Modeling and Simulation to Support Dose Selection and Clinical Development of SC-75416, a Selective COX-2 Inhibitor for the Treatment of Acute and Chronic Pain

Pharmacokinetic/pharmacodynamic (PK/PD) models were developed and clinical trial simulations were conducted to recommend a study design to test the hypothesis that a dose of SC-75416, a selective cyclooxygenase-2 inhibitor, can be identified that achieves superior pain relief (PR) compared to 400 mg ibuprofen in a post-oral surgery pain model. PK/PD models were developed for SC-75416, rofecoxib, valdecoxib, and ibuprofen relating plasma concentrations to PR scores using a nonlinear logistic-normal model. Clinical trial simulations conducted using these models suggested that 360 mg SC-75416 could achieve superior PR compared to 400 mg ibuprofen. A placebo- and positive-controlled parallel-group post-oral surgery pain study was conducted evaluating placebo, 60, 180, and 360 mg SC-75416 oral solution, and 400 mg ibuprofen. The study results confirmed the hypothesis that 360 mg SC-75416 achieved superior PR relative to 400 mg ibuprofen (DeltaTOTPAR6=3.3, P<0.05) and demonstrated the predictive performance of the PK/PD models.

Model-based Drug Development

The low productivity and escalating costs of drug development have been well documented over the past several years. Less than 10% of new compounds that enter clinical trials ultimately make it to the market, and many more fail in the preclinical stages of development. These challenges in the "critical path" of drug development are discussed in a 2004 publication by the US Food and Drug Administration. The document emphasizes new tools and various opportunities to improve drug development. One of the opportunities recommended is the application of "model-based drug development (MBDD)." This paper discusses what constitutes the key elements of MBDD and how these elements should fit together to inform drug development strategy and decision-making.

Efficient screening of covariates in population models using Wald's approximation to the likelihood ratio test

We propose an efficient algorithm for screening covariates in population model building using Wald's approximation to the likelihood ratio test (LRT) statistic in conjunction with Schwarz's Bayesian criterion. The algorithm can be applied to a full model fit of k covariate parameters to calculate the approximate LRT for all 2k - 1 possible restricted models. The algorithm's efficiency also permits internal validation of the model selection process via bootstrap methods. We illustrate the use of this algorithm for both model selection and validation with data from a Daypro pediatric study. The algorithm is easily implemented using standard statistical software such as SAS/IML and S-Plus. A SAS/IML macro to perform the algorithm is provided.

Design evaluation for a population pharmacokinetic study using clinical trial simulations: a case study

Clinical trial simulations were conducted to assess power and sample size requirements for a population pharmacokinetic (PK) substudy of a phase III clinical trial. The simulations were based on a population PK model developed from phase I healthy volunteer data. A sparse sampling design was employed taking into account the practical considerations regarding the desire not to keep patients at the study sites for extended periods of time for blood sampling. It was expected that the sparse sampling design would not support fitting the same model developed in healthy volunteers due to the narrow range of sampling times. Therefore, a model with fewer parameters and variance components was fit to simulated data from the proposed design to assess the bias in the estimates of the population mean PK parameters and variance components. Results indicate that the proposed design employing the simple model can provide accurate mean estimates of oral drug clearance (CL) and the apparent steady-state volume of distribution (V(ss)). However, the simulation results also suggest that the size and power of the likelihood ratio test for subpopulation differences in CL are inflated when using the simple model.

A semicompartmental modeling approach for pharmacodynamic data assessment

A new method is proposed for modeling the temporal aspects of the pharmacodynamic-pharmacokinetic relationship of drugs. A semicompartmental solution to the effect-site link model of Sheiner et al. (1) formed the basis for this new approach. This semicompartmental solution does not require the specification of a compartmental model for the pharmacokinetic response and may offer an advantage when model misspecification is present in using standard compartmental models. A Monte Carlo simulation study was conducted to evaluate the performance of the semicompartmental modeling approach. This method is easily implemented in standard nonlinear regression packages.

An algorithm for estimating the terminal half-life in pharmacokinetic studies

An algorithm is developed for automating the process of identifying the terminal phase in plasma concentration-time curves for non-compartmental estimation of the terminal elimination rate and half-life. The manual process of plotting each curve, selecting the observations in the terminal phase, performing the regressions and plotting the fitted regression lines can be very time-consuming in cases where many curves are to be processed. The algorithm makes use of single-row regression diagnostics to determine the first observed time point in the terminal phase. An example is used to illustrate the calculations, and a simulation study was conducted to evaluate the performance of the algorithm. The algorithm has been implemented in a computer program.

Absorption and disposition of a new antiarrhythmic agent bidisomide in man

Absorption and disposition of bidisomide were studied in 12 healthy male subjects after a 20-min iv (1 mg/kg; N = 6) infusion and oral (2 mg/kg; N = 6) administration of the 14C-labeled drug. The oral absorption profile of unlabeled bidisomide was also studied after administration of a solution by a nasoenteric tube to different sites of the gastrointestinal tract (stomach, duodenum, jejunum, and ileum). The systemic availability was 61%. Absorption was slow initially and then rapid, achieving peak plasma concentrations between 2 and 4 hr. Less than complete systemic availability was attributed to incomplete absorption rather than first-pass metabolism. When the drug solution was delivered directly to the stomach, two distinct peak plasma levels were found. This was attributed to the more rapid absorption of bidisomide in the duodenum and ileum (and/or possibly colon). Following an iv dose, plasma levels of the drug declined with mean half-lives of 0.11, 2.0, and 12 hr for alpha, beta, and gamma phases, respectively, and a plasma clearance of 380 mL/min. The percentages of the dose recovered as bidisomide in urine and feces were 19 +/- 1 and 29 +/- 4 for the iv dose and 9.1 +/- 0.9 and 48 +/- 5 for the oral dose. Bidisomide did not exhibit substantial enantioselective pharmacokinetics in plasma regardless of the route of administration. The mean urinary excretion of the (-) enantiomer was, however, slightly higher than that of the (+) enantiomer, with (-)/(+) enantiomeric ratios of 1.2 and 1.3 after iv and oral administration, respectively. The enantiomeric ratio of bidisomide recovered in the feces was approximately 1.

Bidisomide (SC-40230), a new antiarrhythmic agent: initial study of tolerability and pharmacokinetics

Forty-nine healthy male volunteers received the test article for bidisomide (SC-40230) in a double-blind, placebo-controlled, dose-ranging study. Intravenous doses ranged from 0.03 to 2.5 mg/kg. There was a close relationship between the dose and the peak plasma concentration. The PR, QRS, QT, RR, and QTc intervals each demonstrated a statistically significant response to the dose administered. The PR and QRS intervals lengthened and the other intervals shortened (although to a lesser degree). The compound was well tolerated, with mild symptoms only at higher doses. Bioavailability was studied in 12 male volunteers, with each receiving 2.0 mg/kg of bidisomide, both orally and intravenously, in an open-label crossover trial. After a 10-minute zero-order intravenous infusion, bidisomide plasma levels could best be described in terms of a three-compartment pharmacokinetic model with the mean half-life values of alpha, beta, and gamma phases of 0.12, 1.77, and 12.3 hours, respectively. The mean absolute oral bioavailability was 43%.

Effects of food and antacid on oral absorption of misoprostol, a synthetic prostaglandin E1 analog

Misoprostol, a synthetic PGE1 analog with mucosal protective and antisecretory properties, is rapidly and essentially completely metabolized to its active metabolite, misoprostol acid. Relative to fasting conditions administration of misoprostol with antacid resulted in reduced bioavailability, as manifested by lower AUC(0-4) values from (mean +/- SD; n = 12) 417 +/- 135 to 349 +/- 108 pg.hr/ml (P less than 0.05), without significant changes in the rate of absorption (tmax = 14 +/- 8 [fasting] vs. 20 +/- 14 min [with antacid]). The observed Cmax values were also reduced (from 811 +/- 317 to 689 +/- 315 pg/ml), however, the difference was not statistically significant. Drug administration with a high-fat content meal resulted in a marked slowing in the absorption rate without a substantial decrease in the extent. Relative to fasting conditions food decreased misoprostol acid Cmax from 811 +/- 317 to 303 +/- 176 pg/ml (P less than 0.05) and increased tmax from 14 +/- 8 to 64 +/- 79 min (P less than 0.05): the AUC changes remained statistically insignificant from 417 +/- 135 to 373 +/- 111 pg.hr/ml. Administration of misoprostol with food could potentially decrease the incidence of systemic side effects by reducing the early high peak plasma concentrations of misoprostol acid.