Incorporating clinical utility and Pharmacokinetics (PK)/Pharmacodynamics (PD) in clinical development of oncology drug candidate with a Bayesian decision analysis perspective in the context of OPTIMUS initiative
Elisabeth Rouits(1), Arnaud Monseur (1), Maud Hennion (1), Bruno Boulanger (1)
(1) Cencora Pharmalex
Objectives: In oncology, poor dose optimization can have negative consequences for patients,most commonly because of toxicity, including poor quality of life, reduced effectiveness because ofinability of patients to stay on current therapy or receive subsequent therapy because of toxicities, and difficultyin developing combination regimens.
With the launch of the FDA OPTIMUS initiative (1), drug developers are encouraged to explore optimized dosing ranges for anticancer drug instead of focusing on the old paradigm of Maximum Tolerated Dose (MTD) evaluation. In phase I trials, effectively treating patients and minimizing the chance of exposing them to subtherapeutic and overly toxic doses are clinician’s top priority. Motived by this practical consideration, Bayesian Optimal Interval Design (BOIN) seems to be very popular and somehow used as a standard design to address the optimal dosing range selection. Indeed, BOIN design can be easily implemented in a simple way similar to the traditional “3+3” design. Compared to the well-known continual reassessment Method (CRM), the BOIN design yields comparable average performance to select the MTD but has a substantially lower risk of assigning patients to subtherapeutic and overly toxic doses (2). If these latter characteristics addressed by the BOIN design answer part of the requirements of OPTIMUS, it does not really help or support the evaluation of the Exposure-Response characteristics which is critical to identify a dose range and schedule of administration expected to optimize the Benefit/Risk ratio.
Methods: To that aim, a Bayesian Decision Analysis (BDA) based on Clinical Utility Index (CUI) could be set-up to strike the optimal balance between FDA’s need to limit adverse effects and patients’ need for expedited access to a potentially effective therapy (3). Instead of predefining a standard threshold of acceptable toxicity (eg 30%), a CUI reflecting the unmet medical need would be set-up. The CUI would be based not only on standard safety parameters evaluation but also on Health Related Quality of Life (HRQoL) and on Pharmacokinetics (PK), Pharmacodynamics (PD) and Pharmacogenomics (PGx) markers evaluation. PK, PD, PGx, toxicity and proximal efficacy data are collected all along the phase I study thus supporting the Exposure (i.e. PK)-Response characterization either in terms of safety (toxicity clinical endpoints +/- PD markers of toxicity) and efficacy (proximal efficacy endpoints +/- PD markers of efficacy). PGx is generally informative to weight and fine-tune the translational and clinical relevance of PK and PD by identifying and highlighting specific targets and patients populations characteristics.
Results: The PK-PD or Exposure-Response relationship would be fully fine-tuned at the end of phase I to select at least two dosing regimens assumed to support a clinically relevant probability of efficacy and signal-to-noise ratio of treatment effect. At the end of phase II, Exposure-Response relationship would be characterized, thus identifying the most relevant dosing regimen assumed to maximize chance of success of the phase III.
Conclusions: BDA based on CUI provides a useful framework to incorporate subjective perspectives of cancer patients and objective burden-of-disease metrics to evaluate the therapeutic effects of anticancer drugs under development. This is even more powerful in the context of the development of the different mechanistic approaches of anticancer agents which are no longer limited to cytotoxic chemotherapies but mainly targeted therapies, immunotherapies or combination of those.
References:
[1] FDA OPTIMUS project FDA OPTIMUS
[2] Bayesian Optimal Interval Design: A Simple and Well-Performing Design for Phase I Oncology Trials. Yuan Y et al., Clin Cancer Res 2016
[3] Incorporating Patient Preferences via Bayesian Decision Analysis. Chaudhuri E et al., Clin J Am Soc Nephrol 2017