As a result of the Biological Weapons Anti-Terrorism Act of 1989 and the Antiterrorism and Effective Death Penalty Act of 1996, the United States Department of Health and Human Services (DHHS) formed an interagency group that developed a list of 31 infectious biological agents and 12 toxins that could pose a threat to public health. This was in part due to an incident where an individual inappropriately ordered Yersinia pestis strains from a commercial source and officials recognized they had no legal mechanism to charge this individual with a crime other than mail fraud. Criteria for inclusion on the biological select agent and toxins (BSAT) list included consideration of the effect of exposure on human health; the degree of contagiousness of the agent, or toxin, mode of transmission, availability and effectiveness of therapeutics and vaccines; and any other criteria that the Secretary of DHHS and/or Secretary of the Department of Agriculture (USDA) considered appropriate. By addition of language to the code of federal regulations and publication in the Federal Register on October 24, 1996, regulatory oversight mechanisms were put in place for registration of facilities transferring or receiving BSAT, establishing requirements for transfer, establishing verification procedures and accountability mechanisms, and setting disposal requirements. Regulations provided exemptions for specific BSAT work activities in an effort to maintain the availability of the agents for research, education and other legitimate purposes. Both DHHS and the USDA were delegated with oversight of compliance with regulations and the BSAT list was eventually subdivided into agents specific to DHHS, agents specific to USDA, and overlap agents and toxins. Select agent regulations and BSAT are codified within 3 areas of the Code of Federal Regulations (CFR), 42 CFR Part 73, 9 CFR Part 121, and 7 CFR Part 331. Further oversight of BSAT was extended beyond regulation of transfers to cover the full range of activities surrounding possession, use and transfer of BSAT as part of the USA PATRIOT Act in 2001 and the Public Health Security and Bioterrorism Preparedness and Response Act of 2002 and the Agricultural Bioterrorism Protection Act of 2002, with the later legislation requiring biennial review of the list of BSAT. Implementation of the legislation within these bills is accomplished through a series of regulations and policies known as the select agent regulations (SAR).
For biennial review of listed BSAT, the Intra-agency Select Agents and Toxins Technical Advisory Committee (ISATTAC) was established. For DHHS agents and overlap agents, the committee included representatives from the Centers for Disease Control and Prevention (CDC), the National Institutes of Health (NIH), the Food and Drug Administration (FDA), the Biomedical Advanced Research and Development Authority (BARDA) within the Assistant Secretary for Preparedness and Response (ASPR), the Department of Defense (DOD), the Department of Homeland Security (DHS) and within the United States Department of Agriculture (USDA), the Animal and Plant Health Inspection Service (APHIS), and the Agricultural Research Service (ARS). For agents/toxins listed as USDA only, the ISATTAC committee included only representatives from the agencies within USDA. In compliance with requirements set within the Public Health Security and Bioterrorism Preparedness and Response Act of 2002, deliberations concerning inclusion (or exclusion) of an agent or toxin on the BSAT list required consideration of the degree of pathogenicity (ability of an organism to cause disease), communicability (ability to spread from infected to susceptible hosts), ease of dissemination, route of exposure, environmental stability, ease of production, ability to genetically manipulate or alter the agent or toxin virulence, long-term health effects, acute morbidity (illness), acute mortality (death), available treatment, host-pathogen interactions (ie, status of host immunity), vulnerability of special populations, and the burden or impact on the health care system. As a result of biennial reviews, modifications to the list of regulated BSAT were published in the Federal Register on October 16, 2008, October 5, 2012, and January 19, 2016.
With President Obama’s signature on Executive Order 13546 on July 2, 2010, a Federal Experts Security Advisory Panel (FESAP) was established to provide technical recommendations to the Secretaries of DHHS and USDA and the US Attorney General related to (bio)security measures for working with BSAT. This committee was also tasked with recommending mechanisms to coordinate government oversight programs to minimize adverse impacts on the legitimate use of BSAT. During initial deliberations in 2010, in addition to reviewing agents and toxins on the BSAT list, the FESAP was also directed to identify and establish security standards for BSAT with the highest risk of misuse. As reported to USDA and DHHS on June 13, 2011 (published in Federal Register on October 5, 2012), the FESAP recommended removal of some BSAT and stratified the remaining BSAT into 2 tiers. Agents with the greatest risk for deliberate misuse causing significant adverse effects were placed into Tier 1, and the report included recommendations related to physical security, cyber security, and personnel suitability requirements for work with these BSAT. In addition to periodic review of BSAT agents and associated security, FESAP was rechartered in July of 2014 and assigned to broadly evaluate approaches to enhance biosafety and biosecurity in the United States.
In addition to ISATTAC and FESAP reviews, the July 2, 2010 executive order also directed the National Science and Technology Council (NSTC) to establish a Fast Track Action Committee (FTAC) which published a report in October 2015 with recommendations related to regulatory issues and biosafety concerns related to work with BSAT. Related reports addressing biosafety/biosecurity and the benefits and costs of SAR have been provided from a number of federal task forces and panels including the 2009 report from the Executive Order 13486 Working Group on Strengthening the Biosecurity of the United States, and the 2009 report of the Trans Federal Task Force on Optimizing Biosafety and Biocontainment Oversight (TFTF). It is also important to recognize that concerns related to research that could have both beneficial and harmful purposes (dual-use research of concern; DURC) has precipitated a report,
1
and the chartering of a National Science Advisory Board for Biosecurity at NIH. That board, in collaboration with the National Academies of Science and a Trans Federal Task Force composed of representatives from DHHS and USDA were tasked with making recommendations related to DURC to the Select Agent Program.2,3
Three species of Brucella, B. abortus, B. melitensis, and B. suis were included on the initial list of BSAT in 1996 and were eventually listed as overlap BSAT due to their zoonotic capabilities. Sheep and goats are the preferred hosts for B. melitensis, cattle for B. abortus, and swine for B. suis. Their inclusion on the BSAT list is most likely related to weaponization of B. suis by the United States in the 1950s as one of its first biological weapons.4
–6 Concerns regarding their ability to be BSAT agents may also be related to efforts by the 731 Manchuria Unit of Japan before and during World War II, to weaponize Brucella, and reports that Russia developed antibiotic-resistant forms for weaponization that were eventually dropped in favor of use of Burkholderia pseudomallei, a pathogen considered more potent.
