“We’ve been served” are words that no one wants to hear, especially when it concerns a new product. There are steps, however, that can be taken prior to launch that may help alleviate some concern and mitigate some exposure. The United States Department of Health and Human Services, Food and Drug Administration, and Center for Drug Evaluation and Research (collectively referred to as “FDA”) published in March 2005 a Guidance for Industry: Premarketing Risk Assessment. In it, FDA sets out several nonbinding recommendations for assessing risks associated with new medicines. Although adherence is not required, it generally is quite helpful to be able to show a jury that you followed the practices recommended by FDA prior to submitting your New Drug Application (“NDA”).
Risk management involves essentially two processes: risk assessment and risk minimization. Management of risk is an evolving function that continues throughout the lifecycle of the product but, naturally, should be initiated during development. It begins with evaluating the new medicine’s risk/benefit ratio. Once risks and benefits are separately identified, tools may be designed to lower the risks and preserve, or even elevate, the benefits. After implementing those tools, the sponsor must determine whether they are effective. In other words, the first step should be repeated: re-evaluate the risk/benefit ratio to determine any improvement. Finally, based on the follow-up evaluation, the sponsor should adjust the tools as appropriate and, again, evaluate the ratio). While this process should be conducted prior to marketing the medicine in order to secure an acceptable risk/benefit ratio, it should continue after approval. New safety concerns always appear after approval — when larger and more diverse populations are exposed for longer periods of time. As such, the ratio may change, which in turn, may require the implementation of new risk management tools. Risk assessment and minimization is a never-ending process.
Sponsors are required by FDA’s regulations to assess risk and attempt to minimize it during development and after approval. FDA pointed out in the Guidance, however, that these regulations “establish requirements for routine risk assessment and risk minimization.”1 Perhaps due in part to recent withdrawals, FDA noted that the Guidance’s recommendations focus on the non-routine — the unusual type of risk or the unusual level of risk.
II. Risk Assessment
So, what is meant by “risk assessment,” and how is it accomplished? In a nutshell, “assessment” means “identification.” The sponsor must identify the nature of the risk associated with the medicine. What is the frequency of the risk? How often does the risk occur? When does it occur? In which populations does it occur? Additionally, part of the nature of the risk is its severity. Is it life-threatening, or will its effects be felt for life? Again, although this process must continue throughout the product’s lifecycle, an adequate premarketing assessment is crucial for at least two reasons: First, FDA requires it prior to approval. Second, a jury requires it prior to a defense verdict.
Deciding whether a risk assessment is “adequate” — specifically during Phase 3 studies — depends on the amount of information gathered, how the information is presented and analyzed, and the nature of the information. Adequate risk assessment is a matter of quantity and quality. Quantity is the number of patients studied. Quality is the design of the clinical trials to include appropriate (broad, diverse) patient populations, consideration of the appropriate reactions in the populations, and reasonable analysis of the results.
III. Gathering the Information
Before an assessment begins, there must be something to assess. Information must be generated. But, as FDA notes, “[p]roviding detailed guidance on what constitutes an adequate safety database for all products is impossible.”2 In other words, the specific drug will dictate the necessary contents of the safety database. Prior to approval, FDA considers known risks and unanswered questions versus demonstrated benefits. Thus, in addition to gathering information about efficacy, it behooves sponsors to design clinical trials so as to maximize the amount of safety information. For instance, if preclinical data suggests a potential problem, a trial should be designed to target that problem. Or, if related drugs already on the market have generated suspicious post-market safety information, a trial to address those suspicions should be designed. FDA addresses four main areas to guide sponsors with respect to developing an adequate safety database.
a. population size matters
New safety concerns always arise after approval because the medicine generally is used by more people, for longer periods of time, with different medical histories, and with different concomitant drug use. Even the largest clinical trial simply cannot mimic real world use. However, the larger the preapproval database and the more comprehensive the information in that database, the better. Certain factors might dictate when additional efforts should be made:
- The therapy is novel
- The effects of the proposed medicine are already safely available
- The proposed population is especially vulnerable
- The proposed duration of use is long-term
Generally, smaller safety databases may be appropriate when the disease being treated is life-threatening and there is no adequate alternative. In such situations, it is more acceptable to have less certainty concerning safety because the disease itself is so serious. The converse is true as well: Use larger databases for not-so-serious illnesses. Additionally, safety signals gathered from preclinical data also may warrant a larger trial. If concerns are raised and no in-house consensus formed, FDA suggests discussing the matter with the appropriate review division.
Although FDA does not offer much guidance with respect to the appropriate size of the safety database for new medicines developed for acute use, the agency specifically recommends 1,500 patients for products intended for use six months or longer (cumulative or continuous treatment). Three to six hundred should be exposed for six months, and at least one hundred should be exposed for a year. In addition to exposure time, sponsors should design trials to include different dosing regimens, including doses above the amount sought for marketing.
As always, certain signals may exist that should prompt sponsors to propose larger trials. These may include: indications that the medicine is associated with adverse events that develop later or that increase in frequency or severity; serious adverse events may have been observed in earlier trials; an adequate alternative may exist; the overall benefit achieved from the medicine is small; or the condition sought to be treated has a high rate of mortality or morbidity. A larger database may be necessary to distinguish between the baseline rate and that seen with the medicine. Larger databases likewise may be appropriate when the intended population is healthy (i.e., vaccines).
b. design matters
Once the sponsor determines the size of the safety database, several other considerations come into play. Oftentimes, the total database is comprised of multiple clinical trials. In such situations, it is imperative to coordinate terminology (so that investigators from each trial describe the same events similarly and the statistical rates of adverse events are not masked by differing lingo) and methods of assessment (so that investigators from each trial actually record similar events).
