Contributors:
- Panuwat Trairatphisan
- Diana Clausznitzer
What are your main research interests?
In the Quantitative Translational Systems Pharmacology and Toxicology group, we aim to apply quantitative modeling and simulation methods to support the development of new medications and treatment modalities to improve patients’ lives.
PK/PD modeling for drug pharmacokinetics and efficacy is routinely and successfully integrated into our processes during preclinical and clinical development and informs the translation for first-in-human studies. Our interest is to understand where it makes sense to apply quantitative modeling also for (translational) drug safety assessments. Quantitative models that integrate systemic and tissue-level drug and metabolite exposure, mechanistic data for drug-induced toxicity, as well as organ physiology and sensitive, monitorable biomarkers associated with organ injury, seem to be ideal resources to best interpret available data and making predictions, which would also help to support our 3R commitment (replace, reduce, and refine) for animal use.
The organ systems considered in TransQST are relevant for significant attrition in drug development, and dedicated representatives from different functions have been contributing to the consortium across all work packages with their scientific knowledge and drug development experience, driving the execution of dedicated studies for the consortium, and with hands-on development a proof-of-concept model for drug-induced kidney injury (DIKI).
How has participating in TransQST impacted your work?
Joining TransQST has allowed our organization to access novel experimental and computational tools, quantitative models, and modeling concepts which are being explored within this research field. Applying quantitative models to preclinical safety data is still relatively new to our organization. Disseminating TransQST science and proof-of-concept quantitative systems toxicology (QST) models helped to illustrate the potential utility of QST models and applicability to safety data sets. Being part of TransQST also brought scientists from different functions together. This helped developing a mutual understanding and gaining traction to identify internal projects, which seed the broader application of QST approaches in our organization.
We explored the utility of “minimal QST” models, which utilize typical data that is currently generated during preclinical development. Adapting TransQST proof-of-concept models for DIKI and drug-induced liver injury (DILI) we now have internal examples which show that integrating drug PK/exposure and in vivo toxicity data from early non-GLP studies of short duration and models that describe minimal tissue physiology, enables extrapolating the progression of tissue injury and yields reasonable predictions for the outcome of longer-term toxicity studies. This opens the potential for model-informed preclinical study designs based on all available early toxicity data, e.g., to set optimal dose range and better chance of identifying a non-adverse observable event level (NOAEL) from animal studies. We also explored such “minimal QST” models for preclinical-to-clinical translation to simulate expected clinical tissue toxicity and associated biomarker responses. Working through retrospective and current portfolio cases, as well as data for key reference compounds, helps to characterize the utility and limitations of QST models in internal practice.
What challenges did you face while supporting the project across organ systems?
One major hurdle we experienced is that commonly generated data for safety assessments in drug development do not have the time resolution which is ideal for constructing QST models. In addition, transcriptomics data – a key data type explored in TransQST to inform mechanism of toxicity – is not routinely generated in practice, and if so, typically terminal samples only after some durations of dosing are available. Hence, more additional data than anticipated at the project start had to be generated to feed QST model development.
What are your main accomplishments in TransQST?
Our main achievement in TransQST is the seeding of a new mindset and showing through proof-of-concept examples that divergent, mechanistic data for mechanism of toxicity can be integrated into QST models to describe organ-level responses. These proof-of-concept QST models allow to explore mechanistic hypotheses and extrapolate beyond a particular toxicity study to make prospective predictions for drug toxicity and safety pharmacology.
What do you consider as the principal project outcomes?
TransQST was able to make progress on several proof-of-concept models for different organ toxicities, which illustrate that quantitative models can be successfully developed for toxicity and safety pharmacology, e.g., for mechanistic exploration in the virtual assay for cardiac arrhythmia, linking organ toxicity to monitorable biomarker responses in the DILI and DIKI models, and preclinical to human translation for gastrointestinal (GI) toxicity.
QST models force us to specify the hypothesis how drug exposure links to toxicity or safety pharmacological mechanisms, which links to tissue-level endpoints. By quantitatively connecting drug-, mechanism- and species-specific input data such models allow to test whether a biological hypothesis is consistent with all observed data.
We identified the relevance of incorporating (species-specific) physiological regeneration or turnover into QST models. This component was integrated in the QST models for GI tract, as well as in the proof-of-concept QST models for kidney and liver. Adding compound-specific drug-induced organ toxicity then allows to predict toxicity endpoints mechanistically and quantitatively.
Equally important, QST models highlight where we still miss adequate human-relevant input data to translate preclinical observed toxicity into the expected clinical toxicity, e.g., for mechanism of organ toxicity or projecting systemic and organ exposures of specific drug metabolites. Therefore, while we see some great utility of QST models already now, recognizing and communicating current limitations will be important to build confidence in QST as preclinical-to-clinical translational approach.
How do believe TransQST has contributed to advance the field?
Toxicity assessments in pharmaceutical practice dominantly depend on preclinical animal toxicity studies. Animal doses are subsequently used to calculate the clinical maximum recommended starting dose (MRSD) by applying empirical safety factors. TransQST has developed or advanced QST proof-of-concept models for several organ toxicities relevant in pharmaceutical development and laid out a vision how QST models could change the current empirical approach by integrating data from pre-clinical studies and species-specific tissue physiology, to project a human NOAEL/MRSD based on dynamical and quantitative model simulations. While we cannot expect to have fully validated QST models after the 5-year consortium period, the concept of applying mechanistic and quantitative QST models in safety assessments has been strengthened through the learn/confirm cycles in TransQST.
*The provided commentaries are based on the interviewee’s personal opinions towards the consortium which do not reflect the views from AbbVie as an organization.