The BOIN or the Bayesian Optimal Interval design is a statistical approach quite commonly used in early phase clinical trials. It is particularly effective when the goal is to determine the maximum tolerated dose (MTD) of the investigational new drug or and/or to determine the recommended dose for the Phase II study (RP2D).

In early-phase clinical trials, researchers often need to explore a range of dose levels to assess both safety and efficacy. The BOIN Design provides a framework for efficiently allocating patients to different dose levels based on accumulated data from the trial. It combines elements of Bayesian methods and interval estimation to make informed decisions for dose escalation or de-escalation.

Key steps in implementing BOIN design for an early-phase trial:

  1. Dose selection:

The trial starts with a small number of dose levels, typically a wide range spanning from low to high doses. The initial doses are selected based on existing information, preclinical data, or previously conducted studies, if available.

  1. Cohort size:

A fixed cohort size is assigned to each dose level, usually ranging from 1 to 3 participants. The cohort size can be adapted, allowing adjustments based on accumulating data or predefined stopping rules.

  1. Bayesian updating:

The BOIN design employs Bayesian statistical methods to update the dose-toxicity and dose-efficacy models as data accumulates.

  1. Specify prior distributions for the dose-toxicity relationship and any other relevant parameters. This information can be based on preclinical data, previously conducted clinical trials or expert knowledge.
  2. After each patient's toxicity outcome is noted, the prior distribution is updated using Bayesian methods to obtain a posterior distribution. This updated distribution helps determine the dose assignment for subsequent study participants.
  1. Dose escalation and De-escalation:

Typically, the trial begins with the lowest dose. After each cohort, the safety and efficacy data are evaluated to determine the next dose. If any toxicity observed is within acceptable limits and efficacy is promising, the dose is escalated to the next level. If toxicity seen is beyond acceptable limits, the dose is de-escalated to a lower level. The specific rules for dose escalation and de-escalation are predefined and based on detailed statistical models.

  1. Stopping criteria:

This design defined stopping rules based on toxicity and efficacy outcomes. For example, if the highest dose level reaches a predefined acceptable toxicity level or if a recommended dose is identified based on efficacy criteria, the trial may be stopped.

Advantages

“Statistical Method: Bayesian Optimal Interval (BOIN) design” is designated as a drug development tool (DDTs) accepted by the FDA under the Fit-for-Purpose (FFP) Initiative which provides a pathway for regulatory acceptance of dynamic tools in drug development programs.

The BOIN design is highly adaptive, allowing for dose adjustment based on accumulating trial data. This proves to be an efficient dose-finding mechanism to identify the optimal dose or dosing regimen much faster with a smaller sample size. This helps reduce patient exposure and expedites study completion with marked efficiencies while delivering statistically robust outcomes.

To talk to our early-phase biostatistics and programming experts, reach out to our team at hello@algorics.com

References: