In vitro and in vivo human metabolism and pharmacokinetics of S‐ and R‐praziquantel

Abstract Racemic praziquantel (PZQ) is the drug of choice for the treatment of schistosomiasis. R‐Praziquantel (R‐PZQ) has been shown as the therapeutic form, whereas S‐PZQ is less efficacious and responsible for the bitter taste of the tablet. This study aimed at investigating the metabolism of R‐ and S‐PZQ as this could have implications on efficacy and safety of racemate and R‐PZQ specific formulations under development. In vitro CYP reaction phenotyping assay using 10 recombinant CYP (rCYP) isoenzymes showed hepatic CYP1A2, 2C19, 2D6, 3A4, and 3A5 were the major enzymes involved in metabolism of PZQ. Enzyme kinetic studies were performed by substrate depletion and metabolite formation methods, by incubating PZQ and its R‐ or S‐enantiomers in human liver microsomes (HLM) and the rCYP enzymes. The effect of selective CYP inhibitors on PZQ metabolism was assessed in HLM. CYP1A2, 2C19, and 3A4 exhibited different catalytic activity toward PZQ, R‐ and S‐enantiomers. Metabolism of R‐PZQ was mainly catalyzed by CYP1A2 and CYP2C19, whereas metabolism of S‐PZQ was mainly by CYP2C19 and CYP3A4. Based on metabolic CLint obtained through formation of hydroxylated metabolites, CYP3A4 was estimated to contribute 89.88% to metabolism of S‐PZQ using SIMCYP® IVIVE prediction. Reanalysis of samples from a human PZQ‐ketoconazole (KTZ) drug‐drug interaction pharmacokinetic study confirmed these findings in that KTZ, a potent inhibitor of CYP3A, selectively increased area under the curve of S‐PZQ by 68% and that of R‐PZQ by just 9%. Knowledge of enantioselective metabolism will enable better understanding of variable efficacy of PZQ in patients and the R‐PZQ formulation under development.


| INTRODUC TI ON
Schistosomiasis is the second most important parasitic infection in the world after malaria with an estimated 250 million people being infected and approximately 800 million people at risk of infection. 1 Schistosomiasis or Bilharzia is caused by schistosomes, which are parasitic trematode worms of the genus Schistosoma. Two thirds of the cases of schistosomiasis in sub-Saharan Africa are attributed to S haematobium infection which causes severe urinary tract diseases 2 and is a significant cause of clinical morbidity and disability.
Schistosoma mansoni is also common in Africa 3 while Schistosoma japonicum is confined to Asia. 4  Current formulations of PZQ are racemates consisting of Rand S-enantiomers. R-PZQ showed superior properties than its S-enantiomer in biological activities 5 and cytotoxicity, 6 with much higher antischistosomal activity and worm reduction rates as indicated by much lower values in IC50 (approx. 500-fold lower) and ED50 (5-fold lower) against worms than its S-enantiomer, as well as lower cytotoxicity in several cell lines. However, pharmacokinetics studies of PZQ showed that R-PZQ is cleared 2.5 times faster than its S-antipode from the circulating system with a maximum plasma concentration (C max ) levels of only one third of S-PZQ. 7 PZQ is metabolized extensively in the liver to the main metabolite, 4-hydroxy praziquantel (4-OH PZQ) 8 and numerous other mono-, di-, and tri-hydroxyl metabolites and conjugates. 9 The CYP3A4, CYP1A2, CYP2C9, and CYP2C19 have been identified as the main enzymes metabolizing the drug. 10,11 The plasma half-life of PZQ is estimated to be between 1 and 3 hours 12 and more than 80% of the drug is excreted within 24 hours in man. 13 The systemic bioavailability of PZQ is, therefore, very low at <20%. 14 In vivo, PZQ is well absorbed. More than 80% of the oral dose (taking into account unchanged drug and metabolites) is eliminated renally indicating a high fraction absorbed. 13 PZQ demonstrates rapid first pass metabolism and high interindividual variability. 7 The drug, however, relies on metabolism for excretion with <0.02% of unmetabolized PZQ being detected in urine. 15 Therefore, any modulation of enzymes metabolizing PZQ is likely to have a major impact on the pharmacokinetics of the drug. PZQ metabolism is also affected by drug-drug interactions (DDI) caused by either inhibition or induction of CYPs 1A2, 2C9, 2C19, and 3A4. For example, co-administration of PZQ with cimetidine causes a 100% increase in its bioavailability 16  Other examples include inhibitory effects of grapefruit juice 17 and ketoconazole (KTZ) on CYP3A4. 18,19 Co-administration with food has also been shown to increase bioavailability. Induction has also been shown with rifampicin (CYP2C9, CYP2C19, and CYP3A) 20 and dexamethasone (CYP3A4). 21 Plasma PZQ concentrations decreased to undetectable levels in patients who had been pretreated with rifampicin for 5 days. A 50% reduction in PZQ plasma concentration was observed in patients with parenchymal brain cysticercosis who had been treated with dexamethasone a few days earlier to prevent PZQ related anti-inflammatory side effects.
It was reported that CYP3A, CYP2C9, and CYP2C19 exhibited different catalytic activity toward PZQ enantiomers in vitro to PZQ metabolites observed in vivo in mice, 11 however, there are no studies which have determined enantiospecific metabolism of R-and S-PZQ in humans. In this study, we investigated stereoselective metabolism of PZQ in vitro and in vivo in man, to better understand and rationalize the mechanism behind the variable pharmacokinetics and selective clearance of R-and S-PZQ.

