Critical Remarks on Reference-Scaled Average Bioequivalence.

PURPOSE: More than a decade ago the option to assess highly variable drugs / drug products by reference-scaled average bioequivalence was introduced in regulatory practice. Recommended approaches differ between jurisdictions and may lead to different conclusions even for the same data set. According to our knowledge, implemented methods have not been directly compared for their operating characteristics (Type I Error and power). METHODS: We performed Monte Carlo simulations to assess the consumer risk and the clinically relevant difference for the recommended regulatory settings. RESULTS: In all methods for reference-scaled average bioequivalence the Type I Error can be inflated with a consequently compromised consumer risk. Furthermore, the clinically relevant difference could vary between studies performed with the same reference product. CONCLUSIONS: Only average bioequivalence with fixed - widened - limits would both maintain the consumer risk and offer an unambiguously defined clinically not relevant difference. As long as such an approach is not implemented in regulatory practice, we recommend adjusting the level of the test a.

Assessment of pharmacokinetic linearity after repeated drug administration.

The prediction of drug concentration time courses after different dosing scenarios is greatly facilitated if the pharmacokinetics (PK) can be assumed linear. The assumption of linear PK thus needs careful evaluation for any new drug in development. Under linear PK, exposure is proportional to dose (linear PK across doses) and exposure at steady state can be predicted from a single dose based on the superposition principle (linear PK over time). While investigation of dose-proportionality is common practice, evaluation of time dependent PK has received less attention in the literature. In particular, the superposition principle can be used to assess whether the observed extent of accumulation after repeated administration is expected under the premise of linear PK. This work emphasizes the importance of the time related aspect of linear PK by introducing the predictability ratio (PR). Linear PK over time can be concluded if PR = 1. Accumulation is higher than expected if PR >1, and lower if PR <1. If PK data from multiple dose cohorts are available, the PR is assessed for each dose cohort and a supportive hypothesis test can be applied to test for potential differences between doses in PR.

Assessing goodness-of-fit for evaluation of dose-proportionality.

For the clinical development of a new drug, the determination of dose-proportionality is an essential part of the pharmacokinetic evaluations, which may provide early indications of non-linear pharmacokinetics and may help to identify sub-populations with divergent clearances. Prior to making any conclusions regarding dose-proportionality, the goodness-of-fit of the model must be assessed to evaluate the model performance. We propose the use of simulation-based visual predictive checks to improve the validity of dose-proportionality conclusions for complex designs. We provide an illustrative example and include a table to facilitate review by regulatory authorities.

The potential of the estimands framework for clinical pharmacology trials: Some discussion points.

The recently finalised and published guideline ICH E9 (R1) introduced a new framework for the statistical analysis of clinical trials, namely that of "estimands". While the framework was originally developed for the analysis of late-phase trials, it could also provide a rigorous basis for the analysis of clinical pharmacology trials. We illustrate potential applications on two examples: a multiple dose pharmacology trial and the interpretation of confirmatory bioequivalence (BE) trials according to the current FDA and EMA BE guidelines.

Adjustment of endogenous concentrations in pharmacokinetic modeling.

PURPOSE: Estimating pharmacokinetic parameters in the presence of an endogenous concentration is not straightforward as cross-reactivity in the analytical methodology prevents differentiation between endogenous and dose-related exogenous concentrations. This article proposes a novel intuitive modeling approach which adequately adjusts for the endogenous concentration. METHODS: Monte Carlo simulations were carried out based on a two-compartment population pharmacokinetic (PK) model fitted to real data following intravenous administration. A constant and a proportional error model were assumed. The performance of the novel model and the method of straightforward subtraction of the observed baseline concentration from post-dose concentrations were compared in terms of terminal half-life, area under the curve from 0 to infinity, and mean residence time. RESULTS: Mean bias in PK parameters was up to 4.5 times better with the novel model assuming a constant error model and up to 6.5 times better assuming a proportional error model. CONCLUSIONS: The simulation study indicates that this novel modeling approach results in less biased and more accurate PK estimates than straightforward subtraction of the observed baseline concentration and overcomes the limitations of previously published approaches.

