Pharmacokinetic interactions among imatinib, bosentan and sildenafil, and their clinical implications in severe pulmonary arterial hypertension

Abstract

Aims

This study characterized the population pharmacokinetics (PK) of imatinib in patients with severe pulmonary arterial hypertension (PAH), investigated drug–drug interactions (DDI) among imatinib, sildenafil and bosentan, and evaluated their clinical implications.

Methods

Plasma concentrations of imatinib, bosentan and sildenafil were collected in a phase III study and were used to characterize the PK of imatinib in this population. DDIs among the three drugs were quantified using a linear mixed model and log-transformed drug concentrations.

Results

The population mean estimates of apparent clearance (CL/F) and volume (V/F) were 10.8 l h^–1 (95% CI 9.2, 12.4 l h^–1) and 267 l (95% CI 208, 326 l), respectively. It was estimated that sildenafil concentrations increased, on average, by 64% (95% CI 32%, 103%) and bosentan concentrations by 51% (95% CI 12%, 104%), in the presence of imatinib. Despite increased concentrations of co-medications, treatment differences between imatinib and placebo for change in 6 min walk distance and pulmonary vascular resistance were relatively constant across the entire concentration range for sildenafil and bosentan. Overall, higher concentrations of imatinib and bosentan were not associated with increasing liver enzymes (serum glutamic oxaloacetic transaminases [SGOT]/serum glutamic-pyruvic transaminase [SGPT]).

Conclusions

Population PKs of imatinib in patients with severe PAH were found comparable with those of patients with chronic myeloid leukemia. Imatinib was found effective regardless of the co-medications and showed intrinsic efficacy beyond merely elevating the concentrations of the co-medications due to DDIs. There was no evidence of increased risk of liver toxicity upon co-administration with bosentan.

What is Already Known about this Subject

• Population PK of imatinib have been characterized in patients with chronic myeloid leukemia (CML) and gastrointestinal stromal tumours (GIST).

• Imatinib improved exercise capacity and haemodynamics in patients with advanced PAH who remain symptomatic on at least two drugs of the currently available drug classes.

• PK interactions between bosentan and sildenafil have been reported.

What this Study Adds

• Population PK of imatinib in severe PAH were comparable with CML.

• Bosentan and sildenafil concentrations were elevated on co-administration with imatinib.

• Imatinib has intrinsic efficacy beyond merely elevating plasma concentrations of bosentan and sildenafil.

• There was no evidence of increased liver-related toxicity with co-administration of bosentan and imatinib.

Introduction

Pulmonary arterial hypertension (PAH) is a progressive disease with poor prognosis, characterized by marked and sustained elevation of pulmonary arterial pressure (PAP), pulmonary vascular resistance (PVR) and incremental pulmonary vasculopathy that ultimately leads to premature death 1–4. The endothelin, nitric oxide and prostacyclin pathways are three physiological pathways that play an important role in the pathophysiology of PAH 5. These pathways are primarily associated with vasodilation and unregulated proliferation of pulmonary artery vascular smooth muscle cells 2.

Currently three classes of drugs have been approved for the treatment of PAH in Europe and the United States, namely, prostacyclin analogues, endothelin receptor antagonists (ambrisentan, bosentan, tadalafil and epoprostenol) and phosphodiesterase (PDE₅) inhibitors (avanafil, sildenafil, lodenafil, mirodenafil, tadalafil, vardenafil, udenafil and zaprinast). There is limited improvement in pulmonary haemodynamics as the current therapeutic interventions available, including bosentan and sildenafil, primarily target pulmonary vasodilatation, whereas PAH is a proliferative disease of small pulmonary resistance vessels. Thus, mortality remains high among patients with PAH. The median survival of patients with idiopathic or heritable PAH is <3 years 6, despite current therapy, which highlights the need for more treatment options 7–9.

Imatinib is a tyrosine kinase inhibitor (TKI) targeting the Abelson tyrosine kinase (ABL1), together with the Abelson-related kinase (ABL2) and the oncogenic BCR-ABL fusion protein, platelet-derived growth factor receptor (PDGFR)-α and β, discoidin domain receptor (DDR) and the KIT receptor 4. Imatinib is approved for the treatment of various malignant disorders including Philadelphia (Ph) chromosome-positive chronic myeloid leukemia (CML), acute lymphoblastic leukemia and gastrointestinal stromal tumours (GIST).

Imatinib has been shown to inhibit certain cytochrome P450 (CYP450)-metabolizing enzymes, and thus drug–drug interactions (DDIs) may occur. Imatinib is a substrate for CYP3A4/5 and has been shown, in vitro, to be a competitive inhibitor of CYP3A4/5, CYP2C9 and CYP2D6 4. Drugs that inhibit or induce the CYP3A4 isozyme have been shown to alter imatinib pharmacokinetic (PK) exposure 10. Sildenafil is metabolized predominantly by CYP3A4 and to a minor extent by CYP2C9 11. Bosentan is metabolized in the liver by CYP3A4 and CYP2C9 12,13. Mutual PK interactions between bosentan and sildenafil have been reported in healthy volunteers, and the dosage of each drug in a combination treatment may have to be adjusted accordingly. Thus, reciprocal PK interactions on co-administration of these three drugs warranted investigation.

IMPRES 14, a multicentre, randomized, double-blind, placebo-controlled, 24 week trial, evaluated imatinib in patients with severe PAH. The primary objective of this study included evaluation of safety and efficacy of imatinib in patients with PAH. One of the secondary objectives of this study was to assess the PK of imatinib in this patient population and the potential for interaction of imatinib on sildenafil and bosentan.

The present analysis was performed primarily to characterize the population PK of imatinib in severe PAH and to determine the PK interactions among imatinib, bosentan and sildenafil. After this analysis confirmed the presence of significant interactions among the three drugs, the effect of the intrinsic efficacy of imatinib, as well as the potential for increased risk of hepatotoxicity when it is co-administered with bosentan, was assessed.

Methods

Study participants

The study population consisted of 202 adult males and females aged ≥18 years, with a diagnosis of severe PAH, defined as those who remained symptomatic, i.e. had WHO functional class II-IV status, were on at least two PAH-specific therapies and had a baseline PVR of ≥800 dyn s cm^-5. PAH could be either idiopathic or heritable (familial or sporadic). It could be associated with (a) collagen vascular disease including systemic sclerosis, rheumatoid arthritis, mixed connective tissue diseases and overlap syndrome, (b) the use of appetite suppressants or toxic compounds or (c) congenital heart disease (≥1 year post-complete repair of atrial septal defect, ventricular septal defect or posterior descending artery).

Study design and treatments

This was a multicentre, randomized, double-blind and parallel group study in patients with PAH. After informed consent was obtained, the patients were screened to evaluate the pulmonary haemodynamics, and if found suitable, were randomized in 1: 1 ratio to receive imatinib (in 100 mg film-coated tablets) or placebo once daily. The therapy was initiated with 200 mg imatinib for 2 weeks, followed by 400 mg imatinib, if tolerated well, until 24 weeks. If 400 mg of imatinib was not tolerated, the dose was down-titrated to 200 mg. Details of the study design and main results are described elsewhere 14. The study protocol was approved by ethics committees and/or institutional review boards at each study centre and each patient provided written, informed consent to participate in the study.

PK assessments and monitoring

A sparse PK sampling approach was taken and plasma samples were typically obtained at the following times in the study: day 0 (first 200 mg once daily dose, at pre-dose and between 0.5 and 3 h post-dose), day 14 (first 400 mg once daily dose, at pre-dose and between 0.5 and 3 h post-dose), day 28 (at pre-dose and between 0.5 and 3 h post-dose) and day 168 (at pre-dose, between 0.5 and 3 h post-dose, between 3 and 6 h post-dose and between 6 and 8 h post-dose).

All samples were taken by either direct venipuncture or via indwelling cannula inserted in a forearm vein. For each plasma sample, 6 ml of blood were collected into a tube containing heparin, inverted several times and centrifuged at 1100 g for at least 10 min. Plasma samples were separated into polypropylene screw-cap tubes and frozen at –20 °C. All tubes were kept frozen until shipment. All samples were carefully packed in suitable packing material containing sufficient dry ice to keep them frozen during shipment.

The parent compound imatinib and its metabolite, CGP74588, were measured in plasma by validated liquid chromatography-mass spectrometry (HPLC-MS/MS) assay 15. The limit of quantification for imatinib and its active metabolite assays was 20 ng ml^–1. The parent drug bosentan and its major active metabolite, Ro 48-5033, were determined by validated HPLC-MS/MS assays. The limit of quantification for bosentan and its active metabolite was 1 ng ml^–1. The parent drug sildenafil and its active N-desmethyl metabolite were determined by validated HPLC-MS/MS assays. The limit of quantification for sildenafil and its active metabolite was 1 ng ml^–1.

