Prediction of human pharmacokinetics--evaluation of methods for prediction of hepatic metabolic clearance

Methods for prediction of hepatic clearance (CL(H)) in man have been evaluated. A physiologically-based in-vitro to in-vivo (PB-IVIV) method with human unbound fraction in blood (f(u, bl)) and hepatocyte intrinsic clearance (CL(int))-data has a good rationale and appears to give the best predictions (maximum approximately 2-fold errors; < 25% errors for half of CL-predictions; appropriate ranking). Inclusion of an empirical scaling factor is, however, needed, and reasons include the use of cryo-preserved hepatocytes with low activity, and inappropriate CL(int)- and f(u, bl)-estimation methods. Thus, an improvement of this methodology is possible and required. Neglect of f(u, bl) or incorporation of incubation binding does not seem appropriate. When microsome CL(int)-data are used with this approach, the CL(H) is underpredicted by 5- to 9-fold on average, and a 106-fold underprediction (attrition potential) has been observed. The poor performance could probably be related to permeation, binding and low metabolic activity. Inclusion of scaling factors and neglect of f(u, bl) for basic and neutral compounds improve microsome predictions. The performance is, however, still not satisfactory. Allometry incorrectly assumes that the determinants for CL(H) relate to body weight and overpredicts human liver blood flow rate. Consequently, allometric methods have poor predictability. Simple allometry has an average overprediction potential, > 2-fold errors for approximately 1/3 of predictions, and 140-fold underprediction to 5800-fold overprediction (potential safety risk) range. In-silico methodologies are available, but these need further development. Acceptable prediction errors for compounds with low and high CL(H) should be approximately 50 and approximately 10%, respectively. In conclusion, it is recommended that PB-IVIV with human hepatocyte CL(int) and f(u, bl) is applied and improved, limits for acceptable errors are decreased, and that animal CL(H)-studies and allometry are avoided.     

On the observed large interspecies overprediction of human clearance ("vertical allometry") of UCN-01: further support for a proposed model based on plasma protein binding

The prediction of a human clearance (CL) value for UCN-01, an extreme example of vertical allometry (a large overprediction by allometric scaling), was examined using commonly used simple allometry and the "rule of exponents," as well as a newly proposed model, which quantitatively incorporates plasma protein-binding information from rats and humans. Simple allometry and the rule of exponents were shown to overpredict the human CL value of UCN-01 by about 5000- and 1750-fold, respectively. The new model incorporating the ratio of fraction unbound between rats and humans improved the prediction by about 20-fold compared to the rule of exponents. The model is expected to improve if a more accurate measurement of the unbound fraction in human plasma is obtained. The prediction of volume distribution for UCN-01 by allometric scaling was also shown to be dependent on the difference of fraction unbound between animal species and humans. In summary, plasma protein binding has been demonstrated to be an important measure for interspecies scaling of pharmacokinetics.     

Interspecies scaling of a camptothecin analogue: human predictions for intravenous topotecan using animal data

As a class, camptothecin analogues via market entry of topotecan and irinotecan, have shown promise for the treatment of various solid tumours. Topotecan, in particular, was chosen as the substrate for allometric scaling and prediction of human parameter values for both total clearance (CL) and volume of distribution (V(ss)). The availability of published data in mouse, rat, dog, and monkey paved the way for interspecies scaling via allometry. Although it appeared that at a minimum mouse, rat, and dog would reasonably fit in a three-species allometry scale-up, the inclusion of monkey data enabled a better prediction of the human parameter values for total topotecan-e.g., CL: allometric equation: 1.5234W(0.7865); predicted value = 43.04 l h(-1): observed CL = 24-53 l h(-1); V(ss): allometric equation: 1.1939W(1.0208); predicted value = 91.29 litres: observed V(ss) = 66-146 litres. The proximity of the allometric exponent values of CL (0.7885) and V(ss) (1.0208) to the suggested values of 0.75 and 1.00 was not only encouraging, but also confirmed the applicability of interspecies scaling approach for topotecan. The data suggest that allometric scaling approaches with suitable correction factors could potentially be used to predict the human pharmacokinetics of novel CPT analogues prospectively.     

