Appendix A


Key Issues for Risk/Benefit Assessment for Gain-of-Function Research

The purpose of this appendix is to compile key points from the presentations and discussions at the symposium. Each of the points is attributed to the person(s) who made it, or to the discussions from which it emerged, along with the page number(s) where it may be found. More detailed versions of these points and the rationales for them may be found in the foregoing chapters.

CHAPTER 2: ASSESSING RISKS AND BENEFITS

1. Although the major steps in risk assessment were first enunciated in a National Research Council report titled Risk Assessment in the Federal Government: Managing the Process, the basic steps in the process remain the same today:
2. • •   Hazard Assessment: The determination of whether a particular chemical (or microbiological agent) is or is not causally linked to particular health effects.
   • •   Exposure Assessment: The determination of the extent of human exposure and the probability of occurrence of the health effects in question.
   • •   Dose-Response Assessment: The determination of the relation between the magnitude of exposure and the probability of occurrence of the health effects in question.
   • •   Risk Characterization: The description of the nature and magnitude of human risk, including attendant uncertainty.

• •   Risk Management: The reduction of risks and the increase of expected benefits.

2. Risk Communication and Appropriate Involvement of Stakeholders (Haas, 2014:7).The major focus of attention with regard to GoF research has been on hazard assessment, which has largely been focused on occupational health risks, but it is important to go beyond this to consider risks to the members of the public near research sites as well as global risks from pandemic organisms. Scarce attention has been paid to either exposure assessment or dose response assessment (Haas, 2014:7-8).
3. There are a number of questions to be addressed in a risk assessment such as whether the safety records of high-containment laboratories provide an appropriate basis for quantifying the risks of lab accidents that lead to worker or public exposures and how deliberate misuse of either the pathogens themselves or the information obtained through the research on these pathogens is to be incorporated into the risk assessment (Haas, 2014:8).
4. Risk assessment can inform decisions, but it is not determinate per se. Determining what is “acceptable” risk is a policy decision (Haas, 2014:9).
5. In addition to following the framework given above, the risk/benefit analysis for GoF research requires socially acceptable, technically sound definitions of “risk” and “benefit;” a strategic focus on design or decision, with proper disciplinary breadth and treatment of uncertainty; ongoing, scientifically sound two-way communication with stakeholders; and organization for transparency and learning (Fischhoff, 2014b:9-11, 17).
6. All analyses embody values that favor some interests above others. Thus, when transparent, the underlying assumptions can be controversial, and, therefore, an analytical and deliberative process is required to create socially acceptable definitions that acknowledge subjectivity (Fischhoff, 2014a:9).
7. Risk assessments on low-probability/high-consequence events are not new, and because the roles of uncertainty and human factors are crucial in risk assessment, acknowledging and incorporating them are an important goal (Fischhoff, 2014a:11).
8. “Human factors” research is the study of the interrelationships between humans, the tools they use, and the environment in which they live and work. Eighty percent of motor vehicle accidents, 80 percent of medical errors, and 60-80 percent of aviation accidents are estimated to be attributable to human factors. Studies have shown that physical (e.g., working in personal protective equipment) and cognitive (e.g., working under conditions

9. of fatigue) stresses undermine human reliability and that not only can human error not be eliminated, but it has also actually increased as a contributor to accidents in some arenas. Analyses of human reliability and errors identify the critical areas that are incompatible with human capabilities and the areas where a system is vulnerable to human error (Huntley-Fenner, 2014:11-13).

10. Key questions regarding human factors include
    • •   Are task demands compatible with human capabilities and characteristics?
    • •   Has the system been designed to cope with the inevitability of human error?
    • •   Does the system take advantage of unique human capabilities? (Huntley-Fenner, 2014:12).
11. Other areas of limitations in risk assessment include variability among observations, quality of the studies the analysis is based on (internal validity), whether these studies are generalizable (external validity), and how good the underlying science is (“pedigree”) (Fischhoff, 2014a:13).
12. In the case of GoF research, “public” engagement may require dealing with the local public, the national public, and even the global public given that the consequences of a failure might be a global pandemic of infectious disease (Schoch-Spana, 2014:13-14).
13. There are three different kinds of public engagement: communication, consultation, and collaboration:
    • •   In the communication mode, an official or an agency conveys information to members of the public in a one-way fashion, often with the intent of educating and informing the public. Public feedback is not required and not necessarily sought.
    • •   The consultation mode is an interaction in which authorities solicit opinions through surveys, polls, and focus groups or during public comment periods. Again, this communication is one-way, but it is from the citizens to the authorities.
    • •   Collaboration is a two-way flow of information and influence between citizens and authorities; it is about dialogue fostering better understanding of very complex problems from all sides and perspectives and allows an opportunity for collective learning as part of honest and respectful interaction among the authorities and diverse constituents. Such iterative exchanges are necessary to approach policy concerns that are technically and ethically complex (Schoch-Spana, 2014:14-15).
14. Engaging the public improves product quality, enhances legitimacy of decisions, and builds a foundation of trust and mutual

14. understanding as well as practical experience with dialogue (Schoch-Spana, 2014:15-16).