Statistical methods

Population PK analysis

The population PK of imatinib was described by a one compartment disposition model with zero order input and inter-individual variability (IIV) on CL/F and volume of distribution (V/F). The covariate search included age, gender, race, haemoglobin, white blood cell (WBC) count and co-medications (CYP3A4 inhibitors such as sildenafil and bosentan). Two covariates, presence/absence of bosentan and haemoglobin concentrations, were included in the final model. More details on methods for covariate searching and validation of the population PK model are separately provided as an online appendix. Population PK analyses were performed with NONMEM (version VI, Icon Development Solutions).

Drug–drug interaction assessment

For graphical exploration, dose-normalized concentrations of one drug were plotted vs. absolute concentrations of the second drug in the presence or absence of the third drug. Only concentrations at steady-state for all three drugs were included, thus excluding day 0 (i.e. first dose of imatinib).

To quantify interaction effects more precisely, concentrations of sildenafil (respectively, bosentan) were log transformed and analyzed using a linear mixed model that included total daily dose and baseline concentration of sildenafil (respectively, bosentan) as continuous covariates, and indicator variables for the presence/absence of bosentan (respectively, sildenafil) and imatinib. Drug concentrations of sildenafil and bosentan at baseline were calculated as the average of the two concentrations obtained at day 0. The model further included subject as a random effect. Geometric mean ratios with 95% confidence intervals (CI) were derived to quantify the mean fold difference in sildenafil concentrations in the presence vs. absence of bosentan or imatinib. A similar approach was used to investigate the effects of bosentan and sildenafil on imatinib concentrations, with the analysis model including imatinib dose (log transformed) as a covariate and indicator variables for the presence/absence of bosentan and sildenafil. Those analyses were performed in R version 2.10.1 using the LME function (NLME library).

The results were retrospectively contrasted with those from a dedicated DDI study (unpublished data, NCT01392469; http://clinicaltrials.gov/show/NCT01392469), which was conducted at the request of health authorities. This DDI study focused on characterizing the PK effect of imatinib on the co-administered drugs bosentan and sildenafil. Changes in exposure (AUC over dosing interval) of sildenafil and bosentan, before and after administration of imatinib (200 mg for 2 weeks followed by 400 mg for 2 weeks), were used for this purpose.

Clinical implications of drug–drug interaction findings

Graphical exploration to investigate relationships between plasma concentrations of each drug to key efficacy and safety variables was undertaken.

Results

Two hundred and two adult patients with severe PAH were randomized to receive either imatinib (n = 103) or placebo (n = 99). In total, 69 patients (67%) in the imatinib treatment arm and 81 (81.8%) patients in the placebo treatment arm completed the study. We refer to the original publication 14 for additional details related to the study population.

The overall PK analysis dataset consisted of 751 measurable concentrations of bosentan, 1024 of sildenafil and 572 of imatinib collected from 191 PAH patients. Among these, 101 patients received at least one dose of imatinib, 165 received sildenafil and 114 received bosentan during the study. Details of imatinib-treated patients who received sildenafil and bosentan are presented in Table1.

Table 1.

Number of patients per drug combination in population PK dataset.

Placebo (n = 90)			Imatinib (n = 101)
	Bosentan			Bosentan
Sildenafil	No	Yes	Sildenafil	No	Yes
No	0	10	No	5	11
Yes	36	44	Yes	36	49

Population PKs of imatinib

The population PK dataset of imatinib is represented in Figure 1, where dose-normalized concentrations at steady-state are plotted vs. time after last dose administration. Superimposed on this plot is a historical prediction from the population PK model of imatinib in CML 16. As it can be seen, the curve and corresponding 90% prediction interval provide a reasonable description of the measured concentrations of imatinib in PAH, suggesting that PKs in these two different patient populations are quite similar.

Figure 1.

Comparison of dose-normalized imatinib concentrations in PAH with historical prediction in CML. Dose-normalized concentrations measured in the study (circles) were overlaid with a historical population prediction (solid line, population median; coloured area, 90% prediction interval) from the CML population model of imatinib.