| CYP selective diagnostic inhibition studies
Reactions were carried out in triplicate in 96 well plates. Each reaction was measured at 2 time points, that is at 0 and 30 minutes. Each reaction mixture consisted of the 0.5 mg/mL pooled HLM, substrate (1.0 µmol/L R-, S-or racemic PZQ), inhibitor (10 µmol/L), and 0.1 mol/L potassium phosphate buffer pH 7.4 in a final volume of 200 µL. Inhibitor concentration of 10 µmol/L has been shown to be selective enough to show the contribution of some CYPs. 23 The substrate was diluted to 40% ACN so that the percent organic component in the final solution was maintained at 1%. Furafylline, KTZ, ticlopidine, and quinidine were used as diagnostic inhibitors for CYP 1A2, 3A4/5, 2C19 and 2D6, respectively. The reactions were initiated by addition of NADPH after a pre-incubation of 5 minutes at 37°C. All reactions were terminated by the addition of 150 µL ice-cold ACN. Internal standard (0.2 µmol/L DPZ) was added to ACN prior to termination of the assay. This was followed by centrifugation at 4500 g for 20 min- where CL int is the intrinsic clearance, T1/2 is the half-life of PZQ, incubation volume is the final volume of each reaction mixture and protein, or enzyme amount is the amount of protein in the reaction mixture.
Each reaction was monitored over 7 time points.

| Enzyme kinetics
The reactions were carried out in duplicate at each time point. % remaining compound = 1 − Peak area at T 30min Peak area at T 0min × 100.
% contribution = amount remaining with inhibitor − amount remaining without inhibitor amount remaining without inhibitor × 100.

| In vitro-in vivo extrapolation
In vitro metabolism data for R-, S-, and racemic PZQ with rCYPs were used for in vitro to in vivo extrapolation (IVIVE) using SIMCYP ® (ver- within the Simcyp ® program: where there are j CYPs with corresponding CL int values calculated from enzyme kinetic parameters for the different pathways in each recombinant system, CL int = in vitro clearance, MPPGL = microsomal protein per gram liver, ISEF j is the scaling factor for the corresponding CYP and fumic is fraction of drug unbound in microsomes which was predicted as 0.771. The CYP abundance for each of CYP enzymes used is given in Table A1. where K M and V max are obtained from the in vitro assay, ISEF j is the scaling factor for the corresponding CYP and MPPGL = microsomal protein per gram liver. The CYP abundance for each of CYP enzymes used is given in Table A1. The well-stirred model was used to estimate the hepatic clearance (CL H ) due to metabolism. The model was chosen over the parallel tube and the dispersion model because of its simplicity and the fact that very small differences in predicted values by the 3 models have been observed. 27 The hepatic clearance was expressed as 26 : where Q H is the hepatic blood flow (1500 mL/min), CL uH,int is clearance scaled to the whole liver, and reflects the actual metabolic capacity of the enzyme system and fuB is the free fraction of drug in blood calculated as fraction unbound in plasma (fu) divided by the blood to plasma drug concentration ratio (B/P). The value of fu used for PZQ, R-PZQ, and S-PZQ was 0.2 14 and B/P for R and S PZQ was 0.8 and 0.78, respectively, 28 obtained from studies reported in literature. This was based on the similar fragmentation pattern and behavior insource. Trans-OH-PZQ undergoes extensive MS in-source dehydration (loss of water), its dehydrated ion, m/z 311.

| Determination of R-and S-PZQ in human plasma samples
Protein precipitation was used to extract PZQ from human plasma samples from a previous study, 19 Briefly, internal standard (10 µL DPZ) was added to 100 µL of plasma followed by extraction with 690 µL of ice-cold ACN. The samples were vortexed for 30 seconds  Figure 1. The data were gathered and processed using Agilent OpenLab CDS for Chemstation software version 1.9.
Linear calibration curves were plotted as concentration vs peak area ratio for analyte to IS with no weighting. R-PZQ and S-PZQ were linear in the range of 50-10 000 ng/mL with an R value of 0.9997. The calibration curve was used to quantitate plasma PZQ concentrations.

| Data analysis
The pharmacokinetic (PK) analysis was performed using the

| Statistical analysis
Statistical analysis was performed using the GraFit software (ver-  where x T and x R are the geometric means of the ln transformed values for the test treatment (T) and the reference treatment (R); S 2 is the error variance obtained from the ANOVA; n is the number of sub- 1 v is the t-value for 90% of the t-distribution and v is the degree of freedom of the error variance from the ANOVA. The anti-ln of the above CI values was then computed to give the 90% CIs of the ratios of the test to the reference treatment geometric means.

| Inhibition by CYP selective inhibitors
The inhibition assay was performed in HLM to confirm the contribution of the identified CYP enzymes in the metabolism of PZQ and its individual enantiomers. The diagnostic inhibitors used, that is, furafylline (CYP1A2), ticlopidine (CYP2C19), quinidine (CYP2D6), and KTZ (3A4, 3A5) have been shown to be potent diagnostic inhibitors. Concentrations of PZQ and its enantiomers in HLM as well as the inhibitory effects of diagnostic inhibitors on metabolism are shown in Table 1. Due to the lack of high selectivity at the high inhibitor concentration, only relative contributions can be deduced from the results presented in Table 1. The greatest contribution to R-PZQ metabolism was mainly by CYP2C19, CYP1A2, and CYP2D6 which had an almost equal contribution.
The contribution of CYP3A4 was low for R-PZQ. The inhibitory effects on CYP3A4, CYP2C19, and CYP1A2 activity by the selective inhibitors showed metabolic stereoselectivity. Metabolic stereoselectivity was observed with CYP1A2 where R-PZQ was metabolized more as compared to S-PZQ. CYP3A4 mainly contributes to S-metabolism (Table 1). CY2C19 showed stereoselectivity by contributing more to the metabolism of S-PZQ than of R-PZQ.
The activity was, however, lower for racemic PZQ for CYP1A2 as compared to R-and S-PZQ.

| Determination of intrinsic clearance (CL int ) using the substrate depletion approach
The depletion rates of R-and S-PZQ were determined in HLM and rCYPs 1A2, 2C19, 2D6, 3A4, and 3A5. The calculated intrinsic hepatic clearance and hepatic clearances are shown in Table 2. No comparisons were made with the racemate since the concentration used was half for each enantiomer. No major differences were, however, expected since we did not observe any enantiomer-enantiomer interaction. HLM-and CYP1A2-mediated R-PZQ was high.
CYP3A4-mediated R-PZQ clearance was low, whereas the remainder of the compounds was medium clearance.