A hybrid method to estimate the minimum effective dose for monotone and non-monotone dose-response relationships.

This article proposes a new multiple-testing approach for estimation of the minimum effective dose allowing for non-monotonous dose-response shapes. The presented approach combines the advantages of two commonly used methods. It is shown that the new approach controls the error rate of underestimating the true minimum effective dose. Monte Carlo simulations indicate that the proposed method outperforms alternative methods in many cases and is only marginally worse in the remaining situations.

A comparison of methods for classifying samples as truly specific with confirmatory immunoassays.

Biotechnology-derived therapeutics may induce an unwanted immune response leading to the formation of anti-drug antibodies (ADAs) which can result in altered efficacy and safety of the therapeutic protein. Anti-drug antibodies may, for example, affect pharmacokinetics of the therapeutic protein or induce autoimmunity. It is therefore crucial to have assays available for the detection and characterization of ADAs. Commonly, a screening assay is initially used to classify samples as either ADA positive or negative. A confirmatory assay, typically based on antigen competition, is subsequently employed to separate false positive samples from truly positive samples. In this manuscript we investigate the performance of different statistical methods classifying samples in competition assays through simulation and analysis of real data. In our evaluations we do not find a uniformly best method although a simple t-test does provide good results throughout. More crucially we find that very large differences between uninhibited and inhibited measurements relative to the assay variability are required in order to obtain useful classification results questioning the usefulness of competition assays with high variability.

Estimation in AB/BA crossover trials with application to bioequivalence studies with incomplete and complete data designs.

Crossover studies are frequently used in clinical research as they allow within-subject comparisons instead of the between-subject evaluation of parallel group designs. Estimation of interesting parameters from such designs is, however, not trivial. We provide three methods for estimating treatment effects and associated standard errors from an AB/BA crossover trial. Assuming at least asymptotic normality, we can obtain the confidence intervals for single parameters as well as for differences or ratios of treatment effects. The latter is particularly useful in a pharmacokinetic context to establish bioequivalence using area under the concentration versus time curves (AUCs). In this work, we will illustrate how Fieller-type confidence intervals can be constructed for the ratio of AUCs estimated using a noncompartmental approach in a sparse sampling setting from a two-treatment, two-period, two-sequence crossover trial. In particular, we will discuss a flexible batch design, which includes traditional serial sampling and complete data designs as special cases. Via simulation, we show that the proposed intervals have nominal coverage and keep the type I error even for small sample sizes. Moreover, we illustrate the methodology in a real data example.

Distinct characteristics of antibody responses against factor VIII in healthy individuals and in different cohorts of hemophilia A patients.

Neutralizing antibodies against factor VIII (FVIII) remain the major complication in the replacement therapy of hemophilia A patients. To better understand the evolution of these antibodies it is important to generate comprehensive datasets which include both neutralizing and nonneutralizing antibodies, their isotypes, and IgG subclasses. We developed sensitive ELISAs to analyze FVIII-binding antibodies in different cohorts of hemophilia A patients and in healthy individuals. Our data reveal the prevalence of FVIII-binding antibodies among healthy individuals (n = 600) to be as high as 19%, with a prevalence of antibody titers > or =1:80 of 2%. The prevalence of FVIII-binding antibodies was 34% (5% for titers > or =1:80) in patients without FVIII inhibitors (n = 77), 39% (4% for titers > 1:80) in patients after successful immune tolerance induction therapy (n = 23), and 100% (n = 20, all titers > or =1:80) in patients with FVIII inhibitors. We found significant differences for IgG subclasses of FVIII-binding antibodies between the different study cohorts. IgG4 and IgG1 were the most abundant IgG subclasses in patients with FVIII inhibitors. Strikingly, IgG4 was completely absent in patients without FVIII inhibitors and in healthy subjects. These findings point toward a distinct immune regulatory pathway responsible for the development of FVIII-specific IgG4 associated with FVIII inhibitors.

Non-compartmental estimation of pharmacokinetic parameters for flexible sampling designs.

