June 15, 2001 COLUMBUS, Ohio -- Drug companies may evaluate new generic drugs more thoroughly than ever before, with software developed at Ohio State University.
Drug companies must prove that the generic form of a drug functions like the original before they can receive approval from the Food and Drug Administration (FDA), explained Jean Powers, professor emeritus of statistics and veterinary clinical sciences at Ohio State.
Powers and her colleagues formulated a new, improved statistical method for comparing drugs, and Edward Herderick, manager of the Biomedical Engineering Computer Center at Ohio State, wrote the software.
The Ohio State software compares sets of curves that chart drug characteristics, and takes into account all the data relating to a particular drug function.
"The FDA already has an excellent procedure for assuring that the generic drug matches the original, but we believe our procedure represents a potential improvement," she said. She added that the same procedures she, Herderick, and their colleagues developed could be adapted to compare any two sets of curves.
Powers and Herderick collaborated with Robert Bartoszynski, a former professor of statistics at Ohio State who died in 1998, and Joseph A. Pultz, Bartoszynski's former graduate student, who is now a statistician at DuPont Pharmaceuticals. A description of the software and the statistical methodology appeared in a recent issue of the journal Pharmacological Research.
In the article, the researchers show how the software compares dissolution curves -- graphs that map how quickly a drug is released into the body. This is one of several criteria that drug companies must address when seeking approval for a generic drug. Dissolution is an important factor for the consumer of a drug, since the drug must be released into the body before it can begin to take effect, Powers said.
"You wouldn't want to buy two different brands of aspirin and find out that one can begin to relieve your headache in half an hour and the other can't begin to relieve it for two hours," said Powers.
The procedure to compare the dissolution of two drugs is computationally intensive, Powers and Herderick explained. The current approach is for the drug company statisticians to chart the dissolution of each drug graphically, take averages of multiple graphs, and compare points along the averaged graphs to see if they match up.
With the Ohio State software, statisticians could compare all points of the graphs at once -- a much more comprehensive analysis.
For instance, in a typical dissolution analysis, a statistician would examine six graphs from the original drug, and six from the generic drug, for 12 graphs total. Then he or she would combine the 12 graphs into two average graphs, and compare selected points along both.
The Ohio State software can do it in only one-tenth of a second of computer time, and still examine all the points along the curves, instead of just comparing selected points on the two average curves.
If a statistician wanted to examine a larger data set -- say, 24 curves instead of 12 -- the new software could do that computation in only 10 minutes, Herderick said. He thinks that even larger computations will probably become practical in the future, as faster computers are developed.
Drug makers who are interested in testing the software can contact Herderick. The results in the Pharmacological Research paper are based on simulated dissolution curves, so he and Powers are eager to test the software in a real-world manufacturing situation. Several companies have already expressed interest.
The FDA partially funded this work. The Ohio State researchers are currently seeking more funding to extend the software to include the analysis of other drug functions besides dissolution, as well as additional statistical designs that drug companies commonly use. For instance, Powers said, the software could compare time-concentration graphs, which represent the length of time a quantity of a drug stays in the body.
The software holds applications outside of drug development, she added.
"Suppose you were designing a new fertilizer," she said. "If you changed the formula, you could use our software to see if a particular function changed."
Herderick wrote the software in the FORTRAN computer language for a Windows PC, but said it would work just as well on other operating platforms. He also said a drug company statistician wouldn't need any special training to use the software.
Other social bookmarking and sharing tools:
Note: Materials may be edited for content and length. For further information, please contact the source cited above.
Note: If no author is given, the source is cited instead.