Intelligence, Detection, Surveillance, and Diagnosis
A comprehensive approach to coping with bioterrorism must incorporate efforts to prevent the proliferation of biological weapons; methods for detecting covert biological weapons programs; strategies for deterring their use if biological weapons do proliferate; and mechanisms for protecting civilian and military populations if deterrence fails. The emphasis in this multitiered approach should be on defense, simply because the proliferation of biological weapons is difficult to control (biotechnology equipment and expertise are now available globally), covert biological weapons programs (e.g., those of the former Soviet Union and Iraq) are difficult to detect, and deterrence will likely be less effective against suicidal terrorist groups than against states. Consequently, in addition to improving intelligence and information management, the S&T community should be focused on improving defenses against biological weapons. The means to do so include environmental detection of biological agents together with preclinical, clinical, and agricultural surveillance and diagnosis.
INTELLIGENCE AND INFORMATION MANAGEMENT
Increased awareness in the S&T community could reduce the inadvertent spread of knowledge that may aid terrorists, although there is a fine balance that must be achieved so as to not quash legitimate exchange of scientific information. Voluntary international and national efforts to share biotechnology information could improve security and safety in the handling, storage, and transport of sensitive biological material and equipment. Information technology could help monitor international trafficking in biotechnology products.
Detection of covert programs will involve technical intelligence (e.g., remote
sensing and environmental sampling) as well as human intelligence, which has special importance because it can distinguish the benevolent use of biotechnology from the malevolent. Understanding intent in the area of biotechnology, which requires familiarity with S&T culture, processes, and procedures, is an expertise that scientists and technologists can offer the intelligence community. Meanwhile, there is a need to teach, reinforce, and strengthen ethical standards of the S&T community against the production and use of biological weapons; this will reduce the likelihood of scientists working in covert programs and increase the chance of them helping to abort malevolent efforts.
Although much has been written about the potential efficacy (or inefficacy) of ways to deter biological attacks, the S&T community has yet to fully explore means for strengthening deterrence. An obvious option is biological forensics (discussed later), because without reliable attribution, most deterrence strategies are likely to fail. Nucleic acid sequence databases for pathogen strain types and advances in chemical-trace analysis and the use of taggants will help the process of attribution, thus discouraging terrorism, but they will by no means guarantee that perpetrators can be identified.
The greatest potential benefit of a counterterrorism strategy might derive from preemptive efforts at earlier points in the bioterrorism-attack timeline—that is, the evolution of a bioweapons program from inception through weapon deployment, before any biological agent is released. The S&T communities have had relatively little input into detection and characterization of terrorist activities during this early stage, yet they could offer significant untapped resources. Opportunities for their involvement in the area of human intelligence should be explored (see Box 2.1).
Recommendation 1: All agencies with responsibility for homeland security should work together to establish stronger and more meaningful working ties between the intelligence, S&T, and public health communities.
IDENTIFICATION OF BIOLOGICAL AGENTS IN THE ENVIRONMENT
At the present time, efforts to identify biological agents in air, soil, and water samples have had only limited success. Ideally, one would hope to be able to collect air samples, for example, and identify a pathogen in those samples in near real time, allowing the population to be warned of the pathogen’s presence. However, existing technologies for rapid and reliable detection (collection and identification) of bioagents have not been widely evaluated or well validated in real-world settings. Much greater attention must therefore be given to the transition between basic laboratory research and field application.
Traditional laboratory approaches include microbial cultivation, immunological (e.g., antibody-based) assays, and nucleic acid detection schemes, espe-
cially amplification methods such as the polymerase chain reaction (PCR). The last two approaches seek molecular evidence of agent components, such as characteristic immunological markers and genome sequences. A fourth broad approach relies upon the response of a surrogate host—such as cultivated cells from humans, animals, or plants.
Each of the four approaches has its advantages and disadvantages. It is important to note, however, that even though cultivation is slow, limited in scope (by ignorance of appropriate growth conditions in the test tube and in human tissues for many pathogens), and the least technologically sophisticated approach, it provides the most ready assessment of complex microbial phenotypes (behaviors), such as drug resistance. It also is the most widely used approach in laboratories throughout the world, especially in developing nations, and hence is currently the most common identification method for international surveillance.
