Perinatal Azithromycin Provides Limited Neuroprotection in an Ovine Model of Neonatal Hypoxic-Ischemic Encephalopathy

BACKGROUND

Hypoxic-ischemic brain injury/encephalopathy affects about 1.15 million neonates per year, 96% of whom are born in low- and middle-income countries. Therapeutic hypothermia is not effective in this setting, possibly because injury occurs significantly before birth. Here, we studied the pharmacokinetics, safety, and efficacy of perinatal azithromycin administration in near-term lambs following global ischemic injury to support earlier treatment approaches.

METHODS

Ewes and their lambs of both sexes (n=34, 141-143 days) were randomly assigned to receive azithromycin or placebo before delivery as well as postnatally. Lambs were subjected to severe global hypoxia-ischemia utilizing an acute umbilical cord occlusion model. Outcomes were assessed over a 6-day period.

RESULTS

While maternal azithromycin exhibited relatively low placental transfer, azithromycin-treated lambs recovered spontaneous circulation faster following the initiation of cardiopulmonary resuscitation and were extubated sooner. Additionally, peri- and postnatal azithromycin administration was well tolerated, demonstrating a 77-hour plasma elimination half-life, as well as significant accumulation in the brain and other tissues. Azithromycin administration resulted in a systemic immunomodulatory effect, demonstrated by reductions in proinflammatory IL-6 (interleukin-6) levels. Treated lambs exhibited a trend toward improved neurodevelopmental outcomes while histological analysis revealed that azithromycin supported white matter preservation and attenuated inflammation in the cingulate and parasagittal cortex.

CONCLUSIONS

Perinatal azithromycin administration enhances neonatal resuscitation, attenuates neuroinflammation, and supports limited improvement of select histological outcomes in an ovine model of hypoxic-ischemic brain injury/encephalopathy.

Pharmacokinetics of high-titer anti-SARS-CoV-2 human convalescent plasma in high-risk children

BACKGROUNDWhile most children who contract COVID-19 experience mild disease, high-risk children with underlying conditions may develop severe disease, requiring interventions. Kinetics of antibodies transferred via COVID-19 convalescent plasma early in disease have not been characterized.METHODSIn this study, high-risk children were prospectively enrolled to receive high-titer COVID-19 convalescent plasma (>1:320 anti-spike IgG; Euroimmun). Passive transfer of antibodies and endogenous antibody production were serially evaluated for up to 2 months after transfusion. Commercial and research ELISA assays, virus neutralization assays, high-throughput phage-display assay utilizing a coronavirus epitope library, and pharmacokinetic analyses were performed.RESULTSFourteen high-risk children (median age, 7.5 years) received high-titer COVID-19 convalescent plasma, 9 children within 5 days (range, 2-7 days) of symptom onset and 5 children within 4 days (range, 3-5 days) after exposure to SARS-CoV-2. There were no serious adverse events related to transfusion. Antibodies against SARS-CoV-2 were transferred from the donor to the recipient, but antibody titers declined by 14-21 days, with a 15.1-day half-life for spike protein IgG. Donor plasma had significant neutralization capacity, which was transferred to the recipient. However, as early as 30 minutes after transfusion, recipient plasma neutralization titers were 6.2% (range, 5.9%-6.7%) of donor titers.CONCLUSIONConvalescent plasma transfused to high-risk children appears to be safe, with expected antibody kinetics, regardless of weight or age. However, current use of convalescent plasma in high-risk children achieves neutralizing capacity, which may protect against severe disease but is unlikely to provide lasting protection.Trial registrationClinicalTrials.gov NCT04377672.FundingThe state of Maryland, Bloomberg Philanthropies, and the NIH (grants R01-AI153349, R01-AI145435-A1, K08-AI139371-A1, and T32-AI052071).

Population pharmacokinetics and pharmacodynamics of Tranexamic acid in women undergoing caesarean delivery

AIMS

The population pharmacokinetics (PK) and pharmacodynamics (PD) of tranexamic acid (TXA) have not been studied to prevent postpartum haemorrhage (PPH) in pregnant women. It is unclear which TXA dose assures sufficient PPH prevention. This study investigated population PK/PD of TXA in pregnant women who underwent caesarean delivery to determine the optimal prophylactic doses of TXA for future studies.

METHODS

We analysed concentration (PK) and maximum lysis (PD) data from 30 pregnant women scheduled for caesarean delivery who received 5, 10 or 15 mg/kg of TXA intravenously using population approach.