Prediction of human pharmacokinetics--improving microsome-based predictions of hepatic metabolic clearance

Physiologically based methods generally perform poorly in predicting in-vivo hepatic CL (CL(H)) from intrinsic clearance (CL(int)) in microsomes in-vitro and unbound fraction in blood (f(u,bl)). Various strategies to improve the predictability have been developed, and inclusion of an empirical scaling factor (SF) seems to give the best results. This investigation was undertaken to evaluate this methodology and to find ways to improve it further. The work was based on a diverse data set taken from Ito and Houston (2005). Another objective was to evaluate whether rationalization of CL(H) predictions can be made by replacing blood/plasma-concentration ratio (C(bl)/C(pl)) measurements with SFs. There were apparently no or weak correlations between prediction errors and lipophilicity, permeability (compounds with low permeability missing in the data set) and main metabolizing CYP450s. The use of CL(int) class (high/low) and drug class (acid/base/neutral) SFs (the CD-SF method) gives improved and reasonable predictions: 1.3-fold median error (an accurate prediction has a 1-fold error), 76% within 2-fold-error, and a median absolute rank ordering error of 2 for CL(H) (n = 29). This approach is better than the method with a single SF. Mean (P < 0.05) and median errors, fraction within certain error ranges, higher percentage with most accurate predictions, and ranking were all better, and 76% of predictions were more accurate with this new method. Results are particularly good for bases, which generally have higher CL(H) and the potential to be incorrectly selected/rejected as candidate drugs. Reasonable predictions of f(u,bl) can be made from plasma f(u) (f(u,pl)) and empirical blood cell binding SFs (B-SFs; 1 for low f(u,pl) acids; 0.62 for other substances). Mean and median f(u,bl) prediction errors are negligible. The use of the CD-SF method with predicted f(u,bl) (the BCD-SF method) also gives improved and reasonable results (1.4-fold median error; 66% within 2-fold-error; median absolute rank ordering error = 1). This new empirical approach seems sufficiently good for use during the early screening; it gives reasonable estimates of CL(H) and good ranking, which allows replacement of C(bl)/C(pl) measurements by a simple equation.     

Pharmacokinetics and interspecies scaling of a novel VEGF receptor inhibitor, SU5416.

The pharmacokinetics and allometric relationships of SU5416, a novel small anti-angiogenesis agent, were studied. The pharmacokinetics of SU5416 were examined in mice, rats, dogs, and cancer patients. The in-vitro intrinsic clearance (CLint) was estimated from the in-vitro metabolism study in mouse, rat, dog, monkey and human liver microsomes. The parameters of interest were correlated across species as a function of bodyweight using an allometric approach. The steady-state volume of distribution (Vd(ss)), plasma clearance (CLs), and CLint of SU5416 were well correlated across species. The exponent of the allometric relationship (b) of the corresponding parameters was 0.92, 0.80 and 0.66, respectively. The elimination half-life (t1/2) was consistent across species and independent of bodyweight. The prediction of CLs, Vd(ss), CLint, and t1/2 in humans using the data from mouse, rat, and dog, and monkey (for CLint) was reasonably good (within 4-fold of the observed values). However, an improved prediction (within 2-fold of the observed values) of the corresponding parameters in humans was obtained when extrapolation from only the rodent data was performed, suggesting that the rodent data are sufficient for the scale-up of SU5416 pharmacokinetic parameters in humans. Using allometry, it was possible to achieve reasonable predictions of the pharmacokinetic parameters of SU5416 in cancer patients with various solid tumours.     

Prediction of human pharmacokinetics using physiologically based modeling: a retrospective analysis of 26 clinically tested drugs.

The aim of this study was to evaluate different physiologically based modeling strategies for the prediction of human pharmacokinetics. Plasma profiles after intravenous and oral dosing were simulated for 26 clinically tested drugs. Two mechanism-based predictions of human tissue-to-plasma partitioning (P(tp)) from physicochemical input (method Vd1) were evaluated for their ability to describe human volume of distribution at steady state (V(ss)). This method was compared with a strategy that combined predicted and experimentally determined in vivo rat P(tp) data (method Vd2). Best V(ss) predictions were obtained using method Vd2, providing that rat P(tp) input was corrected for interspecies differences in plasma protein binding (84% within 2-fold). V(ss) predictions from physicochemical input alone were poor (32% within 2-fold). Total body clearance (CL) was predicted as the sum of scaled rat renal clearance and hepatic clearance projected from in vitro metabolism data. Best CL predictions were obtained by disregarding both blood and microsomal or hepatocyte binding (method CL2, 74% within 2-fold), whereas strong bias was seen using both blood and microsomal or hepatocyte binding (method CL1, 53% within 2-fold). The physiologically based pharmacokinetics (PBPK) model, which combined methods Vd2 and CL2 yielded the most accurate predictions of in vivo terminal half-life (69% within 2-fold). The Gastroplus advanced compartmental absorption and transit model was used to construct an absorption-disposition model and provided accurate predictions of area under the plasma concentration-time profile, oral apparent volume of distribution, and maximum plasma concentration after oral dosing, with 74%, 70%, and 65% within 2-fold, respectively. This evaluation demonstrates that PBPK models can lead to reasonable predictions of human pharmacokinetics.     