A number of challenges must be addressed in order to develop and implement effective methods of environmental identification. An improved understanding of natural background is needed, regarding both the agent (including genetic, antigenic, geographical, and temporal variations) and the setting (including related agents and inhibitors). Additionally, standards must be established by which sampling and detection methods can be rigorously evaluated, validated, and standardized (see Recommendation 16 and surrounding discussions). Cen-
tralized repositories of diverse, high-affinity binding and detection reagents (e.g., antibodies, peptides, oligonucleotides) should be established, as well as repositories of genomic material and control samples. There are dozens of ways to identify bioterrorism agents that are sensitive and accurate. However, agreement on how a few well-developed platforms are implemented would allow the data to be broadly understood and make the limitations of the test used apparent to all. For example, whether one is identifying anthrax on the farm, from the environment, or in a patient’s blood stream, the identification can be quickly made using a fairly easily agreed upon set of standard genomic and immunological reagents. Subsequently, there must be cultures of microorganisms grown in the laboratory using agreed upon standard methods. The identification should be based on uniform standards and not a free-for-all depending on program officers or agencies with differing views.
To date, a disproportionate amount of the effort in the bioagent detection arena has been focused on the development of technology platforms. Efforts on standardization or validation of sample collection and sample processing procedures, as well as on test validation in a real-world setting, have had much lower priority. But the use of genomic and proteomic information, as well as the development of robotic sensing devices that can communicate signals from many environmental sites, offers new possibilities for the early detection of biologic agents in the environment. It also increases the risk of false alarms when sophisticated analysis and decision-making systems are lacking.
Another challenge involves creating broad-spectrum detection tools and methods. Currently a large number of tests rely on a small number of specific antibodies or microbial genomic sequences. This reliance creates vulnerabilities—for example, with respect to bioagents having modified antibody epitopes (binding sites) or sequences. Rather than relying on methods that target specific, known organisms, one would like to have detection methods that target groups of organisms (i.e., all members of these groups) and that can identify specific members of the group, including recognition of those that may not yet have been characterized. Although there are experimental challenges, the expertise exists to immediately begin addressing these problems (Cummings 2000, 2002; Nikkari et al., 2002).
A further challenge is the need for highly sensitive systems, as some highly infectious pathogens require the inhalation of only 1 to 10 organisms to cause disease. In general, much greater attention is needed to translate basic laboratory research into field applications and clinical validation (standards will play an important role; see Recommendation 16 and surrounding discussion). Finally, because no test is perfect, it is important to be able to anticipate false-positive test results in a reliable and quantitative fashion. One potential strategy for minimizing the impact of false-positive test results is to create a system of multiple, parallel, independent technical platforms so as to avoid dependence on any one testing procedure. This requires crosscutting, interdisciplinary science (e.g., com-
bining environmental microbiology, cell biology, biophysics, electronics, materials science and microfabrication, microfluidics, and bioinformatics/statistics) and would require collaboration between several federal agencies and industry. However, even the currently available tests could be made significantly more useful by adopting a quality assurance index that would be applied to any positive test result. For example, single positives in tests with high false-positive rates, such as ELISA, would receive a low ranking, whereas successful culture of a known biological agent from a sample would receive the highest ranking. Informed decisions on public action could be made based on the quality of the result rather than simply on the presence of a positive result.
Recommendation 2: Federal agencies should work cooperatively and in collaboration with industry to develop and evaluate rapid, sensitive, and specific early-detection technologies.
The types of identification systems needed are likely to be developed by industry, not in an academic laboratory. Federal funding agencies can speed this process by supporting the early stages of the work. The same kind of milestones should be applied to this kind of work as are used in industry to ensure that the technology is valid and meets the expected specifications. There is a role for the mobilization of established detection procedures and for those that might be second-generation detecting devices sometime in the future. The immediate need is acute and very attainable.
SURVEILLANCE AND DIAGNOSIS OF INFECTION AND DISEASE
Early diagnosis of patients infected with potential biological warfare (BW) agents is complicated by the lack of relevant medical experience with most of these agents in the United States and by the nonspecific symptoms of their associated diseases (e.g., many cause flulike symptoms in the early stages). Systems for effective surveillance and diagnosis of biothreat agents, as well as of many naturally occurring and emerging pathogens, are either unavailable at present or inadequate.
Many of the current challenges in surveillance and diagnosis are quite similar to those described above for identification of pathogens. Surveillance and diagnosis must also address the important distinction between infection and disease—that is, between the colonization or contamination of a host with a potential biothreat agent and the actual manifestation of pathology (disease). Sensitive and specific diagnostic tests are important adjuncts to clinical diagnosis; however, such tests cannot substitute for astute clinical recognition of symptoms to raise the suspicion of a particular diagnosis. Equally vital is the role of classical epidemiological analysis in assessment and recognition of human- and animal-disease patterns.