Sponsors also must ensure that they test the medicine in a sufficiently diverse, yet adequately representative, population. The safety database, to the extent feasible and ethical, should mimic real-world use. A wide population spectrum allows for safety information to be developed from individuals like the elderly, those with concomitant disease states, or persons ingesting concomitant drugs. In fact, clinical trials could be designed to specifically target interactions that may occur with expected concomitant drug use or expected concomitant disease states or even common concomitant dietary supplements.
There may also be instances when sponsors should consider designing trials to generate comparative safety information. In other words, it may be advantageous to compare the new medicine to an active control. For instance, if the class of drugs is such that a high rate of adverse events is expected, then a direct comparison may demonstrate that the new drug does not have a higher rate than the class. There may be certain scenarios where it is unethical to use a placebo instead of a comparator. And, of course, where the sponsor intends to claim better safety or efficacy, the sponsor should have the data from a direct comparison to back up that claim.
Finally, Phase 3 trials should incorporate differing dosing regimens. This is especially true when Phase 2 trials have not established a dose level. Data gathered from differing dose responses may be quite helpful in establishing both safety and efficacy or in demonstrating that exceeding the recommended dose raises the risk without raising the benefit. FDA recommends that sponsors meet with the review division at the completion of Phase 2 to discuss studying dose levels.
IV. Special Considerations
Specific concerns can help tailor the type of assessment strategies sponsors should employ, the type of data that should be sought. For instance, if the product has an unusually long half-life, trials with differing doses may generate helpful information. If an expected adverse event is not likely to be reported by the patients, the investigators may need instructions to pay special attention to certain facts or to make certain inquiries of the patients. Sponsors may wish to retain patient samples (blood, tissue) from Phase 3 for later examination if signals warrant.
Data from trials also can be used to assess medication errors — errors caused by similar names or similar packaging. Errors may occur from the presentation of the product (i.e., if the medicine should be diluted prior to use but is presented in a form capable of direct ingestion prior to dilution). Analysis of medication errors and their causes should be assessed and addressed in the premarket stage. Sponsors are encouraged to use experts, direct observation during trials, interviews of consumers and pharmacists, and focus groups to determine how best to minimize these risks.
Although FDA makes few specific across-the- board recommendations in this Guidance, the agency specifically sets out a list of certain adverse events that should be assessed in NDAs for small molecule drugs: QTc prolongation, liver toxicity, nephrotoxicity, bone marrow toxicity, drug interactions, polymorphic metabolism.3 Most of these adverse events arose in a constellation of withdrawn drugs. Sponsors and FDA alike suffer when withdrawal becomes a reality, so if analysis of specific adverse events can lessen the likelihood of a withdrawal, it should be performed.
V. Presenting the Information
Most Phase 3 studies are directed towards efficacy. FDA’s point with this Guidance, however, is that safety should not be relegated to a back seat. In one sense, sponsors should place a big net under their Phase 3 trials to catch all of the safety data available. One way to ensure accurate identification of safety signals is to ensure that investigators describe and code adverse events consistently. Throughout Phase 3, sponsors should use one dictionary and one coding convention. Additionally, sponsors should perform audits prior to analysis of the safety database to determine the extent of any variability with respect to coding. Acknowledging that product development may be years in duration, subsequent versions of dictionaries and coding conventions should be avoided as much as possible. However, the same version should be used for analysis and for proposed labeling.
FDA recommends that sponsors prospectively develop definitions and group expected adverse event terms. All such definitions and groupings, of course, must be adequately explained in the NDA so that the reviewers clearly understand the information. Sponsors also should avoid characterizing syndromes and withdrawals from trials with single terms. Further explanation is required. Was the withdrawal due to a safety concern or simply because the patient moved from the area? Sponsors should take adequate follow-up measures to ascertain specific information.
Temporal associations must be critically considered as well and accurately reported. This includes the time between exposure and the adverse event but also involves the total duration of the adverse event itself. Analyzing changes in both over time (i.e., long-term intermittent use leads to shorter duration of AE) is crucial to a full understanding of the total safety profile. Study of concomitant drug use also should be considered temporally. Does a concomitant drug decrease the length of time between exposure and AE? Or does a concomitant therapy increase the actual length of the AE?
The use of pooled data can be problematic as well when sponsors consider how to report the information gleaned from pooled trials. For instance, if a single trial detected a serious adverse event but the total pooled analysis lessened the risk below statistical significance, is it proper to ignore the single trial? It depends. Sometimes, pooled analysis protects against too much weight being given to chance happenings. At the same time, if the single trial is superior in design or if it considered a distinct population, it may be worthwhile for the sponsor to separately report the findings. Factors to consider when deciding whether to pool data include any differences in duration or dose and distinct differences in population groups. FDA specifically recommends “[w]hen there is clinical heterogeneity among trials with regard to the safety outcome of interest […], sponsors should present risk information that details the range of results observed in the individual studies, rather than producing a summary value from a pooled analysis.”4
FDA presents several recommendations to
sponsors concerning steps that should be taken to make an adequate premarketing
risk assessment and how to present that assessment in the NDA. Abiding by FDA’s
recommendations or creating a thorough audit trail otherwise generally will be
helpful in not only obtaining an approval letter but also in obtaining a
 Guidance, at 4. (Emphasis in original.)
 Guidance, at 5.
 Guidance, at 16.
 Guidance, at 22.