Sampling and Preservation Methods
THE BROAD PERSPECTIVE
Smith (2011) references a Department of Homeland security meeting in 2005 from which a report on sample collection, handling, and preservation was issued. The authors of this report stated that the collection and preservation of vital microbial forensic evidence is a critical element of successful investigation and ultimate attribution subsequent to a biological event.
A primary goal of collection is to obtain sufficient biological agent to support both species/strain or toxin identification for critical public health decisions and complete signature characterization for valuable lead information. Also, the collection of other relevant traditional forensic evidence must not be overlooked. Trace evidence, fingerprints, and other traditional evidence should be collected and preserved in order to support the attribution mission. (Smith, 2011:380)
This chapter explores the principles and practices that guide the collection, handling, and preservation of microbial forensic evidence.
Mr. Adam Hamilton, President and CEO of Signature Science LLC, gave an overview of sampling and preservation, reviewing potential applications of microbial forensics both inside and outside of law enforcement. He also discussed the challenges of collecting and preserving immensely diverse evidence samples for multiple purposes, and made suggestions for specific areas of opportunity for international collaboration to advance the discipline.
In all cases, the bottom line in sampling is that samples must be collected and preserved in a manner that prevents or minimizes degradation or contamination. This requirement makes sampling and preservation as important to the microbial forensic process as is scientific analysis. The key messages of his presentation were that microbial forensics is a multidisciplinary field developed to serve law enforcement needs and provide criminal attribution. However, although it was initially motivated by biocrimes and bioterrorism, there are many new applications for microbial forensics that create both opportunities and challenges. Expanding roles for microbial forensics also provide opportunities for leadership on the development of international guidelines for sampling and handling practices. “Standardized flexibility” is required to accommodate myriad applications. In addition, while quality control is a necessity, so is transparency. Anyone who is potentially affected by the outcome of microbial forensic analyses should be provided insight into what transpired and how conclusions were derived.
The expansion in potential applications of microbial forensics is being driven by the availability and accessibility of new technologies, such as NGS and bioinformatics. As the cost of using these technologies drops, more communities will use them to address the problems they face, as has already been well illustrated by the public health cases discussed in a previous chapter. Security and surveillance are possible new areas of application for microbial forensics. Microbial forensics could be used to identify strategic locations for collecting monitoring samples and also could inform approaches for monitoring systems, such as those needed in agriculture. In the clinical and public health realm, microbial forensics could play a major role in pathway identification of health care–associated infections. The combination of an increase in multidrug-resistant pathogens, shrinking therapeutic pipelines, and enhanced access to health care has the potential to produce a health crisis. Phylogenetics determined through sequencing and other technologies can complement traditional epidemiology. Microbial forensics also could aid the development of diagnostics and interventions for transboundary diseases, particularly animal and zoonotic diseases.
As with other forensics, the collection and preservation of microbial evidence are critical for efficient and successful investigation and attribution. Moreover, a relatively small number of samples may be the basis for highly significant strategic and policy-level decisions. Evidence sampling approaches must include (1) planning and design; (2) protocols for quality assurance and quality control; (3) logistics and preparation; (4) collection, which is a fairly small part of the overall process; and (5) documentation, which is particularly important for maintaining the chain of custody. Evi-
dence handling must take into account storage conditions, packaging and labeling, shipment/transportation, and maintaining the chain of custody.
Best practices need to be developed for sampling and handling systems to create a viable microbial forensics system. This is well recognized, as exemplified by the 2008 recommendation made to the U.S. Department of Defense that “[t]he broad group of government agencies…should develop and promote best practices for microbial sample collection that best preserve the genetic and epigenetic signatures of interest” (JASON Advisory Panel, 2009). Because the need for microbial forensics is international, there should be internationally accepted practices. Efforts to evaluate and improve collection methods and standards are under way at the U.S. Department of Homeland Security’s National BioForensic Analysis Center1 (NBFAC) and National Biodefense Analysis and Countermeasures Center2 (NBACC), but similar efforts are needed globally in order to advance the science of microbial forensics as a whole.
The goals of sampling and handling are shown in Box 5-1. Hamilton stated that the first goal is to ensure the health and safety of the evidence collectors, any resident populations near the location of an outbreak event, and those who may have already been exposed in a real attack. The design for evidence collection must sufficiently demonstrate the hypothesis or hypotheses that investigators propose. The capabilities of sample analyses must be evaluated with the knowledge that results will be fed not only into a microbial forensics system, but also into other forensic systems. And flexibility must be incorporated throughout, owing to the variety of matrixes and samples that can be collected.