Preclinical Surveillance and Diagnosis
It would be critical, in the event of a biothreat agent attack, to be able to recognize or identify infected persons, animals, or plants before they develop overt disease. Great benefit could be achieved by rapid intervention in those persons, animals, or plants known to be infected, while avoiding unnecessary intervention in those who are not. It is at this stage that the difficulties and challenges of diagnosis are greatest as well. In recent years, novel biotechnological and biological approaches have opened up new opportunities in this area.
In the interim, while new approaches are developed and refined, assessment of white blood count, fever, and relatively simple observations will remain the first line of defense in protecting human health. A primary focus of diagnostic strategy will continue to be the continuing education of physicians and healthcare workers.
An example of a plausible new technological approach is the host-genome-wide gene-expression profile. The availability of a nearly complete human-genome sequence and the power of DNA microarray technology have been harnessed to create an approach for surveying the responses of nearly all known human genes to various infectious agents. Cells are programmed to recognize pathogenic agents and foreign life forms, and they respond with changes in host-gene expression; microbial agents, meanwhile, have evolved strategies for manipulating and subverting these programmed responses. The result is an intricate, choreographed, and time-dependent set of induced and repressed gene-expression patterns that can be detected in small blood samples (Cummings and Relman, 2000).
Although the dominant features of these patterns are common to virtually all infections, regardless of the particular infectious agent, other features may be more specific to the agent or disease. With further research and refinement, one might actually be able to distinguish infections by different pathogens and generate signatures that allow early identification. These patterns reflect how the host “sees” the pathogen, and they also reflect (and perhaps predict) the outcome of the host-pathogen interaction. Research exploring the potential usefulness of this approach is still in its early phases, however.
Host-gene expression patterns are just one complex biological pattern that might lend itself to this kind of diagnostic and prognostic approach. Others include patterns of secreted proteins in host fluids, volatile compounds in breath (analyzed, for example, with mass spectroscopy), and spectral features of host cells and fluids (studied using spectrometers and hyperspectral analysis). The enormous advantage of such technology, should it be able to fulfill researchers’ expectations, is that it could distinguish genuine infection from hysteria or terror, either at the emergency room or in the clinic.
Human Disease Surveillance and Diagnosis
In this country and elsewhere, the recognition of almost all emerging infectious diseases—both naturally occurring and intentional—has depended on an astute clinician contacting a public health agency after suspecting an unusual serious illness (e.g., hantavirus in the Southwest or anthrax in Florida). This traditional system of notifiable human disease surveillance depends on the training of physicians and other health care providers, in terms of both disease awareness and their responsibilities to public health. In addition, the important systems linking hospitals around the country with CDC, known as sentinel surveillance systems, need to be enhanced; they can establish whether a common cause of disease is being seen simultaneously in multiple regions. Research should be conducted on the strategies likely to be most useful in enhancing the notifiable human disease reporting system for the broad range of potential threat agents (strategies such as education, animal sentinels, changes to the surveillance systems, and the use of infection control specialists). Mathematical models of disease transmission and distribution using simulations of a covert release of various agents could be helpful in assessing the potential and relative value of different surveillance systems. An integrated national system that can report diseases electronically in real time is needed to support these networks. Information technology advances should be explored both to automate required reporting (e.g., laboratory reporting of pathogens) and to develop new surveillance tools (e.g., the automated scanning of electronic media, such as that utilized by the Global Public Health Information Network).
Systems of syndrome surveillance—that is, screening for changes in the frequency of cases of flulike illness seen in hospital emergency rooms across a city or town—should be developed to identify outbreak patterns. Relevant computer programs are being developed, but there are known fluctuations in emergency room admissions from season to season and day to day, and it will be important to determine their potential predictive value, specificity, and usefulness. Syndrome surveillance has allowed early recognition of some respiratory and diarrheal disease outbreaks, but it is not clear whether it will be useful for early detection of key threat agents such as smallpox, anthrax, and tularemia.
Because infectious diseases do not respect national borders, international cooperation is vital in the sharing of epidemiological and clinical data, both on emerging infectious diseases and on outbreaks caused by potential bioterror agents. A global network for surveillance of infectious diseases in humans and animals would be strengthened by augmenting the numbers and capabilities of U.S. overseas laboratories and by providing enhanced support for current initiatives on international surveillance (e.g., DOD’s Global Emerging Infectious Diseases program and corresponding Department of Health and Human Services (HHS) initiatives).