This was further confirmed by fitting the same structural compartmental PK model (one compartment disposition model with zero order absorption) to the PAH dataset. Parameter estimates of the final model for imatinib are shown in Table2. The apparent clearance (10.8 l h^–1, 95% CI 9.2, 12.4 l h^–1) in the absence of bosentan was similar to values previously reported in CML (13.8 ± 0.5 l h^–1) and GIST (9.3 ± 1 l h^–1) patients 16,17. The V/F (267 l, 95% CI 208, 326 l) was similar to that in CML patients (252 ± 8 l) and approximately 45% greater than in GIST patients (184 ± 14 l). Bosentan was estimated to increase apparent imatinib clearance and V/F by 46%, corresponding to a decreased exposure (AUC) of approximately 30%.

Table 2.

Parameter estimates of the final population pharmacokinetic model for imatinib

Parameter	Estimate(standard error)
CL/F (L/h)	10.8 (0.83) IIV: CV = 43%
V/F (L)	267 (30.0) IIV: CV = 64%
Fractional increase of CL/F and V/F due to bosentan	0.46 (0.15)
Effect (power coefficient, b) of haemoglobin on V/F and CL/F, i.e. (Hb/128)b with Hb in g/L	0.49 (0.25)
Duration of first order input (h)	1.52 (0.15)

CL/F, apparent clearance of drug from plasma

CV, coefficient of variation

Hb, haemoglobin

IIV, inter-individual variability

V/F, apparent volume of distribution at steady-state

Drug–drug interactions

Sildenafil concentrations tended to be reduced on co-administration with bosentan and increased with imatinib (Figure2). The statistical analysis (Figure3) estimated that sildenafil concentrations, on average, increased by 64% (95% CI 32%, 103%) in the presence of imatinib and decreased by 44% (95% CI 30%, 56%) in the presence of bosentan. The estimated combined effect of bosentan and imatinib was null (ratio relative to no drug co-administered = 0.92, 95% CI 0.68, 1.23). Figure3 also shows the estimated increase in sildenafil exposure (AUC) after administration of imatinib (red triangle), as determined in the dedicated DDI study. This effect was consistent with the estimate from our own analysis.

Figure 2.

Effects of imatinib and bosentan on sildenafil concentrations. A, B: Log–log plot of dose-normalized sildenafil concentrations vs. (absolute) bosentan concentrations conditioned by absence (A)/presence (B) of imatinib. The median is indicated by a line summarizing dose-normalized concentrations of sildenafil in the absence (left cluster) or over a range of concentrations (right cluster) of bosentan. Bos, bosentan. C, D: Log–log plot of dose-normalized sildenafil concentrations vs. (absolute) imatinib concentrations conditioned by absence (C)/presence (D) of bosentan. The median is indicated by a line summarizing dose-normalized concentrations of sildenafil in the absence (left cluster) or over a range of concentrations (right cluster) of imatinib. Ima, imatinib.

Figure 3.

Estimated effects of imatinib and bosentan on sildenafil concentrations. Geometric mean of sildenafil concentrations with 95% confidence intervals in the presence/absence of imatinib and/or bosentan (A) and mean relative effects on sildenafil concentrations for co-administration of imatinib and/or bosentan vs. no co-administration (B). None, neither imatinib nor bosentan co-administered with sildenafil; Ima alone, imatinib co-administered; Bos alone, bosentan co-administered; Ima + Bos, imatinib and bosentan co-administered. ◂, DDI study (geometric mean ratio for AUC over dosing interval)

Increased bosentan concentrations were observed on co-administration of bosentan with sildenafil or imatinib (Figure4). The statistical analysis (Figure5) estimated that bosentan concentrations, on average, increased by 51% (95% CI 12%, 104%) in the presence of imatinib and by 53% (95% CI 9%, 115%) in the presence of sildenafil. The estimated combined effect of sildenafil and imatinib was an increase in bosentan concentrations of 132% (95% CI 46%, 269%). Figure5 also shows the estimated increase in bosentan exposure (AUC) after administration of imatinib (red triangle), as determined in the dedicated DDI study. This effect was consistent with the estimate from our own analysis.

Figure 4.

Effects of imatinib and sildenafil on bosentan concentrations. A, B: Log–log plot of dose-normalized bosentan concentrations vs. (absolute) sildenafil concentrations conditioned by absence (A)/presence (B) of imatinib. The median is indicated by a line summarizing dose-normalized concentrations of bosentan in the absence (left cluster) or over a range of concentrations (right cluster) of sildenafil. Sil, sildenafil. C, D: Log–log plot of dose-normalized bosentan concentrations vs. (absolute) imatinib concentrations conditioned by absence (C)/presence (D) of sildenafil. The median is indicated by a line summarizing dose-normalized concentrations of bosentan in the absence (left cluster) or over a range of concentrations (right cluster) of imatinib. Ima, imatinib.

