How Mechanistic Modeling Informs First-in-Human Dose Selection

What is the role of mechanistic modeling in determining first-in-human starting doses, and how does it relate to traditional approaches like NOAEL and MABEL?

Mechanistic modeling plays a crucial role in predicting first-in-human doses, enhancing our understanding of the therapeutic process. When selecting a safe starting dose for human trials, we need to consider factors like pharmacology and toxicology. Traditional approaches, such as NOAEL (No Observable Adverse Effect Level) and MABEL (Minimum Anticipated Biological Effect Level), help guide this process. NOAEL represents the maximum dose at which unacceptable toxicity is observed in preclinical studies, and MABEL indicates the level at which a therapeutic begins to have a biological effect. These values are typically determined through experiments or in vitro assays. Mechanistic modeling takes these concepts further by incorporating scientific mechanisms and mathematical predictions. It allows us to correlate in vitro efficacy data with in vivo results, considering factors like the number of cells, sites per cell, protein synthesis rates, feedback loops, and cytokine concentrations. This approach provides a more comprehensive and holistic understanding, enabling us to make better-informed first-in-human dose predictions.

“[A] safe starting dose in humans is really driven by the pharmacology, and the toxicology.”

How do you guide customers through the process of selecting a first-in-human starting dose using mechanistic modeling?

Guiding customers through the selection of a first-in-human starting dose involves creating a mechanistic model that integrates various data sources, including in vitro assays and preclinical disease-relevant animal models. This model aims to represent the dynamics of the therapeutic's mechanism of action (MOA), such as its impact on protein synthesis and cell responses. Because this is a mechanistic model and considering various factors such as MOA, affinities,  synthesis rates, numbers of cells, and sites per cell, we can do backward and forward translation. The model helps us predict the human dose required to achieve the desired therapeutic effects within the safe and efficacious range. Importantly, mechanistic modeling allows us to quantify uncertainty in the predictions and use this information to design preclinical experiments more effectively, reducing uncertainty as we move forward.

What happens after a company files its IND (Investigational New Drug application), and how does mechanistic modeling come into play in the clinical trial phase?

After filing an IND, a clinical pharmacologist may conduct initial analyses to understand the effects of dose, frequency, and concentration on pharmacokinetics (PK) and pharmacodynamics (PD) using empirical approaches. However, with limited patient data in early clinical phases, it can be challenging to draw meaningful conclusions.

Mechanistic modeling allows for real-time analysis and predictions as data accumulates during clinical trials. Even before phase 3, you can run mechanistic, virtual patient analyses to consider how uncertainty is impacting your dosing, in terms of safety and efficacy. By continuously updating the mechanistic model and estimating model parameters, you can gain insights into how uncertainty affects dosing and make informed decisions. This approach facilitates the early identification of effective and safe dose levels, potentially accelerating the clinical trial process.

Try out a 2-week trial with predefined models or request a demo today

How does uncertainty and variability impact the design of clinical trials in the context of mechanistic modeling?

Mechanistic modeling embraces uncertainty and variability, considering that no two patients are identical. Patients exhibit differences in the number of cells, cytokine levels, protein synthesis rates, and other critical factors. Mechanistic modeling helps account for this variability by simulating various patient subgroups with different parameter values.

By simulating how different patient characteristics affect responses to the therapy, mechanistic modeling can guide the selection of doses, frequency, and patient subgroup definitions. The variability and uncertainty can impact patient selection, clinical trial design, dose escalation considerations, and even applications across indications. This approach allows for more personalized and precise clinical trial design, taking into account individual patient variability.

“[W]here a lot of people I think are fearful of uncertainty and variability, mechanistic modeling embraces that.”

Is there a situation where companies might opt for a different approach later in the drug development pipeline, rather than mechanistic modeling?

While mechanistic modeling is highly valuable throughout the drug development process, there are situations where companies may opt for different approaches in later stages. For example, as you approach the Biologics License Application (BLA) or New Drug Application (NDA) phase, population pharmacokinetics (popPK) modeling becomes increasingly important. PopPK modeling involves fitting models to data from a larger patient population, allowing for more comprehensive statistical analysis. For gene and cell therapies, standard popPK models are not available. Even as the discipline of popPK continues to evolve, a combination of mechanistic modeling and traditional approaches will likely be used to ensure the success of drug development programs.

Mechanistic modeling offers a powerful and holistic approach to guiding first-in-human dose selection, improving clinical trial design, and addressing patient variability. It plays a crucial role in the development of safe and effective therapeutics, especially in emerging fields like gene and cell therapies.

A more detailed discussion of mechanistic modeling for clinical development can be heard on this podcast.

Listen to our podcast

Related Resources

• Mechanistic Modeling and its Role In Drug Discovery
• Transforming Preclinical Research with Mechanistic PK/PD Modeling Approaches