The US Food and Drug Administration (FDA) is preparing to advance the use of structured benefit-risk assessments in its decision making under its commitments in the recently reauthorized Prescription Drug User Fee Act (PDUFA VI), and agency officials on Monday laid out key considerations for sponsors.
In 2009, FDA began work to develop a structured approach to conducting benefit-risk assessments as part of an effort to bring more clarity and consistency to its regulatory decisions.
Under PDUFA V, FDA committed to developing a five-year plan to implement a structured benefit-risk assessment in its review processes, and in 2013 the agency released a draft implementation plan describing its approach.
In general, the structured benefit-risk assessment looks at information about the condition a drug is meant to treat, currently available treatments, benefits, risks and risk management. FDA then weighs the evidence and uncertainties in each area to inform its conclusions and rationale behind regulatory decisions.
The implementation plan also discusses quantitative benefit-risk assessments, which involve assigning numerical weights to benefits and risks and using quantitative modeling to inform decision making.
But while FDA acknowledges that quantitative analyses are an important aspect of its review process when paired with qualitative methods, the agency expresses skepticism about fully quantitative approaches.
"The subjective judgements and assumptions that would inevitably be embodied in such quantitative decision modeling would be much less transparent, if not obscured, to those who wish to understand a regulator's thinking," FDA writes.
Considerations for Quantitative Benefit-Risk Models
Speaking at a public meeting on FDA's implementation of its benefit-risk framework, Richard Moscicki, deputy center director for science operations at the Center for Drug Evaluation and Research (CDER) said that formal quantitative and semi-quantitative approaches to benefit-risk assessment "may add further value to FDA's most challenging regulatory decisions."
Richard Forshee, associate director for research at the Office of Biostatistics and Epidemiology within the Center for Biologics Evaluation and Research (CBER) said FDA will need to ensure that such models are "accurately conveying the uncertainty and the variability of the system."
"If the data going in is not good, you're not going to be able to get a model that's believable or useful," Forshee said.
To address concerns that fully quantitative approaches may obscure regulatory decision making, Forshee added, "No one believes that quantitative benefit-risk assessments are going to replace risk management and the judgment that's necessary for making these very difficult decisions."
But Forshee said, "There's certainly a possibility for more quantitative approaches to be considered, especially now that the ICH guidance has been out," referring to ICH's M4E(R2) guideline that was adopted last June.
Specifically, the M4E(R2) guideline says there are multiple approaches available for conducting a benefit-risk assessment, and while descriptive approaches are "generally adequate…applicant[s] may choose to use methods that quantitatively express the underlying judgments and uncertainties in the assessment."
According to Forshee, sponsors should conduct many sensitivity analyses when developing their models.
A best practice, Forshee said, would be to test the model against an external dataset "to see if the results are going to still be valid when you move beyond the data that you originally used."
Forshee also cautioned that quantitative models may need to be updated to reflect changing circumstances such as new scientific discoveries, but noted that this is true for less quantitative methods as well.
Editor's note: This article was edited and updated with more background information.