The 2010 site-specific risk assessment (SSRA) of the proposed National Bio- and Agro-Defense Facility (NBAF) did not adequately address the unique issues and challenges associated with work in a maximum-biocontainment environment, and the previous National Research Council committee was therefore led to conclude that the 2010 SSRA did not adequately characterize risks associated with biosafety level 4 (BSL-4) containment activities. An overall risk assessment of the NBAF BSL-4 would need to include an evaluation of the additive risks posed by all BSL-4 work, including the risk of a release from the BSL-4 associated with use of large animals in the BSL-4 suite, the risk of a release associated with non-animal-related activities in the BSL-4 suite, and the risk of a release from the BSL-4 suite associated with natural disasters. The epidemiological and economic impacts of such a release would then need to be evaluated as part of the risk assessment. The committee recognizes the inherent limitations in the available information regarding henipaviruses (see below) that formed the basis of the BSL-4 review and regarding other agents that may be studied in the BSL-4 suite. However, the risk assessment presented focuses on only one component of the overall risk, namely, the unique risks of release from the BSL-4 suite associated with the use of large animals. During the March 2012 meeting of the present committee, the Department of Homeland Security (DHS) indicated that it assessed only BSL-4 risks associated with large animals on the basis of its interpretation of the previous committee’s evaluation, which was misunderstood. The uSSRA does not consider the overall risk and presents a limited qualitative assessment, and therefore the evaluation likely underestimates overall risk related to the BSL-4 suite.
The uSSRA responded to the congressional mandate by conducting a semi-quantitative risk assessment on the only two exclusively BSL-4 agents that are on the priority list for work at the NBAF: Hendra virus (HeV) and Nipah virus (NiV). HeV and NiV are paramyxoviruses (Henipavirus genus) that were recognized in the 1990s and produce high-mortality disease in animals and humans (Eaton et al., 2006; Field et al., 2007).
The uSSRA states that the primary objective of the BSL-4 risk assessment is “to identify and characterize the unique risks associated with working with large animals in BSL-4 conditions.” The analysis therefore focuses exclusively on risks associated with handling infected large animals in BSL-4 containment. The uSSRA modeled four potential release pathways (aerosol, solid, liquid, and transference) and developed scenarios in consultation with an international panel of experts in high-containment settings and pathogens (including representatives of the Australian Animal Health Laboratory in Geelong and the Canadian Science Centre for Human and Animal Health in Winnipeg). The committee commends DHS for consulting an international expert panel to delineate the major and unique risks of the BSL-4 environment.
Some risks are inherent to working in a BSL-4 environment, which include the use or manipulation of dangerous pathogens that are highly lethal to humans or animals and for which there are no preventive or therapeutic interventions. Added to those risks are the combination of the presence of large animals in the maximum-containment environment coupled to the difficulty of maneuvering in biocontainment suits and with separate air supplies. These difficulties raise the risk of injuries, disruptions in air supply, and compromised suit integrity from holding pens, animal bites, inoculations, and use of sharp implements during experiments and necropsies. These hazards highlight the importance of having administrative measures in place—including buddy requirements for BSL-4 systems—to ensure recognition and reporting of such breaches and occupational health programs to ensure proper management of personnel.
The committee concurs with the finding in the uSSRA that transference represents the major risk of inadvertent escape for BSL-4 pathogens relative to other release pathways. However, the committee has many concerns about the analysis and found that the uSSRA does not adequately address the overall risks related to work with BSL-4 pathogens; it elaborates on those below.
Lack of Consideration for Full Array of BSL-4 Activities
The uSSRA focuses exclusively on risks associated with handling of large animals in BSL-4 containment and neglects risks posed by other activities in BSL-4 containment. Work in a BSL-4 laboratory that does not house large animals is not risk-free as implied by the uSSRA. Due to the nature of BSL-4 pathogens, work that would normally be conducted in BSL-3 special procedure or BSL-3 Enhanced areas of the facility are required to be conducted in BSL-4 if it involves live virus. The 20-year record for safe operation of major BSL-4 laboratories in the United States is excellent (Johnson, 2003) and the use of biosafety cabinets and other biosafety measures in the BSL-4 laboratory setting can reduce the risk of release from non-animal-related activities, but it does not reduce the risk to zero. Consequently, by omitting the risk of activities in non-large-animal BSL-4 space, the uSSRA does not address “overall risks” as this committee had expected because it ignores these risks.
Magnitude of Risks Associated with Unknown or Uncharacterized Pathogens
The uSSRA notes that NiV and HeV are not the only pathogens that would be examined in the BSL-4 laboratory. As part of its mission to serve as a world-class laboratory, the NBAF is expected to conduct essential and cutting-edge research both on known BSL-4 agents and on unknown and uncharacterized pathogens. There is a possibility that experiments would involve large animals for hemorrhagic fever agents, variant influenza viruses, or pox viruses, and work on these pathogens will pose risks. As the microbiological and epidemiological features of these agents differ from those of henipaviruses, it is likely the risks associated with working on these agents would also be different from those of henipaviruses. The committee concurs that it is difficult, if not impossible, to model the risks associated with unknown agents. However, the uSSRA provides only a minimal risk estimate, and the present committee echoes the previous committee’s concern that the risk assessment did not adequately discuss “the magnitude of risk and the strategy or process flow to identify and mitigate risk in future research areas” (NRC, 2010, p. 94).
Scenarios Not Fully Developed or Characterized
The uSSRA modeled 109 scenarios in the BSL-4 assessment, but the risks associated with some of them were not fully developed. For example, the necropsy scenario does not consider what the procedure entails with regard to livestock, including the use of knives, saws, rib cutters, and various
other surgical instruments for collecting fluids and tissues. Those activities pose among the highest risks of exposures in the BSL-4 laboratory, but these events are covered in only a single scenario in the uSSRA.
Failure to Consider Natural Disasters
For FMDv, the risk analysis includes potential releases associated with tornadoes and earthquakes. The uSSRA concludes that the overall risk of release for FMDv, although low, is higher for these natural disasters than for any non-disaster scenarios. Although the relative risks in the BSL-4 environment are likely to differ because of the potential for escape via human infection (transference), if one accepts the contention that natural hazards create the greatest risk of releases from the BSL-3 laboratories, the risks associated with natural disasters are also likely to be higher for the BSL-4 laboratories than those associated with the other pathways modeled in the risk assessment. The potential for natural disasters to affect the BSL-4 portions of the facility is not mentioned in the assessment of BSL-4 risks, and the committee wonders why the uSSRA fails to consider natural disasters as part of the BSL-4 risk assessment. Although the facility would be designed to withstand many natural disasters, there is a potential for loss of containment because of pressure fluctuations that can occur during a tornado or loss of structural integrity during an earthquake; this constitutes a significant omission and leads to an understatement of the risks associated with the BSL-4 containment suite.
Because the BSL-4 semi-quantitative assessment used an approach similar to the epidemic modeling of foot-and-mouth disease virus (FMDv), the modeling concerns expressed in Chapter 6 of the present report regarding the assessment of FMDv also pertain to the BSL-4 assessment. For example, the uSSRA calculated probabilities as the simple product of individual risks, which fails to recognize the potential interdependence of risks; for example, a single mechanism of failure may simultaneously impact multiple nodes, meaning these nodes are not completely independent and cannot be illustrated as a simple product. In addition, the uSSRA treats all mechanical errors and human errors as equivalent by using a single numeric value for each in all the risk calculations in the BSL-4 section of the uSSRA. Taken together, these factors may have artificially lowered the calculated risk probabilities.
Inappropriate Extrapolation of Data and Assumptions
The committee recognizes that limited data are available on NiV and HeV and that no studies have been performed for some of those data gaps. The uSSRA attempts a semi-quantitative approach, which would be commendable; however, some assumptions and extrapolations are inappropriate, including the calculations of infectious dose and relative impact.
