Incorporating lower grade toxicity information into dose finding designs.

Area of Science:

• Clinical Trial Design
• Pharmacological Research
• Biostatistics

Background:

• Phase I clinical trials often use binary outcomes for dose-limiting toxicities, neglecting valuable information from individual toxicity grades.
• The continual reassessment method (CRM) is a standard adaptive design, but its efficiency can be limited by its dichotomous approach to toxicity.

Purpose of the Study:

• To investigate if incorporating individual toxicity grade information can enhance the operating characteristics of the continual reassessment method.
• To compare the performance of standard CRM with modified designs that utilize lower-grade toxicity data.

Main Methods:

• Simulations were used to compare the original CRM with two-stage CRM designs that incorporate lower-grade toxicity information.
• Designs included utilizing lower grades in the first stage only or throughout the trial via explicit models.
• Models were developed to relate lower-grade toxicity rates to dose-limiting toxicity rates.

Main Results:

• Significant improvements in accuracy and precision were observed when toxicity grades were used in the first stage of a two-stage CRM design.
• Modest gains were seen when lower-grade toxicity information was integrated throughout the trial using explicit models.
• The use of toxicity grades increased the number of patients treated at the maximum tolerated dose by approximately 5%.

Conclusions:

• Incorporating toxicity grades into phase I trial designs, especially in early stages, can offer performance benefits for the continual reassessment method.
• While not always substantial, using graded toxicities did not lead to poorer performance compared to standard methods.
• Investigators are encouraged to consider using graded toxicities during the design phase of phase I trials.