To ensure the broad applicability of preserving and protecting the integrity of samples, the methods that are efficient and low cost are most
1 NBFAC is part of the National Biodefense Analysis and Countermeasures Center (NBACC), which applies science to challenges critical to defending the nation against bioterrorism. The Department of Homeland Security’s Directorate for Science and Technology established the NBACC to be a national resource to understand the scientific basis of the risks posed by biological threats and to attribute their use in bioterrorism or biocrime events. NBFAC’s mission is not primarily research. Rather, NBFAC conducts bioforensic analysis of evidence from a biocrime or terrorist attack to attain a “biological fingerprint” to help investigators identify perpetrators and determine the origin and method of attack. NBFAC is designated by Presidential Directive to be the lead federal facility to conduct and facilitate the technical forensic analysis and interpretation of materials recovered following a biological attack in support of the appropriate lead federal agency (http://www.dhs.gov/national-biodefense-analysis-and-countermeasures-center).
2 NBACC’s National Biological Threat Characterization Center (NBTCC) conducts studies and laboratory experiments to fill in information gaps to better understand current and future biological threats; to assess vulnerabilities and conduct risk assessments; and to determine potential impacts to guide the development of countermeasures such as detectors, drugs, vaccines, and decontamination technologies. Neither NBFAC nor NBTCC has as its primary responsibility to conduct research on microbial forensics.
Sampling and Handling Goals
- Health and safety
- Sufficiency of design
- Analytical compatability
- Preservation and integrity
- Low cost and high efficiency
- Documentation and training
- Quality and transparency
SOURCE: Hamilton presentation, 2013.
desirable. Spending $10 per collection kit for every sample is unrealistic; ideally it should cost pennies. Although compared to the cost of sequencing, $10 per collection kit may not seem unreasonable. However, it is possible that many samples will be collected on a routine basis, and it would be desirable to be more cost-effective by reducing the cost to a few cents. Lower cost collection devices would allow more samples to be collected and preserved for subsequent analyses, if warranted.
Documentation and training in how to conduct sampling missions are needed so that collection efforts can be delegated to people with other responsibilities or who may not be involved in the microbial forensic field at all (e.g., first responders). Cerys Rees of Porton Down in the United Kingdom noted that because first responders are often police or military personnel, results of sampling are much improved if the microbial forensic experts can advise the samplers before they begin to collect evidence. When the scientists are asked in advance for their recommendations for the sampling process, analytical results tend to be better. Also, when samplers are told what the scientists plan to do with the samples in the laboratory, they are better equipped to sample appropriately. Matts Forsman of the Swedish Defense Research Agency also related that in Sweden, first responders may include police, fire, and rescue squad personnel, but a specially trained bomb squad is then called in to sample, pack, and transport samples to the Swedish National Laboratory for Forensic Science.
Some generic sample matrices for evidence collection appear in Box 5-2. Microbial forensics incorporates both clinical and environmental samples, which in turn vary in size, shape, matrix, procedure, and method used to collect them. Portions or the entire sample or item may be collected. For example, the entire house of the Unabomber, Ted Kaczynski,
Sample Matrices for Evidence Collection
SOURCE: Hamilton presentation, 2013.
was moved to a controlled area for evidence analysis. Standardization is needed for handling bulk as well as minute samples. As pointed out by Dana Kadavy, because microbial forensic investigators deal with such a broad range of sample matrices, they should maximize cleanup procedures to enhance typing success.3 A laboratory may need more than one sample processing method, particularly if it receives a variety of samples that reside in different matrices. One technology may be particularly effective for processing clinical samples, for example, but not soil, effluents, or plant material. Automated technologies can reduce labor, minimize hands-on time, and reduce human error in extraction and preparing PCR plates and sequencing libraries. However, an automated technology is not necessarily a superior methodology. Validation studies at each laboratory must determine whether the technology reduces error and/or contamination, affects yields, and/or improves the analysis.
Bruce Budowle also touched upon sampling objectives, which may address elements such as real-time monitoring, screening, random sampling, targeted sampling, collecting bulk material and suspicious items, and the conditions of the crime scene. Each element is accompanied by
3 This is required to remove most of the interfering substances from the matrix and to promote the concentration of the analyte.
a set of criteria that must be considered. Each of the different sampling approaches will require incorporating different considerations when planning the collection process. (For further detail, see Chapter 6.)