RESULTS

TXA PK was best described by a two-compartment model with first-order elimination and the following parameters: clearance (between-subject variability) of 9.4 L/h (27.7%), central volume of 10.1 L (47.4%), intercompartmental clearance of 22.4 L/h (66.7%), peripheral volume of 14.0 L (13.1%) and additive error of 1.4 mg/L. The relationship between TXA concentration and maximum lysis was characterized by a sigmoid Emax model with baseline lysis of 97%, maximum inhibition of 89%, IC50 of 6.0 mg/L (65.3%), hill factor of 8.5 (86.3%) and additive error of 7.3%. Simulations demonstrated that 500 and 650 mg of TXA maintained therapeutic targets for 30 minutes and 1 hour, respectively, in 90% of patients.

CONCLUSION

This is the first population PK and PD study of TXA in pregnant women undergoing caesarean delivery. Our analysis suggests that a 650 mg dose provides adequate PPH prophylaxis up to 1 hour, which is less than the currently used 1000 mg of TXA in pregnant women.

Optimal use of intravenous tranexamic acid for hemorrhage prevention in pregnant women

BACKGROUND

Every 2 minutes, there is a pregnancy-related death worldwide, with one-third caused by severe postpartum hemorrhage. Although international trials demonstrated the efficacy of 1000 mg tranexamic acid in treating postpartum hemorrhage, to the best of our knowledge, there are no dose-finding studies of tranexamic acid on pregnant women for postpartum hemorrhage prevention.

OBJECTIVE

This study aimed to determine the optimal tranexamic acid dose needed to prevent postpartum hemorrhage.

STUDY DESIGN

We enrolled 30 pregnant women undergoing scheduled cesarean delivery in an open-label, dose ranging study. Subjects were divided into 3 cohorts receiving 5, 10, or 15 mg/kg (maximum, 1000 mg) of intravenous tranexamic acid at umbilical cord clamping. The inclusion criteria were ≥34 week's gestation and normal renal function. The primary endpoints were pharmacokinetic and pharmacodynamic profiles. Tranexamic acid plasma concentration of >10 μg/mL and maximum lysis of <17% were defined as therapeutic targets independent to the current study. Rotational thromboelastometry of tissue plasminogen activator-spiked samples was used to evaluate pharmacodynamic profiles at time points up to 24 hours after tranexamic acid administration. Safety was assessed by plasma thrombin generation, D-dimer, and tranexamic acid concentrations in breast milk.

RESULTS

There were no serious adverse events including venous thromboembolism. Plasma concentrations of tranexamic acid increased in a dose-proportional manner. The lowest dose cohort received an average of 448±87 mg tranexamic acid. Plasma tranexamic acid exceeded 10 μg/mL and maximum lysis was <17% at >1 hour after administration for all tranexamic acid doses tested. Median estimated blood loss for cohorts receiving 5, 10, or 15 mg/kg tranexamic acid was 750, 750, and 700 mL, respectively. Plasma thrombin generation did not increase with higher tranexamic acid concentrations. D-dimer changes from baseline were not different among the cohorts. Breast milk tranexamic acid concentrations were 1% or less than maternal plasma concentrations.

CONCLUSION

Although large randomized trials are necessary to support the clinical efficacy of tranexamic acid for prophylaxis, we propose an optimal dose of 600 mg in future tranexamic acid efficacy studies to prevent postpartum hemorrhage.

Clinical Pharmacology-Driven Translational Research to Optimize Bedside Therapeutics of Sotalol Therapy

Oral sotalol, used in adults for sinus rhythm control, is initiated at 80 mg b.i.d. and titrated to a maximum safe dose. The US Food and Drug Administration recommends monitoring the corrected QT interval (QTc ) for at least 3 days, until steady-state exposure of the drug is reached, before patient discharge, which can significantly impact the total cost of treatment. The objectives of this research were to design an accelerated intravenous sotalol loading and maintenance therapy that will reduce the hospital length of stay and to also evaluate the pharmacoeconomic impact in a hospital setting. Pharmacokinetic simulations of sotalol plasma concentrations vs. times profiles were performed to determine the optimal intravenous/oral transition regimen. A cost minimization analysis from the health sector perspective was conducted to assess the cost savings for these proposed accelerated regimens. For a chosen target dose of 120 mg b.i.d., two infusions of 40 mg over 1 hour and 20 mg over 0.5 hour, each followed up by an evaluation of QTc , can be administered followed immediately by the target oral maintenance dose of 120 mg at the end of the second infusion. Consequently, steady-state exposure and, therefore, steady-state QTc are obtained on the first day of therapy, facilitating an earlier hospital discharge. Two and 1-day mean total cost of -$3,123 (95% confidence interval (CI), -$3,640, -$2,607) -$4,820 (95% CI, -$5,352, -$4,288) were observed for this strategy, respectively. We are proposing an intravenous to oral transition strategy for sotalol that has the potential to significantly reduce cost and increase patient convenience.

