Model-Informed Drug Discovery and Development (MID3): Harnessing Knowledge through Collaboration to Inform Decision-Making

Connecting the dots

In any scientific experiment, the ultimate goal is to decrease uncertainty in the result by evaluating the totality of evidence coming from within and outside the experiment. In drug development – given the obvious and critical importance of the results for patients and society – enabling principled decision-making, improving efficiency, increasing probability of success and ultimately delivering a treatment that offers a significant benefit to patients is essential.

Model-informed drug discovery and development (MID3) involves the use of mathematical models for knowledge management, predictions, and decision-making in pharmaceutical R&D. ^{1, 2, 3} These models can “connect the dots” across various inputs related to drug properties, disease biology, patient characteristics and trial designs to relate them to the outputs of different stages of R&D using mathematical frameworks.⁴ Different types of models are used depending on the stage of R&D, intended purpose and the specific questions being addressed.

MID3 is rapidly becoming an essential tool in R&D decision-making and drug development.

How MID3 enables R&D to make timely decisions

While candidate failures are never the objective, it is a necessary and unavoidable part of any successful research and development program. Modelling can help pharmaceutical companies to discover potential failures earlier in the R&D process and make those termination decisions faster. This will ultimately reduce costs and ensure that only candidates with the greatest chance of success enter late-stage clinical development for large scale evaluations in valuable patient participants – ultimately also reducing the time to market for those successful new therapies.

These models may be related to dosing, trial design or patient selection, and bring critical insights on these dimensions into clinical development.

Modelling for efficiency and success
Pharma companies can utilize MID3 in a variety of ways,^1,2,3 including but not limited to:

• Improving efficiency
  
  • Models can help researchers design shorter, more targeted and informative trials.
    
• Increasing probability of success
  • Once a drug-disease model^a is built, simulations based on it can evaluate various design options in silico [via computational methods] without having to conduct clinical experiments.
  • Virtual trials can allow exploration of various options in the design space and enable a rational and objective decision on final design that is aimed at maximizing probability of success.

We are consciously embedding MID3 strategies to help advance our pharmaceutical R&D pipeline. An example is one of our immune checkpoint inhibitors, where we used population pharmacology modelling and simulation to enable post-approval switching from the originally studied and approved body weight-based dosing to simpler fixed dosing. In this instance, modelling allowed us to forgo prospective clinical trials to evaluate the efficacy and safety of the fixed dosing regimen.

A model of success  

While some may think MID3 is purely the domain of modelling experts, a successful modelling approach is built based on collaboration by building understanding and gaining valuable input from stakeholders across organizations. Broader adoption across R&D is then key for MID3 to continue evolving and provide tangible successes. Importantly, integrating quantitative pharmacology models to inform decision-making and clinical trial designs via simulations relies on strong partnerships between clinical experts, quantitative clinical pharmacologists and statisticians in cross-functional team settings.⁵

Specific ways pharma can better integrate MID3 within the organization could include:

• Increased education and communication
  
  • Introduce team members to concepts such as the Bayesian approach^b and dose/exposure-response^c.
    
  • Focus on the big picture and communicate with impact how the results helped to answer key questions.
    
• Sharing success stories and case studies
  
  • Illustrate the influence and impact of MID3 approaches in drug development and therapeutic optimization.
    
• Teamwork and purpose orientation
  
  • Modelling should pivot to a shared purpose, requiring early alignment with the broader team on the questions that are most relevant for stakeholders, to maximize uptake of MID3 and return on investment.

External inputs

Models fundamentally rely on a wide range of data and expertise. As we have seen through the recent pandemic and the numerous models developed to predict, treat and contain coronavirus – input from a variety of experts and data sources supports more accurate modelling, informs and builds confidence in decision-making.

We work with several organizations and in consortia to inform, improve and develop our modelling capabilities. For example, we are members of the Certara Immuno-Oncology QSP^d consortium, whose work is crucial in helping inform rational dose selection, biomarker strategies, drug combinations, and patient selection.⁶

Additionally, we are members of the TransCelerate Biopharma Inc.’s Historical Trial Data Sharing Initiative, enabling access to patient-level control arm data in clinical trials⁷ that are valuable for disease model development for progress in model-informed precision medicine^e.

Modelling, the future of R&D?

As MID3 becomes increasingly important for pharmaceutical R&D and continues to permeate every field of science, it will be vital for organizations to collaborate both internally and externally to understand the latest knowledge, approaches and methods. Modelling can then help the pharmaceutical industry to fulfill its heartfelt commitment to patients to only bring forward those therapies that make significant and much needed improvements to healthcare.  

Glossary of Terms
a. Drug-disease models –Pharmacostatistical models that describe the longitudinal trajectory of disease burden, impact of patient-specific factors on the time course of disease progression, and the effect of drug dose/exposure on the underlying disease dynamics.
b. Bayesian approach – Bayesian decision making involves basing decisions on the probability of a successful outcome, where this probability is informed by both prior information and new evidence the decision maker obtains.
c. Dose/exposure-response – Mathematical relationships between the administered dose of a drug, resulting systemic exposure, and the drug’s effects (desired efficacy and undesirable toxicities) to define optimal dosage that maximizes benefit versus risk.
d. Quantitative systems pharmacology (QSP) – High-resolution models that provide a mathematical description of the mechanism of action of the drug at the molecular, cellular and organismal levels by accounting for drug-target interactions, downstream biochemical and cellular cascades of drug response, and ultimately the modulation of disease activity in the context of relevant biological sources of variability.
e. Model-Informed Precision Medicine - The application of quantitative pharmacology models to generate hypotheses and/ or substantiate evidence regarding the contribution of patient-specific factors (e.g., genetic variation resulting in overexpression of the molecular target of a drug) to the efficacy and/ or safety of drugs, thereby enhancing precision medicine development.

¹ Lalonde RL, Kowalski KG, Hutmacher MM, et al. Model-based drug development. Clin Pharmacol Ther. 2007;82(1):21-32. doi:10.1038/sj.clpt.6100235.
² EFPIA MID3 Workgroup, Marshall SF, Burghaus R, et al. Good Practices in Model-Informed Drug Discovery and Development: Practice, Application, and Documentation. CPT Pharmacometrics Syst Pharmacol. 2016;5(3):93-122. doi:10.1002/psp4.12049.
³ Gupta N, Bottino D, Simonsson USH, et al. Transforming Translation Through Quantitative Pharmacology for High-Impact Decision Making in Drug Discovery and Development. Clin Pharmacol Ther. 2020;107(6):1285-1289. doi:10.1002/cpt.1667.
⁴ Terranova N, Venkatakrishnan K, Benincosa LJ. Application of Machine Learning in Translational Medicine: Current Status and Future Opportunities. AAPS J. 2021;23(4):74. doi:10.1208/s12248-021-00593-x.
⁵ Venkatakrishnan K, Zheng S, Musante CJ, et al. Toward Progress in Quantitative Translational Medicine: A Call to Action. Clin Pharmacol Ther. 2020;107(1):85-88. doi:10.1002/cpt.1687.
⁶ Chelliah V, Lazarou G, Bhatnagar S, et al. Quantitative Systems Pharmacology Approaches for Immuno-Oncology: Adding Virtual Patients to the Development Paradigm. Clin Pharmacol Ther. 2021;109(3):605-618. doi:10.1002/cpt.1987.
⁷ Yin PT, Desmond J, Day J. Sharing Historical Trial Data to Accelerate Clinical Development. Clin Pharmacol Ther. 2019;106(6):1177-1178. doi:10.1002/cpt.1608.