The calculation of infectious dose (which is a critical factor in this modeling) uses data based on intraperitoneal injections into Syrian hamsters to estimate infectious dose in large animals and humans. The uSSRA extrapolates the infectious dose in large animals and humans on the basis of weight. Allometric scaling is used in chemical risk assessment but is not an accepted practice in microbial risk assessment, in which the initial inoculum can replicate. The extrapolation fails to recognize that the infectious dose–response curve may not be linear. In addition, the uSSRA uses lethal dose (LD50) instead of an infectious dose–response analysis because of unavailability of data on which to model the latter. The committee recognizes that because these agents are associated with high mortality, the LD50 may not vary substantially from the infectious dose, but the use of LD50 is likely to cause artificially high estimates of the dose necessary to produce infection in humans and large animals. As a result, the threshold Q values included in Table 9.9.2-2 of the uSSRA appear to be speculative at best and are probably inappropriate.
The current practice in microbial risk assessment uses dose–response modeling when data are available, as opposed to the minimum infectious dose (MID) approach used in the uSSRA. As previously mentioned in Chapter 6 (see the section “Dose-Response Modeling and Minimum Infectious Dose”), the committee finds the use of MID to be unacceptable. Data-driven dose–response modeling (including model selection via goodness-of-fit testing) may be limited by the availability of suitable data, but a discussion of uncertainty would cover the accepted one-hit (no-threshold) concept (NRC, 2005) with extrapolation at very low doses for infectious dose–response.
The relative impact section (Section 22.214.171.124 of the uSSRA) uses data from the 30 recognized outbreaks of HeV in Australia and the 14 recognized NiV outbreaks in Malaysia, Bangladesh, and India. The uSSRA acknowledges that conditions in those locations are substantially different from conditions in Kansas, especially animal husbandry practices and the availability and quality of human health care. However, the committee is perplexed that the uSSRA would use the mean value to make calculations, such as the number of animals affected in a possible release and the number of human illnesses. For NiV, the initial outbreak in Malaysia resulted in the culling of more than a million pigs, whereas none were culled in any
of the other outbreaks, because these outbreaks occurred in settings where pigs were not present and were due to different modes of transmission. The uSSRA uses an average number of 71,400 animals per outbreak in subsequent calculations, which disregards the statistical principle of excluding outliers. Similarly, the uSSRA uses a value of 0.23 humans affected per outbreak of HeV (the average of 7 human cases in 30 outbreaks). The committee finds that these values are not soundly derived and thus not valid.
Given these significant limitations, the calculations in the more quantitative sections of the BSL-4 assessment seem highly speculative. And the committee reaches this conclusion without even further addressing the limited epidemiological and economic information that was included in the uSSRA relative to BSL-4 pathogens.
In assessing the potential consequences of a BSL-4 pathogen release from the NBAF, the impact assessment briefly mentions the possibility of introducing the agent into a natural reservoir that may sustain transmission. The committee finds this possibility was understated and is a concern regarding non-endemic agents that produce high mortality in animals and humans. The uSSRA discusses work that will be conducted at the NBAF to determine whether henipaviruses can infect North American bat species, and the committee concurs that such work is important. However, if henipaviruses affect native bat populations, this would affect the overall risk assessment and elevate the risk. There is also a potential for NiV to be introduced and transmitted in feral swine populations, which could be virtually impossible to control. Concerns about impact on native wildlife reservoirs should be part of a comprehensive risk assessment, but the uSSRA fails to fully address these risks to native wildlife reservoirs and their potential impact on animal and human health.