Ideally, ready-for-prep samples arrive at the microbial forensics laboratory, are fed into the system, and research and analysis are performed. But in operational terms, collecting samples is complicated. First, the investigator must determine how to optimize the collection of the clinical or environmental samples from the various sources and how to preserve them in an appropriate manner. At the same time, the collection procedures should not disturb other forensic processes. In fact, the goal is to leverage the microbial samples for use for other forensic applications. Finally, the evidence must be preserved so it is available for future analysis, be it for a secondary analytical or validating process, or for analysis one day using a not-yet-invented technique.
Randall Murch added that sampling from a facility or location identified as worthy of investigation but from which law enforcement may wish to gather evidence unobtrusively while building a case can be a challenge. Because they do not wish to draw attention to themselves, investigators may be constrained in how long they are on site, what they carry, including protective gear and sampling equipment, and how many samples they collect. This constraint will affect strategies, design features of sampling kits all the way through the analytics, and conclusions drawn from the effort. How conclusions should be caveated, given the uncertainty that may accompany collection under these circumstances, is a topic that needs further consideration.
Hamilton summed by stating that there is a global need for microbial forensics. Areas of opportunity for international collaboration to advance the discipline include the following:
- Standardized terminology for the discipline;
- Novel solutions (e.g., for sampling from multiple sources and to enable multiple forensic finishes);
- Best practices;
- Validation expectations;
- Documentation standards;
- Quality assurance and quality control (e.g., blanks, proficiency tests, QC samples, spikes)—At what point is each introduced, and in what standardized manner?
- Experimental design—judgmental (targeted) and statistical;
- Information management (e.g., an expert should be able to easily access and analyze a file remotely), capabilities for which should be in place before an event;
- Training and certification—collaboration should be multidisciplinary (e.g., public health, veterinary, food safety, environmental science) as well as multisectoral (academia, industry, law enforcement, and defense); and
- Data records that should accompany a sample (e.g., temporal, spatial, custodial, storage) to aid epidemiological and forensic analysis.
Hamilton noted that there are a number of potential starting points for collaboration. These include the public health, veterinary, agricultural, food safety, environmental, defense, law enforcement, industry, and academic communities; international organizations with accrediting agencies and governing bodies; and sovereign states. He noted that two important questions are
- What standards or guidelines should be used as the starting point for microbial forensics?
- What sample characteristics are most critical to preserve for current and anticipated analytical methods?
Food and Agriculture
The National Research Council report Countering Agricultural Bioterrorism summarized the historical evidence of planned or actual use of biological agents against livestock and crop plants as follows:
Historical evidence suggests that a number of nations have considered using or have even used biological agents against plants and animals. During World War I, German agents infected horses with bacteria that cause glanders, a fatal equine and human disease. The Soviet Union military also used glanders in the early 1980s during war in Afghanistan. In World War II, the United States, Great Britain, and others all had offensive biological-warfare programs directed against plants or animals, and these continued after the war until three years before the adoption of the Biological and Toxic [sic] Weapons Convention of 1972 (BWC). Some countries continued their offensive programs after 1972. Iraqi bioweapon development in the late 1980s and early 1990s included anticrop efforts based on wheat smut, a disease caused by a fungus in the genus Tilletia. There is also evidence that the Soviet Union had an extensive agricultural-terrorism program aimed at animals and plants. (NRC, 2003:18)
Fletcher et al. (2006) also pointed out that agriculture makes an attractive target for terrorism using plant pathogens because it (1) is a
$1 trillion/year sector of the U.S. economy, (2) employs 17 percent of the U.S. workforce, and (3) is not subject to regular surveillance so that there could be long lag times between the deliberate introduction of a plant pathogen and its detection. Given the economic and health-related importance of agriculture, Fletcher et al. advocated for “the integration of the traditional discipline of plant pathology and the specialized field of forensic science.…”
With these thoughts in mind, Dr. Bruce Budowle discussed sampling and preservation in the food and agriculture context. His purpose was to provide a sense of the range of evidence sampling and collection possibilities that a microbial forensic investigation might confront and the outlines of a comprehensive plan for approaching such varied circumstances and challenges. A review of the major issues dealing with plant pathogen forensics can be found in Fletcher et al. (2006), which builds a case for the integration of plant pathology and forensic science. It describes the potential threats to plants and addresses the development of a program in microbial forensics and criminal attribution that addresses crops and other plant targets.
The challenges in microbial forensics might be greater than simply working with an isolate from an individual. The evidence may be a degraded sample on a laboratory floor or in soil or effluent from a clandestine laboratory. In cases where evidence is highly degraded, NGS technology may not work well whereas a real-time PCR assay may work better.