7
It should be noted that these 3 species of Brucella were not recommended for classification as Tier 1 agents with the initiation of the 2-tier classifications during the biennial review finalized in 2012. Although not recommended for removal in prior reviews of the BSAT list, as published in the Federal Register as an advanced notice on February 27, 2015 and a revised version on January 19, 2016, DHHS and USDA proposed amending the BSAT list to remove B. abortus and B. suis, with retention of B. melitensis but only as a USDA BSAT and not as an overlap agent. As a result of the publication in the Federal Register recommending changes in classification of Brucella as BSAT, the FESAP committee was tasked by the White House National Security Council with reviewing the proposed recommendation and providing a recommendation to DHHS and USDA after reviewing technical and policy implications of the proposed action. Although FESAP recommendations by themselves do not change any current policies, guidance or regulations regarding the Federal Select Agent Program, they must be addressed by DHHS and USDA during the next round of review and revisions of regulated BSAT. Ultimately, the recommendation by FESAP in 2017, and accepted by DHHS and USDA by parallel publication in the Federal Register on January 19, 2017, was that all 3 species of Brucella would remain on the BSAT list.
During the FESAP review process in 2016 which considered removal of B. abortus and B. suis from the BSAT list, differences of opinion were identified related to the biological properties and natural life cycle of Brucella, and the implications of these properties on weaponization and security. The discrepancies of opinions between FESAP representatives in regards to properties of Brucella or other concerns related to the agent suggested a need to conduct an in depth review of relevant scientific publications. Material presented in this paper is an attempt to meet this need and provide published scientific data relevant to possession, use, and/or transfer of Brucella spp. and pertinent to discussions of the agent as a BSAT. It is intended that this review will promote informed discussion, be beneficial to personnel interested in biological properties of Brucella, and encourage further research and publication of relevant data.
Economic Impacts of Brucellosis on Livestock Production and Regulatory Programs
Brucella spp. cause reproductive losses in its animal hosts and is of importance to public health because of potential zoonotic infections which generally correlate to disease prevalence in natural hosts. It has been estimated that between 1934 and 1998, $3.5 billion in federal, state, and private funds were spent on brucellosis eradication programs in domestic livestock (predominantly cattle) with an estimated benefit of greater than $18.3 billion. This investment dramatically lowered the economic losses of brucellosis in the livestock industry from an estimated $400 million in 1952 and led to reductions in costs associated with human infections. However, if brucellosis eradication program efforts were stopped, it is estimated that annual costs to beef and dairy industries could increase to $80 million annually in less than 10 years (http://www.aphis.usda.gov/animal_health/animal_disease/brucellosis/downloads/bruc_facts). Currently, APHIS has an annual budget of approximately $92.5 million to conduct its cattle health program and $22 million for its swine health programs with brucellosis being a focus of eradication in both programs.
However, the endemic infections of wildlife in the areas surrounding Yellowstone National Park with B. abortus and the expanding populations of feral swine (Sus scrofa) infected with B. suis pose new threats for preventing brucellosis in domestic livestock and protecting public health. Impacted bison (Bison bison; approximately 5000 head in the fall of 2017) and elk (Cervus elaphus; approximately 120 000 to 140 000) populations in the Greater Yellowstone Area are currently demonstrating disease prevalence of between 35 to 60%. It is currently estimated that Montana spends approximately $2.4 million annually related to preventing transmission of brucellosis from infected wildlife to cattle (personal communication, M. Zaluski, Montana State veterinarian). Costs in Wyoming related to cattle brucellosis are estimated at $310 000 but do not include costs on mitigation activities by USDA/APHIS, or expenditures by the Wyoming Game and Fish Department (personal communication, J. Logan, Wyoming State veterinarian). Brucellosis costs in Montana are increasing as a result of the disease problem escalating in wild populations—not only in regard to increasing infection rates (a group of elk have been identified with an infection rate of over 50%), but also in the expanding geographic range in which infected animals are being found. For example, the surveillance area for brucellosis in Montana has expanded 3 times in the last 4 years. New vaccines and/or diagnostics that are applicable to free-ranging wildlife are needed to mitigate the brucellosis issues associated with disease in bison and elk. Containment facilities available within the United States that can house sufficient numbers of livestock or wildlife hosts for experimental studies with Brucella are extremely limited. Use of space is also impacted by the fact that bison and elk are seasonal breeders meaning that vaccine efficacy studies can only be conducted once per year.
In the United States, feral swine continue to expand their range with established herds in 41 states and estimated populations exceeding 6 million. Not only have feral swine transmitted brucellosis to domestic cattle and swine, but people involved in hunting feral swine or wild boar have been infected.8,9 Of particular concern is lack of a serologic test with adequate sensitivity and specificity for detecting brucellosis in swine, and transmission of infection from feral swine to cattle leading to shedding of B. suis in milk. Recent interest in allowing sale of unpasteurized milk is of concern as others have demonstrated that large outbreaks of B. suis in people have been associated with raw milk consumption.10,11 As with brucellosis in large ungulates, new vaccines and diagnostics are needed that target B. suis infections in feral swine, but also prevent and detect transmission to domestic livestock. In a similar manner to limitations for conducting studies with BSAT in large ungulates, capacity of US containment facilities to conduct studies related to swine brucellosis is extremely limited.
Over the last 7 decades, billions of dollars have been invested in brucellosis eradication programs for domestic livestock with the purpose of providing public health benefits. As transmission of brucellosis from free-ranging wildlife reservoirs to domestic livestock has increased in frequency, this investment may be at risk. New vaccines and diagnostics are needed that cannot be developed in the absence of in vivo work in the species of interest. Without advancements in intervention tools, states with endemic disease will continue to bear the cost of trying to prevent reintroduction of brucellosis into domestic livestock for the foreseeable future. The possibility cannot be excluded that numbers of human infections may grow if disease prevalence increases in domestic or wildlife hosts, or expands into new areas.
Some Issues Related to Brucella of Importance to FESAP Discussions
Characteristics as a Bioweapon and Availability
Some Brucella spp. (B. melitensis, B. suis, and B. abortus) have traditionally been considered to be potential biological weapons due to their ability to be transmitted by aerosol, cause chronic, debilitating zoonotic disease, and induce vague clinical symptoms that prevent rapid diagnosis. However, they also have zoonotic traits that make them less attractive as bioweapons including low mortality, long incubation periods, relatively mild clinical disease symptoms (“flu-like”), and being clinically responsive to treatment with a number of antibiotics including rifampicin or doxycycline. The incubation period is particularly prolonged for brucellosis, most cases ranging from 1 to 4 weeks but in some instances extending to as long as 6 months.
12
It has been suggested that the emergence of new, more virulent potential biological weapons make inclusion of Brucella among agents of bioterrorism to be mainly of historical significance.
6
Brucella spp. included on the BSAT list are readily available worldwide in domestic livestock and other reservoir hosts. In the United States, B. suis is endemic in feral swine, an invasive species present in at least 41 states with a population estimated to be greater than 6 million animals.
13
Prevalence of brucellosis in feral swine may differ by region but in a study in 2 abattoirs in Texas, Brucella was isolated from 13% of feral swine randomly sampled (Olsen, S.C. unpublished data). Brucella abortus is endemic in bison and elk in Yellowstone National Park and in surrounding areas. Although comparative data are not available for elk, culture data suggest that B. abortus can be recovered from approximately 46% of seropositive bison.
14
As hunting of feral swine and elk is allowed in endemic areas, procurement of tissues from infected animals would be a legal activity that would not require registration, oversight, or any other sort of regulatory burden.
Human Morbidity and Mortality
An area of concern is the hypothesis that weaponization of Brucella spp. could cause significant levels of human mortality when used as a bioweapon. Related to this concern, it is important to consider disease pathogenesis, clinical symptoms, and disease prevalence and epidemiology of brucellosis in humans worldwide. Brucellosis is the most common zoonotic infection worldwide with more than 500 000 new cases occurring annually.
15
In humans, B. melitensis is considered to be more virulent than B. suis or B. abortus. Natural infection in humans is almost exclusively related to exposure to infected animal reservoirs and/or products from infected animals. Humans essentially function as dead-end hosts for Brucella in that human-to-human transmission is rare and of minimal epidemiologic significance. Worldwide, human infection correlates with disease prevalence within populations of reservoir animal hosts in that region. The highest prevalence of human infection with B. melitensis occurs in countries in the Mediterranean basin and Central Asia, whereas human infection with B. suis is more prevalent in areas with high populations of infected domestic or feral swine.
In humans, Brucella is described as a disease of protean manifestations
16
as bacteria can localize in almost any tissue and cause clinical symptoms based on sites of infection. Localization in the osteoarticular system is most frequent (25-35% of patients), followed by infection in the genitourinary, central nervous, and cardiovascular systems.
17
Endocarditis is probably the most significant clinical manifestation in human brucellosis with studies estimating its occurrence in 0.3 to 2% of human cases.
18
Localization in the aortic arch can also result in mycotic, pseudo-, or dissecting aneurysms or abscesses.
19
The association between endocarditis and brucellosis has been known for at least 80 years.
20
Brucella melitensis infection is linked to the majority of cases of endocarditis in brucellosis patients, although B. abortus has been associated with a small percentage of cases.
17
The most frequently affected valve is the aortic valve17,21
–23 and histology has indicated the large vegetations on valves are the result of direct invasion of the endocardium by bacteria.
17
Brucella endocarditis can also be associated with intercardiac abscesses (20.5% in 1 review) and a high incidence of congestive heart failure (59.1%).
17
Before the introduction of open heart surgery, most cases of Brucella endocarditis were treated medically and mortality was more than 80%.
17
Medical advancements since the 1950’s, such as availability of surgical valve replacement, have dramatically reduced mortalities associated with Brucella endocarditis. A 40-year meta-analysis of reports of endocarditis treatment (primarily cases from Middle Eastern countries or Mediterranean regions) found a 6.7% mortality (16 of 239) in patients in which medical and surgical treatments were combined with mortalities appearing to be more frequent in earlier publications.
17
It should be noted that the literature indicates that mortalities related to endocarditis have become increasingly rare in more recent studies. Recent reviews of Brucella endocarditis patients have reported the following: 6 mortalities in a review of 14 brucellosis endocarditis cases over an 18-year period in Iran, 2 mortalities in 13 patients over 15 years in Turkey, 2 mortalities in 31 patients over an 11-year period in Turkey, and 7 mortalities in 45 Brucella endocarditis patients over a 5-year period in Turkey.24
–28 Although not identified as related to endocarditis, a review of 513 034 cases in China since 1955 reported 170 deaths in patients diagnosed with brucellosis (0.03% of cases).
29
In addition, a meta-review of published studies since 1985 on antimicrobial treatment of brucellosis found 6 deaths reported in over 1500 patients (<0.4%).
30
In a further effort to evaluate the risk of fatalities in infected humans, a PubMed search was conducted to identify more recent group or review studies for data on mortalities in patients with brucellosis. A meta-analysis review of clinical manifestations of published human brucellosis studies
31
of 2385 articles published between 1990 and 2010 did not include mortality in the table of clinical manifestations. Recent reviews of clinical cases in which no mortalities were reported included: 88 childhood brucellosis cases in Turkey,
32
14 B. melitensis patients in Qatar,
33
164 cases in Turkey,
34
317 childhood brucellosis cases in Macedonia,
35
52 pediatric cases in Turkey,
36
347 patients with acute brucellosis in Turkey,
37
700 hospitalized cases in Turkey,
38
2630 patients in Bosnia,
39
79 patients in Malaysia,
40
128 childhood brucellosis patients in Israel,
41
496 cases of childhood brucellosis in Turkey,
42
7 cases of neurobrucellosis in Turkey,
43
28 patients with Brucella epididymo-orchitis in Turkey,
44
72 hospitalized cases in Qatar,
45
53 cases in Northern Ireland,
46
144 treated patients in Iran,
47
390 cases of genitourinary brucellosis in Turkey,
48
370 hospitalized patients in Turkey,
49
325 brucellosis patients with liver involvement in Turkey,
50
and 14 patients with endocarditis treated with antibiotics alone.
51
It should be noted that most of these reports were related to patients in developing countries, and the majority of patients were from areas in which B. melitensis, the most virulent Brucella species, is endemic in domestic livestock. Overall, published reports suggest that risk of mortality has dramatically declined with advancement in medical techniques and, in context with the approximately 500 000 new cases worldwide annually, the mortality risk imposed by brucellosis in humans is extremely low.
One overlooked aspect of brucellosis that should be considered is risk of exposure to pregnant women. A report in 1980 documented fetal loss due to B. melitensis septicemia in the second trimester.
52
Studies of maternal brucellosis in areas endemic for B. meltensis infection (southern Israel and Saudi Arabia) reported increased incidence of spontaneous abortion in infected patients (36% and 27%) with one of the studies also reporting an increase in intrauterine fetal death (12.7%).53,54 In a review of 101 cases of pregnant women infected with B. melitensis in Peru, 12.8% experienced spontaneous abortion, 8.1% fetal death and 1.1% congenital malformations.
55
A 6-year review of 29 cases of brucellosis in pregnant women in Turkey found that the incidence of spontaneous abortion (24.1%) was increased when compared to the calculated incidence in noninfected patients (7.6%).
56
Other reports have tied brucellosis to premature delivery,
57
and transplacental transmission,58,59 including subsequent infant myocarditis or pulmonary infection.60,61 In a review of laboratory-acquired brucellosis, 4 of 7 pregnant workers in the identified publications aborted and were positive for brucellosis by serology or culture.
62
All identified reports of abortions in pregnant women were related to infection with B. melitensis infection, or are in areas known for endemic infection of domestic livestock with this Brucella spp. It should be noted that not all reviews have found increases in spontaneous abortion or fetal loss associated with brucellosis in pregnant women. In a study of 39 pregnant women diagnosed with brucellosis in an area endemic for B. melitensis, there was an increase in the incidence of preterm delivery and low birth weight, but no effect on birth defects, anomalies, or mortalities.
63
Also, antibiotic treatment (trimethoprim-sulfamethoxazole with or without rifampicin) of Brucella-infected women during pregnancy in Saudi Arabia was reported to be 90% effective in achieving a normal delivery.
64
For additional information on Brucella, see http://www.usamriid.army.mil/education/bluebookpdf/USAMRIID%20BlueBook%208th%20Edition%20-%20Sep%202014.pdf and https://www.cdc.gov/brucellosis/pdf/brucellosi-reference-guide.pdf.
Infectious Dosages of B. abortus, B. melitensis, and B. suis
Weaponization of a BSAT may be more successful if the agent has a low infectious dosage and discussions indicated that the infectious dosage for BSAT species of Brucella was an area of disagreement. For at least 40 years it has been reported in published literature, usually without citation, that the minimal infectious dose for Brucella was 10 to 100 bacteria. Although there may be unpublished data generated by the military that was unavailable for this review, statements on infectious dosage in the scientific literature may be based on work published in the 1950s. A 1956 study reported that the 50% infective dose for B. suis in guinea pigs was 5 organisms when delivered by a subcutaneous route
65
and 36 bacteria when delivered via aerosol with infectious dose by the aerosol route being influenced by particle size.
66
To interpret the basis for the estimated minimal dose of 36 bacteria when delivered by aerosol, it is important to understand that the estimate was based on work conducted in a closed system, and that dosage was not the amount of bacteria delivered into the closed system. Rather, infectious dose was calculated based on estimation of tidal volume and respiratory rate of guinea pigs, aerosol concentration, spray factor and flow, total flow, and duration of exposure. By the authors own calculations, an estimated inhaled dose of 60 bacteria required preparation of a bacterial suspension with a bacterial concentration of 2.74 × 106 bacteria/ml.
66
Therefore, the estimate was not of bacteria delivered to the animal or the animal’s environment,
67
but rather an estimation of the total numbers of bacteria distributed into pulmonary tissues that resulted in a 50% infection rate (ID50). It should be emphasized that this calculation of infectious dose fails to consider the possibility that infection may have occurred through mucosal routes other than the pulmonary system during aerosol exposure; such as the upper respiratory (nasal and tonsil), oral cavity (with subsequent ingestion), and conjunctival. With the exception of the tonsil, all of these other routes have been successfully used for experimental infection of laboratory animals and/or reservoir hosts with B. suis, B. melitensis, and/or B. abortus. By reporting a calculated respiratory dosage from an artificial environmental exposure system, these studies have calculated an infectious dosage of Brucella that overestimates the true infectivity of the genus under natural conditions.
As estimation of minimal infectious dose was done in a guinea pig model, it is important to emphasize that cumulative data indicate that guinea pigs are uniquely susceptible to infection with virulent Brucella at very low dosages. Guinea pigs may be the most susceptible species known for infection with Brucella
68
and are susceptible by virtually any route of delivery. Guinea pigs are so susceptible to infection they can be used to amplify Brucella when microbiological methods for culture isolation are not successful.69
–71 The high susceptibility of guinea pigs is supported by a meta-analysis of experimental infection studies which estimated median infectious dosages (N50) by aerosol delivery to be 30 colony-forming units (CFU) for B. suis and 94 CFU for B. melitensis.
72
In comparison, the estimated N50 dosage was 1924 CFU for aerosolized B. melitensis in rhesus monkeys, 1840 CFU for aerosolized B. melitensis in mice, and 1885 CFU for intradermal infection of humans with B. melitensis Rev1. When susceptibility through the subcutaneous route was compared, the N50 for B. melitensis was estimated at 1840 CFU in mice as compared to 82 CFU for guinea pigs.
Others have estimated infectious dosages of aerosolized Brucella when delivered to mice. One study estimated
73
a minimal dosage of 102 CFU for B. suis and 103 CFU for B. melitensis when aerosolized for 10 minutes to BALB/c mice. However, as with early studies in guinea pigs, rather than reporting actual dosages delivered to the closed chamber, the authors used published methods
74
to calculate infectious dose based on estimated tidal volume for mice and assuming a lung retention of 50% of respiratory volume. Using calculated retained dosages of 102, 103, and 104 CFU of B. suis resulted in an average mean splenic colonization (CFU/gm) of 1, 2, and 3 logs, respectively, in naive mice at 4 weeks postchallenge.
73
In comparison, average splenic colonization was approximately 0.4, 4.5, and 4.0 logs, respectively, for similar aerosol dosages of B. melitensis. A separate study estimated that retained aerosol dosages of 415 CFU (2.62 logs) for B. abortus strain 2308 and 7988 CFU (3.90 logs) for B. melitensis strain 16 M were required for consistent infection in BALB/c mice.
