Skip to main content
Premium Trial:

Request an Annual Quote

FDA Advisory Panel Mulls Genomic-Driven Strategies for Predictive Safety, Sample Collection


By Turna Ray

ATLANTA — The US Food and Drug Administration last week sought input from an independent panel of experts on the agency's preliminary efforts to develop a predictive safety system that integrates pharmacogenomics, chemical structural data, and systems biology approaches to determine drug-induced adverse reactions before they happen.

At the same meeting, members of the FDA's Advisory Committee for Pharmaceutical Science and Clinical Pharmacology discussed the circumstances under which the agency should encourage DNA data collection by drug sponsors, and ultimately determined that the FDA should "strongly urge" DNA sample collection in all drug development phases.

"With all the focus on drug safety we've developed a concept for a new safety program that is rooted in clinical pharmacology principles," Lawrence Lesko, director of the Office of Clinical Pharmacology and Biopharmaceutics at the FDA's Center for Drug Evaluation and Research, said to the advisory committee. "We think of this as a complementary approach to drug safety, complementing all the other programs the FDA has such as the Sentinel program, the Safety First program, and post-marketing surveillance."

The Adverse Events Reporting System, or AERS, is the agency's computerized system to monitor the ill effects of drugs. But in the post-market setting, ADRs may not be detected for several years after the drug has been approved. In recent years, however, as genomics firms have begun to amass large databases of genetic risk association data, the FDA has expressed a desire to use this information to improve ability to detect adverse events [see PGx Reporter 11-19-2008].

Another step in the direction of developing a proactive product safety system arrived in 2008 with the launch of the Sentinel Initiative, under which the FDA plans to cull adverse event data associated with drugs and devices from electronic healthcare records held by cooperating institutions. In recent months, the FDA has been soliciting input from stakeholders on the technical, policy, and scientific challenges associated with developing the Sentinel Initiative.

"But one thing those complementary programs don't have is a systematic way to get to the mechanism of adverse events," Lesko said. "We're trying to create that, and it is relatively new. So it's a work in progress."

'Mechanistic Approach' To Drug Safety

According to Darrell Abernethy, associate director for drug safety at the FDA's Office of Clinical Pharmacology, the FDA has been working internally and with external partners on this new safety initiative since last fall. The idea is to develop approaches to identify adverse reactions associated with certain drugs before the agency approves them for marketing. According to Abernethy, gaining "a mechanistic understanding of drug safety" will allow the agency to institute risk-mitigation strategies with sponsors to avoid drug-related harm in the post-market setting.

"Are there ways in which we can fully utilize the information to look for information in each step of the way to look forward in predictive safety assessment?" Abernethy posited during a presentation to the committee. "The unifying theme is to identify variability in response. This is not a new theme for a clinical pharmacologist," according to Abernethy, but the challenge is to "figure out the patient subgroups, however the subgroups are segmented out. By genetics, by environmental exposures, or what have you."

The agency is progressing in the development of this proactive adverse events detection system through collaborations with various consortia, gathering data from different databases and published literature, applying pharmacogenomics and system biology approaches, and integrating efforts within the agency's divisions. Some of the strategies the agency will apply to this effort include data-mining algorithms, systems biology approaches, pharmacogenomics, decision trees, and model simulations.

In order for this system to come to fruition several pieces need to be developed, such as a systematic database for molecular toxic targets linked to organ-level toxicity, as well as a database linking chemical systems biology to biological pathways. Also, as a first step to test out its hypothesis for such a system, the FDA is conducting a pilot study to gauge the risk of diabetes mellitus in people on chronic statin therapy, by integrating information from different databases.

According to Abernathy, in order for the FDA to conduct drug safety analysis in a prospective (rather than the current retrospective) fashion, drug sponsors will need a pre-specified data analysis plan and will have to collect DNA samples consistently from study participants. Then in the post-market setting, the challenge will be to gather phenotypic information associated with genomic data.

Drug developers would be incentivized to meet these data requirements, since the early capture of adverse events could significantly improve the success rate of investigational drugs from discovery to the marketplace. Specifically, this type of a system could make late-stage drug development trials move more efficiently, and lower post-marketing safety liability for drug developers, Abernathy said.

Urging Routine DNA Sample Collection

In a separate but related presentation to the advisory committee, Issam Zineh, associate director of genomics in the Office of Clinical Pharmacology at the FDA's CDER, estimated that the attrition rate for investigational drugs going into human studies for the first time is around 90 percent. The three major reasons for the failure of drug candidates in clinical trials are lack of efficacy, lack of differentiation between the drug of interest and a comparator, and safety, he said.