Natural outbreaks in food make good models for microbial forensics. During the passage from “farm to fork” there are many potential targets for investigation and evidence collection. The list in Box 5-3 gives an idea of the range of access for a potential attack. To investigate contaminated sprouts, for example, one may need to go back to the farm or to any of the many places the food has traveled along the way. There have been many examples of deliberate contamination of food: salad-bar tampering in The Dalles, Oregon; cyanide-tainted grapes; Shigella-spiked muffins. Each event comes with its own set of circumstances and questions. In the cyanide-tainted grapes case, for example, only two tainted grapes were ever identified. What happened to the others, or were there others?
Contamination can occur unintentionally through bad planning. A country may use mycoherbicides for biocontrol—to halt cocaine production, for example—but inadvertently cause more far-reaching toxicity. In Australia, a virus was introduced to reduce the population of introduced and invasive rabbits. Terrorists may be interested in such biocontrol agents or how they were effectively administered.
Evidence-gathering tasks can also be diverse. In 2003, the USDA Animal and Plant Health Inspection Service had to determine whether an out-
Potential Food and Agriculture Targets
- Grain elevators
- Water supplies
- Food in grocery stores
- Food and agriculture transportation systems
- Farm workers
- Livestock producers
- Food processors
- Food handlers
- Processing facilities
- Trucks, railroads, ships
- Restaurants, and more
SOURCE: Budowle presentation, 2013.
break of mad cow disease in Canada originated with U.S. cows. Kinship tests were performed on cattle to trace ancestry, and evidence collection included tracking down the hides of parent cattle for DNA trace-back.
Plants tend to make good vehicles for an attack because the perpetrator is unlikely to be at risk when preparing the agent, and plants are poorly protected. Picture the huge areas of unguarded farmland around the world on which it is nearly impossible to impose security. This lack of security (which is impossible to impose) represents both a safety and a forensic challenge. Box 5-4 details some reasons why plants make such an excellent target for agroterrorism.
Various factors must be considered depending on the circumstances when collecting and preserving evidence. For example, in the case of a wheat streak mosaic virus, the mite vector may be implicated. A farm’s wheat may be tainted, but it may not be recognized until 8 months later, after the harvest has been cleared. Does the virus exist underground in roots in the winter? Is it in other reservoirs that are in the vicinity, such as grasses? Sample collection under these circumstances is very difficult, and the difficulties do not end with evidence recovery. If samples are maintained at room temperature, for example, other microbes may take over and mask the one of interest. Under certain conditions, the agent may actually be destroyed.
A general approach to sampling appears in Box 5-5A, and a more focused approach—in this case, for wheat streak mosaic virus—appears in Box 5-5B. When considering approaches to evidence collection and
Plants Are Vulnerable and Efficacious Targets for Agroterrorism
- Many agents are readily available in nature, from low-security laboratories, even from commercial sources, that require little effort or risk to obtain.
- Most agents pose no risk to human health.
- There is less risk in handling and dispersing the pathogen.
- Once released, an agroterrorism event may go unnoticed for days to weeks or longer.
- By that point, it may be nearly impossible to determine if the event was deliberate or occurred naturally.
- Tracking the perpetrator is more difficult.
- Vulnerable targets have low security.
- Moral barrier that perpetrator must cross is lower.
- Maximum effect may not require many cases.
- Mimicking natural introduction can be effective.
- Multiple point-source outbreaks can be initiated by contaminating imported feed or fertilizer, without even entering the country.
- Agronomic practices reduce the genetic variability and create conditions (large, dense populations) that facilitate disease spread.
SOURCE: Partially adopted from Burden (2010); Budowle presentation, 2013.
sampling, the technology that is available and appropriate must be incorporated into the logic of the process. A sampling and collection strategy should be structured, yet the structure cannot be overly rigid. One may decide, for example, to systematically collect a sample every 3 meters in a field. But if a tainted sample appears half a meter away, rigid rules would prohibit collecting the sample. A comprehensive plan should guide efforts (see Box 5-6), but owing to myriad possibilities that investigators may encounter, there is not a single sample collection and preservation strategy that is suitable in all situations. A key step in the comprehensive plan is to identify experts in advance so that when an event occurs, investigators can quickly develop consultation plans before they begin collection. Again, the chain of custody must be maintained throughout the process.
