Principal Director, QSP Scientific Lead
A new ICH guidance on how to assess MIDD evidence integrated into drug development decisions has recently reached stage 4, bringing a common framework that regulatory agencies and drug developers around the world can begin to adopt.
One of the communities that will benefit the most of the M15 guidance are the QSP practitioners and users. The documentation, model development methods, data sources, model assessment and simulation settings for QSP analyses have been long-standing topics of uncertainty and discussion. This struggle was well documented in the proceedings of an FDA-Industry Scientific Exchange that took place virtually in 2020, where the need for a harmonized credibility assessment framework adapted to QSP practices was well captured.
The discussion identified the following key drivers for this need:
QSP models are increasingly used for “high decision risks,” such as clinical waivers (replacing a clinical study with a model prediction). Without a credibility framework, using a model calibrated on, for example, low-dose data to predict high-dose safety can result in significant safety risks.
There is currently no alignment on the strategies or technical criteria used to evaluate model “truth.”
A framework is needed to:
Align on how to evaluate the calibration of state variables and parameters.
Establish statistical rigor that is context-specific.
Assess the validity of extrapolated predictions, which remains a “mainstream question” in the field.
Unlike simple models (rate-in/rate-out), QSP models represent complex human biology with feedback loops and “cross-talk” between pathways.
This results in:
Different companies use different algorithms to create Virtual Patient cohorts. A framework is required to determine if these various algorithms provide an adequate level of scientific rigor and to understand their relative strengths and compromises.
There is a significant challenge in linking a biological, mechanistic model to clinical composite scores (how a patient actually feels or functions). A credibility framework would help standardize how these models capture “observed response variability” in real-world patients.
In the meantime, and as QSP has gained more weight in drug clinical development, the QSP community has turned to the celebration of symposia and workshops (e.g. ‘Establishing QSP Virtual Population (VPOP) Credibility Assessment Guidelines Workshop‘), white papers and contributions to regulatory guidelines released for sister communities.
White papers:
Regulatory guidelines for sister communities:
It could be said that the QSP community has felt ‘disowned’, ‘orphan’ or even ‘stranded’ for a while.
Part of this probably stems from the fact that QSP emerged as a solution for exploring hypotheses and organizing information generated during the early stages of drug development. As the younger sibling of Systems Biology and later Systems Pharmacology, it’s no wonder that most practitioners came from a Bioinformatics or Mathematical background, thereby with a limited clinical pharmacology line of sight. The lack of experience with the sort of challenges, questions, and decisions that clinical drug developers face made it very difficult for these early QSP practitioners to grasp the rigorous level of credibility their simulations had to demonstrate. This new ICH M15 guidance provides a starting point for the community to find common ground for their best practices, with a clinical line of sight in mind, for regulatory purposes, but, perhaps most importantly, for general QSP workflows in support of MIDD.
In addition, given QSP’s ability to inform a wide range of decisions in drug development, harmonizing practices and developing a credibility assessment framework for QSP analyses is not easy. QSP can be applied from Target ID to Life Cycle Management stages. A QSP analysis for a given monotherapy can be expanded into double or triple therapeutic combinations by adding human in vitro data and mechanistic knowledge about the additional pathways. QSP simulations can address questions about dose, patient inclusion/exclusion criteria, companion diagnostic development, sampling times, protocol design, etc. Given the many contexts of use, the versatility of a QSP analysis is an advantage. Still, it can also become a handicap – unless the scope of applicability of a QSP analysis is defined from the planning stages, the modeling process can expand in multiple directions and dilute the real value it can bring to decisions in drug development. With this new M15 guideline, the focus of each QSP analysis should be clear even before the model is built, and the QSP practitioners should shift from the model to its application framework. The emphasis will be on MIDD (i.e., the power to inform decisions) rather than on the effort put into model development.
The challenge for drug development programs now is to understand how to implement the M15 guidance when QSP analyses are part of their MIDD plans, something big pharma companies, biotechs, and Contract Research Organizations (CROs) have yet to do so to a large extent. The Credibility Assessment Working Group put together by the ISoP QSP SIG in 2024 is already looking into it. Pharmetheus has updated its ways of working, analysis plans, and report templates to incorporate the new wave of QSP models, meeting the expectations of the M15 guidelines. The community also requires the voice of QSP practitioners who are very familiar with MIDD at the clinical and regulatory stages. All the QSP consultants at Pharmetheus have experience in clinical development projects. Our Regulatory Strategy Lead, Emma Hansson, recently joined the ISoP QSP Credibility Assessment Working Group, to bring in the perspective of a regulatory affairs expert…
The M15 guidelines have brought the much-needed water to the QSP credibility assessment framework desert – now it’s time for the drug development programs to start collecting it.