Inhibition of efflux transporters by poly ADP-ribose polymerase inhibitors

Poly ADP-ribose polymerase (PARP) inhibitors have been approved for the treatment of various cancers. They share a similar mechanism of action but have differences in pharmacokinetic characteristics and potential for drug – drug interactions (DDI). This study evaluated the potential ATP-binding cassette transporter-mediated interactions between PARP inhibitors (niraparib, olaparib and rucaparib) and statins (atorvastatin and rosuvastatin). We studied the inhibitory activity of PARP inhibitors on breast cancer resistance protein (BCRP), multidrug resistance-associated protein 3 (MRP3) and P-glycoprotein (P-gp) using vesicular transport assays and determined the concentrations required for 50% inhibition (IC 50 ). Then, we predicted the increase of statin exposure followed by the administration of PARP inhibitors using a mechanistic static model. Rucaparib was the strongest inhibitor of BCRP-mediated rosuvastatin transport (IC 50 13.7 μ M), followed by niraparib (42.6 μ M) and olaparib (216 μ M). PARP inhibitors did not affect MRP3. While niraparib appeared to inhibit P-gp, the inhibition showed large variability. The inhibition of intestinal BCRP by rucaparib, niraparib and olaparib was predicted to elevate rosuvastatin exposure by 52%, 37% and 24%, respectively. The interactions between PARP inhibitors and rosuvastatin are probably of minor clinical significance alone, but combined with other predisposing factors, they may increase the risk of rosuvastatin-associated adverse effects.


| INTRODUCTION
Poly (ADP-ribose) polymerases (PARPs) are a superfamily of enzymes that function in diverse biological processes including DNA repair, chromatin modulation and transcription signalling. 1 Due to their central role in DNA repair pathways, they have emerged as an appealing therapeutic target in cancer.PARP inhibitor drugs suppress DNA repair and tumour growth by trapping PARP protein onto the DNA, and thus prevent the repair of DNA singlestrand breaks and stall the DNA replication fork. 2 Olaparib was the first PARP inhibitor to get regulatory approval at the end of 2014 for maintenance therapy in patients with relapsed platinum-sensitive BRCA-mutated ovarian cancer. 3Since then, other PARP inhibitors, such as rucaparib and niraparib, have been approved for clinical use while therapeutic indications have extended to first-line maintenance therapy and other subtypes of ovarian, pancreatic, prostate and breast cancer. 4,5ARP inhibitors differ in their pharmacokinetic characteristics and potential for drug-drug interactions (DDI). 4,6Olaparib has been shown in vitro to inhibit several drug transporters including P-glycoprotein (P-gp), breast cancer resistance protein (BCRP) and organic anion transporting polypeptide 1B1 (OATP1B1). 7iraparib and rucaparib have also been shown in vitro to inhibit P-gp and BCRP. 8,9Cholesterol-lowering statins are increasingly used during cancer therapies.Because their absorption and hepatic uptake are regulated by the P-gp, BCRP and OATP transporters, [10][11][12][13][14] concomitant use of statins and PARP inhibitors could lead to elevated statin concentrations.For instance, the European public assessment report (EPAR) product information documents of olaparib and niraparib recommend that caution should be exercised if PARP inhibitors are administered with statins because olaparib and niraparib can inhibit drug metabolizing enzymes and transporters relevant for statin pharmacokinetics. 8,15owever, the interpretation and translation of earlier findings are limited by the fact that previous studies were performed with varying experimental systems and did not use statins as probe substrates.Therefore, this study aims to evaluate and compare the potential efflux transporter-mediated interactions between PARP inhibitors and statins using a vesicular transport assay.We focused on the inhibitory activity of the three most commonly used PARP inhibitors, olaparib, rucaparib and niraparib, on statin efflux transporters BCRP, P-gp and MRP3 using atorvastatin, rosuvastatin and fluorescent efflux transporter probes as probe substrates.

