US regulators are planning to hold a meeting on emerging modeling methods that could be used to fine-tune dose selection studies and decision-making.
Dose optimization is one of the most crucial parts of drug development, and surprisingly one of the most frequent contributors to the US Food and Drug Administration (FDA) rejecting a drug.
A recent study in the Journal of the American Medical Association (JAMA) found that 15.9% of all drugs rejected by FDA during the first round of reviews were due to "uncertainties related to dose selection." That percentage was higher than any other factor considered in the study, failure to select an endpoint with a clinically meaningful effect (13.2%), inconsistent results (13.2%), inconsistent results among trials or study sites (11.3%) or poor efficacy related to the current standard of care (13.2%).
Consider the case of a 2013 advisory committee meeting for Merck's sleeping pill, suvorexant. The company had requested approval at a 15 mg dose for elderly patients and 20 mg for non-elderly patients. The advisory committee reviewing the drug, however, said there was little evidence to support higher doses of the drug, and pushed instead for a 10 mg dose to be approved. The confusion over appropriate dose levels ultimately caused FDA to issue Merck a Complete Response Letter (CRL), effectively rejecting the drug for the time being.
Clearly, then, there is a high need for industry as a whole to focus on ways of reducing failure, especially as the cost of re-doing trials at the correct dose-level is extraordinary.
New Effort Aimed at Dose Selection
FDA now seems poised to take its own crack at addressing that need. In a 10 February 2014 Federal Register notice, it said it will soon hold a public workshop on applying physiologically based pharmacokinetic (PBPK) modeling to dose-selection methods.
"The purpose of the workshop is to obtain input on scientific approaches for the conduct and assessment of PBPK modeling within the framework of drug development and regulatory decision making," FDA wrote. The end goal, it said, was to take input from the workshop and use it to refine the agency's thinking as it related to the "various applications" of PBPK.
The meeting is reportedly an initiative borne from the Food and Drug Administration Safety and Innovation Act (FDASIA), which calls on FDA to improve regulatory science using industry-paid user fees. Per FDA's commitment letter to industry, one of the regulatory science-related areas set to receive special attention is PBPK.
So what is PBPK, exactly? As FDA explains:
"PBPK modeling is a mathematical modeling technique for predicting drug behavior in humans. A PBPK model takes information about a drug's physical, chemical, and other properties, as well as information about processes in the body, and turns them into mathematical equations to predict what will happen when a patient takes the medication."
PBPK isn't new-it's been around since at least the 1970s-but its applications and methods are becoming increasingly complex, allowing FDA and industry to rely less on clinical trials and more on PBPK modeling. But as with all regulatory science methods, they must first be validated and approved by FDA.
Agency Questions and Aims
The agency said it is currently "looking to adopt a rigorous approach to the review of PBPK submissions and the conduct of de novo PBPK analysis to support regulatory review" using a transparent process to allow the public to understand its evidentiary standards and how they weigh on regulatory decision making.
Key questions at FDA's 10 March 2014 meeting are set to include:
- What gaps in knowledge about PBPK exist?
- When should PBPK simulations be included in drug labeling?
- What is the best format for presenting PBPK simulations in different sections of the labeling?
- How should uncertainty in simulations be presented in the labeling?
The workshop will also be used to prepare a concept paper on best practices and principles of PBPK in drug development and regulatory reviews, FDA said.
Federal Register Notice