The committee finds that the uSSRA qualitatively understates aspects that deal with potential human illness. Particular concerns include (1) the ability and capacity of Mercy Regional Health Center to recognize and handle human illness associated with zoonotic pathogens held in the NBAF (such as NiV, HeV, and Rift Valley fever virus), and (2) the capabilities of the Kansas Department of Health and Environment (KDHE) and local health departments. It will be critical to quickly recognize and diagnose an infection caused by a BSL-4 agent, and the surveillance, diagnostic, and response capabilities of local and state partners may be insufficient for the NBAF in Manhattan, Kansas.
Similar to the previous committee, the present committee is concerned that the medical capabilities that are present at other locations with BSL-4 laboratories (e.g., Galveston and Atlanta) seem to be absent in Manhattan, Kansas. Although Mercy Regional Health Center has isolation rooms, the uSSRA states that there is only a single infectious disease physician and does not assess other medical capabilities in the area. Early recognition of a human zoonotic disease is crucial for proper treatment and could be the key in preventing a potential outbreak. For example, in 2000, a U.S. Army Medical Research Institute of Infectious Diseases (USAMRIID) scientist who was conducting research on Burkholderia mallei, the causative agent for glanders, was presented to a primary care physician with signs and symptoms consistent with glanders, but it took nearly two months for the proper diagnosis to be made (CDC, 2000). Similarly in 2009, it took at least two weeks for tularemia to be diagnosed in a USAMRIID researcher working with Francisella tularensis (Eckstein, 2009).
The ability of providers in the Manhattan area or other rural areas of Kansas to recognize the diseases caused by the suite of pathogens at the NBAF will need to be assessed and will likely need to be improved. The single infectious disease physician and the nursing staff and allied health professionals would be challenged to rapidly diagnose and provide the necessary care for patients with level 3 and 4 infections. HeV and NiV can have relatively long incubation periods, and NiV is known to be transmitted from person to person (Chadha et al., 2006; Gurley et al., 2007; Blum et al., 2009; Homaira et al., 2010). Although some BSL-4 agents are not stable in the environment and are difficult to transmit from person to person, the agent modeled in the uSSRA (NiV) has been transmitted through environmental exposures (from palm sap on trees) and has been transmitted from person to person in family and healthcare settings (Luby et al., 2009). Furthermore, humans are highly mobile (as noted in the FMDv portion of the uSSRA in which Biosecurity Research Institute personnel were questioned about mobility); therefore, patients could be present almost anywhere in Kansas or surrounding states. There may also be secondary cases that are not directly connected to the NBAF. For example, in 2006, a patient was presented to a major Washington, DC, hospital-affiliated primary care center with signs and symptoms of tularemia (a BSL-3 organism) but was not tested for the disease and was eventually discharged without further testing even though the patient informed the medical staff of the characteristic symptoms of tularemia infection (Dudley, 2010).
The uSSRA fails to fully consider the capabilities of KDHE and local health departments. Available information suggests that only emergency management officials in KDHE were contacted for the uSSRA. However, an unrecognized transference event involving human infection would require disease surveillance and diagnostic capacity that depend on KDHE epide-
miology and laboratory personnel more so than on emergency response personnel; this fact was overlooked in the uSSRA. Such capacities should have been assessed and modeled in the overall risk assessment.
Outbreaks involving either human or animal disease caused by a BSL-4 agent would almost certainly be a national problem rather than a regional one. Experience with other high-profile incidents—such as the incidents with anthrax in 2001 and severe acute respiratory syndrome (SARS) in 2003—demonstrates that although the overall number of domestic human cases was small, virtually all locations in the United States were required to investigate and test for potential illness once one case emerged (Perkins et al., 2002; CDC, 2003). If a domestically acquired case of HeV or NiV infection were recognized, there may be a nationwide investigation for disease prevalence in horses and pigs, respectively, but the qualitative portions of the uSSRA do not mention this. The uSSRA uses simple calculations to determine the cost of a human life or the cost of a pig or horse to estimate economic impact, but the costs associated with even a single case would be far greater than suggested due to the nature of the pathogen and the national attention that would ensue. The committee thus finds that the outbreak impact scoring (which is a relative weighting given the lack of a full quantitative analysis) provides false impressions of the impact of an accidental BSL-4 pathogen release from the NBAF.
