Overcoming Challenges to Develop Countermeasures Against Aerosolized Bioterrorism Agents: Appropriate Use of Animal Models (2006)

In this chapter, we discuss the advantages and disadvantages of existing methods for delivering a dose of an aerosolized bioterrorism agent in animal models.

The first method is inhalation delivery, which often entails the whole-body approach. Other approaches involve head-only, nose-only, or mouth-only exposures, where the aerosol is generated in a smaller volume and the animals are individually constrained such that their heads, noses, or mouths project into the test environment of interest.

In the case of whole-body exposure, test animals are placed singly or together in cages into which the desired aerosol environment is introduced. Attention needs to be paid to maintaining a diurnal light cycle in multi-day studies. Whole-body exposure chambers are usually of the type in which a continuous flow of throughput air is maintained; and they tend to be constructed of stainless steel, mainly because stainless steel does not build up localized electrical surface charge (which can affect dosing) and it is sterilizable. This type of exposure has several advantages. It can accommodate a large variety and number of animals for long periods of time, and it does not require restraint or anesthesia during exposure. However, this type of exposure can lead to highly variable dosing because animals can come into contact with the aerosolized material in a variety of ways. In addition to inhalation, exposure can occur through the skin, mouth, and eyes as well as from the animals’ licking their fur or cage material and eating their food. Animals can also receive an additional

aerosol dose from their own fur or that of other animals. To reduce variability, the Committee recommends that food should be removed during aerosol exposure.

In chambers without individual cages, animals can avoid high dose exposures by huddling together and covering their noses with their neighbors’ fur. To avoid this possibility, the Committee recommends the use of chambers with individual cages. A good example of this approach is a Hinners-type of exposure chamber, which allows for the exposure of a small number of animals in subdivided individual sections (Steinbach and others 2004; Moss and Cheng 1995b). Another disadvantage of the whole-body approach is that the desired level of exposure may take some time to stabilize in the chamber (Phalen and others 1984). When using a whole-body chamber, the Committee also recommends that the environmental temperature and humidity be regulated and the spatial and temporal distribution of the aerosolized material be uniform. Uniformity can be achieved by fitting the chamber with a cone or pyramid-shaped entry and exit and by either mixing the throughput air or by rotating the cages during exposure. The Rochester-type of chamber has a tangential inlet at the top that rotates the chamber air and provides good uniformity of exposure (Leach and others 1959). The Committee further recommends that samples for characterization of the exposure environment should be taken from the breathing zone nearest to the animal during an actual exposure. For more information regarding the use of whole-body exposure chambers, see Phalen and others (1984) and Moss and Cheng (1995b).

An alternative to whole-body exposure is head-only exposure, which requires that the head or neck region of the animal be firmly restrained. In contrast to the whole-body chamber, this type of exposure reduces the number of ways the aerosolized material can enter the animal and reduces variability in dosing. For this type of exposure, the Committee recommends that a good neck seal, which does not interfere with blood flow or ventilation, be utilized. In addition, environmental air temperature, humidity, and levels of carbon dioxide and oxygen need to be properly regulated. Other alternatives to whole-body exposures are nose-only or mouth-only exposure systems, which limit the entry of the aerosolized material to either the nose or oral cavity. An advantage to head-only, nose-only, or mouth-only exposure is that the amount of aerosolized material per animal is reduced compared to whole-body exposure and the concentration of the exposure material can be rapidly changed. Inhalation exposures by nose or mouth can be achieved using masks or cylindrical tubes (with a conical end to accommodate the head and one end open to the exposure environment) (Phalen 1984). The Committee points out that it is important to design and validate the tubes and masks properly so that exposure using these systems does not lead to stress on the part of the animal and altered respiration, which can affect responses to test agents.

Aerosol can also be delivered directly to the lungs via the mouth by introducing the aerosol through an endotracheal tube (Phalen and others 1984). This approach has been shown to eliminate losses in the upper airways and

significantly increase aerosol delivery to the lungs of rhesus macaques compared to nebulization-only delivery to their mouths (Beck and others 2002). However, losses still occur in the tubing, and normal protective mechanisms such as deposition of particles in airways of the head are bypassed. Other disadvantages include the need for general anesthesia, mechanical trauma to the larynx and trachea, interference with normal airflow characteristics, and loss of normal humidification and thermal regulation of the inspired air (Phalen and others 1984). Loss of humidification can be overcome by warming and humidifying the aerosol (without increasing particle size) to near-physiological values.