Pharmacokinetics of HLD200, a Delayed-Release and Extended-Release Methylphenidate: Evaluation of Dose Proportionality, Food Effect, Multiple-Dose Modeling, and Comparative Bioavailability with Immediate-Release Methylphenidate in Healthy Adults

OBJECTIVES

HLD200, an oral, once-daily, evening-dosed, delayed-release, and extended-release methylphenidate (DR/ER-MPH), was designed to provide efficacy from the early morning, throughout the day, and into the evening to individuals with attention-deficit/hyperactivity disorder. The objectives were to evaluate DR/ER-MPH pharmacokinetic (PK) properties in healthy adults, including dose proportionality, food effect, the potential of accumulation using multiple-dose modeling, and bioavailability compared to an immediate-release MPH (IR MPH).

METHODS

Three open-label, single-dose, crossover studies were conducted, all with a 7-day washout between treatments. In Study I, 20 subjects received evening-dosed DR/ER-MPH (20 and 100 mg) followed by a medium-fat breakfast; 13 subjects received a subsequent 100-mg dose of DR/ER-MPH followed by a low-fat breakfast. In Study II, 18 subjects were evaluated after receiving evening-dosed DR/ER-MPH (100 mg) under 3 conditions: immediately after a high-fat meal, sprinkled on applesauce, and in a fasted state. In Study III, 11 and 12 subjects received evening-dosed DR/ER-MPH (100 mg) and morning-dosed IR MPH (20 mg), respectively.

RESULTS

DR/ER-MPH demonstrated dose proportionality between 20- and 100-mg doses. DR/ER-MPH PK parameters were not significantly affected by breakfast content or by sprinkling capsule contents. A high-fat meal immediately preceding evening dosing did not affect total MPH exposure but lowered peak MPH exposure by 14% and 11% versus fasted and sprinkled states, and time to peak exposure was delayed by ∼2.5 hours; these PK differences are unlikely to be clinically significant. Based on multiple-dose simulations using data from Study I, negligible accumulation of DR/ER-MPH was predicted. The relative bioavailability for DR/ER-MPH compared to IR MPH was 73.9%. No serious adverse events (AEs) were reported, and the observed AEs were consistent with MPH. There were no discontinuations in Studies I and III, but three participants withdrew in Study II due to AEs.

CONCLUSIONS

Evening-dosed DR/ER-MPH demonstrated dose proportionality and can be administered with or without food. Significant accumulation is unlikely with multiple dosing.

Pemetrexed in Malignant Pleural Mesothelioma

Purpose: This report describes the data and analysis leading to the approval of pemetrexed (LY 231514, MTA, Alimta, Eli Lilly and Co., Indianapolis, IN) by the U.S. Food and Drug Administration (FDA) of a New Drug Application for the treatment of malignant pleural mesothelioma (MPM).

Experimental Design: The FDA review of the efficacy and safety of pemetrexed assessed in a randomized clinical trial of 448 patients with unresectable MPM comparing pemetrexed plus cisplatin with cisplatin alone, as well as preclinical pharmacology and chemistry data, are described. The basis for marketing approval is discussed.

Results: In one randomized, single-blind, multicenter international trial, 226 patients were randomized to the pemetrexed and cisplatin arm and 222 patients were randomized to cisplatin alone. Median survival times were 12.1 months for pemetrexed and cisplatin and 9.3 months for cisplatin (P = 0.021; hazard ratio, 0.766; 95% confidence interval, 0.61-0.96). Myelosuppression, predominantly neutropenia, was the most common toxicity of pemetrexed plus cisplatin. Other common adverse events were fatigue, leucopenia, nausea, dyspnea, vomiting, chest pain, anemia, thrombocytopenia, and anorexia.

Conclusions: Pemetrexed in combination with cisplatin was approved by the FDA on February 4, 2004 for the treatment of patients with MPM whose disease is either unresectable or who are otherwise not candidates for curative surgery. The recommended dose of pemetrexed is 500 mg/m2 intra venous infusion over 10 minutes on day 1 of each 21-day cycle in combination with 75 mg/m2 cisplatin infused over 2 hours beginning 30 minutes after the pemetrexed infusion. Patients must receive oral folic acid and vitamin B12 injections before the start and during therapy to reduce severe toxicities. Patients should also receive corticosteroids with the chemotherapy to decrease the incidence of skin rash. Approval was based on a demonstration of survival improvement in a single randomized trial. Response rates and time to tumor progression were not included in product labeling because of inconsistencies in assessments among the investigators, independent radiologic reviewers, and the FDA, reflecting the difficulty of radiographic assessments in malignant mesothelioma. Complete prescribing information is available on the FDA Web site at http://www.fda.gov/cder/approval/index.htm.

Utilisation of pharmacokinetic-pharmacodynamic modelling and simulation in regulatory decision-making

Modelling and simulation (M&S) play an important role in regulatory decision-making that affects both the public and industry. Technological advances in various fields related to drug development call for more focus on ways to optimise current drug development practices. Recognition of the potential of M&S by regulatory agencies inevitably has a substantial impact on drug development. The objective of the current review is to present the various regulatory initiatives for application of M&S to clinical drug development. The relevant parts of the various recommendations issued by the US Food and Drug Administration (FDA), via guidance documents and advisory committee meeting proceedings, are highlighted. Application of M&S to a variety of activities, such as integrating pharmacokinetic-pharmacodynamic knowledge across a new drug application and designing efficient trials, is discussed. Some of the challenges that pharmaceutical institutions currently face when implementing M&S projects, such as team structure, communication with regulators, training and time constraints, are also presented, and solutions are proposed.

Pharmacokinetic-pharmacodynamic modeling of rivastigmine, a cholinesterase inhibitor, in patients with Alzheimer's disease

Rivastigmine is a cholinersterase inhibitor approved recently for the treatment of Alzheimer's disease (AD). The objective of this study is to characterize the pharmacokinetics-pharmacodynamics of rivastigmine in patients with AD. Eighteen AD patients received doses ranging from 1 to 6 mg bid for about 11 weeks. Rivastigmine and its active (major) metabolite (ZNS 114-666, also called NAP 226-90), plasma, and cerebrospinal fluid (CSF) concentrations were determined together with the AChE activity and computerized neuropsychological test battery (CNTB) scores. Nonlinear mixed-effects modeling of pharmacokinetic and pharmacodynamic data was conducted using NONMEM. Rivastigmine and its metabolite exhibited dose-disproportional pharmacokinetics. The apparent clearance and volume of distribution (plasma) of rivastigmine were estimated to be 120 L/h and 236 L, respectively. The relative bioavailability at the 6 mg dose was about 140%. The metabolite had a clearance of about 100 L/h and a volume of distribution of 256 L. The kinetics of the parent and metabolite in CSF showed an equilibration half-life of about 0.2 and 0.5 hours, respectively. The metabolite levels in CSF correlated very well with the acetylcholinesterase inhibition, with a ZNS 114-666 concentration of about 5.4 microg/L required for half-maximal inhibition of acetylcholinesterase activity. No statistically significant correlation of the CNTB scores with enzyme inhibition, parent or metabolite concentration (plasma/CSF), or rivastigmine dose could be established. The PK-PD model presented in this study can provide valuable information to optimize the drug development of rivastigmine and other related compounds and also in rationalizing dosing recommendations.

Role of dosage regimen in controlling indirect pharmacodynamic responses

The role of drug delivery in controlling indirect pharmacodynamic responses was assessed via computer simulations and literature review. Simulations of responses related to basic indirect response mechanisms were performed for various drug input rates which allowed the importance of drug delivery rate on the overall pharmacodynamic response to be evaluated. Response versus time profiles of integrated or net responses and efficiency were examined. Rate of drug input has the greatest influence on the area under the effect curve when doses are larger and target drug concentrations are above the IC(50)/SC(50). The pharmacodynamics of drugs which elicit indirect pharmacologic responses such as corticosteroids, diuretics, growth hormone, erythropoietin and insulin indicate that sustained drug delivery enhances the therapeutic efficiency and pharmacodynamic availability.

Carbamazepine level-A in vivo-in vitro correlation (IVIVC): a scaled convolution based predictive approach

A method is presented for prediction of the systemic drug concentration profile from in vitro release/dissolution data for a drug formulation. The method is demonstrated using four different tablet formulations containing 200 mg carbamazepine (CZM), each administered in a four way cross-over manner to 20 human subjects, with 15 blood samples drawn to determine the resulting concentration profile. Amount versus time dissolution data were obtained by a 75 rpm paddle method for each formulation. Differentiation, with respect to time, of a monotonic quadratic spline fitted to the dissolution data provided the dissolution rate curve. The dissolution curve was through time and magnitude scaling mapped into a drug concentration curve via a convolution by a single exponential, and the estimated unit impulse response function. The method was tested by cross-validation, where the in vivo concentration profiles for each formulation were predicted based on correlation parameters determined from in vivo-in vitro data from the remaining three formulations. The mean prediction error (MPE), defined as the mean value of 100% x(observed-predicted)/observed was calculated for all 240 cross-validation predictions. The mean values of MPE were in the range of 10-36% (average 22%) with standard deviations (S.D.s) in the range of 9-33% (average 13%), indicating a good prediction performance of the proposed in vivo-in vitro correlation (IVIVC) method.

Pharmacokinetic-pharmacodynamic modeling of ipamorelin, a growth hormone releasing peptide, in human volunteers

PURPOSE

To examine the pharmacokinetics (PK) and pharmacodynamics (PD) of ipamorelin, a growth hormone (GH) releasing peptide, in healthy volunteers.

METHODS

A trial was conducted with a dose escalation design comprising 5 different infusion rates (4.21, 14.02, 42.13, 84.27 and 140.45 nmol/kg over 15 minutes) with eight healthy male subjects at each dose level. Concentrations of ipamorelin and growth hormone were measured.

RESULTS

The PK parameters showed dose-proportionality, with a short terminal half-life of 2 hours, a clearance of 0.078 L/h/kg and a volume of distribution at steady-state of 0.22 L/kg. The time course of GH stimulation by ipamorelin showed a single episode of GH release with a peak at 0.67 hours and an exponential decline to negligible GH concentration at all doses. The ipamorelin-GH concentration relationship was characterized using an indirect response model and population fitting. The model employed a zero-order GH release rate over a finite duration of time to describe the episodic release of GH. Ipamorelin induces the release of GH at all dose levels with the concentration (SC50) required for half-maximal GH stimulation of 214 nmol/L and a maximal GH production rate of 694 mIU/L/h. The inter-individual variability of the PD parameters was larger than that of the PK parameters.

CONCLUSIONS

The proposed PK/PD model provides a useful characterization of ipamorelin disposition and GH responses across a range of doses.

Influence of gender on prednisolone effects on whole blood T-cell deactivation and trafficking in rats

Prednisolone (5 mg/kg intravenous) was administered to adrenalectomized male and female Sprague-Dawley rats (250-350 g) to assess the effects of gender on disposition and pharmacoimmunodynamics. Plasma concentrations of prednisolone were determined by high-performance liquid chromatography. Incorporation of [3H]thymidine (3H-TDR) was used to determine whole blood T-cell (WBTC) trafficking and deactivation following stimulation with Concanavalin-A. Whole blood T-cell trafficking was determined indirectly by using the glucocorticoid receptor antagonist RU-40555 (250 ng/mL) added to ex vivo cultures of whole blood from animals dosed with prednisolone. Mean (+/- SD) prednisolone clearance values were 3.22 +/- 0.88 and 3.46 +/- 0.96 L/h/kg in males and females, respectively. After administration of prednisolone, relative T-cell counts decreased slowly with time to reach a nadir at 3-5 h and returned to baseline levels by 8 h. Fitting data using an indirect response model yielded mean prednisolone 50% inhibitory concentration for inhibition of WBTC trafficking (IC50T) that was lower in males compared with females (0.14 +/- 0.16 versus 1.03 +/- 0.06 ng/mL; p < 0.05). In the absence of RU-40555, an immediate and complete inhibition of 3H-TDR incorporation into WBTC was observed (deactivation) and baseline levels were recovered slowly as prednisolone was cleared from blood. The mean 50% inhibitory concentration for inhibition of WBTC deactivation (IC50D) based on an inhibitory Imax model was similar in males and females (0.20 +/- 0.24 versus 0.18 +/- 0.12 ng/mL). Although male and female rats have similar exposure to prednisolone after 5-mg/kg doses, males are more sensitive to the inhibition of WBTC trafficking, whereas no gender effects on deactivation of WBTC exist.

Pharmacokinetics/dynamics of 5c8, a monoclonal antibody to CD154 (CD40 ligand) suppression of an immune response in monkeys

The pharmacokinetics and pharmacodynamics (PK/PD) of chimeric (Ch5c8) and humanized (Hu5c8) 5c8, a monoclonal antibody that binds CD154 (CD40 ligand), thus blocking the interaction between CD40 and CD154, were investigated in cynomolgus monkeys. Single-dose groups (n = 3 animals per dose) received saline, 0.2, 1, 5 or 20 mg/kg i.v. doses of Hu5c8. The repeat-dose groups (n = 4 animals) received 0 or 5 mg/kg i.v. doses of Ch5c8 or Hu5c8 on days 1, 2, 3, 5, 7 and 9. The single-dose PK parameters showed dose proportionality, with a terminal half-life of 300 h, a volume of distribution at steady state of 73 ml/kg and clearance of 0.2 ml.h-1.kg-1. The repeat-dose regimen produced a longer terminal half-life (500 h) and lower clearance (0.13 ml.h-1.kg-1) than in the single-dose groups. The antibody titer to tetanus toxoid (ATT) challenge served as the immunodynamic marker. The primary ATT response consisted of a latent phase of approximately 10 days, during which the immune system was processing antigen but not yet producing antibody, a rise to an antibody maximum titer at approximately 18 days and a decline toward baseline by approximately 40 days in controls. The 5c8 produced a log(dose)-proportional reduction in the area under the curve of ATT. An indirect PK/PD model based on the kinetics of tetanus toxoid exposure and inhibition of ATT production in relation to 5c8 concentrations was developed. A median inhibitory concentration of 0.84 microg/ml and a efficacy of 0.84 reflected marked inhibition of ATT response by 5c8. The model provides quantitation of reduced ATT responses after 5c8 and was applicable to primary and secondary immune responses and to both single-dose and multiple-dose treatments. The monoclonal antibody 5c8 blocks the CD40 and CD154 interaction, producing consistent and substantive reduction in antibody formation after administration of tetanus toxoid, which can be characterized with PK/PD modeling. It is anticipated that 5c8 may have utility in the treatment of antibody-mediated autoimmune disease.

Artificial neural networks as a novel approach to integrated pharmacokinetic-pharmacodynamic analysis

A novel model-independent approach to analyze pharmacokinetic (PK)-pharmacodynamic (PD) data using artificial neural networks (ANNs) is presented. ANNs are versatile computational tools that possess the attributes of adaptive learning and self-organization. The emulative ability of neural networks is evaluated with simulated PK-PD data, and the power of ANNs to extrapolate the acquired knowledge is investigated. ANNs of one architecture are shown to be flexible enough to accurately predict PD profiles for a wide variety of PK-PD relationships (e.g., effect compartment linked to the central or peripheral compartment and indirect response models). Also, an example is given of the ability of ANNs to accurately predict PD profiles without requiring any information regarding the active metabolite. Because structural details are not required, ANNs exhibit a clear advantage over conventional model-dependent methods. ANNs are proved to be robust toward error in the data and perturbations in the initial estimates. Moreover, ANNs were shown to handle sparse data well. Neural networks are emerging as promising tools in the field of drug discovery and development.

Quantitative structure-pharmacokinetic relationships (QSPR) of beta blockers derived using neural networks

This study demonstrates the application of neural networks to predict the pharmacokinetic properties of beta-adrenoreceptor antagonists in humans. A congeneric series of 10 beta-blockers, whose critical pharmacokinetic parameters are well established, was selected for the study. An appropriate neural network system was constructed and tested for its ability to predict the pharmacokinetic parameters from the octanol/water partition coefficient (shake flask method), the pKa, or the fraction bound to plasma proteins. Neural networks successfully trained and the predicted pharmacokinetic values agreed well with the experimental values (average difference = 8%). The neural network-predicted values showed better agreement with the experimental values than those predicted by multiple regression techniques (average difference = 47%). Because the neural networks had a large number of connections, two tests were conducted to determine if the networks were memorizing rather than generalizing. The "leave-one-out" method verified the generalization of the networks by demonstrating that any of the compounds could be deleted from the training set and its value correctly predicted by the new network (average error = 19%). The second test involved the prediction of pharmacokinetic properties of compounds never seen by the network, and reasonable results were obtained for three out of four compounds tested. The results indicate neural networks can be a powerful tool in exploration of quantitative structure-pharmacokinetic relationships.