Blum, L.S., R. Khan, N. Nahar, and R.F. Breiman. 2009. In-depth assessment of an outbreak of Nipah encephalitis with person-to-person transmission in Bangladesh: Implications for prevention and control strategies. Am J Trop Med Hyg 80(1):96-102.
Chadha, M.S., J.A. Comer, L. Lowe, P.A. Rota, P.E. Rollin, W.J. Bellini, T.G. Ksiazek, and A.C. Mishra. 2006. Nipah virus-associated encephalitis outbreak, Siliguri, India. Emerg Infect Dis 12(2):235-240.
CDC (Centers for Disease Control and Prevention). 2000. Laboratory-acquired human glanders—Maryland, May 2000. MMWR 49(24):532-535.
CDC. 2003 Severe Acute Respiratory Infection (SARS) and coronavirus testing—United States, 2003. MMWR 52(14):297-302.
Dudley, J.P. 2010. Tularemia: A case study in medical surveillance and bioterrorism preparedness. J Med C Def 8:1-14.
Eaton, B.T., C.C. Broder, D. Middleton, and L.F. Wang. 2006. Hendra and Nipah viruses: Different and dangerous. Nat Revi Microbiol 4(1):23-35.
Eckstein, M. 2009. Fort Detrick researcher may be sick from lab bacteria. Frederick News-Post. December 5, 2009. Available online at http://www.fredericknewspost.com/sections/news/display.htm?StoryID=98629 (accessed April 25, 2012).
Field, H.E., J.S. Mackenzie, and P. Daszak. 2007. Henipaviruses: Emerging paramyxoviruses associated with fruit bats. Curr Top Microbiol Immunol 315:133-159.
Gurley, E.S., J.M. Montgomery, M.J. Hossain, M. Bell, A.K. Asad, M.R. Islam, M.A. Molla, D.S. Carroll, T.G. Ksiazek, P.A. Rota, L. Lowe, J.A. Comer, P. Rollin, M. Czub, A. Grolla, H. Feldmann, S.P. Luby, J.L. Woodward, and R.F. Breiman. 2007. Person-to-person transmission of Nipah virus in a Bangladeshi community. Emerg Infect Dis 13(7):1031-1037.
Homaira, N., Rahman, M.J. Hossain, J.H. Epstein, R. Sultana, M.S. Khan, G. Podder, K. Nahar, B. Ahmed, E.S. Gurley, P. Daszak, W.I. Lipkin, P.E. Rollin, J.A. Comer, T.G. Ksiazek, and S.P. Luby. 2010. Nipah virus outbreak with person-to-person transmission in a district of Bangladesh, 2007. Epidemiol Infect 138(11):1630-1636.
Johnson, K.M. 2003. Biosafety at National Institute of Allergy and Infectious Diseases: 1982-2003. In: Final Environmental Impact Statement, RML Integrated Research Facility(2004), Appendix D. Review of Biocontainment Laboratory Safety Record. National Institutes of Health, Ravalli County, Hamilton, MT. Available online at http://www.nems.nih.gov/Documents/FEIS%20IRF.pdf (accessed April 28, 2012).
Luby, S.P., E.S. Gurley, and M.J. Hossain. 2009. Transmission of human infection with Nipah Virus. Clin Infect Dis 49(11):1743-1748.
NRC (National Research Council). 2005. Reopening Public Facilities After a Biological Attack: A Decision-Making Framework. Washington, DC: The National Academies Press.
NRC. 2010. Evaluation of a Site-Specific Risk Assessment for the Department of Homeland Security’s Planned National Bio- and Agro-Defense Facility in Manhattan, Kansas. Washington, DC: The National Academies Press.
Perkins, B.A., T. Popovic, and K. Yeskey. 2002. Public health in the time of bioterrorism. Emerg Infect Dis 8(10):1015-1018.
This page intentionally left blank.