| Enzyme kinetics of hydroxylated metabolites formation
The metabolism and formation of the main metabolites of racemic PZQ have been previously described. 10 The formation of the 2 main in vitro metabolites namely cis-4-OH-PZQ and X-OH-PZQ showed stereoselectivity. This was also reflected in the clearance ( Table 3).
The enzyme kinetics of the formation of the major hydroxylated metabolites, cis-4-OH-PZQ and X-OH-PZQ, in HLM showed that the intrinsic clearance (CL uH,int ) via the formation of X-OH-PZQ from racemic PZQ is 5 times more than that via the formation of 4-OH-PZQ (Table 3). The obtained CL uH,int in rCYPs indicates the metabolic enantioselectivity of CYP1A2 for R-PZQ than S-PZQ with values of 7.55 and 0.83 mL/min/kg, respectively. Clearance by CYP2C19 with R-PZQ was 2 times more than S-PZQ with CL uH,int of 4.60 and 2.44 mL/min/kg respectively. Clearance by CYP2D6 was low and did not show any significant difference between R-and S-PZQ.
There was higher affinity (low K M ) for racemic PZQ and both enantiomers with CYP1A2 and CYP2C19 compared to CYP3A4 which showed low affinity as indicated by the higher K M values. Although CYP2D6 contributed to metabolism of PZQ and its enantiomers, it had very low affinity characterized by a K M greater than 100μM.
The formation of X-OH-PZQ is mainly attributed to S-PZQ metabolism rather than R-PZQ metabolism with a V max of 6.3 nmol/min/ mg protein for S-PZQ vs 1.2 nmol/min/mg protein for R-PZQ. The Michaelis-Menten kinetic plots for PZQ, R-PZQ, and S-PZQ metabolism by rCYP3A4 for the formation of X-OH PZQ are shown in Figure 3. The high V max for CYP3A4 with S-PZQ shows that CYP3A4 metabolizes S-PZQ with a higher turnover rate than R-PZQ. This data agreed with our observations from the substrate disappearance assay ( Figure 2) and previously determined clearance data (Table 2) as well as the results obtained from the inhibition assays (Table 1).

TA B L E 1
Relative contributions of the CYP1A2, CYP2C19, CYP3A4/5, and CYP2D6 to the metabolism of PZQ, R-PZQ, and S-PZQ in human liver microsomes (HLM) by inhibition studies to identify important contributing isoforms Note: Furafylline, quinidine, ticlopidine, and ketoconazole were used as diagnostic inhibitors of CYP1A2, CYP2D6, CYP2C19, and CYP3A4, respectively; the contribution of each isoform is shown in parenthesis. a Blank--Amount of PZQ before incubation.
b Control --Amount of PZQ after incubation in the absence of an inhibitor.

| IVIVE using SIMCYP ®
The relative percentage contribution of each of the enzymes to PZQ, R-PZQ, and S-PZQ metabolism was simulated using SIMCYP ® based on the in vitro metabolic Cl int with rCYPs using substrate depletion method as well as the metabolite formation assays. From the substrate depletion assay, S-PZQ was predicted to be mainly metabolized by CYP1A2, CYP2C19, and CYP3A4 (  The FDA no effect boundary for clinical DDI studies for the 90% CI of the geometric mean ratio ranges between 80% and 125%. 29 The reference and the treatment were significantly different for