75
Estimated retained dosages were based on culture of lungs from 4 or 5 mice euthanized immediately after aerosol delivery of 5 × 107, 5 × 108, or 5 × 109 CFU total dosages to 4 or 5 mice in an enclosed chamber. As with the previous studies, infectious dosage was estimated based upon pulmonary delivery and excluded consideration of bacteria deposited onto nasal, oral, or conjunctival mucous membranes. For comparative purposes, the low, middle, and high challenge dosages of B. melitensis in that study resulted in mean colonization in spleens of approximately 2.2, 4.5, and 5.2 logs, respectively at 4 weeks postinfection. A fourth study evaluated aerosol delivery of 103 to 1010 CFU of B. abortus and B. melitensis to BALB/c mice in a closed chamber and found that for aerosol delivery of B. melitensis, inconsistent infection occurred between 103 to 105 CFU with high levels of pulmonary, liver and spleen infection occurring at 106 CFU and higher dosages.
76
For B. abortus, the threshold for consistent infection after aerosolized delivery was 107 CFU.
76
In summary, regression analysis concluded
72
that at least 3.26 logs of aerosolized Brucella are needed to infect mice but this is based on estimated numbers of bacteria deposited only within the lung parenchyma while housed within a closed chamber. It would seem logical to extrapolate that aerosol dosages delivered to the closed chamber would need to be 1 to 2 logs higher than pulmonary deposition estimates. This conclusion is supported by published data demonstrating that calculated retained aerosol dosages used for estimating minimal infectious dose are several logs below the dosages of Brucella actually delivered to the enclosed chamber.
75
Therefore, if 3.26 logs of Brucella need to be delivered to the lung parenchyma, minimal infectious dosage delivered to mice in an aerosol chamber would need to be approximately 104 CFU or greater. This is higher than the estimated infectious dosage obtained by meta-analysis,
72
but may reflect the use of papers based on calculated retained pulmonary dosages in the meta-analysis as compared to evaluating bacterial concentrations actually delivered into aerosol chambers.
Although the work is not as extensive as in mice and guinea pigs, aerosol dosages of B. melitensis and B. suis have been evaluated in primates with some authors concluding they are good models for human brucellosis.66,77
–79 Using a closed system, 6 × 102 to 1 × 106 CFU of B. melitensis was delivered via aerosol to rhesus monkeys (Macaccus rhesus).
80
Reported methods used in this study were based on previous work,
66
which infers that estimates of bacteria reaching the lung parenchyma were calculated as in the previous work, rather than the total bacterial dose aerosolized into the chamber, and also failed to consider infection through oral, nasal, or conjunctival routes. At the 2 lowest calculated dosages (6 × 102 and 9.5 × 102 CFU), 3 of 10 and 4 of 10 monkeys were found to be infected at 6 weeks after exposure. In comparison, at 1.5 × 103 infection rates reached 50% (5 out of 10) and at 1.45 × 104 CFU all monkeys (10 of 10) were infected. The calculated ID50 for this study was 1.3 × 103 CFU. In a separate set of studies, a head only model was used to deliver via aerosol 103 to 105 CFU of B. melitensis to rhesus monkeys.
78
As with previous studies, estimates of challenge dosage were based on respiratory minute volume and concentrations of bacteria in the exposure chamber. In the initial study, only 1 of 2 animals receiving calculated aerosol dosages of 103 to 105 CFU had B. melitensis recovered from weekly ultrasound-guided liver biopsy samples but all were reported as culture positive in lung, liver, and/or spleen tissue at necropsy at 7 weeks postinfection. In subsequent efforts to characterize the pathophysiology of B. melitensis infection in rhesus monkeys, the authors elected to use an aerosol dosage of 105 CFU for their second experiment. In another report using a primate model, dosages of 105 to 106 of B. melitensis were effectively used to infect rhesus monkeys through an aerosol delivered into a head-only exposure chamber.
79
Although not clearly stated in the report, citations used in the methods section suggest that this was a calculated dosage based on aerosol concentration and respiratory parameters. Although a minimal infectious dosage was not calculated, B. melitensis was recovered from the blood of all macaques (n = 3) exposed to the lower aerosol dosage (105) at 1 day after exposure. Brucella was not recovered from blood of macaques receiving the 106 CFU aerosolized dosage at 15 and 45 days postinfection, but was recovered from other tissues. An additional study used a calculated dose of 108 CFU of B. suis delivered via aerosol to rhesus macaques in a head-only chamber.
77
As with previous studies, respiratory minute volumes and aerosol concentrations were used to estimate exposure dose. Although B. suis could not be recovered from blood, systemic infection in lung, liver, and spleen ranged between 0.75 and 5.3 logs at necropsy 1 to 7 days after infection. This study reported that aerosolized B. suis essentially did not cause clinical illness over the short course of the study as reflected by changes in physical activity or blood chemistry parameters, although fever spikes were noted in a small number of monkeys at 1 to 3 days after infection. As mentioned previously, a meta-analysis approach estimated the minimal infectious dosage for B. melitensis at 1885 CFU for humans and primates based on published data from aerosolized infection of macaques and human vaccine trials with B. melitensis strain Rev1.
72
In a similar manner to the discussion estimating minimal infectious dosages in mice, it could be easily concluded that reproducible infection of primates require delivery of at least 104 to 105 CFU of Brucella bacteria into the immediate environment even under controlled conditions and that fewer aerosolized bacteria and a lower exposure dose would lead to unreliable or absence of infection in exposed individuals.
It is also important to consider the effects of dosage on infection of natural reservoir hosts of Brucella. Under experimental conditions, conjunctival exposure is the most common route for infection of reservoir hosts and experimental challenges are usually done in midgestation when animals are most susceptible to infection. Dose response studies in cattle found that conjunctival delivery of 3.5 × 105 CFU of virulent B. abortus strain 2308 resulted in infection of 78% of naïve cattle, whereas 1.5 × 107 CFU resulted in 100% infection.
81
The ID50 for infection in any tissue of naïve goats with B. melitensis was estimated at 2.62 × 104 CFU for conjunctival delivery, whereas the ID50 for generalized infection was estimated at 6.13 × 104 CFU.
82
Others have estimated the ID50 of B. melitensis to be 2.6-5 × 104 CFU in naïve goats,
83
2 × 104 CFU for Swedish goats, and 4 × 105 CFU for Swedish sheep.
84
It should also be noted that studies in cattle and goats typically use conjunctival delivery of approximately 107 CFU of a virulent strain of B. abortus or B. melitensis for experimental challenges.85,86
The authors of the meta-analysis of aerosolized Brucella infections
72
used their regression analysis to extrapolate an infectious dosage of 10 CFU per person related to a laboratory exposure in Italy in which 31% of identified laboratory workers seroconverted after B. abortus biovar 1 exposure (due to a broken centrifuge tube).