Predicting oral clearance in humans: how close can we get with allometry?

BACKGROUND: Oral clearance (CL/F) is an important pharmacokinetic parameter and plays an important role in the selection of a safe and tolerable dose for first-in-human studies. Throughout the pharmaceutical industry, many drugs are administered via the oral route; however, there are only a handful of published scaling studies for the prediction of oral pharmacokinetic parameters. METHODS: We evaluated the predictive performances of four different allometric approaches -- simple allometry (SA), the rule of exponents, the unbound CL/F approach, and the unbound fraction corrected intercept method (FCIM) -- for the prediction of human CL/F and the oral area under the plasma concentration-time curve (AUC). Twenty-four compounds developed at Johnson and Johnson Pharmaceutical Research and Development, covering a wide range of physicochemical and pharmacokinetic properties, were selected. The CL/F was predicted using these approaches, and the oral AUC was then estimated using the predicted CL/F. RESULTS: The results of this study indicated that the most successful predictions of CL/F and the oral AUC were obtained using the unbound CL/F approach in combination with the maximum lifespan potential or the brain weight as correction factors based on the rule of exponents. We also observed that the unbound CL/F approach gave better predictions when the exponent of SA was between 0.5 and 1.2. However, the FCIM seemed to be the method of choice when the exponent of SA was <0.50 or >1.2. CONCLUSIONS: Overall, we were able to predict CL/F and the oral AUC within 2-fold of the observed value for 79% and 83% of the compounds, respectively, by selecting the allometric approaches based on the exponents of SA.     

Determination of volume of distribution at steady state with complete consideration of the kinetics of protein and tissue binding in linear pharmacokinetics

The assumption of an instant equilibrium between bound and unbound drug fractions is commonly applied in pharmacokinetic calculations. The equation for the calculation of the steady-state volume of distribution V(ss) from the time curve of drug concentration in plasma after intravenous bolus dose administration, which does not assume an immediate equilibrium and thus incorporates dissociation and association rates of protein and tissue binding, is presented. The equation obtained V(ss) = (Dose/AUC)*MRT(u) looks like the traditional equation, but instead of mean residence time MRT calculated using the total drug concentration in plasma, it contains mean residence time MRT(u) calculated using the plasma concentration of the unbound drug. The equation connecting MRT(u) and MRT is derived. If an immediate equilibrium between bound and unbound drug fractions occurs, MRT(u) and MRT are the same, but in general, MRT(u) is always smaller than MRT. For drugs with high protein affinity and slow dissociation rate MRT(u) may be of an order of several hours smaller than MRT, so that V(ss) can be considerably overestimated in the traditional calculation.     

Mechanistic Approaches to Volume of Distribution Predictions: Understanding the Processes

PURPOSE: To use recently developed mechanistic equations to predict tissue-to-plasma water partition coefficients (Kpus), apply these predictions to whole body unbound volume of distribution at steady state (Vu(ss)) determinations, and explain the differences in the extent of drug distribution both within and across the various compound classes. MATERIALS AND METHODS: Vu(ss) values were predicted for 92 structurally diverse compounds in rats and 140 in humans by two approaches. The first approach incorporated Kpu values predicted for 13 tissues whereas the second was restricted to muscle. RESULTS: The prediction accuracy was good for both approaches in rats and humans, with 64-78% and 82-92% of the predicted Vu(ss) values agreeing with in vivo data to within factors of +/-2 and 3, respectively. CONCLUSIONS: Generic distribution processes were identified as lipid partitioning and dissolution where the former is higher for lipophilic unionised drugs. In addition, electrostatic interactions with acidic phospholipids can predominate for ionised bases when affinities (reflected by binding to constituents within blood) are high. For acidic drugs albumin binding dominates when plasma protein binding is high. This ability to explain drug distribution and link it to physicochemical properties can help guide the compound selection process.     

