Verifying and Validating Quantitative Systems Pharmacology and In Silico Models in Drug Development: Current Needs, Gaps, and Challenges

Abstract

The added value of in silico models (including quantitative systems pharmacology models) for drug development is now unanimously recognized. It is, therefore, important that the standards used are commonly acknowledged by all the parties involved. On April 25 and 26, 2019, a multistakeholder workshop on the validation challenges for in silico models in drug development was organized in Belgium. As an outcome, a White Paper is foreseen in 2020 on standards for in silico model verification and validation. CURRENT STATUS, GAPS, AND CHALLENGES IN ASSESSMENT OF MODELS FOR REGULATORY SUBMISSIONSDrug research, design, and development has a long-standing tradition in the use of in silico methodologies. In the context of clinical drug development Quantitative Structure-Property Relationship models in general and Quantitative Structure-Activity Relationship (QSAR) methods in particular, as well as pharmacometric approaches like population pharmacokinetics, pharmacokinetics (PKs)/pharmacodynamics, exposure-response, and physiology-based pharmacokinetics (PBPK) models are well-known. However, the in silico toolbox is rapidly expanding beyond these traditional/historical modeling technologies and new ones have emerged the last decades, including multiphysics simulations, the so-called systems medicine/pharmacology models (QSP) and clinical trial simulation tools (in silico clinical trials). In the remainder of this document, the term in silico models will be used to describe the collection of all the aforementioned modeling technologies.The added value of in silico models for drug development is now unanimously recognized by the scientific community. 1,2 Irrespective of the model used and the concerned part of the drug development pipeline, the evidence generated from these models, also called digital evidence, might eventually be included in regulatory submissions. In that case, the incorporation of digital evidence needs to follow standards of data/evidence generation, analysis, and reporting to enable the regulatory bodies to efficiently perform an adequate assessment of the submitted material.It is, therefore, of utmost importance that the standards to be considered are commonly acknowledged by all the involved parties (regulators, health technology assessment (HTA) agencies, academia, industry, regulators, and patients) and are relevant for all the types of models that can be included in regulatory submissions. The endorsement of these standards by regulators is particularly valuable because regulators generally provide guidance for data generation and reporting back to sponsors (industry or academia) thereby accelerating the uptake of the standards in the entire community and in the healthcare systems.Guidance documents have been published for QSAR models, 3 population PK models, 4 PK/pharmacodynamic or exposure-response models, 5,6 and more recently PBPK models, both by the European Medicines Agency (EMA) and the US Food and Drug Administration (FDA). 7,8 However, these guidelines are not fully applicable to all emergi...

Using numerical modeling and simulation to assess the ethical burden in clinical trials and how it relates to the proportion of responders in a trial sample

“…[21] According to new policies issued by the FDA and EMA, performing in silico clinical trials might be a solution to improve the efficiency of R&D by reducing uncertainty across all the stages. [22-24] Individualized computer simulations could become therefore a valuable tool to identify responders ahead of designing RCT. [8,25]…”

A Perspective on Quantitative Systems Pharmacology Applications to Clinical Drug Development

“…Along with this notion, the QSP community needs to collaborate to address (i) what constitutes the clinical credibility of QSP modeling; (ii) what the risk‐based assessment matrix is; and (iii) what the risk‐based quantitative and statistical criteria for individual elements in the assessment matrix are needed for model validation in the context of an intended application. Referencing the Standard by American Society of Mechanical Engineers (ASME) 9,10 for the framework of assessing model credibility regarding context of use, model validation, decision risk, and applicability of a QSP model may well be the first step.…”

A compound attributes-based predictive model for drug induced liver injury in humans

“…However, such efforts have limited utility so far due mainly to the fact many of these models cannot be applied to novel compounds. A number of papers have reported notable performance for predicting ADRs based on similarity calculation by considering compound structural similarities [22][23][24], target structural similarities [25], side effect similarities [26] or a combination of all [7,27]. However, intrinsic limitations of similarity-based methods are that certain evaluations of similarities cannot be fully captured for novel compounds.…”

“…In addition to animal models, alternative translational approaches, especially in silico computational models, are widely used to refine drug development design, and reduce the size and duration of clinical series [4]. Various in silico efforts have been reported with predictive modeling based on drug properties [5], drug-ADR connections [6], drug-drug similarities [7], drug-drug interactions [8], or drug-target effects [9,10], in order to detect pharmaceutical-related toxicities for different phenotypes [11,12], based on different source of safety data [13,14], suggesting its fundamental role in toxicity prediction.…”