Improving Drug Development: Exploratory Clinical Studies
The difficulties in developing abacavir, and the widespread concern about the limitations of the data currently available (see "Ziagen (Abacavir) Approved: Caution Essential," in this issue) show some of the limits of clinical trials, and also suggest ways to improve future research, to better provide the information doctors need when they use a new drug.Even in retrospect it is not clear how abacavir should have been developed. Is there any ethical or practical way today to obtain long-term superiority data (which means that some patients are on long-term inferior treatment)? How can a research program provide the information doctors need, when the needs change faster than trials can be designed, conducted, analyzed, and published?
One change clearly needed in AIDS drug development is more research on mechanisms, on understanding what is happening and why, instead of blind pursuit of an arbitrary level of statistical confidence that, on the average, the drug is better than some standard alternative. Why are some patients responding and others not? What is the role of viral resistance, vs. individual differences in absorption or metabolism, vs. other mechanisms that might explain how a drug works well for some but not others? What can be done to predict who will respond--and to treat the others effectively?
Some regulatory reform may be needed, as the current system is designed mainly for large trials with rigid protocols. But exploratory work needs small trials early in a drug's development--sometimes with only a few patients, or only one. These trials must be flexible to allow changes as more is learned about what does or does not work for an individual. Such studies do occur today, but they can be handicapped by the need to keep a low profile and not be considered "research," and to avoid using experimental drugs. Meanwhile, large trials are designed around statistical concepts to produce a go/no-go for marketing, when the designers do not know what is happening medically. Much of the real advance in medical knowledge happens almost incidentally, for example in further analysis of the data from government trials after they have been run.
Several years ago the FDA methodically analyzed its drug-approval process, identified unnecessary obstacles, and redesigned a more efficient system. But this highly successful effort to create a more rational process was focused on the later stages of drug development. A new analysis should ask whether more flexibility in early research could significantly improve the design of the large clinical trials, so that they will better address the most important issues and produce data more useful for clinical practice.
