Fostering culture and optimizing organizational structure for implementing model-based drug development.

Model-based drug development (MBDD) is a promising approach to improve decision making in drug development. The pharmaceutical industry has made substantial progress from engaging in empirical decision making to increasingly using pharmacometrics (ie, modeling and simulation [M&S]) as a quantitative decision-making tool. Focusing on culture and an organizational structure perspective, this commentary summarizes experiences and vision from industry M&S leaders on implementing MBDD. A culture for MBDD needs to have wide acceptance of MBDD, enhanced decision making with probability-based evidence and transparent rationale, quantitative impact metrics, and a brand that emphasizes cross-disciplinary collaboration and ownership. An organizational structure for MBDD needs to have a dedicated pharmacometrics function, fine balance between quick wins and impact on long-term R&D goals, and collaborative MBDD teams among clinical pharmacologists, statisticians, pharmacometricians, and clinicians. Pharmaceutical companies with these characteristics are prepared to fully embrace and implement MBDD.

Exposure-response analysis in patients with schizophrenia to assess the effect of asenapine on QTc prolongation.

An exposure-response (E-R) analysis using linear mixed effects modeling was conducted on data from a thorough QTc trial for asenapine in 148 patients with schizophrenia. In a parallel design, patients received asenapine 5 mg twice daily (BID) for 10 days (10d) followed by 10 mg BID (6d), asenapine 15 mg BID (10d) followed by 20 mg BID (6d), quetiapine 375 mg BID (for assay sensitivity; 16d) or placebo (16d). Triplicate 12-lead electrocardiograms and concentration measurements were obtained on day -1 (baseline), 1, 10, and 16 at 8 scheduled times on each day. At mean C(max) for all asenapine doses, the E-R model predicted that the mean QTcF increase was less than 5 milliseconds, the International Conference on Harmonisation-established threshold for clinical concern. The model predicted a mean increase of 7 to 8 milliseconds for quetiapine. The corresponding upper bounds of the 95% confidence intervals were 7.5 milliseconds and 11.2 milliseconds for asenapine and quetiapine, respectively.

Performance characteristics for some typical QT study designs under the ICH E-14 guidance.

The International Conference on Harmonization (ICH) guidance for clinical evaluation of QT prolongation (E14) affected drug development by advocating that a thorough QT study (TQT) be conducted during development to assess the QT prolongation liability of a compound. The ICH E14 Statistics Group shortly thereafter recommended that a noninferiority intersection-union test (IUT) be used to exclude a clinically worrisome QT prolongation. Recent analyses have indicated that the IUT might be overly conservative with respect to excluding QT prolongation. This report assesses the IUT false positive rate for 4 recently conducted TQT trials using simple simulation experiments. Positive TQT study rates ranged from negligible to nearly 60% depending on study design, sample size, and patient status, despite no drug effect. Addition of clinically nonmeaningful QT prolongations (up to 5 milliseconds) increased the positive study rate to 80% for 1 particular study design. Ultimately, these results reveal significant limitations of the IUT with respect to excluding an effect and study interpretation for certain trial designs.

How modeling and simulation have enhanced decision making in new drug development.

The idea of model-based drug development championed by Lewis Sheiner, in which pharmacostatistical models of drug efficacy and safety are developed from preclinical and available clinical data, offers a quantitative approach to improving drug development and development decision-making. Examples are presented that support this paradigm. The first example describes a preclinical model of behavioral activity to predict potency and time-course of response in humans and assess the potential for differentiation between compounds. This example illustrates how modeling procedures expounded by Lewis Sheiner provided the means to differentiate potency and the lag time between drug exposure and response and allow for rapid decision making and dose selection. The second example involves planning a Phase 2a dose-ranging and proof of concept trial in Alzheimer's disease (AD). The issue was how to proceed with the study and what criteria to use for a go/no go decision. The combined knowledge of AD disease progression, and preclinical and clinical information about the drug were used to simulate various clinical trial scenarios to identify an efficient and effective Phase 2 study. A design was selected and carried out resulting in a number of important learning experiences as well as extensive financial savings. The motivation for this case in point was the "Learn-Confirm" paradigm described by Lewis Sheiner. The final example describes the use of Pharmacokinetic and Pharmacodynamic (PK/PD) modeling and simulation to confirm efficacy across doses. In the New Drug Application for gabapentin, data from two adequate and well-controlled clinical trials was submitted to the Food and Drug Administration (FDA) in support of the approval of the indication for the treatment of post-herpetic neuralgia. The clinical trial data was not replicated for each of the sought dose levels in the drug application presenting a regulatory dilemma. Exposure response analysis submitted in the New Drug Application was applied to confirm the evidence of efficacy across these dose levels. Modeling and simulation analyses showed that the two studies corroborate each other with respect to the pain relief profiles. The use of PK/PD information confirmed evidence of efficacy across the three studied doses, eliminating the need for additional clinical trials and thus supporting the approval of the product. It can be speculated that the work by Lewis Sheiner reflected in the FDA document titled "Innovation or Stagnation: Challenge and Opportunity on the Critical Path to New Medical Products" made this scientific approach to the drug approval process possible.

Evaluation of mixture modeling with count data using NONMEM.

Mixture modeling within the context of pharmacokinetic (PK)/pharmacodynamic (PD) mixed effects modeling is a useful tool to explore a population for the presence of two or more subpopulations, not explained by evaluated covariates. At present, statistical tests for the existence of mixed populations have not been developed. Therefore, a simulation study was undertaken to evaluate mixture modeling with NONMEM and explore the following questions. First, what is the probability of concluding that a mixed population exists when there truly is not a mixture (false positive significance level)? Second, what is the probability of concluding that a mixed population (two subpopulations) exists when there is truly a mixed population (power), and how well can the mixture be estimated, both in terms of the population parameters and the individual subjects classification. Seizure count data were simulated using a Poisson distribution such that each subject's count could decrease from its baseline value, as a function of dose via an Emax model. The dosing design for the simulation was based on a trial with the investigational anti-epileptic drug pregabalin. Four hundred and forty seven subjects received pregabalin as add on therapy for partial seizures, each with a baseline seizure count and up to three subsequent seizure counts. For the mixtures, the two subpopulations were simulated to differ in their Emax values and relative proportions. One subpopulation always had its Emax set to unity (Emax hi), allowing the count to approach zero with increasing dose. The other subpopulation was allowed to vary in its Emax value (Emax lo = 0.75, 0.5, 0.25, and 0) and in its relative proportion (pr) of the population (pr = 0.05, 0.10, 0.25, and 0.50) giving a total of 4.4 = 16 different mixtures explored. Three hundred data sets were simulated for each scenario and estimations performed using NONMEM. Metrics used information about the parameter estimates, their standard errors (SE), the difference between minimum objective function (MOF) values for mixture and non-mixture models (MOF (delta)), the proportion of subjects classified correctly, and the estimated conditional probabilities of a subject being simulated as having Emax lo (Emax hi) given that they were estimated as having Emax lo (Emax hi) and being estimated as having Emax lo (Emax hi) given that they were simulated as having Emax lo (Emax hi). The false positive significance level was approximately 0.04 (using all 300 runs) or 0.078 (using only those runs with a successful covariance step), when there was no mixture. When simulating mixed data and for those characterizations with successful estimation and covariance steps, the median (range) percentage of 95% confidence intervals containing the true values for the parameters defining the mixture were 94% (89-96%), 89.5% (58-96%), and 95% (92-97%) for pr, Emax lo, and Emax hi, respectively. The median value of the estimated parameters pr, Emax lo (excluding the case when Emax lo was simulated to equal 0) and Emax hi within a scenario were within +/- 28% of the true values. The median proportion of subjects classified correctly ranged from 0.59 to 0.96. In conclusion, when no mixture was present the false positive probability was less than 0.078 and when mixtures were present they were characterized with varying degrees of success, depending on the nature of the mixture. When the difference between subpopulations was greater (as Emax lo approached zero or pr approached 0.5) the mixtures became easier to characterize.