Figure 5.

Estimated effects of imatinib and sildenafil on bosentan concentrations. Geometric mean of bosentan concentrations with 95% confidence intervals in the presence/absence of imatinib and/or sildenafil (A) and mean relative effects on bosentan concentrations for co-administration of imatinib and/or sildenafil vs. no co-administration (B). None, neither imatinib nor sildenafil co-administered with bosentan; Ima alone, imatinib co-administered; Sil alone, sildenafil co-administered; Ima + Sil, imatinib and sildenafil co-administered. ◂, DDI study (geometric mean ratio for AUC over dosing interval)

On co-administration with bosentan, imatinib concentrations tended to decrease (Figure6). In addition, there were no clear changes observed in imatinib concentration on co-administration with sildenafil. The statistical analysis (Figure7) confirmed that imatinib concentrations, on average, decreased by 33% (95% CI 18%, 45%) in the presence of bosentan and did not change in a statistically significant manner in the presence of sildenafil (ratio present : absent = 0.96, 95% CI 0.76, 1.22). The estimated combined effect of sildenafil and bosentan was a decrease in imatinib concentrations of 35% (95% CI 10%, 53%).

Figure 6.

Effects of bosentan and sildenafil on imatinib concentrations. A, B: Log–log plot of dose-normalized imatinib concentrations vs. (absolute) sildenafil concentrations conditioned by absence (A)/presence (B) of bosentan. The median is indicated by a line summarizing dose-normalized concentrations of imatinib in the absence (left cluster) or over a range of concentrations (right cluster) of sildenafil. Sil, sildenafil. C, D: Log–log plot of dose-normalized imatinib concentrations vs. (absolute) bosentan concentrations conditioned by absence (C)/presence (D) of sildenafil. The median is indicated by a line summarizing dose-normalized concentrations of imatinib in the absence (left cluster) or over a range of concentrations (right cluster) of bosentan. Bos, bosentan

Figure 7.

Estimated effects of bosentan and sildenafil on imatinib concentrations. Geometric mean of imatinib concentrations with 95% confidence intervals in the presence/absence of bosentan and/or sildenafil (A) and mean relative effects on imatinib concentrations for co-administration of bosentan and/or sildenafil vs. no co-administration (B). None: neither bosentan nor sildenafil co-administered with imatinib; Bos alone, bosentan co-administered; Sil alone, sildenafil co-administered; Bos + Sil, bosentan and sildenafil co-administered

Clinical implications of DDI findings

The analysis data set consisted of 186 patients (imatinib n = 94, placebo n = 92), which included measures of efficacy (6 min walk distance (6MWD) and PVR) as well as safety indicators (liver enzymes, serum glutamic oxaloacetic transaminase [SGOT] and serum glutamic-pyruvic transaminase [SGPT]).

To assess the potential impact of increased exposure of co-medications on efficacy, 6MWD and PVR % changes from baseline, evaluated after 24 weeks of treatment, were plotted against individually averaged concentrations of sildenafil and bosentan (Figure8).

Figure 8.

Relationship of 6MWD and PVR per cent change from baseline, evaluated after 24 weeks of treatment, vs. averaged sildenafil (left) and bosentan (right) concentrations. Percent changes from baseline in 6MWD (A) or PVR (B) after 24 weeks of treatment are plotted against the individually averaged sildenafil (left) and bosentan (right) concentrations. Averaged concentrations were obtained as the geometric mean of all measurable plasma concentrations of sildenafil or bosentan in each patient. Circles are for patients in the imatinib group and triangles for patients in the placebo group. Patients not receiving sildenafil (No Sil) or not receiving bosentan (No Bos) were assigned small random values for appearance on the logarithmic axes. In each plot, the dashed line corresponds to a smooth (loess) fit to the placebo data and the solid blue line to the imatinib data. The horizontal boxplots (in the lower part of each figure) refer to the distributions of the individually averaged sildenafil/bosentan concentrations in the placebo (Pbo) or imatinib (Ima) groups. In each boxplot, the bold line is the median value, the edges of the box correspond to the 1st and 3rd quartiles (hence length of the box = inter-quartile range), and the whiskers extend to the most extreme data point that is not more than 1.5 times the inter-quartile range from the box