| Vesicular transport assay
The inhibitory activity of PARP inhibitors on BCRP, MRP3 and P-gp was studied using vesicular transport assay, performed as described previously. 13In short, transporter-expressing membrane vesicles were incubated in transport assay buffer (PharmTox, Radboud UMC, Nijmegen, The Netherlands) supplemented with 10 mM MgCl 2 and probe substrate, to which various concentrations of PARP inhibitors or vehicle (DMSO) was added.The solutions were preincubated at 37 C for 10 min after which the transport was started by adding pre-warmed ATP or AMP solution to a final concentration of 4 mM.Then, the samples were incubated at 37 C for 2 min (NMQ), 5 min (atorvastatin, rosuvastatin) or 10 min (CDCF, LY) depending on the probe substrate.The transport was terminated by adding 200 μL of ice-cold stop buffer (PharmTox, Radboud UMC, Nijmegen, The Netherlands), and the samples were immediately transferred onto a MultiScreenHTS FB Filter Plate 1.0/0.65 μm (Merck KGaA, Darmstadt, Germany).The membrane vesicles were washed twice with 200 μL stop buffer and twice with 200 μL ice-cold washing buffer (40 mM MOPS-Tris pH 7.0 and 70 mM KCl), and dried.Then, the membrane vesicles were broken down and the probe substrate was released by the addition of 0.1 M NaOH (CDCF, LY), 10% SDS (NMQ) or 50% MeOH (atorvastatin, rosuvastatin).For the 50% MeOH lysis solution, stable isotope-labelled atorvastatin or rosuvastatin was added as an internal standard.The lysate was eluted by centrifugation at 2200g for 10 min.To enhance the fluorescence of the lysate, an equal volume of 0.1 M HCl or 0.1 M H 2 SO 4 was added to LY or NMQ samples, respectively.All assays were performed in triplicates on 96-well plates.

| Analytical methods
CDCF, LY and NMQ were fluorometrically quantified with Cytation 5 multi-mode reader (BioTek, Winooski, Vermont, USA).The excitation and emission wavelengths of CDCF, LY and NMQ were 510/535 nm, 430/538 nm and 255/460 nm, respectively.Atorvastatin and rosuvastatin were analysed using a SCIEX 5500 QTRAP LC-MS/MS system (Sciex, Toronto, Ontario, Canada) interfaced with an ESI ion source.The chromatographic separation was performed on a Luna Omega polar C18 analytical column (100 Â 2.1 mm i.d., 1.6 μm particle size; Phenomenex, Torrance, California, USA) protected by a guard column of the same material.The mobile phases A and B consisted of 5 mM ammonium formate (pH 3.9, adjusted with 98% formic acid) and acetonitrile, respectively, and the flow rate and the column temperature were maintained at 300 μl/min and 40 C. The following gradient conditions were applied: 1 min at 20% B on hold, then a linear ramp from 20% B to 40% B over 3 min followed by a second linear ramp to 90% B over 2 and 1 min at 90% B before a re-equilibration step back to the initial conditions (20% B).The mass spectrometer was operated in multiple reaction monitoring (MRM) mode.Characteristic ion transitions for each analyte and internal standard were reported previously. 13

| Experimental design
The inhibition studies were conducted in two phases.First, the effect of 50 μM PARP inhibitors on probe substrate transport was tested in a single experiment containing appropriate positive inhibition controls.Here, 50 μM LY, 5 μM CDCF and 5 μM NMQ were used as probe substrates for BCRP, MRP3 and P-gp, respectively.In addition, the effect of PARP inhibitors on 10 μM atorvastatin (in BCRP, MRP3, P-gp vesicles) and 10 μM rosuvastatin transport (in BCRP vesicles only) was investigated in a single experiment with appropriate positive inhibition controls.In the BCRP experiments, 10 μM ticagrelor and 10 μM Ko143 were used as positive inhibition controls. 16,17For MRP3, 100 μM benzbromarone was used as a positive inhibition control, whereas for P-gp, the positive inhibition control was 100 μM ketoconazole. 18,19Next, the most potent inhibitors of BCRP and P-gp were subjected to concentration dependency studies to determine the concentrations required for 50% inhibition (IC 50 ).In these experiments, 10 μM rosuvastatin and atorvastatin were selected as BCRP and P-gp probe substrates, respectively, and each investigated drug was examined in three separate experiments.The study was conducted in accordance with the Basic & Clinical Pharmacology & Toxicology policy for experimental and clinical studies. 20

| Data analysis
ATP-dependent transport was calculated by subtracting the probe substrate uptake in the absence of ATP from the probe substrate uptake in the presence of ATP.Inhibition is presented as relative transport values (%), which were obtained by comparing the ATP-dependent transport of probe substrate in the presence of PARP inhibitor to that of control containing only the vehicle (DMSO).GraphPad Prism 8.4.2 (GraphPad Software Inc., San Diego, California, USA) was used to calculate the IC 50 by fitting data from the inhibition studies to the fourparameter logistic curve using Equation 1: where relative transport is the relative transport value of probe substrate in the presence of [I] concentration of investigated PARP inhibitor, minimum and maximum correspond to the plateaus of minimal and maximal relative transport (fixed to 0% and 100%, respectively), and h is the Hill slope, which describes the steepness of the slope.

