Keeping Approvals on Track

Recent safety issues associated with several blockbuster drugs, such as the Cox-2 inhibitors and antidepressants, have raised tremendous public and governmental concern regarding the safety of drugs and devices on the market. This publicity is naturally creating a more cautious and reactionary environment around the approval of new drugs by the Food and Drug Administration (FDA). This reaction, unfortunately, comes at the very time when it had been hoped that the approval process would be more streamlined and that promising, innovative therapies would be approved more quickly.

Photography: Comstock Illustration: Paul A. Belci

While the problem of slower approval cycles seems likely to worsen rather than improve, a solution is already at hand. New technologies and methodologies being applied in Phase IV safety studies and registries create an opportunity to change the paradigm to a proactive and systematic postapproval approach to detecting, estimating, and controlling risk as early as possible after approval. This approach is necessary because it is often not reasonable or possible to have all relevant safety information available prior to approval.

A proactive approach to postapproval risk management and safety monitoring has two important components. First, the public should be educated on the limitations of pre-approval testing and the current postmarketing surveillance process of FDA's MedWatch, as well as programs from other initiatives and independent groups. Second, pharmaceutical companies should create and advocate safety registries as well as targeted safety studies and controlled distribution systems for situations where risk is not fully known at the time of approval (e.g., for potential blockbusters) or where risk is known but primarily confined to populations that can be readily identified.

By proactively addressing safety issues through programs that improve on the industry's ability to monitor, detect, and act on adverse events in a broad patient population, pharmaceutical companies can make a stronger case for more rapid approvals and allay concerns for future blockbusters.

Figure 1. Safety registry with controlled distribution program.

Limitations of the current approach

Two related limitations of pre-approval trials, which are difficult to overcome prior to approval, demonstrate why the anticipated reaction of slowing the approval process may be counterproductive.

First, pre-approval studies can only include a tiny fraction of the aggregate number of patients who will eventually be exposed to a drug or device. Second, pre-approval studies are often not fully representative of the population who will eventually receive the product for both scientific and ethical reasons. Taken together, these two limitations virtually guarantee that there will be subpopulations of patients in whom safety has not actually been assessed in the pre-approval phase.

With blockbuster drugs and devices, millions of patients are treated with the product and it becomes more likely than not that some patients, whose genetic constitution or co-morbid health status places them at greater risk than those included in the pivotal trials, will be "exposed." It is a matter of numbers.

While the relative risk in these population subsets may be very low, events will still occur. The general public needs to be educated that it is not possible to predict all potential safety issues prior to approval. Therefore, in order to enable promising therapies to be approved with reasonable amounts of pre-approval data, a new emphasis is required on effective, early detection systems.

Clozapine and Controlled Distribution

Currently, postapproval safety signal detection is primarily managed through voluntary, spontaneous reporting methodologies such as the the FDA's MedWatch program. This program is under intense scrutiny for failing to detect the recent high-profile safety cases. MedWatch, programs managed by its sister agencies like the European Medicines Agency (EMEA), and other United States agencies such as the Centers for Disease Control and the Consumer Products Safety Commission¹ can be effective in identifying unusual adverse events. However, they are less likely to be effective in picking up a safety signal when the signal itself is a common event like heart attack or stroke. This is because the MedWatch system does not have a true "denominator."

In other words, in order to calculate an incidence rate for events, one needs both the true number of events from a given population (numerator = number of patients with event) and the size of the population at risk (denominator = number of patients with event plus number of exposed without event). These calculations assume that the population at risk is being fully reported on (if an event occurs, it will be reported). With spontaneous reporting, the denominator cannot be known (or even estimated) with certainty since not all events that occur are detected or reported. The March 2005 FDA guidance documents on pharmacovigilance have clarified but not changed this spontaneous reporting paradigm.²

In contrast, Phase IV safety registries and studies have a greater likelihood of selecting subjects representative of those who actually receive the drug or device, and in greater numbers than can typically be recruited in an earlier phase trial. They also provide a clear denominator. As a result, they generate a more statistically representative sample for monitoring for the adverse effects likely to be seen in the general population.

Detecting and managing risk

When an adverse safety event is known, specific Phase IV studies can be developed to look for an increased event rate in larger and more diverse populations in a hypothesis-driven way. Specifically, studies are developed to address the safety question in a representative population of patients enrolled from real-world clinical sites. These studies are undertaken in a rigorous fashion with data safety monitoring boards and clear stopping rules. They are launched at or sometimes prior to the time of approval (collecting background safety data prior to the approval) and extend until appropriate sample sizes in several subpopulations are achieved.

When not all adverse events are known, as is the case with all drugs destined for blockbuster status, a safety registry is currently the best means for early detection of unanticipated adverse events. In a well-designed safety registry, a representative subset of patients receiving the drug or device are enrolled, and followed prospectively. Serious adverse events are universally collected and reviewed based on meeting specified definitional criteria, not specific anticipated event types.

Effective early detection should require that the data be collected in a central database that is available in real-time for evaluation by a designated data safety monitoring board. Otherwise, there is a risk of delay in detection that would undermine the intent of the program. In certain systems, this capability is augmented by utilizing real-time algorithms to assess data for each patient for the possibility of an adverse event that may not have been recognized by the practitioner. If one is found, it alerts the sponsor and data safety monitoring board to the possibility. No system can be flawless, but attempting to detect elevated risk systematically rather than in an ad-hoc fashion is certainly a more effective way to monitor safety.