Interpretation of those graphs requires caution as they do not show typical concentration–response relationships, as patients entered the study with some co-medications already prescribed by their treating physician. The interest of such displays primarily lies in contrasting the placebo and imatinib responses over the range of co-medication concentrations. A key point to emphasise is that the treatment difference varies over the range of concentrations for sildenafil and bosentan. If efficacy of imatinib was partly attributable to increased exposure to sildenafil or bosentan, one would expect to see increasing differences between placebo and imatinib with higher concentrations of the respective co-medications. However, as shown in Figure8, the treatment differences tended to remain relatively constant over the range of co-medication concentrations, especially at the highest concentration range.

As hepatic AEs were to be expected based on the AE profile of both bosentan and imatinib, and the combination of both drugs is a potential source for additive or even synergistic interaction in this regard, we investigated the potential impact of increased exposure of co-medications on the risk for hepatotoxicity. For this purpose, SGOT and SGPT concentrations were plotted against the corresponding trough concentrations of imatinib, both in the presence or absence of bosentan (Figure9). Overall, there was no clear tendency of increasing liver enzymes with higher concentrations of imatinib. Similar conclusions hold when plotting liver enzymes vs. averaged bosentan concentrations (results not shown).

Figure 9.

SGOT/SGPT vs. steady-state trough concentrations of imatinib. Only SGOT (A)/SGPT (B) values coinciding with a trough measurement at a steady state are plotted (the number of SGOT/SGPT measurements exceeded the number of steady-state trough concentrations). The presence/absence of bosentan is indicated (shape and colour). The lines refer to loess smoothers through the respective groups. The shaded areas are 95% confidence intervals of the smoothers. , bosentan; , no bosentan

Discussion

Population PK of imatinib in severe PAH were characterized by similar CL/F and V/F compared with CML patients 16. They were also comparable with those of patients with GIST, although the estimated volume of distribution was smaller in patients with GIST. Similar dosing regimens appear to be effective in these different disease areas. The major covariate relevant for imatinib in PAH was co-administration of bosentan, which decreased the exposure to imatinib and, therefore, does not pose an additional safety risk. This leads to lower steady-state metrics in the PAH population. Thus, typical trough values at steady-state in the CML/GIST populations are approximately 1000 ng ml^–1. Our population model would predict a similar value for a typical patient not receiving bosentan. However, when taken across the study population, the averaged value is approximately 700 ng ml^–1, which reflects bosentan usage in approximately 60% of the patients in our study.

In addition to co-administration with bosentan, a more weakly significant relationship between haemoglobin and apparent volume and clearance was identified. This is consistent with findings in CML 16 and supported by the known distribution of imatinib into erythrocytes 18,19. Although other covariates were not identified as statistically significant in the covariate search analysis, it should be stressed that our dataset was considerably smaller than that investigated by Schmidli et al. 16 and, therefore, the results should not be misinterpreted as an absence of effect (e.g. bodyweight on CL/F and V/F). An effect of WBC count on PK parameters was most likely not identified because of the different populations (PAH vs. CML), their median WBC count (5.8 vs. 16.0 10⁹ l^–1) and vastly different distributions. We can further emphasize that there was no evidence for a relevant PK difference between different races, including Japanese and other Asian patients. Obviously, this involved small numbers of patients and such conclusions cannot be considered definitive.

When focusing on potential DDIs among imatinib, sildenafil and bosentan, we can first note that bosentan was found to decrease sildenafil concentrations by approximately 50% and sildenafil to increase bosentan concentrations by approximately 50%, which is consistent with previous reports 11,20. This is a first hint supporting the validity of the sparse PK sampling approach within a phase III study to elucidate such a complex question of PK interactions among the three drugs. In this analysis, it was also found that, on average, co-administration of imatinib resulted in increased concentrations of bosentan by 50% and sildenafil by 66%. These results were confirmed by a retrospective comparison with the results of a dedicated DDI study requested by health authorities to quantify the PK impact of imatinib on co-administration of bosentan and sildenafil. This unpublished study estimated the increase in sildenafil exposure after administration of imatinib to be 70% (90% CI 43%, 103%), and the corresponding increase in bosentan exposure to be 40% (90% CI 23%, 59%). This was again consistent with our own findings, which further testifies to the validity of the conclusions based on a sparse PK sampling approach to characterize PK interactions among the three drugs.