| Calculation of parameters required for mechanistic static model predictions
The theoretical intestinal luminal concentration (I 2 ) of PARP inhibitors was calculated by dividing a single commonly used oral dose by 250 ml.The maximum enterocyte concentration (I ent ) of PARP inhibitors was calculated using Equation 2 adapted from Rostami-Hodjegan and Tucker 21 : where F a is the fraction absorbed, k a is the absorption rate constant, dose is a single commonly used oral dose in moles, and Q ent is the enterocyte blood flow (0.3 L/min).The values used for each PARP inhibitor are presented in Table 1, but in case the F a value has not been reported, the bioavailability of the PARP inhibitor was used instead.Since k a values were not available, they were estimated from plasma concentration profiles assuming one-compartment model pharmacokinetics using Equation 3 27 : where t max is the time of peak drug plasma concentration and k e is the elimination rate constant of the drug.The k e was derived from t ½ , the elimination half-life of the drug in plasma using Equation 4.
AUCR describes the increase of area under the plasma rosuvastatin concentration-time curve (AUC) when rosuvastatin is taken concomitantly with a PARP inhibitor compared to rosuvastatin taken alone.To predict the AUCR, mechanistic static model describing the inhibition of intestinal efflux transporters was employed (Equation 5): where f e is the fraction of rosuvastatin excreted by intestinal BCRP.Here, the value of 0.5 was used for f e . 28| RESULTS

| Screening of niraparib, olaparib and rucaparib inhibitory activity
Rucaparib strongly inhibited the transport of LY, rosuvastatin and atorvastatin in BCRP vesicles (Figure 1).The relative transport of LY and rosuvastatin in the presence of 50 μM rucaparib was 18% and 12%, respectively, of that in the absence of rucaparib (Figure 1G).Furthermore, the relative BCRP-mediated transport of atorvastatin in the presence of rucaparib (44%) was similar to that of a positive BCRP control inhibitor (10 μM ticagrelor).Similarly, niraparib, and olaparib to a smaller extent, inhibited the BCRP-mediated transport of LY and rosuvastatin (Figure 1A and D).The relative transport of rosuvastatin in BCRP vesicles in the presence of 50 μM olaparib and niraparib was 61% and 29%, respectively, of control transport without inhibitors.
In regards to P-gp, niraparib inhibited the transport of atorvastatin (relative transport 39%, Figure 1B).On the other hand, olaparib had a minor inhibitory activity towards P-gp (relative transport 79%, Figure 1E), whereas rucaparib stimulated the atorvastatin transport in P-gp vesicles (relative transport 157%, Figure 1H).Furthermore, none of the PARP inhibitors inhibited the transport of NMQ.Finally, the investigated PARP inhibitors did not affect MRP3 activity (Figure 1C,F,I).

| Quantitative assessment of drugdrug interaction potential using I 2 /IC 50 ratios and mechanistic static model
The theoretical intestinal luminal concentration of PARP inhibitors after a commonly used single oral dose (I 2 ) was calculated and compared to the determined IC 50 values.The I 2 /IC 50 ratios for niraparib, olaparib and rucaparib were 88, 17 and 542, respectively.Additionally, the estimated maximum enterocyte drug concentration (I ent ) was used in a mechanistic static model to predict the increase of rosuvastatin AUC followed by the inhibition of intestinal BCRP.Similarly to the I 2 /IC 50 ratios, the model suggests that rucaparib has the greatest impact on rosuvastatin AUC (+52%), followed by niraparib (+37%) and olaparib (+24%).
T A B L E 1 F a , k a , dose and I ent values of poly ADP-ribose polymerase (PARP) inhibitors used in mechanistic static model.In case F a value was not available, the bioavailability of PARP inhibitor was used instead.

| DISCUSSION
In the present study, we determined the inhibitory activity of niraparib, olaparib and rucaparib on rosuvastatin and atorvastatin transport mediated by BCRP, P-gp and MRP3.We observed that all the investigated PARP inhibitors inhibited BCRP-mediated rosuvastatin transport.Furthermore, the results suggest that niraparib may inhibit P-gp to a small extent.Rucaparib was found as the strongest inhibitor of BCRP with IC 50 of 14 μM.On the other hand, 50 μM of rucaparib did not inhibit MRP3 or P-gp in our experimental setup.Previously, rucaparib has been reported to inhibit BCRP, MRP3 and P-gp with IC 50 values of 55, 900 and 169 μM, respectively. 9Small differences in the IC 50 values can be explained with different experimental systems and probe substrates.For instance, Liao and coworkers used MDCK-BCRP cells instead of membrane vesicles.Interestingly, rucaparib increased the AUC of digoxin, a P-gp substrate, by 20% in a clinical trial. 29ased on our mechanistic static model, rucaparib was predicted to have the greatest impact on rosuvastatin AUC of all investigated PARP inhibitors.The absorption of rosuvastatin is limited by intestinal BCRP, and in the event of poor BCRP function as determined by homozygosity for the c.421C > A (rs2231142) genetic variant, rosuvastatin AUC increases 2.1-2.4-fold. 11,14Based on our data and static interaction prediction, we estimated that rucaparib may increase the AUC of rosuvastatin by 52%.The prediction is close to an observation from a DDI study in humans, where 600 mg of rucaparib taken twice a day for 2 weeks increased rosuvastatin AUC and C max by 34% and 29%, respectively. 30This interaction is unlikely to be of major clinical significance by itself.However, together with other risk factors, for example, the no-function genetic variants of SLCO1B1 (which encodes OATP1B1), advanced age, other interacting drugs, or a high rosuvastatin dose, 16 rucaparib may increase the risk of rosuvastatin-induced adverse reactions.
The impact of rucaparib on atorvastatin has not been studied in vivo in humans, but the magnitude of the effect could be similar to that on rosuvastatin.The The inhibition of BCRP, MRP3 and P-gp by 50 μM niraparib, olaparib and rucaparib in membrane vesicles.In each graph, the y-axis stands for the relative transport of probe substrate in the presence of test compound (black bar), control inhibitor (grey bar) and vehicle control DMSO (white bar).Vesicles and probe substrates used in the experiment are specified on the x-axis.Control inhibitors used in the study were the following: 10 μM ticagrelor and Ko143 (*Ko143 was only used when olaparib was studied using LY as a probe substrate) for BCRP; 100 μM ketoconazole for P-gp; 100 μM benzbromarone for MRP3.Each bar represents the relative transport of probe substrate ± SD from a single experiment performed using three technical replicates.fraction of atorvastatin excreted by BCRP has not been established, and thus the implications of the inhibition of intestinal BCRP were not estimated.While the role of BCRP in atorvastatin absorption is smaller than in that of rosuvastatin, 11 atorvastatin absorption is limited by P-gp as well. 10,13Finally, intestinal and hepatic cytochrome P450 (CYP) 3A4 plays an important role in the metabolism of atorvastatin.In vitro, rucaparib inhibited CYP3A with IC 50 values of 17-23 μM. 9 In vivo in humans, rucaparib has increased the AUC of oral midazolam, a CYP3A4 probe substrate, by 39%. 29Therefore, the possible effect of rucaparib on the pharmacokinetics of atorvastatin cannot be ruled out.
In the present study, niraparib was found to be a moderate inhibitor of BCRP with an IC 50 of 43 μM.This is an order of magnitude higher than reported previously (IC 50 5.8 μM), 8 which might be explained by a different experimental system or probe substrate, but these have not been reported for the previous study.On the other hand, the IC 50 value of niraparib towards P-gp is most likely high, but we were not able to determine that with our experimental setup.Niraparib is considered to have a low DDI potential as it does not exhibit significant inhibition of CYP enzymes or hepatic uptake transporters, 8 However, due to its inhibitory activity towards BCRP, it may affect the absorption of drugs restricted by intestinal BCRP.Here, we predicted that concomitantly taken niraparib may increase the AUC of rosuvastatin by 37%, which is a smaller increase than estimated in the case of rucaparib.Similarly, niraparib may increase the oral bioavailability of atorvastatin, but the extent of interaction should be smaller than that of rucaparib as niraparib is not expected to inhibit hepatic CYP3A4. 8I G U R E 2 IC 50 determination of niraparib, olaparib and rucaparib in BCRP and P-gp vesicles using 10 μM rosuvastatin and atorvastatin, respectively, as a probe substrate.In each panel, the y-axis stands for the relative transport of the statin, and the x-axis for the concentrations of PARP inhibitors.Each point represents a mean relative transport of an individual experiment and the solid lines show the curve fitting of inhibition.In total, three separate experiments were performed (four for niraparib in P-gp vesicles) and in every experiment, three technical replicates were used for each test condition.