87
It should be noted that the original paper of the human exposure does not estimate exposure dosage.
87
Also, the lab exposure was to B. abortus, rather than B. melitensis or B. suis which were the aerosolized pathogens used for development of an infectious dose regression curve in the meta-analysis paper. As the authors’ own personal experiences indicate that a single petri dish cultured under in vitro conditions can easily have 10 to 11 logs of viable bacteria, the estimation of a 10 CFU dosage per person appear to be conservative and also fails to consider that the respiratory system may not be only route responsible for infection during laboratory exposures. We are concerned that extrapolating data obtained in closed chambers to exposure dosages in open environments may exceed parameters used to develop the regression model.
It is a real possibility that data obtained in closed chambers and based on assumptions concerning respiratory physiology, grossly overestimate the likelihood that comparable exposures will lead to infection in an open environment. Some publications that cite a low infectious dose without citing specific studies may be inappropriately applying conclusions from controlled conditions to potential exposures under natural conditions or open environments.
Consequences of Infection Through the Pulmonary Route
Aerosol delivery of Brucella as a bioweapon may be more likely to result in pulmonary localization as compared to tissue distribution after mucosal exposure, the primary route for natural infection. One concern is whether pulmonary infection, which is estimated to occur in 0.6 to 16% of all brucellosis cases,
88
is associated with more severe clinical symptoms or is more refractory to treatment. A retrospective 30-year meta-analysis of 107 papers on pulmonary human brucellosis estimated 0.5% of patients had pulmonary localization with the most common clinical symptoms being fever (80.6%), cough (63.90%) and pleuritic chest pain (33.3%).
89
Treatment with antibiotics was found to result in an excellent prognosis with only 1 patient having a relapse of infection. It should be noted that 2 mortalities were identified in the report including a patient with pleural effusion, pulmonary and pleural nodules, and liver and spinal lesions
90
and another who died of septic complications after a diagnosis of lymphoproliferative syndrome and Brucella pleural empyema.
91
Other studies have also reported excellent responses to antibiotic therapy in patients with pulmonary brucellosis.92,93 A separate review found that 7.3% of brucellosis patients in a Turkish hospital over a 10-year period had pulmonary involvement.
88
Brucella melitensis is endemic in Turkey and all patients responded to antibiotic treatment within 7 to 10 days. In a 6-year study of patients in Kuwait, 0.4% of 1100 children with brucellosis had pulmonary infection.
94
In 17 workers infected with B. suis at a pork processing plant in Argentina, 4 had indications of respiratory involvement and infections were assumed to be the result of aerosols, conjunctival splashes, or direct entry through skin lesions.
95
In similar manner to other reports of pulmonary brucellosis, patients in this study responded well to antimicrobial treatment with no relapses in patients with respiratory involvement.
95
In a survey of human brucellosis caused by B. suis in Australia, 38% (n = 12) had respiratory symptoms but the manuscript did not clarify if patients with respiratory symptoms were among the 9% (n = 3) that had relapse after treatment.
73
In conclusion, data suggest that localization in pulmonary tissues; even in environments in which exposure may have been via aerosol, is not associated with more severe clinical symptoms and responds equally well to antimicrobial treatment as compared to brucellosis patients without pulmonary involvement. Although a newly identified Brucella-like bacteria, strain BO2, was associated with chronic destructive pneumonia in a human patient, it is not a BSAT and is unlikely to be considered as a candidate to be weaponized.
96
Cost of Treatment for Brucellosis Patients
Medical costs associated with a bioweapon attack using Brucella organisms was also an area of concern. One of the first estimates of the cost of a Brucella bioweapon attack was published in 1997 using the following assumptions: 1) exposure to 1000 CFU was equal to 10 ID50, 2) brucellosis would induce a case fatality rate of 0.5%; 3) a calculated epidemic curve with 82.5% of exposed patients being infected with 50% of patients requiring hospitalization for an average of 7 days; 4) 80% efficacy of antibiotic treatment (doxycycline-rifampicin for 42 days at a cost of $285); and 5) an average of 7 and 14 outpatient visits for hospitalized and nonhospitalized patients, respectively.
97
Based on their assumptions and calculations, the authors estimated a bioterrorism attack using Brucella spp. would cost $477.7 million per 100 000 exposed people. Assuming that weaponized strains of Brucella were similar to natural infection, data presented previously within this manuscript would suggest that estimates for infectious dose, epidemic curves, mortality rates, and total cost were high in the 1997 publication. Estimated cost for treatment also appears excessive as a recent review estimated treatment costs for 45 day treatments of doxycycline, tetracycline, and rifampicin were $21.16, $52.20, and $35.10 to 52.65, respectively, with the cost of the recommended doxycycline-rifampicin combination being estimated at $56.70 to 74.26.
30
Clinical data suggest that efficacy of antimicrobial treatments against brucellosis would probably be higher than the estimated 80% in the 1997 paper. A meta-review of clinical trials of antimicrobial treatment of human brucellosis over a 27-year period found relapse rates in clinical patients treated with doxycycline-rifampicin and doxycycline-streptomycin to be 18% and 6.6%, respectively,
30
whereas a separate meta-analysis found relapse rates for the same antibiotic combinations to be of 14% and 4.4%.
98
It should be noted that the publications included in these meta-reviews were clinical patients with symptoms of brucellosis, and that additional studies report very low relapse rates after treatment.99,100
In addition, economic data from an endemic area questions the perception that an outbreak of human brucellosis would be associated with high medical costs as estimated in the Brucella bioweapon attack paper.
97
Annual health care costs for 470 patients with brucellosis in Israel were higher ($1327 versus $380 annually) than average costs of controls (age and sex matched cohorts not infected with Brucella).
101
Contributors to the elevated costs were a 7.9 times higher hospitalization cost, 3.6 higher diagnostic test costs, 2.8 times greater emergency room visits, 1.8 times higher medication costs and 1.3 times greater diagnostic procedure costs. After diagnosis of brucellosis, patients averaged 11 outpatient visits and 27% were hospitalized with an average stay of 6 days. Another study from Spain in 1989 estimated a mean economic cost of $1825 for human brucellosis patients when converted under 2017 exchange rates.
102
The frequency of hospitalization may also be lower than the 1997 estimate as other studies have reported only 28% of brucellosis patients were hospitalized in Turkey
103
whereas 50% of brucellosis patients were hospitalized in a study in Mongolia.