15. Because the expected benefits of GoF research are potentially reduced risks, the same methodologies apply to assessing both the risks and expected benefits (Fischhoff, 2014a:9). Assessing the benefits side of the equation is more difficult and poses more interesting problems that require an investment in formalizing the benefit arguments as well as the arguments for alternative paths. Knowing which numbers are really important and whether they are even relevant to an analysis would assist with that process (Fischhoff, 2014b:17).
16. The risk/benefit assessment may also need to address the consequences of not doing GoF research (Haas, 2014:9).

CHAPTER 3: GAIN-OF-FUNCTION RESEARCH: BACKGROUND AND ALTERNATIVES

1. In virology, GoF research encompasses a broad range of experiments including any selection process involving an alteration of genotypes and their resulting phenotypes (Subbarao, 2014:16).
2. Research leading to the generation of viruses with properties that do not exist in nature could be categorized differently than research on strains that may be more pathogenic and/or transmissible than the wild-type viruses but are comparable to or less problematic than those existing in nature (Kawaoka, 2014:17). Routine virological methods involve experiments that aim to produce a gain of a desired function, such as higher yields for vaccine strains, but often also lead to loss of function, such as loss of the ability for a virus to replicate well, as a consequence (Subbarao, 2014:16).
3. GoF research on SARS-CoV and MERS-CoV is likely to be extremely different than influenza research because of fundamental biological differences and complexity that make these viruses very different (Baric, 2014:18). Unlike flu, there are currently no small animal models suitable for MERS or SARS transmissibility assays (Baric, 2014:30). Research on SARS-CoV infectivity has shown that adaptation of the virus to the mouse ACE2 receptor decreases its interaction fitness with the human receptor (Baric, 2014:18). Given that we are in the midst of a MERS-CoV pandemic, the “pause” on GoF research for MERS can have unintended and severe consequences. There are currently no small animal models to study MERS-CoV and, therefore, MERS-CoV restrictions should be lifted immediately (Baric, 2014:20).

4. The term GoF needs some refinement that will differentiate the type of research typically performed for basic virological research from experiments that clearly raise concerns (see “GoF Research as Defined by the U.S. Government” on pp. 25-27)
5. Research on CoV should be considered using a case-based approach and be subject to an iterative process that incorporates risk, milestones, and identifies problems along the way (Denison, 2014:25-26, 41, 45).
6. Research leading to the combination of increased transmissibility and virulence with the lack of efficient counter-measures would clearly define the line that would prompt the use of alternatives (Relman, 2014:25, 47-51).
7. Although alternative scientific approaches are not only less risky, but also more likely to generate results that can be readily translated into public health benefits (Lipsitch, 2014:42-43), alternatives to GoF research do not always provide the full answers to the key virology questions (Kawaoka, 2014:27).

CHAPTER 4: POTENTIAL BENEFITS OF GAIN-OF-FUNCTION RESEARCH

1. Because it is not possible to predict what breakthroughs may occur as a result of fundamental research, including GoF, it is impossible to quantify the benefits of GoF research for risk/benefit analyses. Long-term research benefits are achievable, but it is not possible to specify what these are when the research is initiated (Atlas, 2014:29).
2. GoF research in the short term can be used to help adapt viruses to growth in culture for vaccines and to develop essential animal models for the study of emerging pathogens and escape mutations with which to understand drug resistance and viral evasion of the immune system (Atlas, 2014:29).
3. GoF research may also allow the generation of information that is not obtainable through other methods, but whether all the long term benefits envisioned for GoF research will actually be realized is still unclear (Atlas, 2014:30).
4. There is significant disagreement about whether GoF methods are essential for vaccine development; therefore evaluating its contributions to benefits could clarify this. These methods appear to have limited utility for current production methods, but there were arguments that the increasing introduction of synthetic methods of production based on genomics and other molecular

5. techniques could increase the contributions of GoF research (see discussion on pp. 42-45).

6. GoF research-derived information is used to develop the Influenza Risk Assessment Tool that looks at the properties of a virus, especially the molecular determinants of virulence and transmissibility that can help identify candidate vaccine viruses in a time of limited resources (Shultz-Cherry, 2014:31).
7. Genotype to phenotype prediction is one of the holy grails of influenza biology research (Russell, 2014:34). Some argue that studies of particular amino acid changes in one strain of virus may not apply to another strain and that it is not possible to calculate the level of risk from the mutational landscape (Fraser, 2014:32). Others think GoF studies must continue so that eventually this inability to predict phenotypes can be overcome (Shultz-Cherry, 2014:31).