Previously, olaparib was found to be a weak BCRP and P-gp inhibitor.In MDCK-BCRP cells, olaparib inhibited the transport of 2-Amino-1-methyl-6-phenylimidazo (4,5-b)pyridine (PhIP) by 47% at 100 μM concentration. 7oreover, olaparib inhibited the P-gp-mediated transport of digoxin with an IC 50 of 76 μM in MDCK-MDR1 cells.The level of reported BCRP inhibition agrees with that found in the present study using LY or rosuvastatin probe substrates.Similarly, we observed that olaparib was a weak inhibitor of atorvastatin transport in P-gp vesicles.Based on our data, olaparib is predicted to have a minor impact at best on the efflux of atorvastatin and rosuvastatin in the small intestine.However, several other transporters and CYP enzymes are involved in the excretion and elimination of atorvastatin and rosuvastatin.Olaparib has a slight impact on CYP2C9 (22% inhibition at 100 μM), CYP2C19 (26% inhibition at 100 μM) and CYP3A4/5 (K i 72 μM), 31 and it inhibits OATP1B1 and organic anion transporter 3 (OAT3) relatively efficiently (IC 50 20-27 and 18 μM, respectively). 7To our knowledge, the possible effects of olaparib on statin pharmacokinetics have not been investigated in humans.Therefore, despite the lack of strong BCRP and P-gp interaction, the possible effects of olaparib on the pharmacokinetics of atorvastatin and rosuvastatin cannot be ruled out.
The strength of the study is the direct in vitro comparison of the inhibitory activity of multiple PARP inhibitors conducted in a single laboratory.Furthermore, clinically relevant probe substrates were used in the in vitro vesicular transport assay.The assay enables the direct and accurate determination of inhibition parameters and elucidates the inhibition of individual transporters.However, the results of inhibition studies and mechanistic static modelling should be treated with caution.Firstly, the transporter inhibition could be probe substrate dependent. 32Even though we studied the inhibition of BCRP with three probe substrates, our findings may not apply to other BCRP substrates.Secondly, the feasibility of a mechanistic static model depends on the model parameters.Provided that they are selected carefully, the models may be capable of yielding accurate predictions as exemplified by previous studies. 28In the present study, rather than using worst-case scenario F a and k a parameters, which may overestimate the extent of the available drug in the site of inhibition, we obtained and derived model parameters from recent publications to provide better insight into the potential of DDI.The accuracy of our predictions is supported by the fact that rucaparibrosuvastatin DDI prediction was relatively close to the in vivo observations.As far as cancer patients are concerned, it must be remembered that they are at a high risk of DDIs and associated adverse effects.The patients typically receive a large number of concomitant medications in the treatment of cancer, and are prone to cancer medicationrelated toxicities and other comorbidities. 33,34We recently studied 232 commonly used drugs for BCRP inhibition and found 13 new potent BCRP inhibitors. 35hree of these inhibitors (vemurafenib, bexarotene and everolimus) are antineoplastics while aprepitant is an antiemetic used in the prevention of chemotherapyinduced vomiting and nausea.In addition, common cardiovascular and non-steroidal anti-inflammatory drugs inhibited BCRP strongly as well.As BCRP regulates the absorption of a wide range of drugs, our previous and present findings underline the DDI risk involved in cancer treatments.
In summary, rucaparib was a modest BCRP inhibitor, while niraparib and olaparib inhibited BCRP to a smaller extent.A mechanistic static model showed that rucaparib may increase rosuvastatin exposure by 52%, which is in agreement with the recent in vivo study.Based on this, the effect of niraparib and olaparib on rosuvastatin exposure is likely less prominent.