The science of bioequivalence and biosimilarity has greatly evolved over the past 3 decades. Current methods for assessing bioequivalence mostly rely on noncompartmental pharmacokinetic (PK) analyses, which have proven to be reliable and robust for most products. However, the development of more complex products is forcing scientists and regulators to consider alternative approaches, including those derived from model-based population PK analyses. This article will examine the strengths and weaknesses of standard noncompartmental methods and compare them to model-based approaches, including a comparison of metrics associated with each method. Specific situations for which model-based approaches could prove to be more suitable will be presented, as well as potential bioequivalence metrics that could be considered for bioequivalence comparisons. The opportunities and challenges that are associated with these novel methods will also be discussed.

SAds are associated with increased perioperative bleeding events, particularly abnormal bleeding and blood transfusions. From a clinical perspective, the potential bleeding risks of SAds in surgical settings need to be carefully weighed against their psychiatric benefits. Future research will need to investigate potential strategies to mitigate SAd-related bleeding risk in the surgical context.

Purpose: As per the US FDA guidance issued on June 2, 1995, the establishment of bioequivalence for topical dermatologic corticosteroids is based on comparing the pharmacodynamic (PD) effects of Test and Reference products at the dose duration corresponding to the population ED50, determined either by naïve pooled data or nonlinear mixed effect modeling (NLME). The guidance was introduced using a study case example where the expectation maximization (EM) NLME algorithm, as implemented in P-PHARM®, was used. Although EM methods are relatively common, other methods such as the First-Order Conditional Estimation (FOCE) as implemented in the NONMEM® software are even more common. The objective of this study was to investigate the impact of using different parametric population modeling/analysis methods and distribution assumptions on population analysis results. Methods: The dose duration-response data from 11 distinct skin blanching blinded pilot studies were fitted using FOCE (NONMEM®) and an EM algorithm (ADAPT5® (MLEM)). Three different Emax models were tested for each method. Population PD estimates and associated CV%, and the agreement between model predicted values and observed data were compared between the two methods. The impact of assuming different distributions of PD parameters was also investigated. Results: The simple Emax model, as proposed in the FDA guidance, appeared to best characterize the data compared to more complex alternatives. The MLEM method in general appeared to provide better results than FOCE; lower population PD estimates with less inter-individual variability, and no variance shrinkage issues. The results also favored ln-normal versus normal distribution assumptions. Conclusions: The population ED50 estimates were influenced by both the type of population modeling methods and the distribution assumptions. We recommend updating the FDA guidance with more specific instructions related to the population approach to be used (EM-like versus FOCE-like methods) and to the normality assumptions that need to be set (ln-normal versus normal distribution).

No abstract