艾昆纬-IQVIA试用设计器的应计建模和自适应试用监控（英）-2023-20页_2mb

Summary of the White Paper

Introduction Overview

The white paper introduces IQVIA Trial Designer (ITD) as a web-based system for adaptive clinical trial designs, addressing challenges like uncertain effect size through statistical models, simulations, and adaptive monitoring. It highlights the importance of incorporating accrual modeling to predict recruitment and event collection, ensuring efficient trial execution and resource allocation. The paper outlines ITD's capabilities in dose-escalation, group-sequential designs, and various statistical methods.

IQVIA Trial Designer Features

ITD is a user-friendly, modular platform providing tools for designing, simulating, and monitoring trials. Key features include:

• Dose-escalation methods (e.g., 3+3, BOIN, mTPI-2) for finding maximum tolerated doses.
• Adaptive group-sequential designs for endpoints like continuous, binary, and time-to-event outcomes.
• Spending functions (e.g., O’Brien-Fleming, Pocock) for managing type I and type II error rates across interim analyses.
• Sample size re-estimation techniques (CDL, CHW) for updating trial size based on interim data.
• Accrual modeling using Poisson-gamma and non-homogeneous processes to handle recruitment variability.
• Technical support via React, Node.js, Docker, and R for robust calculations.

Adaptive Trial Designs

Adaptive designs address uncertainty in effect size through three main approaches:

• Conservative: Start with a large sample size, allowing early stopping for efficacy or futility.
• Agile: Begin with a smaller sample size, facilitate sample size re-estimation at interim analyses.
• Enrichment: Strategically target subgroups based on interim data, using machine learning tools.
  These designs allow flexible decisions based on accumulating data, but accrual assumptions (e.g., constant recruitment rates) can impact timelines.

Spending Functions

Spending functions control error rates across interim analyses:

• O’Brien-Fleming: Conservative, allocates few early boundaries, better for efficacy stopping.
• Pocock: Less conservative, maintains near-constant boundaries, suitable for futility or early termination.
• Other variants (HS, HP) offer flexibility in error spending. ITD supports custom user-defined functions.

Sample Size Re-estimation

Methods like the Chen-DeMets-Lan (CDL) and Cui-Hung-Wang (CHW) adjust sample size increases based on interim analysis, ensuring appropriate power without decrements, aligning with FDA guidelines.

Accrual Modeling and Trial Monitoring

ITD models accrual using Poisson-gamma processes, considering:

• Delayed site openings, piece-wise recruitment rates, and seasonality.
• Simulation to predict recruitment and event accrual timelines, aiding in adapting strategies for delays or inefficiencies.
• Supports monitoring and quantile-based projections to handle operational realities.

Case Study Insights

The case study demonstrates how ITD's accrual simulations and adaptive features enhance trial planning:

• Scenario I: Direct calculation vs. simulation reveals recruitment variability.
• Scenario II: Multiple sites with site-opening delays extend recruitment periods, highlighting needs for additional sites.
• Scenario III: Piece-wise recruitment rate changes impact projections, emphasizing dynamic monitoring.
• Scenario IV: Variable hazard rates under seasonality or trial data influence event accrual timing.

Conclusion

Accrual modeling and simulations in adaptive designs are essential for accurate trial planning, monitoring, and adaptation. IQVIA Trial Designer facilitates collaboration and streamlines processes, helping to optimize resource use, ensure statistical power, and make data-driven decisions in clinical development.