Environmental Contamination: The Public Health View
Dr. Stephen Morse reviewed the needs of the public health sector in the event of environmental contamination, and provided a sense of how the site, what is known or not known about the extent of contamination, and the nature of the contaminant will direct the manner of sample collec-
|A. General||B. Wheat Streak Mosaic Virus|
SOURCE: Budowle presentation, 2013.
Comprehensive Plan for Sampling and Collection
- Develop a mechanism for quickly formulating a “consensus” analytical plan when a new sample (or set of samples) arises.
- Keep and update a set of standard operating procedures and validation data for analyzing case samples.
- Maintain a set of documented guidelines, requirements, and procedures for sample preparation for each analytical procedure.
- Maintain approved procedures for handling and storing samples.
- Develop standardized methods for data analysis, reporting, and presentation.
- Maintain reliable channels for sending and receiving samples.
- Maintain secure conduits for data, information, and discussion.
- Develop a mechanism for formulating an on-the-fly validation plan for a new procedure.
SOURCE: Budowle presentation, 2013.
tion and preservation. He views validating collection methods as a major challenge because the number of potential surfaces from which samples can be taken is unlimited.
From the point of view of high-level decision makers, in the event of an environmental threat, sampling and analysis will enable them to (1) determine who has potentially been exposed, (2) characterize the extent of contamination, (3) remediate indoor sites of contamination, and (4) clear the facility for reoccupation or use. A public health agency wants to establish certain facts, some of which also are sought by law enforcement. In fact, the FBI and public health officers often investigate a site simultaneously, collect specimens together, and share specimens and results.
The type and method of the release or event will determine the impact of sampling. If it is a covert event, and the first indication is sick people appearing in an emergency room, collecting environmental samples 1 to 3 weeks after the fact may not accomplish much. On the other hand, if the release point is determined, trace evidence and dispersal items may still be accessible. If it is an overt release, however, and the site of release can be identified, sampling and modeling can be used to determine the extent of contamination in a particular area. Contamination via food or water presents a further level of challenges.
Whatever the circumstances, identification of the crime scene and sampling are generally the first steps. The entire process—from sample collection, to transport, to analysis—must be validated. Investigators who understand the performance parameters of a process can effect superior analyses. After the Amerithrax letters event, the CDC and other responders received criticism from the U.S. Government Accounting Office about the extent of their ability to interpret performance of sampling methods. For example, does a negative result mean an area is free of contamination? Or could a sampling method simply be inadequate, and contamination still be present? Are the limits of the detection method clearly understood? This becomes very important when making a decision to reopen, for example, the Hart Senate office building after the anthrax attack. What degree of assurance can one give to leaders that a building is actually free of contamination? These questions are necessary for remediation but do not contribute realistically to a microbial forensics investigation.
Methods for sampling can be difficult to validate. The B. anthracis sampling methods have been evaluated, but these methods represented a low hurdle because the organism is stable in the environment. Less is known about sampling methods for other organisms, particularly Gram-negative organisms and viruses that do not survive for long on environmental surfaces.
Swab and wipe methods have been validated for nonporous surfaces,
and validation is under way for a vacuum method for porous surfaces. To further complicate validation, the sampling tool used is validated only for a specific area of surface. A swab, for example, is validated for a 2 × 2-inch square, and a wipe for a 10 × 10-inch square. Samples cannot be taken from a wall using a single swab because the swab may not be adequate for the sampling or at a minimum no validation was performed to guide the collection under best practices.
In the laboratory, one might encounter various types of surfaces, both porous and nonporous. Nonporous surfaces, for example, might be painted wood or wallboard or various types of wood, plastic, glass, and metal. But each of these nonporous surfaces has individual characteristics. A surface may have an electrical charge or may have micropores, which will affect the ability to remove microorganisms. Similarly, porous surfaces, such as fabric and carpeting, have their own characteristics. Although the number of potential surfaces is unlimited, it is only possible now to validate sampling methods for a small number and then extrapolate the experience as necessary.
Environmental conditions, such as humidity, affect the ability to recover organisms from surfaces. Although one can control such factors and validate methods for the laboratory environment, conditions out in the field are uncontrolled and will vary. Another variable is the person collecting the sample. The individual may or may not be well trained. Even if personnel are trained, they may not perform in the same manner every time, especially if sampling guidelines are not available. Moreover, the person is likely wearing personal protective equipment (PPE), which is hot and uncomfortable. Getting the sample quickly becomes the primary goal.