Advancing prediction of tissue distribution and volume of distribution of highly lipophilic compounds from a simplified tissue-composition-based model as a mechanistic animal alternative method

It has been reported that values of tissue-plasma ratios (K(p)) and resulting volume of distribution at steady state (V(ss)) are substantially overpredicted for several highly lipophilic drugs. This effect was observed particularly with the published version of the tissue-composition-based model, which used experimentally determined unbound fraction in plasma (fu(p)) as input for drugs. The reasons for the unreasonably high V(ss) predictions were investigated in this study for 14 highly lipophilic compounds with a log n-octanol-water partition coefficient (log P(ow)) of at least 5.8. Here, we argue that the experimentally determined fu(p) is inaccurate for these compounds, which affected the prediction of K(p) and V(ss). Alternatively, the tissue-plasma ratio of neutral lipids (nl) equivalent was used as the main factor governing K(p), and hence V(ss), in addition to log P(ow). The average fold error of deviation between the predicted and observed human V(ss) is 124 for the published model, whereas it significantly decreased to 1.5 for the proposed model. The sensitivity analysis confirmed the importance of nl content and drug lipophilicity. Overall, this study proposes a generic and simplified tissue-composition-based model for highly lipophilic drugs and chemicals, which is a step forward toward improving prediction of V(ss) into physiologically based pharmacokinetic (PBPK) models.     

Interspecies scaling: prediction of human biliary clearance and comparison with QSPKR

The aim of this study was to evaluate the prediction performance of various allometric scaling methods in predicting human biliary clearance (CL(b)) from data in rats or multiple animal species and to compare the prediction performance with that of quantitative structure pharmacokinetic relationship (QSPKR) models. CL(b) data of parent drugs in rats and humans were collected from the literature for 18 compounds. A simple allometric approach was applied to CL(b) or unbound CL(b) using 0.75 or 0.66 as the allometric exponent. For scaling from rat studies alone, the prediction using 0.66 as the exponent was better than that using 0.75, and a better prediction was obtained for unbound CL(b) than CL(b). For a subset of compounds, six multiple-species scaling methods were compared, with the best prediction achieved with the simple unbound CL(b) approach. However, in the absence of protein binding data, the correction with maximum life-span potential (MLP) or 'Rule of exponent' (ROE) method offered the best prediction. Overall, multiple species had better predictability than scaling with the rat alone. Comparison of predicted human CL(b) values using multiple animal species and QSPKR offered similar prediction performance. In conclusion, the results of the present study, although based on limited data, suggested that the prediction for human CL(b) by allometry was greatly improved by the incorporation of protein binding. Human CL(b) prediction using rat data alone was not satisfactory. Additionally, QSPKR provides an alternative approach to allometry for the prediction of human biliary clearance.     

Pharmacokinetics and allometric scaling of levormeloxifene,a selective oestrogen receptor modulator

The pharmacokinetics of a new selective oestrogen receptor modulator levormeloxifene was investigated in mice, rats, cynomolgus monkeys and humans by compartmental pharmacokinetics. Levormeloxifene was administered as an oral solution in all studies. Allometric scaling was used to predict human pharmacokinetic parameters and the performance of the approach was evaluated. Mean values of clearance confounded by F(CL/F) were 0.073, 0.29, 3.18 and 2.4 l/h in mice, rats, monkeys and humans, respectively. Values of distribution volume at steady state confounded by F(V(ss)/F) were 0.073 and 7.5 l in mice and rats. In monkeys, values of the central volume F(V(c)/F) and volume at steady state F(V(ss)/F) were 28.9 and 57.9 l, respectively. In humans, values of V(c)/F and V(ss)/F were 106 and 587 l, respectively. Predicted CL/F and V(ss)/F showed a linear relationship when plotted vs BW on a log-log scale; for CL/F, r was 0.95-0.98 and for V(ss)/F, r was 0.99. Using allometric scaling the predicted human V(ss)/F deviated 3-fold from the experimentally determined values. Observed values of CL/F deviated 21-25 fold from the predicted, the latter depending on the scaling method. Confidence intervals for the predicted parameters showed major lack of precision for all the allometric scaling methods.     