It is clear that inhalation exposure is the gold-standard approach for studying inhaled agents. However, there may be instances when alternative techniques for delivering a dose of an aerosolized agent may be useful, such as when there is a need for improved quantification of delivered dose. If use of alternative techniques for delivering an aerosolized agent is necessary, clearly documenting and justifying the alternative exposure technique can aid in the interpretation and replication of the study.

Several issues can influence which alternative exposure method is chosen. The Committee has developed this section to provide researchers with the most up-to-date information on alternative exposure methods to facilitate the decision-making process and to maximize the effectiveness of available delivery strategies for current testing.

As mentioned above, inhalation delivery results in significant losses of the exposure material on the head or body of the animal, as well as multiple entryways for exposure, including the eyes, oral cavity, nasal cavity, and gastrointestinal tract. Moreover, because deposition measurements within the lung compartment after exposure is difficult, quantification of the delivered dose is commonly estimated from calculations that include the inhaled particle concentration, or by measurement of biomarkers. These estimates may or may not provide the precise dosing information necessary for determining the relationships that will lead to the development of therapeutic agents that prevent or treat the biological response to inhaled bioterrorism agents. Compared to inhalation delivery, each of the alternative delivery methods described below provide for a higher degree of quantification of the dose delivered to the lower respiratory tract. However, their distributions of material in the respiratory tract may be dissimilar to the distribution during inhalation delivery (Beck and others 2002).

Alternatives to inhalation delivery include intratracheal or transtracheal instillation, during which a solution or suspension of the agent can be placed directly into the lumen of an airway. The dose delivered can be precisely controlled by the use of an injection syringe. Of the two methods, intratracheal instillation is preferred because it avoids the need for surgical penetration of the trachea, which requires humidification of the inspired air and postsurgical care

to prevent infection (Phalen and others 1984). Both approaches suffer from the effect of gravity on the instilled fluid, which distributes unevenly, running down into the dependent areas of the lungs (Brain and others 1976). This can lead to high local concentrations of the bioterrorism agent, local tissue damage, and biological responses that may be unique to this method of exposure.

Another delivery alternative is microspraying, which involves the passing of a small-diameter tube through the oral or nasal cavity and delivery of particles in the form of a spray to the tracheal carina, or into specific lung regions. The spray is produced by means of an injection syringe from a solution or suspension. Microsprayers are commercially available for mice and larger mammals, including nonhuman primates, and can be introduced into the lungs inside of a bronchoscope, which also aids visualization and targeting. Like intratracheal instillation, the delivered dose can be precisely controlled. Unlike instillation, the distribution of the dose is considerably more predictable and uniform (Beck and others 2002). Nevertheless, particles generated during microspraying can be significantly larger than traditionally nebulized particles, with aerodynamic diameters averaging >4 μm for some products and >20 μm for others. These large particles may target different cells and receptors than those that are targeted by smaller-particle generators, leading to differences in the biological responses.

Yet another alternative to inhalation delivery is aspiration. With this approach, a known amount and concentration of a solution or suspension can be pipetted either into one of the nares or into the distal part of the oropharynx of a lightly anesthetized animal, as described previously in mice and rabbits (Steinbach and others 2004; Miller and others 2002; Foster and others 2001; Larsen and others 2001; Gelfand and others 1997). Within minutes, the solution or suspension is aspirated by the animal and deposits in the nasal cavity and/or lungs. There are several differences between the aspiration approach and inhalation delivery in terms of the dose deposited and the distribution of the dose within the lung. Sequential gamma camera images of the lungs of mice, following oropharyngeal aspiration of a radiolabeled liquid, suggest that aspirated particles deposit in the alveolar region of the lungs (Foster and others 2001). In addition, initial images of the lungs indicate that the inhalation method delivers significantly less radiotracer to the lungs compared to aspiration; and images of the stomach and esophageal regions indicate more radiotracer in these regions with the inhalation method compared to aspiration.

One of the major concerns in developing new countermeasures against bioterrorism agents involves the ability to demonstrate their effectiveness: to show that the countermeasure provides a statistically significant level of improvement in some physiologically relevant outcome—often, survival— versus the response in the countermeasure’s absence. The outcome of interest can be dependent on a variety of factors, including the type of agent and its route

of administration, as well as the species in which the tests will be conducted. In the case of potential bioterrorism agents such as anthrax or other infectious diseases, the route of administration—should the agents be used in a conflict setting or terrorist attack— would be by inhalation following aerosolization.

The evaluation of countermeasures will include testing in model animal species to evaluate efficacy, and the value of those experiments will be dependent to a large extent on the experimental design. The first consideration is the dose of agent to use for evaluating the level of efficacy provided. This is often done by first determining the median lethal dose (LD₅₀) of an agent in untreated animals, and then determining a protective ratio—i.e., the ratio of the LD₅₀ in untreated animals versus the LD₅₀ in a population of animals treated with the countermeasure of interest.

Multiples of the LD₅₀ (i.e., 100 LD₅₀) have often been used to set the challenge dose of select agents. In some cases, sufficient data have been presented so that the challenge dose in colony-forming units (bacteria), plaque-forming units (viruses), or mg per liter of air is also known, while in other cases only the multiple of the LD₅₀ given. It is not always clear why a certain multiple of the LD₅₀ is used (i.e., 10 LD₅₀ versus 100 LD₅₀) for some agents in some studies but not for others. This situation underscores the need for well-designed experiments that would allow for reliable inter-laboratory comparisons.

A statistical calculation made from the experimental data, the LD₅₀, is the dose expected to kill 50 percent of the animals from the infection or toxicant within a defined time (often, 30 days). This can be determined by a classic probit-type design (Burn and others 1950), an up-down design (Dixon and Mood, 1948), or generated in sequential stages using the methods described by Feder and others (1991a; 1991b; 1991c). In some instances, the challenge dose is reported to be the LD₉₉ (the dose that would be predicted to kill 99 percent of the animals). An alternate approach is to estimate a dose of the agent that produces the toxic effect in all animals tested, and then test increasing doses of the countermeasure until complete or a statistically significant level of protection is achieved.

It is important to recognize that the LD₅₀ and LD₉₉ (and any other LD value) is estimated from experimental data and has variability associated with it. Confidence limits (e.g., 95 percent confidence limits) express the uncertainty in the estimate. As the studies used to determine the LD values usually use a relatively limited number of animals (especially in the case of nonhuman primates), these upper and lower bounds estimated from a probit analysis are very large. Thus it is important to regard the LD values as estimates that are useful but not absolute. Moreover, even if the LD₉₉ is determined in a way that minimizes error; one potential problem with the approach is that the dose of agent may be so great that it cannot be reversed by the countermeasure under investigation.

It is also very important to recognize that many factors, such as the strain or substrain of the animal model, the strain or substrain of the select agent, and even environmental factors, can markedly affect the LD values for a given

study. Other metrics of dose, including an infectious dose (e.g., ID₅₀) or effective dose (ED₅₀) can also be used to set a challenge dose. Thus it is important to actually determine such a dose, perhaps using the sequential-stages method of Feder and others (1991a; 1991b; 1991c) or fixed-design method and not extrapolate it from a probit analysis designed to evaluate the LD₅₀. Though the sequential-stages and fixed design methods provide a more accurate estimate of dose, there is a level of uncertainty (variability) associated with the estimate, which can be expressed as a confidence interval. Determining whether a sequential-stages or fixed-design method is best depends on the steepness of the dose-response curve and the length of the steps in the sequential method. These concerns emphasize the importance of statistics in the experimental design of aerosol exposures, and the Committee recommends obtaining statistical advice when designing an animal study to develop or test a countermeasure.

Ultimately, two approaches are generally used: (1) fix the concentration of agent at some finite level that produces an adverse response, and then vary (increase) the dose of the countermeasure until protection is achieved; or (2) use a fixed dose of the countermeasure and vary the concentration of the agent across a concentration range (that one expects to encounter) until protection is overwhelmed. The relative advantages of either approach are usually determined by the type of outcome being measured—e.g., for aerosol exposures the former approach may be more practical because of the necessary observation time of the subject animals to make sure that an outcome is valid. In either event, the up-down or sequential-stages approaches will reduce the number of animals needed for a statistically valid response, which is an important consideration.