Unlike the FDA's system, which relies only on spontaneous reporting, a safety registry increases safety signal detection and processing as a data safety monitoring board reviews results on both a scheduled and ad-hoc basis (if increased signal is detected). When data is collected and processed in real-time through Web-based or voice-based systems, the opportunity to detect an increased safety signal and to act upon it rapidly is enhanced. Such programs can serve to alleviate practical and potentially political concerns of regulatory agencies for early detection of potentially unsuspected adverse events. Physicians agree that if pre-approval studies cannot predict the safety risk, then early detection of increased safety signals in combination with a predetermined action is the next best thing.³ A well-designed, real-time, safety registry is currently the best means to accomplish that.

Controlling risk

Some drugs with known adverse events can still have significant (and safe) effectiveness in certain populations. In these cases, a safety registry can be expanded to include a controlled distribution arm. Such programs systematically limit distribution of drugs at the pharmacy level to patients who meet specified demographic and lab test criteria. Using an information technology system, including Web and interactive voice response, these programs function in real-time to provide authorization or denial at the point of prescription or distribution. Examples of such programs include those for thalidomide, clozapine, and others.

These programs are generally mandated by the FDA rather than proactively suggested by pharmaceutical companies. However, that need not be the case if a new drug has a safety profile that would suggest this as an initial course of action. The pharmaceutical company can in fact suggest such a program as part of a risk management action plan. While not ideal from a utilization perspective, a contingent approval is better than none at all.

Therefore, for certain drugs with an overall concerning risk profile, but a subpopulation in whom the drug is both effective and safe, a proactive risk management program using a safety registry with a controlled distribution component should be considered. If the drug or device demonstrates real-world safety after a defined period of time, that contingency can expire and the drug could be made generally available.

Will a "safety registry" scare off prescribers?

Some sponsors have concerns about the implications that the proactive use of a safety registry might have on the general view of the safety of a product after launch. In reality, the recent withdrawals of several products have raised general concern among providers about the safety of new products and the limitations of the pre-approval process. They are welcoming all efforts to improve signal detection.

With an understanding of the inherent potential risks of all drugs and training to constantly evaluate new data regarding prescribed therapies, the physician community will have more confidence in a process that includes systematic postapproval evaluation, rather than blind trust that every possible safety issue was detected in preclinical studies.⁴ In fact, the first major sponsor to announce safety registries on all approved blockbuster products might take the position of the "safety" company, a position highly valued in other industries, but not yet claimed by a major pharmaceutical company.

Furthermore, Phase IV studies and registries offer additional benefits to sponsors. First, they provide useful safety and scientific data to prescribers through publications and, in some cases, directly through the Internet. Second, they provide an abundance of real-world clinical data on disease outcomes, product usage, product value, and market dynamics. Properly managed, data from these studies can generate a plethora of publications which may in fact speed the time to peak sales.

What if something is found?

The hesitancy in routinely developing product or disease registries at the launch of every major product potentially comes from the concern of what to do if something is found. Sponsors and regulatory agencies are aligned on this point. If an unsuspected adverse event is detected as early as possible after product launch, the sponsor can not only mitigate some litigation risk but may potentially identify and contain the subsets of patients who are at risk from receiving the product and avoid a full product withdrawal.

With increasing genetic markers and predictive modeling statistical capabilities, it will be more and more possible to isolate populations at risk. With controlled distribution systems for pharmaceuticals, such as the risk management programs in place for drugs like clozapine, registries can be extended to control delivery of the medication only to patients who are approved to safely receive the drug. Modern information systems allow this at a relatively low cost at the point of care (see sidebar).

Potential barriers to implementing a new approach

Transforming postmarketing surveillance to a systematic "monitor, detect, and act" approach is not without significant challenges. Clear definitional criteria for adverse event recognition need to be applied to help real-world clinicians in safety registries identify truly unexpected adverse events. Further, in some disease states, the background rates of certain adverse events (e.g., sudden death in epilepsy) are not well known. Comparator cohorts of patients without drug exposure would need to be constructed in parallel, thereby increasing expense.

Other data collection issues also need to be considered scientifically and practically. For example, universal registry enrollment of all patients on a newly approved drug would not only be expensive, but less likely to yield accurate detection data than a carefully constructed representative sample of clinicians and patients who agree to track more detailed adverse event information. Costs are another issue. While current electronic data capture techniques make it possible to collect structured data cost-efficiently, participating clinicians would likely require payment or other incentives to take on the data collection burden.

There are of course ethical and privacy issues with real-world data collection, but these are not different than those being effectively addressed in existing safety registries, and the same principles should apply. Finally, with the global nature of many drug launches, the most practical strategy for integrating the surveillance programs across multiple markets and countries is an area for further discussion.

Conclusion

By educating the public on the limitations of current safety detection, and proactively implementing systematic detection and control through postapproval Phase IV safety registries and controlled distribution programs where applicable, pharmaceutical companies can go a long way toward allaying government and public concerns of the inadequacies of our current postapproval safety net. Using currently available technology, companies can proactively recommend safety registries as part of risk management programs and potentially shorten the time to product approval. These programs also allay the concerns of prescribers, which ultimately shortens the time to product adoption. Proactive and strategic use of safety registries and controlled distribution programs may be the best means to keep the approval process on track in a time of increasing caution.

Richard Gliklich, MD, is founder and president of Outcome, 201 Broadway, Cambridge, MA 02139 (www.outcome.com), and an associate professor at Harvard Medical School.

References

1. http://

www.cdc.gov/mmwr/preview/mmwrhtml/mm5415a2.htm

.

2. Guidance for Industry Good Pharmacovigilance Practices and Pharmacoepidemiologic Assessment, http://www.fda.gov/cder/guidance/6359OCC.htm#_Toc48124202.

3. R. Harbaugh, Professor and Chair of the Department of Neurological Surgery, Pennsylvania State University.

4. K. LaBresh, a cardiologist and vice president at the Massachusetts quality improvement organization, MassPRO.