There is growing concern among policymakers and the public about bioterrorism and the emergence of new pathogens. The proposed and imagined scenarios are too numerous to count and raise an impossibly large number of policy issues. An important role of science is to bring some order to these scenarios and issues, both by collecting and analyzing relevant data and by developing models to inform decisions.
In December 2001, the Fogarty International Center (FIC) held a consultation on the utility of modeling to prepare for responses to bioterrorism. This and subsequent meetings made clear that scientists are inadequately prepared to provide good scientific models that might inform policymakers' decisions about how to respond to a bioterrorist threat or an emerging epidemic.
The National Institute of Allergy and Infectious Diseases (NIAID) has a well-developed plan for biodefense research that includes six target areas: biology of the microbe, host response, vaccines, therapeutics, diagnostics, and research resources. The plan does not include computational or mathematical modeling.
Because of its history of supporting modeling and complex systems, the National Institute of General Medical Sciences (NIGMS) has been working closely and collaboratively with FIC and NIAID to develop a modeling component to the biodefense research agenda. Much NIGMS-supported research has benefited from computational and mathematical approaches because these approaches have provided methods for intelligently integrating large amounts of data and for better designing new experiments. Computational and mathematical models may clarify how parts of a system work together and suggest where there is missing information. Models may also suggest the consequences of perturbing a system and point out where a system is vulnerable.
On August 5, 2002, a working group of experts met in Bethesda, MD, to discuss the field of modeling of the emergence and intentional release of infectious diseases. The following summarizes much of the discussion. The experts agreed on three important points:
Consistent with the approach of the NIAID, bioterrorism is a variant of the general problem of emerging infectious diseases. The fact that the mode of transmission or the properties of the microbe may result from a deliberate act does not change the fundamental research approach.
Modeling of infectious diseases must take a broad systems approach. The best models will take into account microbial biology and natural history, host response to infection, ecological and evolutionary information, and the dynamics of intervention strategies (e.g., vaccines, antimicrobials, or antivirals).
Emerging infectious diseases typically come from species-jumping agents or zoonoses. It is important to look beyond current human pathogens at the larger systems from which future pathogens might come.
Models of the Microbe
Genomic and proteomic tools are producing unparalled amounts of data on the basic biology of microbes. It is now feasible to gather information on genetic variation, recombination, and mutation rates for many microbes. Taken with information on natural history, physiology, and molecular biology, computational and mathematical approaches could provide insights into the mechanism of host switching and the number of genes a pathogen could incorporate (its “genetic universe”). One could also determine which microbial genes are under selection and are therefore potential targets for interventions.
Models of Within-Host Dynamics
What happens during an infection comes down to the processes responsible for the population dynamics and evolution of microbes and those of host defense cells. Models of within-host dynamics have been very useful in HIV research and treatment; however, very little modeling has been done with other parasites. We would like to know a great deal more about how parasites interact with other microbes within a host, as well as the dynamics of recombination in the parasite and immune response in the host.
Models are also needed to suggest the impact of various intervention strategies, treatment options, and prevention measures. For example, who should be vaccinated to prevent a flu epidemic? What are effective antimicrobial treatment protocols? In what scenarios would targeted rather than broad-spectrum therapies be desirable? What is the impact of using antibiotics prophylactically?
In conjunction with laboratory and animal experiments, models may suggest promising therapies that would be difficult to test in humans. For example, serum therapy and phage therapy are targeted against specific pathogens and may have a place in the repertoire of interventions.
Models of Evolution and Ecology
Failure to understand the evolution and ecology of parasites is, in itself, a threat. The increase in antibiotic resistance is an example of our failure to understand how rapidly bacteria evolve and how large their genetic universe is. We need to take account of the natural histories of microbes--where they are naturally found, what other organisms they live with, and what they do. We need to build into models the ways that populations are structured, both spatially and genetically. Models also need to account for how selection changes the population size and structure. For example, when a
Borrelia infection is treated with an antibiotic, are all the microbes killed or are some sequestered? If some are sequestered, are they different genetically?