| D ISCUSS I ON
Praziquantel has been shown to be metabolized mainly by cytochrome P450 enzymes in the liver, namely CYP 1A2, CYP2C19, CYP2C9, CYP 3A4, and CYP2D6. 10,11 Previous pharmacokinetic studies have shown R-PZQ to have higher clearance than S-PZQ. 7,8 Studies had earlier shown metabolism to be enantiomer selective, 7,28 but identification of the enzymes involved in R-and S-PZQ metabolism had not been fully explored. Data from this study, therefore, are expected to contribute in providing an understanding of how best patient safety and efficacy can be maintained in the use of PZQ or its enantiomers. This is important in the development of PZQ formulations which contain 1 enantiomer especially when it has to be co-administered with a drug which inhibit the enzymes required for metabolism. It also has implications on potential inducers of the drug of interest as this leads to subtherapeutic levels of PZQ.
Our reaction phenotyping study identified CYP1A1, CYP1A2, CYP2C19, CYP2D6, CYP3A4, and CYP3A5 as significant players in the metabolism of PZQ and its enantiomers (Figure 2). This agreed with our previous finding for PZQ metabolism 10 and findings by Wang and co-workers. 11 There was, however, a discrepancy on CYP2C9 where our assay did not pick it as a major enzyme. The same was observed by Li and co-workers. 10 The Wang study used the metabolite formation approach. Although CYP2C9 showed to be an important contributor of the formation of metabolites that had been identified in mice, it could be contributing to a minor metabolic route in humans hence not resulting in significant substrate depletion of PZQ as observed in this study.
Clearance is an important parameter, a drug is termed a high clearance drug if the hepatic clearance (CL H ) exceeds 14 mL/min/ kg. 24 The difference in CL H between R-PZQ and S-PZQ is not as large as observed in vivo where R-PZQ's clearance is 3 times faster than that of S-PZQ. This could be attributed to the possible involvement of the intestinal metabolism where the CYPs are represented at different proportionalities than the liver. 7,29 The predicted total clearance (CL H ) in Simcyp ® for PZQ, R-PZQ, and S-PZQ was 3.656, 6.923, and 6.265 mL/min/kg, respectively, using the substrate depletion assay and 11.44, 9.777, and 10.144 mL/min/kg, respectively, using the metabolite formation assay. To further investigate the contribution of CYP 1A2, CYP2C19, CYP 3A4/5, and CYP 2D6 to metabolism of PZQ, R-PZQ, and S-PZQ, the SIMCYP simulator was used. The contribution of the tested CYP isoforms to metabolism of R-and S-PZQ metabolism showed CYP1A2 and CYP2C19 to be the main contributors of CYP metabolism using the substrate disappearance assay and the selective inhibition assay where the concentration of the isoforms were 1 μmol/L (Tables 1 and 2). There was significant enantiomer selectivity by CYP1A2 for R-PZQ and CYP3A4/5 for S-PZQ with CYP2C19 contributing to both R and S metabolism (  (Table 5). However, if using the metabolite formation method, CYP3A4 contribute mostly to R-PZQ metabolism, which was not observed in this study in man.
The major disadvantage of the metabolite formation approach is that  We are still facing challenges to unequivocally establish the chemical structure of X-OH-PZQ. However, assuming a 1:1 analytical response of trans-4-OH-PZQ and X-OH-PZQ, the enzyme kinetic data ( Note: Statistical calculations for AUC, and C max were based on ln-transformed data. Bioequivalence criteria are defined as 90% CI of the geometric mean ratios of T/R of between 80.0% and 125.0% for AUC inf , and C max . A single tailed, paired student t-test was used to test for the differences between the means of the critical PK parameters: AUC, C max , T max , clearance, elimination rate constant (K el ), and the apparent volume of distribution.
The significance level was set at α = 0.05.
Abbreviations: AUC inf , AUC from time zero to infinity; AUC t 0 −t last area under the plasma concentration-time curve from time zero to the last sampled time point; CI, confidence interval; C max , peak plasma concentration of the drug; SD, standard deviation; T max , time needed to achieve C max .
R-PZQ ( Figure 5). However, with a limitation of lack of knowledge of the actual concentration of the drug that is presented to the drug metabolizing enzymes in the liver which might be different from the C max observed in plasma and this might not reflect the absorbed drug entering the liver through the hepatic portal vein and the fraction unbound in plasma. The maximum plasma concentration (C max ) and AUC obtained from the in vivo pharmacokinetic study was 0.16 µg/mL and 2.4 µg/mL × h for R-PZQ and 0.54 µg/mL 3.3 µg/mL × h for S-PZQ, respectively. The pharmacokinetic parameters obtained in our study are comparable to those reported in literature. The C max is given as 0.16 µg/mL and 0.52 µg/mL for R-and S-PZQ, respectively, from a previous study. 8,28 The bioavailability of R-PZQ is considerably less than that of S-PZQ. The co-administration with the CYP3A4/5 inhibitor KTZ only increased the exposure of the ineffective S-PZQ (68%) but not R-PZQ (9%). Previous studies have observed that R-PZQ is 100-1000 times more potent in terms of anti-schistosomal activity and hence therapeutic effect. 8,33 The increased exposure of S-PZQ is, therefore, not likely to result in increased efficacy but rather a risk for increased toxicity. Caution should be taken when PZQ is dosed with CYP3A4 inhibitors. Although our in vivo study did not predict any significant DDI with R-PZQ, values for S-PZQ where, however, significant as indicated by the 90% of the geometric mean ratio which was outside the FDA 80-125.
In this study we characterized the enantiomer selective metabolism of PZQ. R-PZQ being mainly metabolized by CYP1A2 and CYP2C19, whereas S-PZQ was mainly metabolized by CYP2C19 and CYP3A4. This finding adds to our knowledge of the potential metabolic basis of inter-individual variation in R-and S-PZQ exposure and resulting efficacy. It also provides a mechanistic basis of observed DDI when other drugs are co-administered with PZQ and the possible implications of such DDI for PZQ efficacy and safety.

CO N FLI C T O F I NTE R E S T
The authors declare that there is no conflict of interest.

DATA S H A R I N G A N D DATA ACCE SS I B I LIT Y
The data that support the findings of this study are available on request from the corresponding author (RT). The data are not publicly available due to privacy or ethical restrictions.