Exploring the Feasibility of a Bracketing Approach Utilizing Modeling for Development of Long-Acting Injectables for Regulatory Approval-A Case Study Using Levonorgestrel.

BACKGROUND

The development of long-acting products of a characterized drug substance is of great interest. It is possible to support the development of these products with available clinical data by matching the exposure to a predefined bracket of a minimal concentration for efficacy and a maximal concentration for safety. This bracketing approach would cut down on the time and cost of new long-acting contraceptive products progressing to market. The current study describes the assessment of the data available to support a bracketing approach to conclude comparable levels of efficacy and safety for a postulated novel long-acting reversible contraceptive (LARC) product of levonorgestrel.

METHODS

Literature evidence of levonorgestrel efficacy, as quantified by the Pearl Index, was utilized and modeled by incorporating three LARC products for the estimation of a minimal concentration required for efficacy. Further literature was reviewed to quantify the maximal concentration required to ensure product safety. Additionally, a review of the regulatory precedence for the approach was conducted using European and UK databases.

RESULTS

There was a reasonable definition of the minimal concentrations for efficacy where the target concentrations of levonorgestrel were in the range of 200-400 pg/mL. Maximum concentrations for safety were less well defined. Although regulatory guidance supports the bracketing approach, there is little precedence for licensing new products based on pharmacokinetic data only, despite much reduced clinical and non-clinical packages being evidenced.

CONCLUSIONS

Understanding of the exposure response is not currently considered sufficient to support a bracketing approach for a new levonorgestrel product. If additional safety data are established, current regulations may allow for a reduced application package. Additional work is needed to support the approach, and this could utilize the wealth of information in real-world datasets combined with systems models.