2), flurbiprofen (Froben®, Abbott Laboratories Limited, Maidenhead, UK) 50 mg (CYP2C9) and omeprazole (Losec MUPS®, AstraZeneca UK Limited, Luton, UK) 20 mg (CYP2C19) .
Subjects were admitted the day before the first dose of investigational product and resided at the unit for 10 days. Subjects received eltrombopag on days 3–9, midazolam on days 1 and 8, and the cocktail of probe compounds on days 2 and 9. Study medication was administered at approximately the same time each day in the morning between 09:00 and 10:50 following an overnight fast of at least 8 h. Study staff checked the mouths of study subjects to ensure compliance. Standard meals were provided approximately 1 h postdose on days 3–7, and at 4–5 h and 9–10 h postdosing on days 1–10. Blood samples were taken by single-stick sampling apart from day 1 and day 8 when they were taken via an indwelling catheter kept patent with saline solution (0.9%).
Cytochrome CYP3A4 assay and pharmacokinetic analysis
Whole blood (5 ml) was collected into ethylenediamine tetraacetic acid (EDTA) anticoagulant tubes at predose and 0.5, 1, 2, 3, 4, 6, 9, 12 and 24 h postdose on days 1 and 8. Plasma samples were analysed for midazolam by York Bioanalytical Solutions, UK, using a validated analytical method based on protein precipitation [using acetonitrile containing triazolam (Sigma-Aldrich Ltd, UK) as the internal standard], followed by high-performance liquid chromatography tandem mass spectrometry (HPLC-MS/MS) analysis using a TurboIonSpray interface and multiple reaction monitoring (m/z 326/244 for midazolam and m/z 343/308 for triazolam). The lower limit of quantification (LLQ) was 0.5 ng/ml, and the higher limit of quantification (HLQ) was 500 ng/ml, using a 100-µl aliquot of human plasma. Method precision and accuracy were determined by replicate analysis of control human plasma samples containing midazolam at six concentrations (from 0.5 ng/ml to 1,000 ng/ml). The imprecision and inaccuracy within assays were less than 20% and 18%, respectively, whereas those between assays were less than 15% and 8%, respectively.
Midazolam Cmax was obtained from concentration-time data, as was the time to Cmax (tmax). Plasma concentration-time data were analyzed by noncompartmental methods using the WinNonlin computer software (version 4.1; Pharsight Corp, USA). Area under the plasma concentration-time curve (AUC) was estimated from the time of dosing to time ‘t’, where ‘t’ was the time of the last quantifiable concentration (AUC0-t), and estimates of AUC from the time of dosing extrapolated to infinite time (AUC0-∞) were determined as the sum of AUC0-t and the last observed concentration divided by the terminal elimination rate constant (λz). Areas under the curve were calculated using a combination of linear (for ascending concentrations) and logarithmic (for descending concentrations) trapezoidal methods. Apparent oral clearance (CL/F) was calculated as midazolam dose/AUC0-∞. The terminal elimination rate constant was derived from the log-linear disposition phase of the concentration time curve using least-squares regression analysis with visual inspection of the data to determine the appropriate number of terminal data points to include in the regression analysis. The elimination half-life (t1/2) was calculated as ln2/λz.
Cytochrome CYP1A2 assay analysis
Whole blood (5 ml) was collected into EDTA anticoagulant tubes predose and 4 and 8 h postdose on days 2 and 9 for determination of plasma concentrations of caffeine and its metabolite paraxanthine. Analysis was conducted at York Bioanalytical Solutions using an established HPLC-MS/MS method that had been previously validated for caffeine and paraxanthine (Sigma-Aldrich Ltd) in EDTA-treated plasma (m/z 195/138 for caffeine and m/z 181/124 for paraxanthine) . For both analytes, the LLQ was 50 ng/ml and HLQ was 5,000 ng/ml, using a 100-µl aliquot of human plasma. Method precision and accuracy were determined by replicate analysis of control human plasma samples containing caffeine and paraxanthine at six concentrations (from 50 ng/ml to 10,000 ng/ml). For caffeine, the imprecision and inaccuracy within assays were less than 8% and 6%, respectively, whereas those between assays were less than 9% and 4%, respectively. For paraxanthine, the imprecision and inaccuracy within assays were less than 12% and 6%, respectively, whereas those between assays were less than 15% and 7%, respectively.
Cytochrome CYP2C19 assay analysis
Whole blood (5 ml) was collected into EDTA anticoagulant tubes at predose and 2 and 3 h postdose on days 2 and 9 for determination of plasma concentrations of omeprazole and its metabolite 5-hydroxy-omeprazole. Plasma samples were analysed by CEPHAC Europe, France, using a validated analytical method  based on solid-phase extraction with omeprazole-D3 (Cluzeau Info Labo, France) and 5-hydroxyomeprazole-D7 (ArtMolecule, France) as internal standards, followed by HPLC-MS/MS using a TurboIonSpray interface and multiple reaction monitoring (m/z 344/194 for omeprazole and m/z 360/194 for 5-hydroxyomeprazole). The LLQ and HLQ using a 100-µl aliquot of human plasma were 1 ng/ml and 1,000 ng/ml, respectively. Method precision and accuracy were determined by replicate analysis of control human plasma samples containing omeprazole and 5-hydroxyomeprazole at four concentrations (from 1.0 ng/ml to 800 ng/ml). For omeprazole, the imprecision and inaccuracy within assays were less than 19% and 6%, respectively, whereas those between assays were less than 16% and 13%, respectively. For 5-hydroxyomeprazole, the imprecision and inaccuracy within assays were less than 14% and 12%, respectively, whereas those between assays were less than 18% and 7%, respectively.
Cytochrome CYP2C9 assay analysis
Urine samples were collected for the determination of flurbiprofen and its metabolite 4-hydroxy-flurbiprofen concentrations over 0–8 h and 8–12 h postdose on days 2 and 9. Concentrations of flurbiprofen and 4-hydroxyflurbiprofen were determined for both untreated urine and urine treated with the enzyme β-glucuronidase, giving estimates of unconjugated and total (the unconjugated and its glucuronides) concentrations, respectively, of these two analytes in human urine. Urine samples were analysed by York Bioanalytical Solutions using a validated analytical method [14, 15] based on solid-phase extraction [fenoprofen calcium (LGC Promochem) as an internal standard] followed by HPLC-MS/MS analysis (m/z 243/199 for flurbiprofen and m/z 259/213 for 4-hydroxyflurbiprofen). Using a 50-µl aliquot of human urine, the LLQs were 0.25 μg/ml and 0.5 μg/ml for flurbiprofen and its metabolite, respectively. The HLQs were 50 μg/ml and 25Z μg/ml, respectively. Method precision and accuracy were determined by replicate analysis of control human urine samples containing flurbiprofen and 4-hydroxyflurbiprofen at five concentrations (from 250 ng/ml to 25,000 ng/ml for flurbiprofen and from 500 ng/ml to 50,000 ng/ml for 4-hydroxyflurbiprofen). For measurement of conjugated + unconjugated flurbiprofen concentrations, the imprecision and inaccuracy within assays were less than 11% and 9%, respectively, whereas those between assays were less than 12% and 6%, respectively. For measurement of the unconjugated flurbiprofen concentrations, the imprecision and inaccuracy within assays were less than 12% and 6%, respectively. For measurement of conjugated + unconjugated 4-hydroxyflurbiprofen concentrations, the imprecision and inaccuracy within assays were less than 9% and 3%, respectively, whereas those between assays were less than 14% and 2%, respectively. For measurement of the unconjugated 4-hydroxyflurbiprofen concentrations, both imprecision and inaccuracy within assays were less than 15%. The imprecision and inaccuracy between assays were not determined for the measurement of the unconjugated analyte.
Physical examination, blood pressure and heart rate measurements, 12-lead electrocardiogram, 1-h continuous cardiac monitoring, physical and ultrasound examination of the spleen, and blood and urine laboratory analysis were performed at the screening visit and throughout the study. Tolerability was monitored throughout the study, and adverse events (AEs) were assessed by the investigator in terms of severity, duration and relationship to investigational products.
