Evaluation of dissolution profile similarity - Comparison between the f2, the multivariate statistical distance and the f2 bootstrapping methods.

A simulation study is presented, evaluating the performance of the f2, the model-independent multivariate statistical distance and the f2 bootstrap methods in the ability to conclude similarity between two dissolution profiles. Different dissolution profiles, based on the Noyes-Whitney equation and ranging from theoretical f2 values between 100 and 40, were simulated. Variability was introduced in the dissolution model parameters in an increasing order, ranging from a situation complying with the European guidelines requirements for the use of the f2 metric to several situations where the f2 metric could not be used anymore. Results have shown that the f2 is an acceptable metric when used according to the regulatory requirements, but loses its applicability when variability increases. The multivariate statistical distance presented contradictory results in several of the simulation scenarios, which makes it an unreliable metric for dissolution profile comparisons. The bootstrap f2, although conservative in its conclusions is an alternative suitable method. Overall, as variability increases, all of the discussed methods reveal problems that can only be solved by increasing the number of dosage form units used in the comparison, which is usually not practical or feasible. Additionally, experimental corrective measures may be undertaken in order to reduce the overall variability, particularly when it is shown that it is mainly due to the dissolution assessment instead of being intrinsic to the dosage form.

Prediction of drug distribution in rat and humans using an artificial neural networks ensemble and a PBPK model.

PURPOSE: To develop a QSAR model, based on calculated molecular descriptors and an Artificial Neural Networks Ensemble (ANNE), for the estimation of rat tissue-to-blood partition coefficients (Kt:b), as well as the assessment of the applicability domain of the model and its utility in predicting the drug distribution in humans. METHODS: A total of 1460 individual Kt:b values (75% train and 25% validation), obtained in 13 different rat tissues were collected in the literature. A correlation between simple molecular descriptors for lipophilicity, ionization, size and hydrogen bonding capacity and Kt:b data was attempted by using an ANNE. RESULTS: Similar statistics were observed between the train and validation group of data with correlations, between the observed values and the predicted average ANNE values, of 0.909 and 0.896, respectively. A degradation of the correlations was observed for predicted values with high uncertainty, as judged by the standard deviations of the ANNE outputs. This was further observed when using the ANNE Kt:b values in a Physiologically based pharmacokinetic (PBPK) model for predicting the Human Volume of distribution of another 532 drugs. CONCLUSIONS: This model (available as a MS Excel® workbook in the Supporting material of this article) may be a valuable tool for prediction and simulation in early drug development, allowing the in silico estimation of rat Kt:b values for PBPK purposes and also indicating its applicability domain.

Tissue-to-blood distribution coefficients in the rat: utility for estimation of the volume of distribution in man.

A compilation of rat tissue-to-blood partition coefficient data obtained both in vitro and in vivo in thirteen different tissues for a total of 309 different drugs is presented. An evaluation of the relationship between several fundamental physicochemical molecular descriptors and these distribution parameters was made. In addition, the ability to predict the Human Volume of distribution by regression analysis and by a Physiologically-based approach was also tested. Results have shown different trends between the drug classes and tissues, consistent with earlier described relationships between physicochemical properties and pharmacokinetic behavior. It was also possible to conclude for the acceptable ability to predict the volume of distribution in Humans by both regression and mechanistic approaches, which suggests that this type of data represents a convenient tool to describe the drug distribution on a new drug development context. These observations and analyses, along with the large database of rat tissue distribution data, should enable future efforts aimed toward developing a full in silico quantitative structure-pharmacokinetic relationships and improving our understanding of the correlations between fundamental chemical characteristics and drug distribution.

Prediction of the human oral bioavailability by using in vitro and in silico drug related parameters in a physiologically based absorption model.

Estimates of the human oral absolute bioavailability were made by using a physiological-based pharmacokinetic model of absorption and the drug solubility at the gastrointestinal pH range 1.5-7.5, the apparent permeability (P(app)) in Caco-2 cells and the intrinsic clearance (Cl(int)) in human hepatocytes suspensions as major drug related parameters. The predictive ability of this approach was tested in 164 drugs divided in four levels of input data: (i) in vitro data for both P(app) and Cl(int); (ii) in vitro data for Cl(int) only; (iii) in vitro data for P(app) only and (iv) in silico data for both P(app) and Cl(int). In all scenarios, solubility was estimated in silico. Excellent predictive abilities were observed when in vitro data for both P(app) and Cl(int) were used, with 84% of drugs with oral bioavailability predictions within a±20% interval of the correct value. This predictive ability is reduced with the introduction of the in silico estimated parameters, particularly when Cl(int) is used. Performance of the model using only in silico data provided 53% of drugs with bioavailability predictions within a±20% acceptance interval. However, 74% of drugs in the same scenario resulted in bioavailability predictions within a±35% interval, which indicates that a qualitative prediction of the absolute bioavailability is still possible. This approach is a valuable way to estimate a fundamental pharmacokinetic parameter, using data typically collected in the drug discovery and early development phases, providing also mechanistic information of the limiting bioavailability steps of the drug.

An alternative single dose parameter to avoid the need for steady-state studies on oral extended-release drug products.

Use of single and multiple-dose studies is required to establish the bioequivalence between two extended-release oral dosage forms under the current European Guidelines. However, FDA is less strict in this regard and only requires a single-dose study. The objective of this work is to use a computer simulation in order to test the two approaches. Three pharmacokinetic models, representing different release mechanisms, were considered, and Monte Carlo simulations with intra- and inter-individual variabilities were performed. Five different bioequivalence protocols were used and a new pharmacokinetic metric -C(τ), the concentration at the end of the intended dosing interval obtained in the single-dose study - is proposed in order to avoid the need for steady-state studies while keeping the ability to detect differences between formulations. Results have shown that the European requirements are more capable to discriminate between two potentially different formulations but at the cost of the multiple-dose study and with an increased number of subjects when compared to the FDA requirements. However, the use of C(max) and AUC(0-)(t) obtained on a single-dose study with the added C(τ) metric equals the discriminatory ability of the current EMA requirements, without the need of a multiple-dose study. This proposed approach results in the reduction in the number of studies and volunteers enrolled in clinical bioequivalence trials, without compromising the quality assurance of a new extended-release oral formulation.

Prediction of the in vitro permeability determined in Caco-2 cells by using artificial neural networks.

Caco-2 cells are currently the most used in vitro tool for prediction of the potential oral absorption of new drugs. The existence of computational models based on this data may potentiate the early selection process of new drugs, but the current models are based on a limited number of cases or on a reduced molecular space. We present an artificial neural network based only on calculated molecular descriptors for modelling 296 in vitro Caco-2 apparent permeability (P(app)) drug values collected in the literature using also a pruning procedure for reducing the descriptors space. LogP(app) values were divided into a training group of 192 drugs for network optimization and a testing group of another 59 drugs for early stop and internal validation resulting in correlations of 0.843 and 0.702 and RMSE of 0.546 and 0.791 for the training and testing group, respectively. External validation was made with an additional group of 45 drugs with a correlation of 0.774 and RMSE of 0.601. The selected molecular descriptors encode information related to the lipophilicity, electronegativity, size, shape and flexibility characteristics of the molecules, which are related to drug absorption. This model may be a valuable tool for prediction and simulation in the drug development process, as it allows the in silico estimation of the in vitro Caco-2 apparent permeability.

Prediction of the in vitro intrinsic clearance determined in suspensions of human hepatocytes by using artificial neural networks.

Use of in vitro suspensions of human hepatocytes is currently accepted as one of the most promising tools for prediction of metabolic clearance in new drugs. The possibility of creating computational models based on this data may potentiate the early selection process of new drugs. We present an artificial neural network for modelling human hepatocyte intrinsic clearances (CL(int)) based only on calculated molecular descriptors. In vitro CL(int) data obtained in human hepatocytes suspensions was divided into a train group of 71 drugs for network optimization and a test group of another 18 drugs for early-stop and internal validation resulting in correlations of 0.953 and 0.804 for the train and test group respectively. The model applicability was tested with 112 drugs by comparing the in silico predicted CL(int) with the in vivo CL(int) estimated by the "well-stirred" model based on the in vivo hepatic clearance (CL(H)). Acceptable correlations were observed with r values of 0.508 and 63% of drugs within a 10-fold difference when considering blood binding in acidic drugs only. This model may be a valuable tool for prediction and simulation in the drug development process, allowing the in silico estimation of the human in vivo hepatic clearance.

Prediction of drug distribution within blood.

Drug distribution in blood, defined as drug blood-to-plasma concentration ratio (R(b)), is a fundamental pharmacokinetic parameter. It relates the plasma clearance to the blood clearance, enabling the physiological interpretation of this parameter. Although easily experimentally determined, R(b) values are lacking for the vast majority of drugs. We present a systematic approach using mechanistic, partial least squares (PLS) regression and artificial neural network (ANN) models to relate various in vitro and in silico molecular descriptors to a dataset of 93 drug R(b) values collected in the literature. The ANN model resulted in the best overall approach, with r(2)=0.927 and r(2)=0.871 for the train and the test sets, respectively. PLS regression presented r(2)=0.557 for the train and r(2)=0.656 for the test set. The mechanistic model provided the worst results, with r(2)=0.342 and, additionally, is limited to drugs with a basic ionised group with pKa<7. The ANN model for drug distribution in blood can be a valuable tool in clinical pharmacokinetics as well as in new drug design, providing predictions of R(b) with a percentage of correct values within a 1.25-fold error of 86%, 84% and 87% in the train, test and validation set of data.