Transportation methods must be carefully selected. B. anthracis is a hardy organism, but some agents require a specific transport medium. Also, after decontamination, particularly via chlorine dioxide or vapor-phase hydrogen peroxide methods, a neutralizing agent must be added to the transport medium to remove the decontaminant, or anything that survives will die before it reaches the laboratory and will not be culturable. Recovery efficiency once in the laboratory is another factor, and again, it will vary from agent to agent.
Sampling approaches are chosen based on the situation. The easiest approach is judgmental, which is typically used in the early characterization phases of a response to establish whether an agent or toxin is present. In a meeting room, for example, an investigator would likely first sample from heating, ventilation, and air-conditioning vents in the wall, and from computer screens, which are great attractants for airborne organisms. When there is a large quantity of organisms or toxin, this sampling method works well. One of the simpler scenarios to handle is a white-
powder incident, where the agent is visible; standard methods exist for collecting white powder for analysis and forensic use. If, however, there is only trace evidence in a room, a random, probability-based method—sampling X number of places in the room in a random fashion, to provide a 95 percent chance or greater of picking up material—is appropriate. In general, the larger the room, the greater the number of samples required.
After decontamination, a combined judgmental and random sampling would be appropriate to ensure that the area is free of contamination. The U.S. Environmental Protection Agency (EPA) uses composite sampling. They collect samples, put them in the same tube, and analyze the contents. They simply want to know if there is contamination present, regardless of its location. This approach could apply to forensic sampling to reduce labor and cost. But it will reduce the ability to reconstruct the event with that/those sample(s).
In the event of a wide-area release, for example, if someone releases B. anthracis over a city, there is a limited time to remediate the area. According to statements made at several exercises, substantial delays in remediating the affected area would reduce the likelihood that people would return. Leaders will want to be able to tell the public when or whether it is safe for them to return, and that decision will also be based on sampling. However, the public could lose confidence in their leadership if any residual infections occur after they move back in.
Morse was asked to comment, in terms of validating collection methods, how much confidence he has that there are no spores when he gets a zero response from sampling a room, and how he establishes probability or statistical confidence limits. Morse responded that a negative result only means that there were no viable spores detected, but one does not know if there are spores below the limit of detection; current sampling methods will detect less than 1 spore per cm2 on a surface. The CDC feels comfortable saying that a negative response would be equal to less than 1 spore per cm2. Although the analytical method may be able to detect a target at very low levels, sampling may or may not be efficient. Therefore, lack of detection may be related to collection method, sampling efficiency, and limit of detection of the analytical assay.
In the United States, the CDC has a network of analytical laboratories, the Laboratory Response Network, comprising 160 laboratories throughout the country. Similarly, the Food Emergency Response Network has protocols for analyzing food for the presence of threat agents, the Animal Health Network has them for agricultural animals, the EPA has an Emergency Response Laboratory Network, and there is a plant network. All use basically the same methods. The labs can process environmental samples using the exact same method; the results in one lab are equivalent to those in another lab. This is important because in the case of a wide-
area threat release, the nearest lab may be contaminated and unavailable. Samples collected using chain of custody are shipped to the closest lab for analysis. Morse pointed out that one problem with giving confidence limits is how to equate that to risk. Because the lowest dose that will infect a human is unknown for many, if not most, pathogens, it is hard to say there is no risk, which is never zero. Also, detecting something on a surface does not necessarily equate to an inhalational risk; it has to be able to detach from the surface and get into the air, to be breathed in. So it is difficult to say, “I am really confident that this room is safe to be in.”
Budowle noted that calculating false-negative rates and the like is a problem because the approach is borrowed from human clinical genetics, in which a true negative can be defined. Minimum criteria should be established for the output of the selected analytical method, such as depth and uniformity of coverage. Defining output thresholds for metagenomic samples may be difficult given the immense quantity of data and microbial diversity; therefore, single-source samples and defined mixtures might be used as a guide. These limitations may be necessary in defining false negatives and false positives. Clearly, there will be ambiguous calls due to sequencing noise and novel genome composition. The specific parameters and settings used to establish thresholds, false-positive, and false-negative rates should be detailed thoroughly to enable sound interpretation and accurate comparison to alternative methods and protocols.
Keim said that the situation he encountered was detecting something above zero, but only in 1 in 1,000 samples. The question was whether that finding was informative or significant given that the sampling had never been validated to that level and the processing labs had never validated the sampling. They concluded it was contamination by the processing system. So one can have a positive result and still not be confident. A single colony of B. anthracis on a plate is still just one, and may or may not be meaningful.