Pharmacokinetic (PK) studies aim to understand the kinetics of absorption, distribution, metabolism and elimination of a drug. Typically, such studies involve measuring the concentration of the drug in the plasma or blood at several time points after drug administration. In studying the PK behaviour, either the non-compartmental approach or alternatively a modelling approach can be utilized. Traditionally, the non-compartmental approach makes minimal assumptions about the data-generating process but requires the data to be collected in a very structured way. Conversely, the modelling approach depends heavily on assumptions about the data-generating process but does not impose a specific data structure. In this paper, we will discuss non-compartmental methods for estimating the area under the concentration versus time curve and other common PK parameters that use minimal assumptions about the data structure making it applicable to a wide range of PK studies. We will evaluate the methods using simulation and give an illustrative example.

A formal comparison of different methods for establishing cut points to distinguish positive and negative samples in immunoassays.

Biotechnology derived therapeutics may induce an unwanted immune response leading to the formation of anti-drug antibodies (ADA). As a result the efficacy and safety of the therapeutic protein could be impaired. Neutralizing antibodies may, for example, affect pharmacokinetics of the therapeutic protein or induce autoimmunity. Therefore a drug induced immune response is a major concern and needs to be assessed during drug development. It is therefore crucial to have assays available for the detection and characterization of ADAs. These assays are used to classify samples in positive and negative samples based on a cut point. In this manuscript we investigate the performance of established and newly developed methods to determine a cut point in immunoassays such as ELISA through simulation and analysis of real data. The different methods are found to have different advantages and disadvantages. A robust parametric approach generally resulted in very good results and can be recommended for many situations. The newly introduced method based on mixture models yields similar results to the robust parametric approach but offers some additional flexibility at the expense of higher complexity.

Two-component polyethylene glycol surgical sealant influence on intraperitoneal infection in a refined rodent model.

OBJECTIVE: This study determined the influence of a 2-component polyethylene glycol surgical sealant (Coseal) as an adhesion prevention device on sepsis-related mortality and/or systemic bacterial translocation to the spleen. STUDY DESIGN: A bacterial inoculum and telemetry probe were implanted in 50 treated and 49 untreated rats. Telemetry probes monitored core-body temperature to determine time of death. Spleens were collected on day 3 for quantitative bacteriology of Escherichia coli and Bacteroides fragilis. RESULTS: Median survival time and mortality of treated rats (37.0 hours, 54.0%) were noninferior to untreated rats (47.5 hours, 55.1%). Median E coli titers in treated rats (2.24 log colony forming units/spleen) were significantly less than untreated rats (4.32 log colony forming units/spleen). B fragilis titers were not different. CONCLUSION: This study demonstrates intraperitoneal administration of a 2-component polyethylene glycol surgical sealant as an adhesion prevention device does not alter time to death or sepsis-related mortality and/or systemic bacterial translocation to the spleen.

Establishing bioequivalence in complete and incomplete data designs using AUCs.

Nonclinical in vivo animal studies have to be completed before starting clinical studies of the pharmacokinetic behavior of a drug in humans. The drug exposure in animal studies is often measured by the area under the concentration versus time curve (AUC). The classical complete data design where each animal is sampled for analysis at every time point is applicable for large animals only. In the case of small animals, where blood sampling is restricted, the batch design or the serial sampling design need to be considered. In batch designs, samples are taken more than once from each animal, but not at all time points. In serial sampling designs, only one sample is taken from each animal. In this article we derive the asymptotic distribution for the ratio of two AUCs and construct different confidence intervals, which are frequently used to assess bioequivalence. The performance of these intervals is then evaluated between the different designs in a simulation study. Additionally, the sample sizes required for the different designs are compared.

Assessing systemic drug exposure in repeated dose toxicity studies in the case of complete and incomplete sampling.

Repeated dose toxicity studies are performed to characterize the toxicological profile of a test compound following repeated administrations. The findings and interpretations from these systemic exposure studies in animals are essential for designing subsequent studies and evaluating the safety of the test item for humans. Blood samples for assessment of systemic exposure are usually collected on day one and at the end of the study with multiple dosings of the compound in between. Restrictions in blood volume often require an incomplete sampling design, in which each animal contributes sample measurements at some but not all time points. In this manuscript we derive an estimator for the ratio of area under the concentration versus time curves (AUCs), a frequently used measure of exposure to a compound, and a corresponding confidence interval to assess differences in exposure as well as equivalence between first and repeated administration that is applicable in such sparse sampling designs as well as complete data situations. An illustrative example is provided and the statistical properties of the proposed estimator, which incorporates the dependencies of measurements between first and repeated dosings as well as the dependency inherent in repeated sampling for each dosing, is studied asymptotically as well as in simulation.

Non-compartmental estimation of pharmacokinetic parameters in serial sampling designs.

Pharmacokinetic studies are commonly analyzed using a two-stage approach where the first stage involves estimation of pharmacokinetic parameters for each subject separately and the second stage uses the individual parameter estimates for statistical inference. This two-stage approach is not applicable in sparse sampling situations where only one sample is available per subject. Nonlinear models are often applied to analyze pharmacokinetic data assessed in such serial sampling designs. Modelling approaches are suitable provided that the form of the true model is known, which is rarely the case in early stages of drug development. This paper presents an alternative approach to estimate pharmacokinetic parameters based on non-compartmental and asymptotic theories in the case of serial sampling when a drug is given as an intravenous bolus. The statistical properties of estimators of the pharmacokinetic parameters are investigated and evaluated using Monte Carlo simulations.

A note on statistical analysis of organ weights in non-clinical toxicological studies.

Statistical comparison of organ weights between treated and untreated animals have traditionally been used to predict potential toxicity for patients. The manner of presentation of organ weight data, and the value of statistical analyses have been topics of discussion. Historically, a decision tree approach has been applied for statistical comparison of organ weights which does not control the overall error rate and can lead to different statistical tests being used by chance for identical settings causing confusion. This paper proposes a simple nonparametric approach for assessing treatment effects on organ weights in terms of ratios based on the Hodges-Lehmann estimator. This allows for simple interpretation of results and aids in the identification of potential target organs as the evaluation is based on effect sizes and not on p-values allowing a robust proof of effect as well as a robust proof of no effect. The proposed estimate and the corresponding nonparametric confidence interval applied to a rank-sum score can be used as a confirmatory test for difference and as a confirmatory test for equivalence. Exploratory analyses can be performed calculating the proposed estimates for each organ separately to be summarized graphically in a confidence interval plot.

Confidence intervals for ratios of AUCs in the case of serial sampling: a comparison of seven methods.

Pharmacokinetic studies are commonly performed using the two-stage approach. The first stage involves estimation of pharmacokinetic parameters such as the area under the concentration versus time curve (AUC) for each analysis subject separately, and the second stage uses the individual parameter estimates for statistical inference. This two-stage approach is not applicable in sparse sampling situations where only one sample is available per analysis subject similar to that in non-clinical in vivo studies. In a serial sampling design, only one sample is taken from each analysis subject. A simulation study was carried out to assess coverage, power and type I error of seven methods to construct two-sided 90% confidence intervals for ratios of two AUCs assessed in a serial sampling design, which can be used to assess bioequivalence in this parameter.

Establishing bioequivalence in serial sacrifice designs.

Nonclinical in vivo animal studies have to be completed before starting clinical studies of the pharmacokinetic behavior of a drug in human subjects. The classic complete data design, where each animal is sampled for analysis once per time point, is usually only applicable for large animals using the traditional two-stage approach. The first stage involves estimation of pharmacokinetic parameters for each animal separately and the second stage uses the individual parameter estimates for statistical inference. In the case of rats and mice, where blood sampling is restricted, the batch design or the serial sacrifice design may be applicable. In batch designs samples are taken more than once from each animal, but not at all time points. In serial sacrifice designs only one sample is taken from each animal. In this paper, three methods are presented to construct confidence intervals for the ratio of two AUCs assessed in a serial sacrifice design, which can be used to assess bioequivalence in this parameter. The presented methods are compared in a simulation study.