Evaluation of irrational dose assignment definitions using the continual reassessment method.

Area of Science:

• Clinical Trials Methodology
• Biostatistics
• Pharmacology

Background:

• Phase I clinical trials aim to determine safe and effective drug dosages.
• Irrational dose assignment is defined as failing to de-escalate a dose despite toxicity events.
• The continual reassessment method (CRM) has been criticized for potentially leading to irrational dose assignments.

Purpose of the Study:

• To critically examine the definition of irrational dose assignment within the context of the CRM.
• To evaluate the frequency and nature of irrational dose assignments generated by the CRM.

Main Methods:

• Simulated 2000 Phase I clinical trials using the CRM with varying dose-limiting toxicity probabilities.
• Assessed dose assignments against definitions of irrationality proposed by Zhou et al.
• Classified irrational dose assignments as underdose, target dose, or overdose based on true toxicity probabilities.

Main Results:

• Irrational dose assignments occurred in 18.4% of decisions flagged by Zhou's criteria, totaling 4.5% of all dose assignments.
• Less than 1% of all simulated dose assignments were classified as irrational.
• A significant portion of these 'irrational' assignments were made at the true target or underdose levels, suggesting they were not truly irrational.

Conclusions:

• The definition of irrational dose assignment by Zhou et al. may not be appropriate for modern methods like CRM.
• Irrational dose assignments are infrequent with CRM and often justifiable upon closer examination of accumulated data.
• The benchmark for evaluating dose assignment methods should not be based on older designs like the 3+3.