Increased support for the development and expansion of public health and
agricultural laboratories in other countries, particularly in their capacity to diagnose threat agents, would yield dividends for recipient and donor alike. This means that CDC and other agencies must reach out to educate, train, and collaborate with scientists from many countries on aspects of surveillance and identification of threats. The World Health Organization could play a critical role in building and strengthening international capabilities.
Recommendation 3: Create a global network for detection and surveillance, making use of computerized methods for real-time reporting and analysis to rapidly detect new patterns of disease locally, nationally, and—ultimately— internationally. The use of high-throughput methodologies that are being increasingly utilized in modern biological research should be an important component of this expanded and highly automated surveillance strategy.
Another important area for applied research is the development of improved clinical diagnostics—rapid assays for the detection of common pathogens and BW agents—that could be used in primary care settings as well as referral laboratories. In addition, the kinds of needs that were described above for preclinical detection also apply to the field of clinical diagnostics. Standards are needed by which diagnostic methods and technology can be rigorously evaluated and validated, and centralized repositories of standardized reagents and samples are needed as well. Because the development and evaluation of diagnostics require interdisciplinary applied research, it is currently difficult to find targeted sources of support for these efforts. NIAID, CDC, and USDA should consider providing extramural funding programs to stimulate research in this area.
Because of the low likelihood of infections with BW agents compared to common, widely circulating agents like influenza viruses, routine application of rapid diagnostics for potential BW agents in a primary care setting in the absence of clinical suspicion will face problems with false-positive and false-negative results, for which rapid adjunctive standards do not exist. A triage system could be applied in which patients with relevant symptoms who test negative for a panel of expected pathogens would be sent to a referral laboratory for a second round of diagnostic tests, which could include suspected BW agents and broad-range methods.
High-throughput automated laboratory technology can now be applied to assist in these efforts. Positive samples could be forwarded to central public health laboratories for more comprehensive characterization. A laboratory designed, for example, to address influenza surveillance (Layne et al., 2001) could be dual use: Not only would it enhance public health by providing more accurate and timely information about the emergence of novel influenza strains, but it could also provide surge capacity to detect other agents if outbreaks occurred as a result of a terrorist attack. Continued development of effective networks of such referral laboratories (private, academic, local, state, and federal) is thus vital.
It should be noted that the first suspicion of the outbreaks of anthrax and of
West Nile virus came not from sophisticated computer technology but from thoughtful and perceptive physicians. Tools to help all health professionals make the appropriate inferences from small numbers of patients must be developed so that the likelihood of missing a new outbreak is markedly reduced. Principal responsibility for this work should rest with CDC, NIH, and DOD.
Recommendation 4: Use knowledge of complex biological patterns and high-throughput laboratory automation to classify and diagnose infections in patients in primary care settings.
Agricultural Surveillance and Diagnosis
The protection of the nation’s food supply presents several unique challenges related to surveillance and diagnosis of disease. The U.S. livestock industry, with revenues of approximately $150 billion annually, is extremely vulnerable to a host of highly infectious and often contagious biological agents (insects and other pests, viruses, and microbes) that have been eradicated from the United States. Unlike traditional biological agents that can be used against humans, many of these animal-targeted agents need not be weaponized to cause an outbreak. Their simple point-introduction into herds could immediately halt all movement and export of U.S. livestock and livestock products.
Although most agents that affect animals are not human pathogens, introduction of any of the agents on the A List of the World Organisation for Animal Health would have wide-ranging and devastating impacts on the U.S. economy— not to mention psychological effects on the country’s human population—from which it could take years to recover. These disease agents are readily available in many countries. Although USDA’s Animal and Plant Health Inspection Service (APHIS), as currently constituted, has proven adequate for naturally occurring disease, it would probably be unable to help eradicate intentional introduction, especially if this were done at multiple sites. There is a need for USDA to develop a research and surveillance capability for plant and animal diseases comparable to the one that CDC oversees for human diseases.
Animal agriculture would seem to be increasingly vulnerable to intentional biological attacks, given recent trends toward concentration and specialization in the livestock industries (MacDonald et al., 1999). For example, tens of thousands of animals can be housed in relatively close quarters in concentrated feedlots prior to slaughter. If the introduced agent is highly contagious, as is the foot-and-mouth disease virus, this concentration creates the potential for greater impact from a single infected animal, as aerosol transmission of pathogens is common within herds. Likewise, animals move across great geographic distances. For example, during September 2001, nearly a million of the swine imported into Iowa came from 24 states and Canada (communication from the Iowa State Department of Agriculture).