[ pagebreak ]

However, industry's growing participation in the FDA's Voluntary Exploratory Data Submissions program indicates a willingness to implement genomic strategies to differentiate drugs and make the development process more efficient. According to Zineh, since 2008 submissions to the VXDS program have grown 250 percent. The FDA launched the program several years ago to encourage sponsors to share genomics data with the agency without any regulatory repercussions [see PGx Reporter 11-29-2006].

VXDS discussions between sponsors and the FDA are increasingly leading to NDA submissions for drugs with companion genetic tests. For instance, Novartis recently acknowledged that it used the VXDS program to get feedback from the agency on the safety biomarker it is using to develop a companion test for the Cox-2 inhibiting painkiller Prexige. The drug was previously rejected by the FDA due to hepatic safety concerns [see PGx Reporter 02-10-2010].

According to FDA officials, however, many of the agency's efforts to improve drug safety and development depend on the willingness of industry to readily collect DNA samples early in clinical trials.

In 2008, the pharmaceutical science advisory committee reached a consensus that DNA samples should be collected from all patients in every drug development clinical trial, between Phase I and Phase III.

There was some pushback from industry regarding this recommendation, however, due to the difficulties associated with routine sample collection, as well as divergent sample collection requirements in different geographic locales.

During discussion of this issue among advisory committee members last week, many expressed reservations about mandating DNA sample collection.

"As an IRB member, I don't think you can mandate collection of that kind," said Jurgen Venitz, acting chair of the pharmaceutical science advisory committee and the vice chairman of the department of pharmaceutics at Virginia Commonwealth University. Venitz further held that "it is not ethically defensible" to mandate sample collection, particularly if there is a benefit from participating in the study.

Similarly, David Flockhart, committee member and chief of the division of clinical pharmacology at Indiana University, felt that while the FDA can "strongly encourage" DNA sample collection, "it is very difficult to argue that it be federalized." Flockhart asserted that sample collection should be scientifically driven when a clear candidate pathway exists. Other members also felt that an a priori hypothesis would be necessary to justify mandatory DNA sample collection in drug trials.

However, by urging sponsors to routinely collect DNA samples, the FDA is trying to lay the groundwork to allow for the early prediction of subpopulations at risk for adverse events and to be able to gauge the potential of such events before they happen.

"If this becomes an exercise in trying to think of a genetic hypothesis before every study, it really just undersells the value of the data," said Jeffrey Barrett, committee member and associate professor of pediatrics at the Children's Hospital of Philadelphia, adding that at the time of sample collection it might not be immediately obvious how this information may be useful later in the life cycle of the drug.

Kathleen Giacomini, committee member and chair of the biopharmaceutical sciences department at the University of California's School of Pharmacy, suggested that while FDA may not be able to mandate that study participants provide DNA samples for studies, the agency can require that sponsors include sample collection as a feature of every study to improve acquisition rates.

Zineh noted that in many phases of drug development sample acquisition rates are fairly high, between 70 percent and 99 percent. However, sample "collection is incomplete," because it is optional.

Optional sample collection could lead to incomplete collection and make a company unprepared for genomic analysis of safety issues that might arise in the post-market setting.

As an example, Lesko noted that in the example of the colorectal cancer drug Vectibix, drug developer Amgen could have avoided regulatory action from European authorities due to low efficacy of the drug in the general population if the firm had completed DNA sample collection in early drug development. "In the early days of trying to develop an association between mutant KRAS and lack of [Vectibix] benefit, the studies were hampered by incomplete data collection," Lesko said.

The FDA updated the labeling of Vectibix and Erbitux to note that patients with mutant KRAS do not respond to the drug, using retrospective genomic data analysis. Due to incomplete DNA sample acquisition, however, the labeling update in the US took several years, as the FDA had to figure out the circumstances under which retrospective DNA analysis would be acceptable [see PGx Reporter 02-24-2010].

Additionally, Lesko said that just by looking at the in vitro data on Plavix, one would have "never predicted" the impact of the CYP2C19 polymorphism. "So it took the post-marketing studies to figure out what was going on with active metabolite levels and polymorphisms, because it wasn't looked at early on during the drug-development phase."

The FDA has updated the labeling for Plavix three times to inform doctors and patients of the impact of CYP2C19 polymorphisms on the metabolism of the drug [see PGx Reporter 03-15-2010].

After listening to the advisory committee's recommendations, Lesko acknowledged that while "it is unlikely that an agency anywhere is going to mandate DNA collection, it could do other things, such as strongly recommend, recommend, what have you."

Thus, the advisory committee was asked to vote whether it felt the FDA should "strongly urge" DNA sample collection in all drug development phases, and the majority of committee members supported such an initiative.