Interestingly, the effects of bosentan and imatinib on sildenafil cancelled out, bringing exposure close to that observed in the absence of both the drugs. Sildenafil did not have an apparent effect on imatinib concentrations. However, bosentan, on average, decreased imatinib concentrations by approximately 30% regardless of the presence of sildenafil. The estimated combined effect of sildenafil and bosentan was a decrease in imatinib concentrations by 35%, indicating that the effect of bosentan on the PK of imatinib is independent of sildenafil presence, and sildenafil does not impact on the PK of imatinib.

Following characterization of the different PK interactions among the three drugs, their clinical implications were assessed. The key finding that imatinib increased co-medication exposure immediately raises the question whether the efficacy of imatinib is caused by enhancing the effect of bosentan and sildenafil via increasing their exposure. In this case, the effect of imatinib would be more pronounced at high concentrations of the respective co-medications. However, treatment differences between imatinib and placebo for change of 6MWD and change of PVR were relatively constant across the entire concentration range for sildenafil and bosentan. Therefore, it is not expected that an increase of either bosentan or sildenafil exposure towards higher concentrations would lead to improved efficacy. Even in the absence of either sildenafil or bosentan, the effect of imatinib is as prominent as at the highest concentrations of the respective drug, lending strong support to its intrinsic efficacy beyond merely elevating the concentration of the co-medications due to DDIs. The ultimate proof would have been mono-administration to patients with PAH, but this would have required either withholding concomitant therapy from patients already treated or substituting first line therapy, which has been proven effective, with a study drug of unknown therapeutic value.

As bosentan concentrations were, on average, roughly doubled in the presence of sildenafil and imatinib, the next question to be answered was the clinical relevance of this increased exposure. With regard to potential increased risk of hepatotoxicity, elevation of transaminase levels has been reported, which occurred in 14% of patients receiving 250 mg twice daily of bosentan, 4% of those receiving 125 mg twice daily of bosentan and 3% of those receiving placebo 13. Overall, in our study, the occurrence of newly occurring or worsening liver function tests (defined as >3 × ULN in SGOT, or >3 × ULN in SGPT or >1.5 × ULN in alkaline phosphatase or >1.5 × ULN in total bilirubin) was 8% (eight of 102 patients) in the imatinib group vs. 10% (10 of 98 patients) in the placebo group. When restricting to patients who were taking bosentan, corresponding numbers were 12% (seven of 59 patients) in the imatinib group and 10% (five of 49 patients) in the placebo group. A trend of increased frequency of liver test abnormalities was therefore seen in this subgroup, but such a difference is hardly conclusive based on the small patient numbers. In addition, an exposure-dependent increase to bosentan or imatinib was not evident for any of the liver transaminases in our study. In conclusion, there is no clear evidence of increased risk of elevated enzymes when co-administering bosentan and imatinib. However, the lack of significant elevated enzymes in our relatively small study should not be taken as evidence for the overall absence of this effect.

In conclusion, parameters of the population PK analysis of imatinib in patients with severe PAH were found comparable with those in patients with CML. Concentrations of bosentan and sildenafil were elevated on co-administration with imatinib. Bosentan, on average, decreased imatinib concentrations, whereas sildenafil had no effect on imatinib concentrations. Beyond merely elevating the concentrations of the co-medications, further exploration of the data supports intrinsic efficacy of imatinib in this patient population. While our DDI assessment based on sparse PK data was quite robust and corroborated by a separate DDI study, further studies would be needed to confirm the lack of clinical relevance of those PK interactions.

Competing Interests

This work was funded by Novartis Pharmaceuticals AG, Basel, Switzerland. All authors have completed the Unified Competing Interest form at http://www.icmje.org/coi_disclosure.pdf (available on request from the corresponding author) and declare no support from any other organization for the submitted work, no financial relationships with any organizations that might have an interest in the submitted work in the previous 3 years and no other relationships or activities that could appear to have influenced the submitted work. All the authors are employees of Novartis and declare no conflict of interest. All the authors participated in the analysis and interpretation of data and critical revision of the manuscript. The authors were assisted in the preparation of the manuscript by Snigdha Santra, Siddharth Vishwakarma, Amit Bhat and Kevin Roche who provided medical writing assistance and subsequent revisions based on authors' feedback.

Supporting Information

Additional Supporting Information may be found in the online version of this article at the publisher's web-site:

Appendix S1

Population PK modeling