Clinical Trials最新文献

Background: Platform trials typically feature a shared control arm and multiple experimental treatment arms. Staggered entry and exit of arms splits the control group into two cohorts: those randomized during the same period in which the experimental arm was open (concurrent controls) and those randomized outside that period (nonconcurrent controls). Combining these control groups may offer increased statistical power but can lead to bias if analyses do not account for time trends in the response variable. Proposed methods of adjustment for time may increase type I error rates when time trends impact arms unequally or when large, sudden changes to the response rate occur. However, there has been limited exploration of the degree of type I error inflation one can plausibly expect in real-world scenarios.

Methods: We use data from the Adaptive COVID-19 Treatment Trial (ACTT) to mimic a realistic platform trial with a remdesivir control arm. We compare four strategies for estimating the effect of interferon beta-1a (the ACTT-3 experimental arm) relative to remdesivir (data from ACTT-1, ACTT-2, and ACTT-3) on recovery and death by day 29: utilizing concurrent controls only (the prespecified analysis), pooling all remdesivir arm data without adjustment (the "unadjusted-pooled" analysis), adjusting for time as a categorical variable, and a Bayesian hierarchical model implementation which adjusts for time trends using smoothing techniques (the "Bayesian time machine"). We compare type I error rates and relative efficiency of each method in simulation settings based on observed ACTT remdesivir arm data.

Results: The unadjusted-pooled approach provided substantially different estimates of the effect of interferon beta-1a relative to remdesivir compared with the concurrent-only and model-based approaches, indicating that changes in recovery and death rates over time were not ignorable across different stages of ACTT. The model-based approaches rely on an assumption of constant treatment effects for each arm in the platform relative to control; error rates more than doubled in settings where this was not satisfied. Relative efficiency of the model-based approaches compared with the concurrent-only analysis was moderate.

Conclusions: In simulation settings where key model assumptions were not met, potential efficiency gains from incorporation of nonconcurrent controls were outweighed by the risk of substantial type I error rate inflation. This leads us to advise against these strategies for primary analyses in confirmatory clinical trials, aligning with current FDA guidance advising against comparisons to nonconcurrent controls in COVID-19 settings. The model-based adjustment methods may be useful in other settings, but we recommend performing the concurrent-only analysis as a reference for assessing the degree to which nonconcurrent controls drive results.

Adaptive seamless Phase 2/3 designs provide possible pathways to expedite drug development by combining dose selection and confirmatory evaluation on the selected dose with the control group in the same trial. Various methods have been developed to demonstrate the potential advantages compared to conventional development plan with separate Phase 2 and 3 trials. More practical and complicated situations occur when we want to achieve the goal of combining dose selection and confirmatory evaluation in clinical trials with multiple endpoints. Examples of multiple endpoints include multiple efficacy endpoints needed in the final stage for regulatory submissions. In this article, a few inferential adaptive seamless Phase 2/3 designs have been proposed which can combine dose selection and confirmatory stage in clinical trials evaluating multiple endpoints, including adaptive graph-based multiple testing procedure, adaptive seamless design with graph-based combination test, and seamless design with rank-based Dunnett-adjusted test. Simulations are conducted to confirm the control of the familywise type I error rate with an illustrated example design and assess the power. These designs can preserve the familywise type I error rate, and adaptive graph-based multiple testing procedure is more powerful than the others.

Introduction: The sequential parallel comparison design has emerged as a valuable tool in clinical trials with high placebo response rates. To further enhance its efficiency and effectiveness, adaptive strategies, such as sample size adjustment and allocation ratio modification can be employed.

Methods: We compared the performance of Jennison and Turnbull's method and the Promising Zone approach for sample size adjustment in a two-phase sequential parallel comparison design study. We also evaluated the impact of allocation ratio adjustments using Neyman and Optimal allocation strategies. Various scenarios were simulated to assess the effects of different design parameters, including weight in the test statistic, initial randomization ratio, and interim analysis timing.

Results: The Promising Zone approach demonstrated superior or comparable power to Jennison and Turnbull's method at equivalent expected sample sizes while maintaining the intuitive property that more promising interim results lead to smaller required follow-up sample sizes. However, the Promising Zone approach may require a larger maximum possible sample size in some cases. The addition of allocation ratio adjustments offered minimal improvements overall, but showed potential benefits when the variance in the treatment group was larger than that in the placebo group. We also applied our findings to a real-world example from the AVP-923 trial in patients with Alzheimer's disease-related agitation, demonstrating the practical implications of adaptive sequential parallel comparison designs in clinical research.

Discussion: Adaptive strategies can significantly enhance the efficiency of sequential parallel comparison designs. The choice between sample size adjustment methods should consider trade-offs between power, expected sample size, and maximum adjusted sample size. Although allocation ratio adjustments showed limited overall impact, they may be beneficial in specific scenarios. Future research should explore the application of these adaptive strategies to binary and survival outcomes in sequential parallel comparison designs.

Metaheuristics are commonly used in computer science and engineering to solve optimization problems, but their potential applications in clinical trial design have remained largely unexplored. This article provides a brief overview of metaheuristics and reviews their limited use in clinical trial settings. We focus on nature-inspired metaheuristics and apply one of its exemplary algorithms, the particle swarm optimization (PSO) algorithm, to find phase I/II designs that jointly consider toxicity and efficacy. As a specific application, we demonstrate the utility of PSO in designing optimal dose-finding studies to estimate the optimal biological dose (OBD) for a continuation-ratio model with four parameters under multiple constraints. Our design improves existing designs by protecting patients from receiving doses higher than the unknown maximum tolerated dose and ensuring that the OBD is estimated with high accuracy. In addition, we show the effectiveness of metaheuristics in addressing more computationally challenging design problems by extending Simon's phase II designs to more than two stages and finding more flexible Bayesian optimal phase II designs with enhanced power.

The adoption of innovative, model-based, and computationally intensive clinical trial designs is challenged by barriers including clinician engagement, regulatory acceptance, dissemination beyond major research institutions, and patient accrual. This session explored strategies to overcome these barriers. Key approaches discussed included the development of user-friendly software and interactive platforms to enhance transparency, open sharing of algorithms, and recognition of software contributions in academic publishing. Building collaborations with stakeholders predisposed to innovation, fostering interdisciplinary communication, and producing complementary methodological and clinical publications were emphasized as essential steps. Practical considerations for trials with small sample sizes included the use of adaptive designs, individualized trials, and alternative optimization strategies when traditional theoretical assumptions are infeasible. A major theme of the discussion was the importance of model assumptions in innovative designs. Questions were raised about the sensitivity of results to these assumptions and the robustness of methods, particularly under limited sample sizes. Addressing this requires extensive simulation studies across varied scenarios to assess operating characteristics. The focus should be on achieving clinically meaningful goals-such as identifying effective dose regions-rather than perfect model specification. Speakers emphasized the need to acknowledge and, when feasible, test assumptions post hoc, integrating such verification as secondary objectives in trial design. An iterative scientific process was encouraged, recognizing that trials not only serve immediate clinical goals but also advance broader scientific understanding. Assumptions provide a principled foundation for methodology, but thoughtful scrutiny of their realism was urged, given the risk of relying on overly strong or untestable premises. The potential of metaheuristic algorithms was highlighted for efficiently identifying optimal designs across different model assumptions, supporting robustness evaluations. Practical implementation should adapt optimal designs to stakeholder needs while preserving acceptable statistical efficiency. In sum, advancing the adoption of innovative designs requires improved communication, infrastructure, and methodological transparency, alongside careful evaluation of model assumptions and robustness.