Given these vulnerabilities, there is a need to recognize an infected animal immediately. At present, however, although there are well-operated state and federal animal diagnostic laboratories, there is no integrated national system that can report diseases and infestations electronically in real time. In addition, there are no rapid field diagnostic assays for most animal pathogens and pests.
Crops, too, are vulnerable. They are grown over very large areas (e.g., some 75 million acres for soybeans) and there is very little surveillance or monitoring. Likewise, plant diagnostic laboratories are scattered across the country and are underresourced and understaffed. In addition, great variability exists in the capabilities of these laboratories from state to state. This situation means that a long time could elapse from the introduction of a crop pathogen to its detection. Remote sensing, particularly satellite imagery, may have value in monitoring crops for disease outbreaks, including those resulting from bioterrorism.
Other factors heighten the vulnerability of U.S. crops: (1) many hybrid crop species exhibit low levels of genetic diversity; (2) there are few restrictions on trade, and large volumes of agricultural products are imported and exported each year; (3) a substantial proportion of the seed used for growing U.S. crops is produced in other countries, presenting a possible route for the introduction of dangerous plant pathogens as well as contaminated fertilizers and pesticides; (4) fungi, viruses, and bacteria cause more than 50,000 diseases of plants in the United States; (5) for any given crop, there are several pathogens that are not yet found in the United States but that cause major losses elsewhere; and (6) the biological agents that could affect crops are more numerous than the pathogens that affect humans, making it more difficult to focus the research funding available for efforts to counter agricultural bioterrorism.
Threats to crops intersect with threats to livestock in the case of animal feed, and there is a particular concern about the timing of ultimate effects. The delay between the time at which a bioterrorist contaminates animal feed and the time the human food product becomes adulterated would cause more uncertainty about the source of the contamination and could minimize the possibility of apprehending the terrorist. The less obvious and the more natural the source of biological contamination, the greater the likelihood that the contamination of the animal feed will be mistaken as a natural phenomenon. Rapid testing of feed and separation of contaminated feed are important steps, followed by the more specific identification of the contaminant to determine the source of adulteration and the possibility of decontamination. The development of specific antibodies for the production of sensitive and specific test kits is the key to identifying contamination. This would allow one to deal effectively with the disposal or decontamination of the animal feed and, ultimately, to prevent the contamination of animal-derived human food products (Von Bredow et al., 1999).
Rapid containment of agricultural pathogens is dependent on an effective system for diagnosis and the coordinated action of various state and federal agencies. Although these agencies, including USDA’s APHIS, have dealt suc-
cessfully in the past with the natural introduction of several foreign pathogens of plants and animals, they are not properly organized to deal with the massive, multiple introductions that terrorists are likely to attempt. In essence, the game has changed, and this requires a substantial restructuring of the nation’s agricultural response systems.
Recommendation 5: USDA should create an agency for control and prevention of plant disease. This agency should have the capabilities necessary to deal effectively with biothreats.
For animal disease, USDA operates several laboratories—Plum Island and Ames among them—that perform diagnoses, carry out research, and provide training for veterinarians. CDC is the central agency for the control and prevention of communicable human disease, but no center currently exists to serve the same function for plant disease. Such a center is desperately needed.2 Departments of plant pathology at various state universities, APHIS, and a wide variety of other agencies, all of which often depend on outside experts, currently deal with new and unusual plant pathogens as best they can.
A major research, development, and training center is called for that would address fungal, bacterial, and viral diseases of plants. Programs would focus on genomics and proteomics, databasing and informatics, forensics, pathogenesis, host-parasite interactions, diagnostics, sensors, food safety, analytical methods, epidemiology, modeling of disease outbreaks, intervention, and management. Other efforts could include outreach, technology transfer, collections of pathogens, and epidemiological intelligence and response. Close linkages could be established with other federal and state agencies, as well as with academic institutions, international agencies with responsibilities for surveillance of plant diseases and bioterrorism, and industrial, extension, and professional organizations. These collaborators could, among other functions, provide advice on containment and control procedures.
A similar recommendation was made in February 2002 by the American Phytopathological Society. The white paper “American Phytopathological Society: The First Line of Defense—Biosecurity Issues Affecting Agricultural Crops and Communities: Genomics, Biotechnology, and Infrastructure” is available for review at <http://www.apsnet.org/media/ps/BiosecurityWhitepaper2-02.pdf>.