Journal of Biopharmaceutical Statistics最新文献

Dose optimization is a critical challenge in drug development. Historically, dose determination in oncology has followed a divergent path from other non-oncology therapeutic areas due to the unique characteristics and requirements in Oncology. However, with the emergence of new drug modalities and mechanisms of drugs in oncology, such as immune therapies, radiopharmaceuticals, targeted therapies, cytostatic agents, and others, the dose-response relationship for efficacy and toxicity could be vastly varied compared to the cytotoxic chemotherapies. The doses below the MTD may demonstrate similar efficacy to the MTD with an improved tolerability profile, resembling what is commonly observed in non-oncology treatments. Hence, alternate strategies for dose optimization are required for new modalities in oncology drug development. This paper delves into the historical evolution of dose finding methods from non-oncology to oncology, highlighting examples and summarizing the underlying drivers of change. Subsequently, a practical framework and guidance are provided to illustrate how dose optimization can be incorporated into various stages of the development program. We provide the following general recommendations: 1) The objective for phase I is to identify a dose range rather than a single MTD dose for subsequent development to better characterize the safety and tolerability profile within the dose range. 2) At least two doses separable by PK are recommended for dose optimization in phase II. 3) Ideally, dose optimization should be performed before launching the confirmatory study. Nevertheless, innovative designs such as seamless II/III design can be implemented for dose selection and may accelerate the drug development program.

Dose selection and optimization in early phase of oncology drug development serves as the foundation for the success of late phases drug development. Bivariate Bayesian logistic regression model (BLRM) is a widely utilized model-based algorithm that has been shown to improve the accuracy for identifying recommended phase 2 dose (RP2D) based on dose-limiting-toxicity (DLT) over traditional method such as 3 + 3. However, it remains a challenge to optimize dose selection that strikes a proper balance between safety and efficacy in escalation and expansion phase of phase I trials. In this paper, we first use a phase I clinical trial to demonstrate how the variability of drug exposure related to pharmacokinetic (PK) parameters among trial participants may add to the difficulties of identifying optimal dose. We use simulation to show that concurrently or retrospectively fitting BLRM model for dose/toxicity data from escalation phase with dose-independent PK parameters as covariate lead to improved accuracy of identifying dose level at which DLT rate is within a prespecified toxicity interval. Furthermore, we proposed both model- and rule-based methods to modify dose at patient level in expansion cohorts based on their PK/exposure parameters. Simulation studies show this approach leads to higher likelihood for a dose level with a manageable toxicity and desirable efficacy margin to be advanced to late phase pipeline after being screened at expansion phase of phase I trial.

In this article, we propose a phase I-II design in two stages for the combination of molecularly targeted therapies. The design is motivated by a published case study that combines MEK and PIK3CA inhibitors; a setting in which higher dose levels do not necessarily translate into higher efficacy responses. The goal is therefore to identify dose combination(s) with a prespecified desirable risk-benefit trade-off. We propose a flexible cubic spline to model the marginal distribution of the efficacy response. In stage I, patients are allocated following the escalation with overdose control (EWOC) principle whereas, in stage II, we adaptively randomize patients to the available experimental dose combinations based on the continuously updated model parameters. A simulation study is presented to assess the design's performance under different scenarios, as well as to evaluate its sensitivity to the sample size and to model misspecification. Compared to a recently published dose finding algorithm for biologic drugs, our design is safer and more efficient at identifying optimal dose combinations.

In oncology dose-finding trials, small cohorts of patients are often assigned to increasing dose levels, with the aim of determining the maximum tolerated dose. In the era of targeted agents, this practice has come under intense scrutiny as treating patients at doses beyond a certain level often results in increased off-target toxicity without significant gains in antitumor activity. Dose optimization for targeted agents becomes more challenging in proof-of-concept trials when the experimental treatment is tested in multiple indications of low prevalence and there is the need to characterize the dose-response relationship in each indication. To provide an alternative to the conventional "more is better" paradigm in oncology dose finding, we propose a Bayesian model averaging approach based on robust mixture priors (rBMA) for identifying the recommended phase III dose in randomized dose optimization studies conducted simultaneously in multiple indications. Compared to the dose optimization strategy which evaluates the dose-response relationship in each indication independently, we demonstrate the proposed approach can improve the accuracy of dose recommendation by learning across indications. The performance of the proposed approach in making the correct dose recommendation is examined based on systematic simulation studies.

We consider a dose-optimization design for a first-in-human oncology trial that aims to identify a suitable dose for late-phase drug development. The proposed approach, called the Pharmacometrics-Enabled DOse OPtimization (PEDOOP) design, incorporates observed patient-level pharmacokinetics (PK) measurements and latent pharmacodynamics (PD) information for trial decision-making and dose optimization. PEDOOP consists of two seamless phases. In phase I, patient-level time-course drug concentrations, derived PD effects, and the toxicity outcomes from patients are integrated into a statistical model to estimate the dose-toxicity response. A simple dose-finding design guides dose escalation in phase I. At the end of the phase I dose finding, a graduation rule is used to assess the safety and efficacy of all the doses and select those with promising efficacy and acceptable safety for a randomized comparison against a control arm in phase II. In phase II, patients are randomized to the selected doses based on a fixed or adaptive randomization ratio. At the end of phase II, an optimal biological dose (OBD) is selected for late-phase development. We conduct simulation studies to assess the PEDOOP design in comparison to an existing seamless design that also combines phases I and II in a single trial.

Combination therapy, a treatment modality that involves multiple treatment agents, has become imperative for improving treatment effectiveness and addressing resistance in the field of oncology. However, determining the most effective dose for these combinations, particularly when dealing with intricate drug interactions and diverse toxicity patterns, presents a substantial challenge. This paper introduces a novel Bayesian dose-finding design for combination therapies with information borrowing, named the DOD-Combo design. Leveraging historical single-agent trials and the meta-analytic-predictive (MAP) power prior, our approach utilizes a copula-type model to connect individual drug priors with joint toxicity probabilities in combination treatments. The MAP power prior allows the integration of information from multiple historical trials, constructing informative priors for each agent. Extensive simulations confirm our method's superior performance compared to combination designs with no information borrowing. By adaptively incorporating historical data, our approach reduces sample sizes and enhances efficiency in selecting the maximum tolerated dose (MTD), effectively addressing the intricate challenges presented by combination trials.

Dose selection is a key decision to make in the early phase of drug development. Classical phase I/II dose-finding trials randomly assign a few doses and select the best among them. Response-adaptive assignment designs are more efficient but are still far from optimal. Recently, some researchers used machine learning (ML) methods such as contextual bandits (CB) to find the "optimal" dose and to investigate the asymptotic properties of the methods. We present a case study for oncology phase I/II dose-finding trial designs using Thompson sampling and Bayesian bootstrap for CB with either modeling clinical utility directly or jointly modeling efficacy and safety. We focus on practical questions such as the number of interim analyses to conduct and whether we should model the utility directly, jointly model efficacy and safety which compose the utility, or use a model independent approach such as multi-armed bandits, but not for a specific compound or tumor type. We also consider how to use weak informative prior information. We conducted an extensive simulation study and compared different combinations of design settings and modeling methods, under several feasible scenarios of the dose-response relationship. Based on simulation results, we make practical recommendations for the use of the proposed ML approach for phase I/II dose-finding trial designs.

We propose a Bayesian optimal phase 2 design for jointly monitoring efficacy and toxicity, referred to as BOP2-TE, to improve the operating characteristics of the BOP2 design proposed by Zhou. BOP2-TE utilizes a Dirichlet-multinomial model to jointly model the distribution of toxicity and efficacy endpoints, making go/no-go decisions based on the posterior probability of toxicity and futility. In comparison to the original BOP2 and other existing designs, BOP2-TE offers the advantage of providing rigorous type I error control in cases where the treatment is toxic and futile, effective but toxic, or safe but futile, while optimizing power when the treatment is effective and safe. As a result, BOP2-TE enhances trial safety and efficacy. We also explore the incorporation of BOP2-TE into multiple-dose randomized trials for dose optimization, and consider a seamless design that integrates phase I dose finding with phase II randomized dose optimization. BOP2-TE is user-friendly, as its decision boundary can be determined prior to the trial's onset. Simulations demonstrate that BOP2-TE possesses desirable operating characteristics. We have developed a user-friendly web application as part of the BOP2 app, which is freely available at https://www.trialdesign.org.

Inverse probability of treatment weighting (IPTW) is a common approach to infer causal treatment effects when covariates are imbalanced at baseline or over time among treatment groups. One limitation of the IPTW is that a few observations with large weights can disproportionately influence inference, leading to dramatically increased variability in estimation. Stabilizing weights were developed to mitigate such a variability caused by excessively large weights. Since then, stabilizing weights have been widely regarded as good practice for IPTW, despite some misunderstandings and misinterpretations of their functionality. For example, a common misconception is that the original IPTW artificially inflates a study's statistical power because the weighted sample size appears to double that of the original. This article clarifies the role of stabilization in IPTW analysis, focusing on linear, logistic, and Cox's Proportional Hazard analyses in baseline binary treatment settings, which are commonly encountered in the regulatory space. Through theoretical derivations and simulation studies, we show that stabilized IPTW models yield identical point estimates to the original IPTW models in saturated linear and logistic regressions but yield slightly different estimates in Cox regressions. Stabilizing IPTW improves variance estimation over the original IPTW only when within-subject correlation due to weighting is ignored (as in model-based variance estimation, which is an incorrect approach), while none to minimal differences are observed when using robust variance estimation. Regardless of stabilization, a robust, sandwich-type variance estimator or resampling-based methods are the more appropriate approach for accurate variance estimation.