Predicted values for human total clearance of a variety of typical compounds with differently humanized-liver mouse plasma data

Prediction of human pharmacokinetics is important in the preclinical stage. Values for total clearance of compounds from plasma should be one of the most important pharmacokinetic parameters for predictions. Although several physiological and empirical methods including single-species allometry for prediction of values for human clearance of compounds using humanized-liver mice have been reported, further improvement of prediction accuracies would be still expected. To optimize these approaches, we proposed methods for unbound intrinsic clearance in virtually 100% humanized-liver mouse by incorporating unbound plasma fractions of compounds in differently humanized-liver mice. Comparisons of prediction accuracies of values for human clearance of 15 model compounds were performed among our current physiological and previously reported models and single-species allometry using humanized-liver mice. Incorporation of the actual unbound plasma fractions of compounds and correction of residual mice hepatocyte in humanized-liver mice showed comparable prediction accuracy to that by single-species allometry. After exclusion of 3 compounds with large species differences in values of clearance and unbound plasma fractions between mice and humans out of 15 compounds, prediction accuracies were improved in the methods investigated. The previously and present reported physiological methods could show the good prediction accuracy of values for clearance of drugs from plasma.     

Prediction of the disposition of midazolam in surgical patients by a physiologically based pharmacokinetic model.

The aim of this study was to predict the disposition of midazolam in individual surgical patients by physiologically based pharmacokinetic (PBPK) modeling and explore the causes of interindividual variability. Tissue-plasma partition coefficients (k(p)) were scaled from rat to human values by a physiologically realistic four-compartment model for each tissue, incorporating the measured unbound fraction (f(u)) of midazolam in the plasma of each patient. Body composition (lean body mass versus adipose tissue) was then estimated in each patient, and the volume of distribution at steady state (V(dss)) of midazolam was calculated. Total clearance (CL) was calculated from unbound intrinsic CL, f(u), and estimated hepatic blood flow. Curves of midazolam plasma concentration versus time were finally predicted by means of a perfusion-limited PBPK model and compared with measured data. In a first study on 14 young patients undergoing surgery with modest blood loss, V(dss) was predicted with an only 3.4% mean error (range -24-+39%) and a correlation between predicted and measured values of 0.818 (p < 0.001). Scaling of k(p) values by the four-compartment model gave better predictions of V(dss) than scaling using unbound k(p). In the PBPK modeling, the mean +/- standard deviation (SD) prediction error for all data was 9.7 +/- 33%. In a second study with 10 elderly patients undergoing orthopedic surgery, hemodilution and blood loss led to a higher f(u) of midazolam. The PBPK modeling correctly predicted a marked increase in V(dss), a smaller increase in CL, and a prolonged terminal half-life of midazolam, as compared with findings in the first study. Interindividual variation in the disposition of midazolam could thus in part be related to the physiological characteristics of the patients and the f(u) of the drug in their plasma.     

Prediction of volume of distribution values in humans for neutral and basic drugs using physicochemical measurements and plasma protein binding data

We present a method for the prediction of volume of distribution in humans, for neutral and basic compounds. It is based on two experimentally determined physicochemical parameters, ElogD(7.4) and f(i(7.4)), the latter being the fraction of compound ionized at pH 7.4 and on the fraction of free drug in plasma (f(u)). The fraction unbound in tissues (f(ut)), determined via a regression analysis from 64 compounds using the parameters described, is then used to predict VD(ss) via the Oie-Tozer equation. Accuracy of this method was determined using a test set of 14 compounds, and it was demonstrated that human VD(ss) values could be predicted, on average, within or very close to 2-fold of the actual value. The present method is as accurate as reported methods based on animal pharmacokinetic data, using a similar set of compounds, and ranges between 1.62 and 2.20 as mean-fold error. This method has the advantage of being amenable to automation, and therefore fast throughput, it is compound and resources sparing, and it offers a rationale for the reduction of the use of animals in pharmacokinetic studies. A discussion of the potential errors that may be encountered, including errors in the determination of f(u), is offered, and the caveats about the use of computed vs experimentally determined logD and pK(a) values are addressed.     

