Would the Recommended Dose Have Been Different Using Novel Dose-Finding Designs? Comparing Dose-Finding Designs in...

Area of Science:

• Clinical Trials
• Biostatistics
• Oncology Drug Development

Background:

• The traditional 3+3 design is widely used in oncology dose-finding trials despite simulation studies suggesting superior alternatives.
• Investigators often cite a lack of real-world validation for novel designs like the continual reassessment method (CRM) and Bayesian optimal interval (BOIN) design as a reason for continued reliance on the 3+3 design.
• Simulation studies indicate that CRM and BOIN designs can identify the maximum tolerated dose (MTD) more frequently, use fewer patients, and allocate more patients to the MTD compared to the 3+3 design.

Purpose of the Study:

• To evaluate whether novel dose-finding designs (CRM and BOIN) would have yielded different results in actual clinical trials compared to the traditional 3+3 design.
• To confirm the advantages of novel designs in real-world settings, addressing the skepticism regarding simulation study outcomes.
• To compare the MTD, patient allocation, and sample size between novel designs and the 3+3 design using published trial data.

Main Methods:

• A random sample of 60 published dose-finding trials was analyzed.
• 22 trials that used the 3+3 design, identified an MTD, had published toxicity data, and included more than two dose levels were selected for comparison.
• The published MTD was compared with MTDs estimated by CRM and BOIN designs using target toxicity rates of 25% and 30%, and the trial's toxicity data.

Main Results:

• Model-based designs (CRM and BOIN) selected higher dose levels than the published MTD in approximately 40% of the analyzed trials.
• Estimated and observed toxicity rates using novel designs were closer to the target toxicity rates (25% and 30%).
• Novel designs resulted in fewer patients assigned to suboptimal doses and allowed for faster dose escalation compared to the 3+3 design.

Conclusions:

• This study provides real-world evidence that novel dose-finding designs (CRM and BOIN) would have recommended different MTDs compared to the 3+3 design.
• The findings confirm the advantages of novel designs in terms of efficiency and accuracy, as previously suggested by simulation studies.
• The results support the adoption of advanced dose-finding methodologies in oncology trials to improve drug development efficiency and patient safety.