Perspective from a U.K. Government Laboratory
Dr. Cerys Rees works with the U.K. Ministry of Defense’s Defense Science and Technology Laboratory based at Porton Down. This facility provides chemical and biological analysis capability on behalf of defense and security customers, including the military and law enforcement. The Laboratory’s law enforcement experience includes the analysis that supported the first U.K. prosecution under the Chemical Weapons Act, which concerned the production of ricin. White supremacists in northeast England were arrested and convicted in 2009 for producing ricin in their home. Porton Down analyzed samples taken from the house, and by
combining several different methods, produced evidence that supported a conviction.
Proficiency testing is particularly lacking in the biological area. Dr. Rees’s colleagues in chemical labs undergo stringent Organization for the Prohibition of Chemical Weapons (OPCW) chemical testing annually to maintain their designated laboratory status. Rees would like to see similar training in the biological area, perhaps starting with confidence-building tests, and then moving on to proficiency testing.
In the United Kingdom, a forensic science regulator who has overall responsibility for activities that support the U.K. criminal justice system oversees evidence sampling. The regulator may soon be granted statutory powers, which would enable him to enforce International Organization for Standardization (ISO) 17025 accreditation on all laboratories that support the criminal justice system. He also is very motivated to require ISO 17020 accreditation of crime-scene examiners and personnel who collect samples. It is the international standard of competence for inspection bodies, which may not fit perfectly with crime-scene examiners, but it is likely the only standard that is potentially applicable.
Rees noted that the Laboratory has frequently worked with defense and security customers to address the requirements for chemical and biological analysis and attribution capability that support the different communities’ needs, which appear to be very similar. All require high-confidence information that enables defense customers to make strategic decisions and security customers to support the criminal justice system.
In their laboratories, validated tests are required to assess staff competence and to ensure that there is standardization in how tasks are performed (e.g., chain of custody) so they can demonstrate that the laboratory has high confidence in the information they produce.
Rees and her colleagues would like to pursue attribution based on all materials, not just the agent alone. They would like to develop methods that enable them to exploit the matrix and anything else associated with the sample, including traditional forensic materials, such as fingerprints, hair, and fibers. Her lab is working with police and forensic colleagues to develop these methods. Ideally they would provide information that, coupled with intelligence and situational information, could lead to a higher degree of attribution of an attack and perhaps identify the perpetrators. Dr. Franca Jones, U.S. Department of Defense, also said that there needed to be more discussion about other signatures in microbial samples that might be used for microbial forensics, such as media components. She encouraged that these nongenomic methods under development receive more consideration.
Weapons Inspections in Iraq
Dr. Rocco Casagrande, founder and managing director of Gryphon Scientific, reviewed the technological challenges he experienced while supporting biological disarmament missions in Iraq for the United Nations. His experience provided an example of the kinds of problems those doing work in the field can encounter, especially in sampling. He emphasized that although his unit was not equipped as well as it could have been, most problems lay in how to apply good technology in an austere environment.
From November 2002 to March 2003 he served as a U.N. biological weapons inspector in Iraq. As chief of the biological analysis laboratory, he collected and tested most of the samples collected by his unit. At peak staffing, the number of inspectors did not exceed 25, and most were field or bioprocess specialists, so he had little assistance in the lab. The lab was small and rudimentary, containing a hood, a real-time PCR machine, and equipment to perform DNA extraction. Real-time PCR was the assay most frequently used owing to its simplicity, low false-positive rate, and excellent sensitivity. The process was labor-intensive because DNA extractions were performed manually. The inspectors were limited to the primer sets provided, and had no input in the choice of reagents. During the first month in Iraq, the equipment was inoperable because they had not been supplied with a transformer adaptable to Iraqi electricity.
Their immunological methods originally included hand-held lateral flow immunoassay devices, but their unacceptable false-positive rates for environmental samples prohibited their use. For toxin detection, they used conventional ELISA kits with antibodies prebound to 96-well plates. In addition to running the lab, Casagrande’s duties included inspections, writing site reports, and interviewing Iraqi personnel. Only 4 hours a day could be dedicated to laboratory work; the remaining time was spent in the field.
In early 2003, equipment was found from a facility that had been destroyed by U.S. bombardment in 1991. There were allegations that the facility had been used for the manufacture of biological warfare agents, but the Iraqis claimed it was a baby-formula factory. After the facility was destroyed, the damaged equipment was moved to another site, and it lay outside for 10 years. Inspectors disassembled the damaged equipment, and samples taken from protected joints showed a borderline detection of Brucella species. The heterogeneous melting temperature of amplified fragments suggested multiple species were present, which would be consistent with natural contamination. A shortfall in inspectors’ equipment, Casagrande noted, was that they were not permitted to pick positive controls, and they had very few. The sample from the damaged equipment proved similar to the Brucella species profile in a control Casagrande
created from the U.N. cafeteria milk. Therefore, the PCR technology was effective in detecting the trace contamination of dual-use items that had been blown up, moved, and then lay unprotected in the weather for a decade.