Comparison of prediction methods for in vivo clearance of (S,S)-3-[3-(methylsulfonyl)phenyl]-1-propylpiperidine hydrochloride, a dopamine D2 receptor antagonist, in humans.

The purpose of this study is to investigate reliable prediction methods for in vivo pharmacokinetics and the likelihood of drug interactions with several cytochrome P450 inhibitors in humans for (S,S)-3-[3-(methylsulfonyl)phenyl]-1-propylpiperidine (PNU-96391). By allometric scaling of in vivo animal data, clearance of PNU-96391 in humans was over-predicted by 4-fold, half-life was under-predicted by 3-fold, and volume of distribution was accurately predicted. High correlation coefficients (>0.99) were observed for these parameters. Neither the in vitro-in vivo correlation approach nor the modified allometric scaling with maximum life span potential or brain weight accurately provided the predicted clearance value. Using an alternative method, based on normalization of in vitro human data with the ratio of in vivo to in vitro animal data, the in vivo clearance in humans was predicted to be 0.39 l/h/kg. This value correlated well with the in vivo value (0.43 l/h/kg). Regarding the interactions of PNU-96391 with cytochrome P450 inhibitors, only quinidine, haloperidol, and ketoconazole showed significant inhibition on the metabolic clearance of PNU-96391 in human hepatocytes. By comparing in vitro K(i) values with in vivo maximum unbound concentrations of the inhibitor, the increases in systemic exposure of PNU-96391 by coadministration of the inhibitors were estimated to be less than 1.5-fold. A preliminary comparison of pharmacokinetics of PNU-96391 between CYP2D6 extensive and poor metabolizers in the clinical study showed only a slight increase in systemic exposure in poor metabolizers (approximately 1.4-fold as area under the concentration-time curve). Therefore, clinically significant drug-drug interactions of PNU-96391 would be unlikely to occur with coadministration of CYP2D6 inhibitors.     

The prediction of drug metabolism, tissue distribution, and bioavailability of 50 structurally diverse compounds in rat using mechanism-based absorption, distribution, and metabolism prediction tools.

The aim of this study was to assess a physiologically based modeling approach for predicting drug metabolism, tissue distribution, and bioavailability in rat for a structurally diverse set of neutral and moderate-to-strong basic compounds (n = 50). Hepatic blood clearance (CL(h)) was projected using microsomal data and shown to be well predicted, irrespective of the type of hepatic extraction model (80% within 2-fold). Best predictions of CL(h) were obtained disregarding both plasma and microsomal protein binding, whereas strong bias was seen using either blood binding only or both plasma and microsomal protein binding. Two mechanistic tissue composition-based equations were evaluated for predicting volume of distribution (V(dss)) and tissue-to-plasma partitioning (P(tp)). A first approach, which accounted for ionic interactions with acidic phospholipids, resulted in accurate predictions of V(dss) (80% within 2-fold). In contrast, a second approach, which disregarded ionic interactions, was a poor predictor of V(dss) (60% within 2-fold). The first approach also yielded accurate predictions of P(tp) in muscle, heart, and kidney (80% within 3-fold), whereas in lung, liver, and brain, predictions ranged from 47% to 62% within 3-fold. Using the second approach, P(tp) prediction accuracy in muscle, heart, and kidney was on average 70% within 3-fold, and ranged from 24% to 54% in all other tissues. Combining all methods for predicting V(dss) and CL(h) resulted in accurate predictions of the in vivo half-life (70% within 2-fold). Oral bioavailability was well predicted using CL(h) data and Gastroplus Software (80% within 2-fold). These results illustrate that physiologically based prediction tools can provide accurate predictions of rat pharmacokinetics.     

Comparative evaluation of US Food and Drug Administration and pharmacologically guided approaches to determine the maximum recommended starting dose for first-in-human clinical trials in adult healthy men

