By Vijayalakshmi Chelliah, Georgia Lazarou, Piet van der Graaf and Andrzej Kierzek
Immuno-oncology (IO) represents an elegant solution; it uses a patient’s own immune system to fight their cancer.
Early IO research used monoclonal antibodies to induce an immune response from patients’ PD1/PD-L1 and CTLA4 receptors, which serve as immune system checkpoints, successfully achieving long-term benefit in cases where the standard of care had failed. But that approach did not work for all patients.
Current IO research focuses on combination therapies because this approach permits administering multiple compounds, with tailored doses and timings, to target an individual’s disease or even a specific tumor.
One drawback with combination therapies is the huge number of possible targets and dosing regimens that need to be tested. There are now more than 2,000 active IO clinical trials1, an unprecedented number that will not only increase drug development costs but may not be attainable with the available patient population. A more efficient strategy is required that will refine this process, allowing only the drug combinations that have a significant probability of success to enter clinical trials.
Employing Quantitative Systems Pharmacology (QSP)
QSP modeling builds on earlier advances in drug development and regulatory decision-making achieved using mechanistic pharmacokinetic/pharmacodynamic (PK/PD) modeling and simulation. PK/PD models, which describe drug absorption, distribution, metabolism, elimination (ADME) and action (how the drug affects the body), have been widely adopted by global pharmaceutical companies and regulatory agencies. QSP introduces systems biology and information about the complex biological processes involved in health and disease into those PK/PD models.
QSP allows compounds’ PK, target binding, and mechanisms of action to be combined with knowledge of the molecular and cellular interactions involved in tumor growth and immune response. The resulting quantitative, dynamic, in-silico models can predict clinical results for novel drug combinations.
Improving the Likelihood of Success
One major reason why IO combination therapy trials fail is they lack quantitative information about the complex dynamic factors determining efficacy. Conventional PK/PD models, which typically use empirical models to connect pharmacological target modulation to clinical outcomes, do not use all the available mechanistic knowledge. That information gap can result in selecting sub-optimal combinations and dosing regimens.
QSP modeling offers an efficient way of refining the drug selection process and reducing both the attrition rate and number of trials required.
Reviewing QSP Applications
QSP models can play a valuable role throughout the drug development continuum. They can support hypothesis generation, and target/combination selection and validation in the early discovery stage. They also provide an excellent alternative to commonly used animal models that have a poor track record in translating to humans.
QSP allows mechanistic assumptions and parameters to be tested. It helps to quantify areas of uncertainty and improve decision-making for combination therapy and personalized therapy selection.
QSP models also facilitate optimal trial design by identifying pertinent patient subgroups, dosing regimens and biomarkers with the help of Virtual Patients.
Recruiting Virtual Patients
It is important to consider patient variability when developing new drugs in any therapeutic area, but that is especially true in oncology due to tumor molecular diversity.
In a recent Clinical Pharmacology & Therapeutics paper2, we explored the benefits of employing a simulated Virtual Patient population within a QSP model. Each Virtual Patient was assigned specific parameters and variables, so that together they reflected the distribution of those values in a real patient population.
The Virtual Patient approach has been widely adopted in PK/PD modeling and simulation, where it is often used to study a specific patient population based on age, gender, disease, or ethnicity. Genetic variability is mimicked by assigning Virtual Patients the allele frequencies for drug transporters and drug-metabolizing enzymes found in the target population.
QSP IO models are expanded to include high-throughput molecular biology data to account for tumor molecular variability. For example, the Cancer Genome Atlas contains genome and transcriptome sequences for 20,000+ primary cancer and matched normal samples for 33 cancer types.3
A Virtual Trial can then be conducted in which the Virtual Patient population receives the same dosing regimen as in a real trial. Individual drug concentration-time profiles can be used to study their variability and impact on clinical outcomes.
Results from actual patients in a clinical trial can also be mapped onto Virtual Patients to determine potential hypotheses for heterogeneous biomarker responses.
In addition, repeated simulations of trials with a specific number of patients can be used to predict between-trial variability and inform power calculations.
Conclusion
QSP models and Virtual Patients are valuable additions to the IO drug development armamentarium. They can help to improve the efficiency and effectiveness of the search for novel combination therapies.
Vijayalakshmi Chelliah (Principal Scientist), Georgia Lazarou (Senior Research Scientist), Piet van der Graaf (Senior Vice President of QSP) and Andrzej Kierzek (Head of Systems Modelling) work at Certara.
References
- Xin Yu J, Hubbard-Lucey VM, Tang J. Immuno-oncology drug development goes global. Nat Rev Drug Discov. 18 899-900. (2019).
- Chelliah V et al. QSP Approaches for Immuno‐Oncology: Adding Virtual Patients to the Development Paradigm. Clinical Pharmacology & Therapeutics. 19 July 2020. https://ascpt.onlinelibrary.wiley.com/doi/full/10.1002/cpt.1987
- Cancer Genome Atlas Research N, et al. The Cancer Genome Atlas Pan-Cancer analysis project. Nat Genet 45 1113-1120. (2013).