"DECONVOLUTION" METHODS FOR MEASURING DRUG ABSORPTION BEING DEVELOPED UNDER FDA CONTRACT; SIMULATIONS MAY BE USED IN DRAFTING BIOEQUIVALENCE GUIDANCES

Third-generation "deconvolution" methods used to quantify drug release, absorption and delivery rates are being developed under a contract from FDA's Office of Generic Drugs, the Generic Drugs Advisory Committee was told Feb. 11. Deconvolution -- a testing method whereby a drug is injected into one side of the body and a blood sample is subsequently taken from the other side of the body -- is useful in situations when it is problematic to measure T[max] (time to maximum concentration) and C[max] (maximum concentration). For example, the method is useful when trying to determine rate of absorption of controlled-release drugs, Peter Veng-Pederson, PhD, University of Iowa, explained. The drug can be delivered by bolus injection or other type of drug delivery method. An elemental principle of deconvolution studies that investigators need to be concerned about is "if there's a stochastic [random], independent transport of the drug from the input side to the sampling side," Veng-Pederson said. Investigators will want to know "whether that transport is independent in the sense that the presence of one molecule is not going to really influence the probability of other molecules of the same kind to arrive at this sampling site," the researcher noted. "On the basis of that, you can quickly by logical means and statistical means arrive at the so-called superposition principle. That's the principle you make use of when you test for disposition linearity." Committee member Gordon Amidon, PhD, University of Michigan College of Pharmacy, said that for "those drugs where we have an I.V. dosage form and where we have linear disposition, [deconvolution] will be the best method for getting at input rate into the systemic circulation." He added that "when it comes to making decisions about regulating modified or controlled-release dosage forms, it may become the preferred technique." FDA Biopharmaceutics Division Director Thomas Ludden, PhD, pointed out that "on the new drug side, we've actually been encouraging sponsors for some time to work with some of the second- and third-generation methods of deconvolution to try to understand their drug products better; actually many sponsors don't need that encouragement." Veng-Pederson said "a lot of people have the misconception that, in order to do deconvolution, you need to base that on some kind of structured models." He noted that "unfortunately, in spite of the generality and usefulness of this methodology, it remains greatly underutilized." Discussing the contract with Veng-Pederson's group, FDA Office of Generic Drugs Director Roger Williams said: "I look forward, perhaps, to them coming back before the committee sometime in the next year or so [to] talk more about this." The group has been working on the project only for a few months. The advisory committee also heard a presentation by Tom Tozer, PhD, University of California at San Francisco, about simulation studies to determine the accuracy of measures of the rate and extent of absorption of drugs. The simulations were conducted under a contract with OGD. FDA wants to reconfirm whether current measures of rate and extent are precise and to determine whether, in some cases, rate may be an unimportant measure. The studies could result in new or modified measures. The research involved simulations of bioequivalence data using various pharmacokinetic and statistical models. Tozer concluded from the simulations that "we find that none of the measures are really robust throughout the conditions in which you're wanting to look at." However, he expounded, "one thing that greatly surprised us was that C[max] appears to be a good measure of extent. This is particularly true when the rate of absorption...is fixed and the variability between...the two preparations is manifest basically in the extent." As for rate measures, Tozer reported that T[max] "overall...was probably the most sensitive of all of the rate parameters." The measure of partial area under the curve (AUC) "seemed to always have a low producer risk," Tozer said. Producer risk is the probability of not detecting bioequivalence in spite of the fact that the test and reference products are bioequivalent. Although committee members were enthusiastic about the simulations, they expressed a desire to see more conducted. Several committee members suggested that Tozer and his colleagues put the characteristics of actual drugs through the simulations. Committee Chairman Terrence Blaschke, MD, Stanford University Medical Center, suggested that Tozer "go to some specific drugs where we may have some data...and use those to help drive the scenarios [of] the simulations." Committee member Kathleen Lamborn, PhD, Director of Statistical Operations, Quintiles Pacific (Palo Alto, Calif.), agreed, saying: "Let's get as close to a couple of real situations as we can." OGD's Williams commented: "I think [Tozer] has given us a really remarkable tool to work with" and added that simulations could determine the best metrics for certain drugs, which could go into the bioequivalence guidances that the Office of Generic Drugs is developing. "I think that's a very realistic possibility and it's really a matter now of implementing it and making it happen, which means we might have to ask [Tozer] to do some more work or try to bring it into the office in some way," the FDAer said. Williams noted that phenytoin and conjugated estrogens would make good candidates for simulations. "Hopefully, maybe in another X months...we might come back to the committee with some of the same kinds of work that Tom talked about today, but with different statistical tests applied."