The authors compared US Food and Drug Administration (FDA) and 9 pharmacologically guided approaches (PGAs; simple allometry, maximum life span potential [MLP], brain weight, rule of exponent [ROE], two 2-sp methods and 3 one-sp methods) to determine the maximum recommended starting dose (MRSD) for first-in-human clinical trials in adult healthy men using 10 drugs. The ROE method as suggested by Mahmood and Balian1 gave the best prediction accuracy for a pharmacokinetic (PK) parameter. Values derived from clearance were consistently better than volume of distribution (Vd)-based methods and had lower root mean square error (RMSE) values. A pictorial method evaluation chart was developed based on fold errors for simultaneous evaluation of various methods. The one-sp method (rat) and the US FDA methods gave the highest prediction accuracy and low RMSE values, and the 2-sp methods gave the least prediction accuracy with high RMSE values. The ROE method gave more consistent predictions for PK parameters than other allometric methods. Despite this, the MRSD predictions were not better than US FDA methods, probably indicating that across-species variation in clearance may be higher than variation in no observed adverse effect level (NOAEL) and that PGA methods may not be consistently better than the NOAEL based methods.     

Pharmacokinetic scaling of bisphenol A by species-invariant time methods

1. The study was performed to predict the pharmacokinetic disposition of bisphenol A in humans using simple allometry and several species-invariant time methods based on animal data. Bisphenol A was injected intravenously to mouse, rat, rabbit and dog (1-2 mg kg(-1) doses). 2. The obtained serum concentration-time profiles were best described by bi-exponential equations in all these animal species, with the mean Cl, V(ss) and t(1/2) of 0.3 l h(-1), 0.1 litres and 39.9 min in mouse, 1.9 l h(-1), 1.3 litres and 37.6 min in rat, 12.6 l h(-1), 7.1 litres and 40.8 min in rabbit, and 27.1 l h(-1), 20.0 litres and 43.7 min in dog, respectively. 3. The human pharmacokinetic parameters of Cl, V(ss) and t(1/2) were predicted by simple allometry as well as by normalization according to species-invariant times of kallynochrons, apolysichrons and dienetichrons. 4. The simple allometric scaling and different time transformation methods predicted the human Cl, V(ss) and t(1/2) ranging from 46.0 to 127.1 l h(-1), 125.3 to 229.7 litres and 43.6 to 196.2 min, respectively. Species-invariant time transformations showed that all animal data from the four species were superimposable. These preliminary parameter values may be useful in interpreting toxicity data in humans on environmental exposure to bisphenol A.     

Evaluation of methods for predicting drug-drug interactions by Monte Carlo simulation

The ratio of the inhibitor concentration to the inhibition constant (K(i)) is used as the index for predicting drug-drug interactions involving metabolic inhibition. The maximum unbound concentration in the circulation (I(p, max, u)) and the maximum unbound concentration at the inlet to the liver (I(u, max)) have been used for the inhibitor concentration. In the present study, the methods for predicting drug-drug interactions using these concentrations were evaluated by Monte Carlo simulation. Information on the pharmacokinetic parameters of drugs and the K(i) values for cytochrome P450(CYP) were obtained from the literature. It was assumed that the pharmacokinetic parameters (intrinsic metabolic clearance, renal clearance and distribution volume for unbound fraction), serum protein binding and K(i) value for substrate and inhibitor are all log-normally distributed. Correlations among the parameters were assessed and were used for further simulations. A change in AUC of the substrate following co-administration of the inhibitor was simulated 1000 times using the physiologically based pharmacokinetic (PBPK) model. The percent of the drug combinations which exhibited a significant increase in the AUC (>125%) was 16.2% of the total combinations. The cases where the I/K(i) using I(u, max) and I(p, max, u) overestimated compared with the actual increased ratio of AUC (false positive prediction) were 41.2% and 16.7%, respectively. The cases where the predicted ratios of AUC from I/K(i) using I(u, max) and I(p, max, u) were comparable with the actual ratio were 3.2% and 8.7%, respectively. The prediction using I(p, max, u) was, thus, more reliable than that using I(u, max). However, in the case of I(u, max), there was no case where the actual increased ratio of AUC was greater than that predicted from I/K(i) (false negative prediction). On the other hand, for I(p, max, u), the rate of false negative prediction was 1.4%. The present study indicates that I(u, max) is better than I(p, max, u) for avoiding false negative predictions and I(p, max, u) is better than I(u, max) for increasing the probability of true positive and true negative predictions and avoiding false positive predictions.In conclusion, it is necessary to use both predictions involving I(u, max) and I(p, max, u) and to use them early on during the development stage of drug candidates. In order to finally choose which compound(s) to take forward to clinical trials, when predicting an interaction, the more quantitative and reliable method based on the PBPK model needs to be used.