CHAPTER 5: POTENTIAL RISKS: BIOSAFETY AND BIOSECURITY

1. There was considerable support among attendees for David Relman’s proposal to focus risk assessments on GoF experiments that involve the deliberate creation of viruses with a high degree of pathogenicity and transmissibility and perhaps with properties that would make the infectious agent impervious to currently available countermeasures. Experiments that produce this combination of properties are of more concern than the experimental approach per se (Relman, 2014:47-51).
2. Assessments of potential biosafety risks will have to deal with serious issues related to availability and quality of data about the frequency and severity of accidents, exposures, and Laboratory Acquired Infections (LAIs). Experience suggests there is substantial underreporting of LAIs, if not primarily in the United States, then in other less developed countries (Johnson, 2014:52). In the United States, there are systematic data available from the Select Agent Program relating to some of the most dangerous pathogens for the period 2005-2012 (Weyant, 2014:54-55). Comparable international data do not exist and there is no central point for reporting all accident and exposure data either inside or outside the United States (see Evans comment on p. 58).
3. The continuing expansion of high containment (i.e., BSL-3 and -4) laboratories affects potential biosafety risks primarily, but certainly not exclusively, in developing countries (Johnson, 2014:54).
4. There are substantial and important differences among the models used and the results obtained in efforts to assess the risks

5. of accidental releases of a dangerous pathogen. Understanding the sources of the differences could inform the risk assessment (Lipsitch, 2014:57; see also discussion between Lipsitch and Fouchier on p. 58).

6. While biosafety and biosecurity are inextricably linked, they mitigate different risks. Consequently, biosafety measures, in and of themselves, cannot fully address biosecurity risks (Linden, 2014:62).
7. Because there have been relatively few bioterrorists incidents, uncertainty is endemic for most factors that would contribute to a biosecurity risk assessment (Koblentz, 2014:60). Including evaluation by both security and scientific personnel in any assessment of a bioterrorist threat risk could expose the limits of our understanding of the following factors:
   • •   organism/pathogen and its weaponization potential;
   • •   capability (including both scientific knowledge, tacit knowledge, and technological know-how) and intent of an adversary; and
   • •   potential consequences of intentional release or misuse and the ability of the targeted population/country to respond (see comment by Ed You, p. 59).
8. Periodic reassessments for bioterrorist threats could take into account:
   • •   advances in science and technical skill;
   • •   changes in actor and targeted country capacity and infrastructure;
   • •   increases in the number of laboratories undertaking research of particular concern; and
   • •   the changing nature of the threat environment (see Hale comment on pp. 59-60).

CHAPTER 6: POLICY IMPLICATIONS

1. It would assist the understanding of the policy trade-offs for GoF research if there was greater clarity about whether the discussion is about marginal benefits versus marginal increases in risks. For example, there was a tendency during the discussions to move unpredictably between the potential value of GoF research overall versus looking at its net benefit or its marginal increase in value when viewed as an addition to all the other types of research or public health efforts that are being done to address the same problem. This appears in the discussions around whether GoF research would help make new vaccines versus GoF research as

2. helpful—in addition to everything done in terms of monitoring—in making more precise predictions of which influenza strain should be worked on next year. (Charo, 2014:66)

3. The choice among policy options will be heavily influenced by the presumptions one brings to the problem. Given that many of the issues will inevitably fall into a “grey zone,” where there can be disagreements about whether their potential benefits outweigh their potential risks, the chosen default position will determine whether they go forward. If the presumption is that one must prove that an experiment is needed, then anything in the grey zone is going to be prohibited. If the presumption is that research is free to proceed until the government, or some other authoritative body, has determined that it is unacceptable for some reason, then everything in the grey zone will proceed because the burden of proof will be on the government. Whoever has the burden of proof of making the case will have the more difficult time, and that is a fundamental issue for any policy (Charo, 2014:66-67).
4. The discussions in the symposium brought out three possible approaches for how a policy for GoF research might operate. One possibility would be to create a threshold beyond which experiments are either prohibited or given special attention. Another would be a purely case-by-case approach, probably at the funding stage, reviewing each of the factors considered important to making the choice whether to allow the experiment to proceed independently and then assessing the overall project holistically. The third would be “risk-based” regulation that reflects experience with a particular type of research, agent, or setting (Charo, 2014:67).
5. Although the symposium focused on a U.S. deliberative process, there will be special challenges for policy because of the global nature of GoF research. For example, one issue for the safety and security risk assessment cited by several participants would be the extent and impact of the diffusion of research capacity on the efficacy of policy options affecting U.S.-funded international research collaborations. The global nature of GoF research poses a series of special challenges:

1. National cultures—To what extent is there agreement about the general balance between risk avoidance and support for innovation and research?
2. Governmental powers—Which powers are traditionally used to regulate or prohibit research: conditions on receipt of funds, direct regulation of personal activity, licensing of

3. institutions, etc.? Does this work primarily through rules with force of law or by advice and voluntary actions?

4. Relative resources—Given that protective measures can be resource intensive (both equipment and personnel) and countries vary in their capacity, how much should be spent to achieve a minimum level of safety? Or an optimal level of safety?
5. International governance—How to manage regulation or prohibition in cases of research collaborations that cross borders? (Charo, 2014:68).

This page intentionally left blank.