Navigating shifting paradigms in dose optimisation and dose selection for oncology therapeutics

In oncology drug development, the maxim “more is better” has long guided dose-selection studies. Sponsors habitually used simple, rule-based designs for dose selection, initially developed to establish the maximum tolerable dose (MTD) for chemotherapeutic agents. However, the MTD is increasingly seen as suboptimal for targeted and immunotherapies, which lack chemotherapy’s generalised cytotoxic effects. These therapies may have a benign toxicity profile with a dose lower than the MTD, providing the highest level of efficacy.

More contemporary dose-selection designs can now identify the lowest dose of an oncology therapy with the highest rate of efficacy that is tolerable for patients — commonly referred to as the optimal biological dose (OBD). These alternative designs are increasingly required by regulators, who expect sponsors to move away from long-entrenched, dose-selection paradigms and instead identify their drug’s optimal dose. To facilitate the shift to OBD-based dose selection, the U.S. Food and Drug Administration launched Project Optimus in 2021.

However, many companies are struggling to adjust to more stringent dose-selection requirements. Diverse dose-selection designs are available, with varied approaches to balancing participant safety, accurate dose selection, and simplicity. Transitioning from simple, rule-based designs to more complex model-based or model-assisted designs requires support and specialised expertise.

CRO partners, such as ICON plc, can assist sponsors in selecting and implementing cost-effective and efficient dose-selection designs based on a drug’s characteristics and study objectives. For example, ICON provided expert consultation, including model simulation, during the early-phase clinical trial of a tri-specific antibody targeting multiple tumor types.

ICON successfully implemented Bayesian designs in a two-part, first-in-human trial that assessed drug safety and toxicity, and then preliminary efficacy in cohorts of patients with different tumor types.

In the first part of the trial, a Bayesian Optimal Interval (BOIN) design was selected for its relative statistical simplicity and intuitiveness to investigators. However, it soon became clear that toxicity did not increase monotonically with dose. ICON then modified the BOIN design to maximise the use of pharmacodynamic information to help select the optimal dose without increasing patient numbers, costs, or study timelines.

Subsequently, a Bayesian Optimal Phase II design was selected for dose-expansion in single-arm patient cohorts with different tumor types because it allowed for efficient monitoring of clinical activity, and resulted in the most efficient sample size. The design was then amended to include a cohort with a control arm using a Bayesian logistic regression model.

For drugs known to have a benign safety profile, alternative designs that explore safety and efficacy simultaneously — such as BOIN 12 or TITE-BOIN 12 — should be considered. Additionally, early-phase oncology studies may benefit from deploying Bayesian basket designs to efficiently investigate efficacy in patients with different types of cancer.

As oncology therapeutics and early-phase design models evolve, sponsors will benefit from partnering with a company experienced in innovative designs for phase I and II oncology trials.

To learn more about dose optimisation and dose selection for oncology therapeutics, please contact us.

This article was first published on Clinical Oncology Daily on 3 June 2024.

In this section

Connect with us