104
As mentioned previously, these publications are for patients with clinical brucellosis which would be anticipated to require more medical care than humans with possible Brucella exposure.
Frequency of Laboratory-acquired Infection
Although brucellosis is frequently reported as the most common laboratory-acquired bacterial infection (LAI) worldwide,
105
this is perceived by some as an indication of bioweapon suitability as evidence of ease of transmission and communicability. It is important to recognize that many reports of laboratory exposures are several decades old and do not reflect current biosafety principles, work practices, and facility design in diagnostic or research laboratories. In contrast to other potential bioweapon agents, it must be recognized that regulatory programs for livestock in many countries, and the high numbers of new human cases occurring worldwide (>500 000 annually) result in greater numbers of samples being submitted and processed annually for diagnostic procedures for brucellosis detection as compared to most other zoonotic pathogens. This results in a much higher potential for laboratory exposure than most other zoonotic pathogens. As culture isolation is the “gold-standard” for diagnosis, clinical laboratories within endemic areas are likely to process millions of diagnostic samples for microbiological isolation and identification of Brucella spp. on an annual basis. Brucella spp. are relatively slow-growing, and diagnostic samples are frequently incubated and observed for at least 7 days during isolation procedures, further increasing the potential exposure of laboratory workers. With implementation and proper use of appropriate biosafety equipment in appropriately designed facilities, laboratory infections with brucellosis have become relatively rare even though the pathogen is still considered by many to be a high-risk for LAI. It must also be considered that laboratory accidents/exposures to Brucella spp. that do not result in human infection or seroconversion are frequently not submitted for publication.
A brief review of some of the publications related to LAI supports the hypothesis that Brucella spp. are not greatly more infectious than other risk group 3 pathogens. The earliest report we were able to identify of a Brucella LAI was a survey in 1941,
106
although a report of 45 cases of brucellosis at the Michigan State College in 1938 that primarily involved nonlaboratory personnel was also identified.
107
In a review of LAI in 1979, brucellosis was identified as a pathogen of special concern and the authors reported that bacterial and rickettsial infections had shown a marked decrease since 1955.
108
It is reasonable to assume that LAI are probably influenced not only by the pathogenicity of the pathogen itself, but also by the frequency of the pathogen being recovered or present within submitted samples. For example, LAI in British clinical laboratories in 1982-1983 included 13 cases of hepatitis (type B or non-A, non-B), 9 cases of tuberculosis, 4 cases of Shigella, and 1 case of brucellosis (B. melitensis) demonstrating that Brucella spp. are just one of many LAI that can and do occur.
109
Although improvements in biosafety equipment and practices over the last 20 years have contributed to a reduction in LAI, a small nonrandomized survey of clinical microbiology laboratory directors estimated a much higher risk for a LAI with Brucella than other bacterial and fungal zoonotic agents.
110
A review of laboratory-acquired brucellosis in 2013 found that 11% of exposures were due to laboratory accidents, 88% were due to aerosols generated during routine identification activities, and 2% were unknown.
62
This review made a number of important observations: (1) 91% of infected workers reported manipulation of Brucella isolates outside of a biosafety cabinet; (2) 80% of laboratory-acquired infections were caused by B. melitensis; (3) median incubation time for onset of symptoms after laboratory exposure was 8 weeks; and (4) none of the subjects who took postexposure antibiotic treatment developed infection. The implication of B. melitensis as the predominant causative agent is consistent with most reports of laboratory-acquired brucellosis.111,112 With a 95% confidence interval of 3% to 38.6% for estimating the likelihood that exposed individuals would become infected, the attack rate in the review appears to be higher than the 1.3% of exposed workers infected after open bench work with a B. melitensis culture,
113
the 3.8% attack rate after laboratory exposure to an aerosol of B. melitensis,
112
or infection in 9% of laboratory workers exposed to B. abortus after failure of a biological safety cabinet.
114
In an endemic country (Turkey), hospital laboratory workers reported an attack rate of 4.8% with risk factors identified as inadequate lab facilities, failing to work in a biosafety cabinet, and not wearing personal protective equipment.
115
Others have found a high correlation between frequency of laboratory infections and high numbers of samples from endemic areas being handled by laboratory personnel (17 500 samples per year in 1 lab).116,117 It would appear that the impressions that risks surrounding work with Brucella spp. are not borne out by data from actual LAI or their root causes.
It has been postulated that one contributing factor may be a failure of personnel to comply with appropriate safety precautions in clinical laboratories that rarely receive patient samples containing highly pathogenic organisms. This assumption is supported by a survey of laboratory-acquired brucellosis in Spain which found that 80% of Brucella infections were associated with failures in biosafety
118
and also consistent with other reports of LAI with B. melitensis.110,113,119 Currently, laboratory-acquired brucellosis is most frequently associated with clinical laboratories working with human diagnostic samples; it is noteworthy that such laboratories would most likely be exempt from registration under the SAR in the United States. It is also interesting that we were unable to identify reports that identified brucellosis as an occupational risk for veterinary diagnostic laboratory personnel, even in endemic countries or in areas within the United States when there was a high prevalence of disease in livestock.
Although not as commonly reported in the literature, published reports are available in which high-risk laboratory exposure to pathogenic Brucella did not result in human infection. Examples include laboratory work outside of a biological safety cabinet with isolates of B. suis biovar 4 (prophylactic treatment and postexposure monitoring)
120
and B. melitensis (postexposure monitoring only).119,121
It has been pointed out that other zoonotic agents that have very low infectious dosages do not have equivalent numbers of LAI. One frequently cited example is Francisella tularensis, a bacteria with an estimated infectious dose of 25 bacteria for aerosol transmission and 100 bacteria for oral transmission.
122
The literature suggests that diagnosis of tularemia in humans is usually based on serology, as culture is not routinely performed in most clinical laboratories as it is difficult, hazardous, requires certain equipment and containment, and is only successful in less than 10% of patients.123
–126 Tularemia has a more limited distribution than brucellosis as it is not considered to be endemic in Africa, Australia, England, and South America. As there are no regulatory programs to address tularemia in animals, diagnostic samples would not be routinely submitted under control or eradication programs for evaluation. However, aerosol exposure to F. tularensis readily led to human infection with one report from Germany reporting that 10 of 39 people became infected.
127
Although publications on laboratory exposures are not as common as with brucellosis, they may reflect reduced numbers of samples submitted for microbiologic culture and the rare ability to isolate F. tularensis from blood.
126
It should be noted that tularemia was the most common laboratory infection in the biological warfare program averaging 15 per year (1955-1959) until the development of a live vaccine in 1959.
128
Despite the vaccine, tularemia infections continued to average one case per year for the remaining 10 years of the program.
It is also relevant to consider the risk of laboratory exposure to Mycobacterium bovis and M. tuberculosis, bacteria that are not listed as BSAT. Both M. bovis and M. tuberculosis can be transmitted via aerosol, and diagnosis in humans is usually by use of skin testing, staining for acid-fast bacteria, and radiography. There are national eradication programs to eliminate tuberculosis from livestock in the United States and other countries with many programs dependent on skin testing using tuberculin. As the organisms take months to grow under in vitro conditions, culture is not attempted in most diagnostic laboratories. However, M. bovis and M. tuberculosis are known to have a very low infectious dosage (6 to 10 bacilli). Because tuberculosis is more widespread in human populations, and human-to-human transmission is possible, confirmation of laboratory-acquired infections can be more complicated. However, there are a number of reports documenting laboratory-acquired infections109,129
–131 despite more limited use of microbiologic techniques in diagnosis.
Impact on Research Activities
Although inclusion of an agent on the BSAT list has been linked to a negative impact on research activities in part by increased security and regulatory requirements,132
–135 the specific impact of BSAT listing can be difficult to quantify and fully characterize. In 2007, 1356 Biosafety Level 3 (BSL-3) laboratories were registered with the Centers for Disease Control and Prevention (CDC) or Animal and Plant Health Inspection Service for work with BSAT.
132
It has been reported that in January 2015, there were 347 entities registered with the CDC or the APHIS Select Agent Program, and that between 2004 and 2008, 92 facilities and 947 high-containment laboratories were added to registrations.
3
However, most of these facilities are not capable of addressing the research needs for brucellosis as related to vaccine development in livestock or wildlife reservoirs. One report evaluating the impact of the USA Patriot Act and the 2002 Bioterrorism Preparedness Act on research related to Bacillus anthracis and Ebola virus found an increased turnover rate of authors in the select agent research community and an estimated 2- to 5-fold increase in the cost of doing research.
133
They also found that BSAT agent work with live pathogens increasingly concentrated within military laboratories between the years 2002 and 2010.
133
A 2009 report recognized the significant financial burden of security compliance for individual entities and recommended that federal agencies provide dedicated funding for research with BSAT beyond standard indirect costs.
134
Although this may have occurred in some areas of research, it does not appear to have occurred in funding for studies with BSAT related to agricultural research. This report also documented that ongoing security and safety sustainment costs for BSAT research far exceed the indirect costs that grantee institutions receive to cover the costs of facilities, maintenance and operations.
There are increased costs for physical and information technology (IT) security related to registration for work with BSAT. Reviews have indicated weaknesses in IT and physical security, and those requirements have subsequently increased, particularly in recent years, to include suggestions that key cards be integrated with intrusion detection systems for security systems.
3
Other surveys have found concerns related to monetary and time costs of security upgrades and procedures, bureaucratic procedures, inventory records, and impact on international collaboration.136,137 The Federal Information Security Management Act of 2002 mandated use of a risk-based approach to information security.
3
However, an Office of the Inspector General review suggests that federal agencies are still struggling to implement this legislation
3
and that information security for BSAT registered facilities does not meet the requirements.
3
Overall, it is common knowledge that SAR is associated with increased indirect costs related to security, personnel screening, and compliance with regulations. The impact of these costs most likely varies by facility, and is dependent upon the basic infrastructure that was in place prior to registration as well as the cost of security improvements required for meeting regulatory requirements. The proportion of base funding required to maintain current requirements, or meet new initiatives, will likely be the tipping point which determines whether a laboratory maintains research programs that include BSAT.
Acquisition of Virulent Strains for Malicious Purposes
Assuming that an attack was planned by an individual rather than a nation state, security personnel postulated that an individual seeking to obtain virulent strains of Brucella for malicious use would most likely obtain them from collections within registered laboratories. Although that option is certainly a possibility, it is based on the mistaken assumption that these agents are rare or difficult to obtain. This assumption stems from much of the formal, professional security work that was conducted prior to the Amerithrax attacks in the nuclear industry, which is based on a rare material in finite and difficult-to-obtain quantities. These assumptions are untrue when considering biological agents. The endemic nature of brucellosis in many developing countries and the high prevalence of B. abortus and B. suis in wildlife reservoirs within the United States means the agent is easily obtainable from nature, and their replicative nature means they do not exist in finite quantity. Collecting isolates from natural reservoir hosts would be associated with minimal risk of detection or scrutiny. As basic microbiologic infrastructure would be required to expand cultures after receipt, it seems logical that such equipment could be as readily used for initial isolation of Brucella from natural samples obtained legally or nefariously within the US. Maintaining listing of Brucella as a BSAT does facilitate prosecution as possession alone (without registration) could result in legal charges, whereas without inclusion on the BSAT list, malicious intent would have to be proven within the judicial system. Although an adequate level of security for Brucella isolates is justifiable, defining Brucella as a BSAT does not appear to be adequate for preventing acquisition of strains by individuals intending to use them for harmful purposes.135,138,139
Most clinical isolations of Brucella in the United States occur within human or veterinary clinical laboratories that are not select agent registered facilities. SARs do specify legal requirements for reporting, tracking, and disposal of Brucella isolates after identification, but it is likely that security in clinical laboratories is most likely reduced as compared to registered facilities. Retention or removal of Brucella from the BSAT list is unlikely to impact clinical laboratories.
It should also be emphasized that the 5th edition of the Biosafety in Microbiological and biomedical laboratories (BMBL) as published by the CDC includes recommendations for work with risk group 3 organisms, such as Brucella, and states that BSL-3 facilities are required for in vivo and in vitro research studies. Recommendations within the BMBL are utilized as baseline guidance by government agencies, institutional biosafety committees, institution safety committees, and/or funding agencies due to liability and safety risks associated with divergence from guidelines and best practices. Because of liability issues, it seems unlikely that work with virulent Brucella would be permitted outside of BSL-3 containment in academic or research facilities. Therefore, it is unlikely that public health would be compromised by research activities with Brucella within these research facilities.