Host shifts are particularly relevant to the science of emergence and can be extended to the study of engineered pathogens. Under what conditions can a microbe move to a new host? Can one predict if it will cause disease? Can one generate human pathogens without human experimentation?
Whether we are talking about the next epidemic or a bioterrorist, we do not have tools at our disposal to rapidly build scenarios, model possible interventions, and provide useful information to policymakers. Often, our best information on how to respond to an outbreak is based on historical data that may not be terribly relevant to a particular situation. We can study the unique characteristics of bioterrorism (e.g., what happens if an infected person deliberately tries to infect as many people as possible, or what happens if there are sequential releases of different pathogens?) using a modeling approach. Models may help us identify threats and vulnerabilities in advance, give us the ability to study “worst-case” scenarios, and help us find good strategies for responding.
During an epidemic or attack, models can also be extremely useful. Models that utilize real-time data and that provide policymakers with current information have proven valuable. For example, Great Britain’s experience with foot-and-mouth disease suggests that several independent modeling groups that share data and information can be an invaluable resource in an emergency.
Improved Statistics and Models
New statistical tools are needed for the design of integrated studies and for analyses of large data sets. We need more research on the design and conduct of epidemiologic studies, with development of the appropriate statistical methods for their analysis. New statistical methods for syndromic surveillance are also needed (e.g., how effectively would surveillance identify an unusual outbreak of influenza?). Infectious disease outbreaks may be influenced by both biological and non-biological factors that could interact in a mind-numbing number of ways.
It is critical that these models work well; thus, a related research activity is studying, comparing, and validating existing models. We need research on how to choose the right parameters for complex models and on how to select dynamic models that will be informative in specific situations. Adding stochastic events such as weather to models is a fertile area of research.
Over the next few years, NIAID biodefense research will expand the amount of data and knowledge considerably. The following scope for a new initiative would complement the NIAID plan and make good use of NIGMS’ expertise in computational and mathematical biology:
These goals suggest two components, one targeting individual research scientists and one focusing on larger collaborative centers. The centers component would have the capacity to respond quickly to national needs by developing scenarios and building models to inform policymakers. These models would take into account the critical information that policymakers need and consider response options. Models such as these may prove their utility if they illustrate critical trade-offs among options, point out vulnerabilities, and suggest consequences of various decisions. At best, they would be one component of many on which final decisions are based.
Centers should also have well-developed training and outreach components. Recruiting and training new scientists is a high priority, and dissemination of information to the public is, likewise, critical.
Simon Levin, Ph.D., ChairmanPrinceton UniversityPrinceton, NJ 08544
Elizabeth Halloran, Ph.D.Department of BiostatisticsEmory UniversityAtlanta, GA 30322
Ron Atlas, Ph.D.University of LouisvilleLouisville, KY 40208
Edward Kaplan, Ph.D.Yale School of ManagementNew Haven, CT 06520-8200
Martin Blaser, Ph.D.New York University School of MedicineNew York NY 10016
Tom Kepler, Ph.D.Duke University Medical CenterDurham, NC 27708
Steven Block, Ph.D.Stanford UniversityStanford, CA 94305-5020
Bruce Levin, Ph.D.Emory UniversityAtlanta, GA 30322
Greg Dwyer, Ph.D.University of ChicagoChicago, IL 60637
Margaret Riley, Ph.D.Yale UniversityOsborn Memorial LaboratoriesNew Haven, CT 06520-8106
Tom Inglesby, M.D.Bloomberg School of Public HealthJohns Hopkins UniversityBaltimore, MD 21205
Staff:Irene Anne Eckstrand, Ph.D.National Institute of General Medical SciencesBethesda, MD 20892Telephone: (301) 594-0943Email: firstname.lastname@example.org
Connect With Us: