Premarketing Risk Assessments

Dear Sir/Madam:

The following comments supplement those presented by the Biotechnology Industry Organization (BIO) at FDA's public meeting to discuss the Risk Management Concept Papers on April 9-11, 2003. BIO represents more than 1,000 biotechnology companies, academic institutions, state biotechnology centers and related organizations in all 50 U.S. states and 33 other nations. BIO members are involved in the research and development of health-care, agricultural, industrial and environmental biotechnology products. The Biotechnology Industry Organization (BIO) appreciates the opportunity to comment on the FDA concept paper #1: Premarketing Risk Assessments.

General Comments

BIO member companies develop innovative therapeutic products, many of which are for serious disorders, with substantial unmet medical needs. These include orphan and ultra orphan diseases. Given the range of indications and study treatments that BIO member companies are involved in, we wish to emphasize the importance of an empirical approach to risk assessment that tailors efforts to the specifics of each development program. Such a paradigm for risk assessment can best fulfill the shared goal of optimizing the benefit risk relation of new therapies, without causing unnecessary delays in access to effective new medicines.

Each of the individual actions outlined in the series of concept papers has great value in specific situations. However, if used in settings in which they are not applicable they have the potential to increase the uncertainty and duration of clinical development without a corresponding benefit to patients. Thus the choice of tools for risk assessment, in all phases of development, should be carefully crafted and optimized for the specific circumstance.

Already the crescendo of scientific advances in molecular biology, immunology, pharmacogenomics, and other disciplines, is being transmuted into a myriad of new therapeutic targets and new biotherapeutic agents. Effective risk management strategies should facilitate the development of novel and effective products. Our obligation to understanding the risks associated with new therapies should not become an unintended obstacle to reaping the benefits of this new knowledge.

Particularly in the biotechnology industry, young and innovative companies make a special contribution to the translation of new science to new treatments. Clearly, the standards of efficacy and safety for a product must be the same regardless of the characteristics of the sponsor. However, risk assessment standards, if they increase the obstacles to approval without ensuring a better product can be especially problematic to such companies. Fitting the risk assessment strategy to the product under development addresses this concern.

The scope of the Concept Paper: "Premarketing Risk Assessment" includes premarketing development of both biologic products and drugs. As can be inferred from the concept paper, some of the most important safety concerns for drugs, such as QT prolongation, are not generally relevant to biologic products. On the other hand immunogenicity may need to be addressed for all biologics. Given such differences, one ought to be cautious about developing generic risk assessment strategies to be applied to all products. BIO recommends that that future guidance further define how differential risk assessment approaches may be developed for different types of products.

Throughout the documents discussed at the recent FDA workshop, the importance of evidence based approaches and the use of validated methods for risk management programs is emphasized. When designing pre-marketing risk assessment strategies, this same principle should apply. Whenever a tool for improved risk assessment is employed there should be good reason to believe that it will provide relevant, interpretable and pertinent information. Focused, differentiated approaches to risk assessment should be designed to delineate the risks of interest for the specific product and indication under development.

Risk assessment strategies are a means to an end and should be continually evaluated from that perspective. They reduce uncertainty about risks by providing relevant information. Well-designed risk assessment strategies can provide reassurance about the product's safety profile. Where clinically significant safety issues are identified, risk assessment provides the basis for designing a risk management strategy.

The premarketing risk assessment paper poses the question "What is an ideal safety database?" It affirms that "the composition of an appropriate safety database for a new product would be determined on a case by case basis." The subsequent discussion, however, of the ideal for what "all programs" would include, proposes three features of an ideal safety database. Each feature has the potential to increase the number of patients included as well as the duration and complexity of clinical development. Some features may not be applicable all programs. It is important that these features not be seen to represent a de facto minimal standard for a product safety database. BIO suggests that the concept of an ideal safety database should be nested in the context of the specific program rather than be developed as an abstract, stand-alone notion.

Three suggestions regarding premarketing risk assessment deserve specific comment. Firstly, long term controlled safety studies are suggested. While the paradigm of the randomized clinical trial may be the ideal to address certain safety concerns, in practice this may not be possible for several reasons. Experience with patient recruitment and compliance has shown that such studies are difficult to conduct. It may not be possible to recruit and maintain a large enough patient population to address the issues of concern. Interpretation of such data is likely to be impaired by dropouts and missing data. In situations where there is a major unmet medical need and where efficacy has been demonstrated, ethical reservations may be raised. Should there be data driven concerns about a specific safety endpoint, an appropriate, targeted post-marketing risk management plan using pharmacoepidemiologic methods may be more feasible than an effort to generate long term controlled safety data prior to marketing.

A second proposed feature of an ideal database for all programs is diversity. A diverse safety database, it is pointed out, is likely to be more representative and more readily generalizable to the post-marketing population. While this is true, generating a diverse safety database has substantial implications for product development. Inclusion of a more diverse population in pivotal studies makes it more difficult to demonstrate efficacy: such groups have more confounding factors such as concomitant diseases and medications. They may also be less compliant, leading to more study dropouts and missing data. The consequence of each of these factors is to increase the number of patients to be studied in order to demonstrate efficacy, or its absence. A consequence is that patient exposure to an ineffective product might under some circumstances be increased.

By contrast, the resulting more "diverse" safety database may not always provide a better basis for assessing risks in the expanded population. The numbers of patients in each sub-group in the diverse population may well be too small to assess sub-group specific risks. Indeed, the risks detected would have to be high for them to be measurable in the premarketing setting. A further potential hazard to this approach is that the heterogeneous population might generate so much "noise" that real safety signals could be obscured. Thus a "diverse" safety database is not necessarily a characteristic of an ideal safety database for all products.

The third suggested characteristic of an ideal safety database is one that would allow comparison of safety profiles at different doses. Where the safety database generated by the clinical development program already includes patients treated at different doses, data analysis strategies exploiting that information can be very valuable. On the other hand, one would generally not design a clinical development program for the purpose of comparing the safety profile of different doses. Since the primary endpoint is efficacy, Phase III studies are neither designed nor powered to distinguish between "levels of risk" at different doses. Adding additional study arm(s) in Phase III increases the number of patients needed to demonstrate efficacy and correspondingly increases the duration of the studies and the required resources. The ability to recruit patients is commonly the limiting step in clinical development. In orphan indications this is an inherent constraint. Thus, while having patients at different dose levels may be useful in a safety database, one should be very cautious about suggesting that this is an ideal to be achieved for all products.

The second part of the concept paper, section (G), on data analysis and presentation, offers creative strategies to identify and delineate specific safety issues. These suggestions open the way to better understand and utilize data collected during clinical development. This approach facilitates early identification of safety concerns. By prospectively defining safety issues during clinical development, available and accruing safety data can be organized and coded to cast light on specific issues. These are efficient and prudent approaches that are tailored to address specific problems. This approach also facilitates the early development of a problem-specific risk management strategy to be implemented post-approval.

In conclusion, BIO looks forward to additional guidance from the FDA on risk assessment based on a paradigm emphasizing empiricism, evidence-based strategies and the development of approaches that are crafted to the unique situation that each product development undertaking represents. This approach will help ensure that innovative, effective and safe biologic products continue to be developed to meet serious unmet medical needs. A more detailed set of comments are attached.

Sincerely,

Gillian R. Woollett, MA, DPhil
Vice President Science and Regulatory Affairs