Using modern scientific tools in conventional clinical trial designs is like running a bullet train on old tracks. It might work, and in some cases there is no alternative, but it isn't efficient. To modernize this railroad, FDA is creating a roadbed for innovative trials that are optimized for genomics, proteomics and other technologies that can provide data with more precision and specificity.

The agency is developing a package of five guidance documents intended to encourage the use of innovative clinical trial designs. If it is successful, companies will be able to more readily expose ineffective compounds and increase the chances that useful products will make it to the market.

While most of the strategies are not novel, they haven't been widely applied because of a lack of awareness on the part of sponsors, or because FDA has not clearly indicated its requirements for using them as the basis for approvals. The guidance documents are an effort to remove these impediments.

First out of the box will be a draft guidance that clarifies FDA's thinking about enriching trials to enhance the probability of detecting efficacy and facilitate the use of genomics, proteomics and other tools. The guidance also will explore the use of adaptive designs to guide enrichment during the course of a trial.

The enrichment guidance "goes to the heart of why people design clinical studies," Robert Temple, director of FDA's Office of Medical Policy, told BioCentury.

The guidance also is timely, he said, because of an explosion in the number and quality of techniques for characterizing an individual's disease, such as identifying the genetic mutation responsible for a tumor, and marking individual differences in the way drugs interact with the body.

"The fact that people differ in their risk and response provides both an opportunity to target therapy better and also to utilize more efficient study designs in enriched populations," said Temple.

Temple, who is writing the guidance, defines enrichment as the "prospective use of any patient characteristic - demographic, pathophysiologic, historical, or genetic, and others - to select patients for study to obtain a study population in which detection of a drug effect is more likely." He has briefed academic and industry scientists on the topic and expects that a draft will be released this year.

There are three main objectives of enrichment designs, Temple noted. One is reducing the "noise" or heterogeneity that interferes with the interpretation of results. Second is improving the ratio of responders to non-responders. Third is increasing the proportion of participants who are likely to demonstrate the outcomes being studied in the trial (see "Enrichment Strategies").

Enriching the signal

The basic idea of enriching trials - selecting study populations so that a therapy's effectiveness can be detected more readily than in an undifferentiated set of patients - is not a new concept. Virtually all clinical trials conducted to support approval are enriched in some fashion to decrease heterogeneity. For example, adjusting a trial size to increase statistical power is one of many methods for enriching trials.

Temple noted that practical techniques for noise reduction are common and non-controversial. These can be as simple as selecting patients with the disease to be treated, or excluding patients with characteristics, such as taking certain drugs, that obscure a treatment effect.

Researchers also have devised techniques to enrich trials for individuals whose behavior indicates they are likely to adhere with a treatment regimen, Temple noted. As an example, he cited a study in which all potential participants were instructed to take medication for a couple of weeks prior to the start of the trial. They were not informed that it included a marker that could be detected in urine. Patients who didn't consistently produce the marker were excluded from the trial.

Conversely, unusually high response rates in a placebo control arm can make it much more difficult to demonstrate efficacy of an active agent, so eliminating placebo-responders during a lead-in period is another common enrichment technique.

Strategies to decrease heterogeneity do not usually raise questions about generalizing the findings, which become more significant when comparing designs focused on isolating responders or eliciting the hoped-for outcome.