A great problem the inspectors faced, as in forensics in general, was where to apply the good technology and practices they had. Selecting sampling sites can be problematic; some sites are very large. An earthen berm, for example, had been built to encircle the Iraqi facility at Tuwaitha to shelter it from Iranian bombardment in the earlier Iran-Iraq war. From which sites of the berm should samples be taken? From which of the facility’s 150 buildings, and from which equipment within those buildings, should samples be taken? The assay technology is effective, but what is needed is better technology to determine the reliability of the intelligence sources who direct inspectors to sampling sites. Better technology is needed to identify activities at a suspect site to determine which of the 150 buildings in Tuwaitha, for example, were of interest to the bioteam. Reliable background information is essential to direct inspectors’ investigations.
In 2002, for guidance the inspectors used a mid-1990s Iraqi declaration of agents and activities putatively in their biological weapons program. This declaration, plus traditional threat agent lists, was used to direct procurement of reagents, such as PCR primers, specific for certain microbes. Inspectors were able to detect B. anthracis and orthopoxvirus, but not any pathogens not contained in the lists, such as Machupo virus and Chlamydia psittaci. Casagrande suggested that usable technology that could detect the presence of threat agents regardless of identity would be revolutionary for this purpose. The technology exists, but in its present form it would be difficult to use in a one-person lab with limited working hours and a single source of electricity. In one case, for example, there were allegations that the BBs within a larger weapon projectile had been coated with Clostriudium perfringens, which causes gas gangrene, but inspectors lacked the technology to test for this bacterium, so the entire weapon was sent away for analysis. Casagrande suggested a need to search for dangerous traits of pathogens, or developing a more sensitive “zoo” chip.
The ability to identify modifications in an agent would also be extremely useful. It might yield clues about the intended target or use of a strain, determine if particular defenses are no longer effective, and indicate a degree of sophistication regarding the perpetrator. An example might be antibiotic resistance genes. For example, there are roughly 36 genes that encode tetracycline resistance; however, even if inspectors look only for that phenotype, the PCR analysis would need to be a highly multiplexed assay. Moreover, frequently only nonviable samples can be
obtained from a site. They may have been sterilized in place, bombarded, left outside, or subject to efforts to “clean” the sites. Samples often are contaminated with—and outnumbered by—environmental microbes, making analyses more difficult. Therefore amplification for genomic or phenotypic analysis is complex.
Inspectors faced this kind of challenge when collecting samples from R400 aerial biological bombs that had been filled with B. anthracis spores, botulinum toxin, or aflatoxin. It was difficult to get a culturable sample even from inside the weapon because the Iraqis had attempted to decontaminate the weapons with potassium permanganate. A challenge also lay in analyzing viruses and toxins. Many threat viruses have RNA as their genetic material, and RNA, even double-stranded RNA, is generally less stable than DNA, and thus is harder to detect. Because it requires an additional step, reverse transcriptase might also reduce detection limits. Another challenge lay in detecting protein and small-molecule toxins, such as botulinum toxin and aflatoxin, which were both on the declaration list. Inspectors used PCR to detect trace contamination of agent DNA because of its superior sensitivity and specificity over the ELISA assays. However, for a number of toxins of interest, such as ricin, inspectors hope to find DNA as well to indicate the presence of inactivated toxins. Although PCR detects DNA or RNA, not protein toxins, analyzing a sample by PCR might detect genetic sequences from the castor bean, even if the ricin had been inactivated or rendered undetectable. The Iraqis had legitimate castor oil plants, for example, and inspectors needed to determine if the ricin being produced was still active, if it was heat-inactivated, or if efforts were being made to extract it. A PCR assay cannot provide that information.
Finally, Casagrande added that one thing that was very important for the weapons inspection/biological disarmament mission was proficiency testing for people collecting samples from dual-use equipment. They needed to understand the equipment, know how to disassemble it and how to sample from the proper places inside the equipment. When inspectors confused very similar looking equipment or sampled from locations that had already been adequately decontaminated, they ran the risk of drawing falsely negative conclusions. Standardization, protocols that are widely disseminated and understood, and testing for the ability to take samples from dual-use equipment would desirable.