This study was designed to estimate the effects of the eltrombopag on the CYP450 substrates. Sample size justification was based on data from recognised studies in which the largest within-subject variability for the CYP450 substrates investigated fell in the region of 81.05% (for omeprazole:5′ hydroxy-omeprazole ratio) . It was estimated that a sample size of 24 subjects would be sufficient to detect a >40% increase (administration of eltrombopag plus CYP450 probe versus CYP450 probe alone). This calculation was based on a two-tailed procedure and a type I error rate of 10%.
The plasma paraxanthine:caffeine concentration ratio was calculated at each time point for each treatment. The plasma paraxanthine:caffeine concentration ratio at the 8-h time point was used as the CYP1A2 metabolic index for statistical analysis . The plasma omeprazole:5-hydroxyomeprazole concentration ratio was calculated at each time point for each treatment. The plasma omeprazole:5-hydroxyomeprazole concentration ratio at the 2- and 3-h time points were used as the CYP2C19 metabolic index for statistical analysis . The urine 4-hydroxy-flurbiprofen recovery ratio was calculated as urine 4-hydroxy-flurbiprofen concentration/ (urine 4-hydroxy-flurbiprofen concentration + flurbiprofen concentration) over the 0–8, 8–12 and 0–12 h collection time points; the 4-hydroxy-flurbiprofen recovery ratio calculation was done separately for unconjugated and conjugated + unconjugated concentrations. The urine 4-hydroxy-flurbiprofen recovery ratio over the 0–8 h urine collection interval was used as the CYP2C9 metabolic index for statistical analysis [14, 15]. For the CYP3A4 assessment, plasma midazolam AUC0-∞, Cmax, and CL/F were evaluated.
Following loge-transformation, the CYP metabolic indices were analysed separately using an analysis of variance (ANOVA) with terms for subject and treatment included in the model (SAS version 6.12, Cary, NC, USA). Point estimates of the treatment mean differences and associated 90% confidence intervals (CI) were then exponentially back-transformed for each CYP450 metabolic index separately to provide point and 90% CI estimates for the ratio of eltrombopag plus CYP450 probe:CYP450 probe. Pharmacokinetic measurements were considered equivalent if the 90% CI for the mean ratios were within 80–125%.
Twenty-four healthy male subjects with a mean (standard deviation) age of 25.3 (4.7) years and BMI of 24.7 (2.9) mg/kg2 entered and completed the study as planned. The ethnicity of 22 of the subjects was Caucasian, one was Black and one was Asian.
Midazolam pharmacokinetics and activity of cytochrome CYP3A4
Summary of midazolam (CYP3A4) pharmacokinetic parameter estimates before and after dosing for 7 days with eltrombopag and treatment comparisons
Geometric mean (CV%)
GLS mean ratio
AUC0-∞ (ng h/ml)
1.03 (0.94, 1.12)
0.98 (0.86, 1.07)
0.95 (0.86, 1.04)
Cytochrome CYP1A2, CYP2C9, and CYP2C19 activity
Summary of CYP probe parent and metabolite concentrations and concentration ratios before and after dosing for 7 days with eltrombopag and treatment comparisons
Geometric mean (CV%)
GLS mean ratio (90% CI)
Day 9 vs day 2
Plasma caffeine (ng/ml)
Plasma paraxanthine (ng/ml)
8-h paraxanthine:caffeine ratio (CYP1A2)
0.97 (0.92, 1.03)
Urine conjugated + unconjugated flurbiprofen (μg/ml)
Urine conjugated + unconjugated 4-hydroxy-flurbiprofen (μg/ml)
Conjugated + unconjugated 0–8 h 4-hydroxy-flurbiprofen recovery ratioa (CYP2C9)
0.95 (0.93, 0.97)
Urine unconjugated flurbiprofen (μg/ml)
Urine unconjugated 4-hydroxy-flurbiprofen (μg/ml)
Unconjugated 0–8 h 4-hydroxy-flurbiprofen recovery ratioa (CYP2C9)
0.93 (0.88, 0.98)
Plasma omeprazole : 5′ hydroxy-omeprazole ratio (CYP2C19)
1.00 (0.93, 1.08)
1.02 (0.88, 1.18)
Safety and tolerability
There were no serious AEs. Overall, 11 subjects (46%) reported a total of 22 AEs during the study; headache was the most frequently reported event (four subjects). Three subjects experienced events after receiving midazolam alone (two episodes of drowsiness and one of headache), and four subjects experienced events after receiving midazolam and eltrombopag (two episodes of headache and two of nausea). There was one mild case of thrombophlebitis, which occurred after the subject had received eltrombopag 75 mg for 6 days. He had also received all the probe drugs during this time. The thrombophlebitis was judged not to be related to eltrombopag but was considered to be a local response to an indwelling cannula. No clinically significant changes in laboratory or cardiovascular safety parameters were observed during the study.
Thrombocytopenia is a complex disease, the aetiology of which can be associated with various disease states. As such, patients are often exposed to multiple therapies. This study was conducted to explore the potential for eltrombopag to interact with common metabolic pathways, including CYP1A2, CYP2C9, CYP2C19 and CYP3A4 probe substrates. The study evaluated an eltrombopag dose of 75 mg once daily because this is the highest dose tested and currently approved for use in patients with ITP. Although plasma eltrombopag concentrations were not measured in this study, historical data for the same capsule formulation show that repeat doses of eltrombopag 75 mg once daily result in a Cmax of approximately 7.3 μg/ml in healthy male volunteers . In this study, we found that eltrombopag 75 mg/day for up to 7 days had no effect on the metabolic indices of probe substrates for CYP3A4, CYP1A2, CYP2C9 and CYP2C19, suggesting that eltrombopag did not inhibit or induce these enzymes at the doses studied. Post- versus pre-eltrombopag treatment ratios exhibited low variability, with the 90% CIs of the ratios falling between 0.88 and 1.18. This was well within the prespecified range of 0.80−1.25 for pharmacokinetic equivalence. Therefore, changes in the point estimates of all isoenzyme indices (−5% to 7%) were not clinically relevant.
Subjects were not genotyped prospectively, and so it was not possible to determine whether or not our study population included CYP2C9 and/or CYP2C19 poor metabolisers. Inclusion of such subjects could skew the observed data leading to an underestimation of the possible drug interactions. However, given the low frequency of poor metabolisers for CYP2C9 (<1%) and CYP2C19 (2−6%) among Caucasian populations such as those studied here [17, 18], it is reasonable to assume that the study was conducted primarily in subjects who were intermediate or extensive metabolisers and that our data, which leads us to conclude that there was no interaction, remains valid even in the absence of genotyping data. Our data (Fig. 1) suggests that two of our subjects may have been poor CYP2C19 metabolisers, as their omeprazole:5-hydroxyomeprazole concentration ratios were much higher than values expressed by the other subjects. Nevertheless, across all of the study subjects, there was no evidence of any change in the omeprazole:5-hydroxyomeprazole concentration ratio between omeprazole alone and omeprazole plus eltrombopag treatments.
Although the combination of caffeine, omeprazole and flurbiprofen has not been fully validated as a cocktail, elements have previously been validated in combination. The use of caffeine and omeprazole in combination with other CYP probes has been characterised and validated in the Karolinska, Cooperstown, and Inje Cocktails where they were confirmed not to have any pharmacokinetic interactions between the probes, [13, 16, 19]. The grouping of caffeine and flurbiprofen, used in combination with other CYP probes, has also been validated, with no pharmacokinetic interactions being observed among any of the probes . In contrast, the combination of flurbiprofen with omeprazole (with or without caffeine) that we employed in this study has not been validated as a CYP probe combination. It was our assumption that no pharmacokinetic interaction would occur between flurbiprofen and omeprazole, as the two compounds are known to be metabolised by different cytochrome isoenzymes; flurbiprofen is metabolised by CYP2C9 and omeprazole by CYP2C19 and CYP3A4. In addition, flurbiprofen has not shown inhibition of CYP2C19 in other probe cocktail studies , nor has omeprazole shown inhibition of CYP2C9 in other probe cocktail studies . This lack of validation places certain limitations on the interpretation of our findings. Interactions can occur even when drugs are metabolised by different pathways; for example, quinidine inhibits CYP2DG, even though it is not metabolised by this isoenzyme . However, the 0–8 h 4-hydroxy-flurbiprofen recovery ratio of 0.66 observed in our study when flurbiprofen was administered with omeprazole and caffeine (without eltrombopag) on day 2 is consistent with values reported in other studies where flurbiprofen was administered alone  or in combination with other probe drugs . The 2-h plasma omeprazole:5′ hydroxy-omeprazole ratio of 0.738 in this study was within the wide range of values reported in other studies when omeprazole was given alone or in combination with other probe substrates [19, 21, 22]. Although these consistencies with previously published studies do not eliminate the need for a validation study of the flurbiprofen/omeprazole combination, they appear to support our assumption of a lack of any detectable pharmacokinetic interactions between these two compounds.
The CYP3A isozymes contribute to the disposition of more than 60 therapeutically important drugs , account for up to 25% of the total CYP present in the adult human liver  and are the major CYP isoform in the human intestine . Midazolam is a recognized marker of the isoenzyme CYP3A4, and its clearance has been studied in detail previously . In our study, the pharmacokinetic profile of midazolam was similar before and after administration of eltrombopag, confirming that eltrombopag 75 mg/day for 7 days does not alter CYP3A4 activity.
The plasma paraxanthine:caffeine ratio used in this study allows for a simple (limited sampling) approach to characterizing the CYP1A2 isoenzyme and has been previously confirmed as a valid marker of CYP1A2 activity . One issue with this method is the potential for subjects to inadvertently be exposed to caffeine. In this study, three subjects had detectable levels (≥50 ng/ml) of caffeine (and two subjects had predose levels of paraxanthine) in the plasma prior to dosing on day 2, and one subject had detectable levels of caffeine and paraxanthine prior to dosing on day 9. Nevertheless, these values were all near the LLQ and therefore did not affect the overall outcome of the investigation. There was no change in the plasma paraxanthine:caffeine concentration ratio at 8 h after dosing when eltrombopag was coadministered, confirming that eltrombopag 75 mg/day for 7 days does not alter CYP1A2 activity.
Coadministration of eltrombopag with flurbiprofen resulted in a small reduction in the 4-hydroxy-flurbiprofen recovery ratio (GLS mean ratio: 0.95; 90% CI: 0.93, 0.97), where the upper limit of the 90% CI was <1.0, but was within the prespecified criteria (90% CI within 0.80–1.25) for no effect of eltrombopag on CYP2C9 activity. This change was considered to be not clinically relevant. Interestingly, there was a change in the amount of flurbiprofen and 4-hydroxy-flurbiprofen eliminated in the urine, where more unconjugated, less conjugated, and less conjugated + unconjugated forms were eliminated. These changes in the urinary elimination profile of flurbiprofen and 4-hydroxy-flurbiprofen suggest that eltrombopag inhibited glucuronidation of both the parent and the metabolite, which is consistent with in vitro findings where eltrombopag showed inhibition of multiple uridine diphosphate-glucuronosyltransferase (UGT) enzymes.
Assuming an average Cmax value of 7.3 μg/ml for the eltrombopag 75 mg once-daily regimen, the Cmax is below the in vitro IC50 for inhibition of CYP2C8 (24.8 μM, 11 μg/ml) and CYP2C9 (20.2 μM, 8.9 μg/ml). In this study, which was conducted primarily in white men, eltrombopag had no effect on CYP2C9 activity. The impact of higher eltrombopag concentrations—such as those observed in women or east-Asian subjects whose Cmax values are 30−70% higher than those observed in white men—on CYP2C8 or CYP2C9 substrates is not known.
At the time this study was conducted, a validated and specific CYP2C8 probe was not available; therefore, the effect of eltrombopag on CYP2C8 could not be assessed. Although CYP2C8 was not evaluated, the in vitro potency for inhibition of CYP2C8 is similar to that for inhibition of CYP2C9 (24.8 μM and 20.2 μM, respectively). Therefore, based on the lack of in vivo interaction with a CYP2C9 probe substrate in this study, eltrombopag is not expected to interact with CYP2C8 substrates in vivo.
The clinical results are consistent with in vitro CYP inhibition studies, where eltrombopag either showed no inhibition or was inhibitory at relatively high eltrombopag concentrations [IC50 for CYP2C8 of 24.8 μM (11 μg/ml) and IC50 for CYP2C9 of 20.2 μM (8.9 μg/ml) for CYP2C9]. The clinical results are consistent with in vitro CYP induction studies, where eltrombopag showed no induction potential based on human PXR at concentrations up to 10 μM (4.4 μg/ml), and based on mRNA levels and CYP activity in human hepatocytes at concentrations up to 30 μM (13.2 μg/ml). The clinical results are also consistent with repeat-dose eltrombopag studies in rats and dogs, where doses of 30−40 mg/kg per day for 14 days did not cause notable changes in hepatic microsomal total CYP levels or the activities of CYP1A, CYP2B, CYP2E and CYP4A.
Eltrombopag demonstrated an excellent tolerability profile, and no major safety concerns were noted. The most frequently reported AEs were headache and gastrointestinal events, but gastrointestinal events are frequently reported with omeprazole and flurbiprofen, so these agents may have been responsible for these episodes. No severe AEs were reported, and there were no discontinuations.
In conclusion, once-daily oral administration of eltrombopag 75 mg for 7 days did not alter the activity of the CYP450 isoenzymes CYP3A4, CYP1A2, CYP2C9 and CYP2C19 in healthy volunteers.
The authors thank the volunteers and staff who participated in this study, as well as Niche Science & Technology Ltd., who provided medical writing services on behalf of GlaxoSmithKline Inc.
- 4.Kuter D, Bussel J, George J et al (2006) Long-term dosing of AMG 531 in thrombocytopenic patients with immune thrombocytopenic purpura: 48-week update. Blood 108: Abstract 476Google Scholar
- 6.Sellers T, Hart T, Semanik M, Murthy K (2004) Pharmacology and safety of SB-497115-GR, an orally active small molecular weight TPO receptor agonist, in chimpanzees, rats and dogs. Blood 104:2063aGoogle Scholar
- 19.Chainuvati S, Nafziger AN, Leeder JS, Gaedigk A, Kearns GL, Sellers E, Zhang Y, Kashuba AD, Rowland E, Bertino JS Jr (2003) Combined phenotypic assessment of cytochrome p450 1A2, 2C9, 2C19, 2D6, and 3A, N-acetyltransferase-2, and xanthine oxidase activities with the "Cooperstown 5+1 cocktail". Clin Pharmacol Ther 74:437–447CrossRefPubMedGoogle Scholar
- 21.Streetman DS, Bleakley JF, Kim JS, Nafziger AN, Leeder JS, Gaedigk A, Gotschall R, Kearns GL, Bertino JS Jr (2000) Combined phenotypic assessment of CYP1A2, CYP2C19, CYP2D6, CYP3A, N-acetyltransferase-2, and xanthine oxidase with the ‘Cooperstown cocktail’. Clin Pharmacol Ther 68:375–383CrossRefPubMedGoogle Scholar
- 22.Yeh RF, Gaver VE, Patterson KB, Rezk NL, Baxter-Meheux F, Blake MJ, Eron JJ Jr, Klein CE, Rublein JC, Kashuba AD (2006) Lopinavir/ritonavir induces the hepatic activity of cytochrome P450 enzymes CYP2C9, CYP2C19, and CYP1A2 but inhibits the hepatic and intestinal activity of CYP3A as measured by a phenotyping drug cocktail in healthy volunteers. J Acquir Immune Defic Syndr 42:52–60PubMedGoogle Scholar