Given animal-welfare concerns, it is hoped that more sophisticated metrics of dose will replace use of the classic LD₅₀ determination in most studies using animal models; and the practice of humanely euthanizing moribund animals prior to their death is recommended to reduce or eliminate pain and suffering (Toth 2000). As previously noted, the outcome chosen need not be lethality; and often it may be preferable to evaluate efficacy by some other criteria (Toth 2000). In the case of infections diseases, for example, delay in time to onset or a reduction in clinical severity may be very acceptable outcomes that would support a claim of efficacy to the regulatory body, particularly if approval is being sought under the newly instituted Animal Rule.

The following are some examples of inhalation LD₅₀ values and their use. A 1966 study (Glassman 1966) reports an inhalation LD₅₀ for anthrax spores of 4,130, with 95 percent confidence limits of 1,980 to 8,630 spores. This was based on 1,236 cynomolgus monkeys (M. fascicularis). The strain of Bacillus anthracis was not given, but it may have been Vollum 1B (Albrink and Goodlow 1959). At the Animal Models for Testing Interventions Against Aerosolized Bioterrorism Agents Workshop, some details were presented on the LD₅₀ that the U.S. Army Medical Research Institute of Infectious Diseases (USAMRIID) has used in publications on spores since 1991. The LD₅₀ for the Ames strain was 54,687 cfu, with 95 percent confidence limits of 44,809 to 8,300,000, in rhesus monkeys (M. mulata). The LD₉₉ was 130,000, with 95

percent confidence intervals of 78,229 to 9,100,000,000. Thus there was a factor of about 2.4 between the LD₅₀ and the LD₉₉ (Pitt 2005). A contemporary report from another laboratory gave an LD₅₀ for the Ames strain of 61,800 (95 percent confidence intervals 34,800 to 110,000) cfu in M. fascicularis (Vasconcelos and others 2003). However, at the workshop an LD₅₀ for a new batch of Ames-strain spores was reported to be 7,221 for rhesus monkeys and 8,294 for African Green monkeys (Pitt 2005). Thus even for anthrax, which has been comparatively well studied, fairly large variations in the LD₅₀ via inhalation in non human primates have been reported. In addition an inhalation dose of 161 to 760 LD₅₀ of Ames-strain spores of B. anthracis has been used as the challenge dose of spores for studies on vaccines (Phipps and others 2004). However, prophylactic use of antibiotics to prevent anthrax in rhesus monkeys was reported following an exposure of about 8 LD₅₀ of Vollum 1B spores (Friendlander and others 1993). This variability makes it very difficult to evaluate apparent differences in protection afforded from vaccines and the efficacy of therapeutics and sort out whether they result from variations in spore lots, spore strains, or monkey strains and substrains. The logic used to select challenge doses for studies on vaccine efficacy and therapeutic agents is also difficult to discern from available information.

Given the extent of variation in published LD₅₀ values for anthrax, the Committee recommends that unclassified data from mortality and natural-history-of-disease studies—including unclassified, unpublished data from USAMRIID—be published in the open literature for all agents¹. In each case, the materials and methods section should include the source of the agent or toxin, the characterization of the agent or toxin (including the number of passages), and the media used to prepare the agent or toxin. It should also include the species, stock, and strain of the laboratory animal used. In the case of nonhuman primates, it is necessary to include the source and country of origin of the animals (Flick-Smith and others 2005). If publication of past, current, and future studies in the open literature in detail is not feasible, an inclusive database should be established by the National Institute of Allergy and Infectious Diseases. All data from unclassified government-sponsored studies should be placed in this database¹.

In the design of efficacy studies for new or potential countermeasures against bioterrorism agents, care needs to be taken to ensure that the outcomes are in some way related to defining, or supportive of defining, the mechanism of action of the countermeasure. Given that the route to approval by the FDA could be the Animal Rule (21 CFR 314 Subpart I and 21 CFR 601 Subpart H), the major requirements are: that efficacy be shown in more than one species; that the mechanism of action of the proposed countermeasure be understood or

defined; that the indication chosen be related to the desired outcome in humans (i.e., survival); and that there be sufficient pharmacokinetic and pharmacodynamic data in animals to extrapolate to a dose for use in humans.

These requirements often constrain the ability of researchers to use “surrogate” markers as end points in the design of their efficacy studies—for example, reduced fever as an endpoint for countermeasure effectiveness would not be a good surrogate marker if the fever was not known to be the direct result of exposure to the agent. In contrast, if the agent was a microorganism and the countermeasure reduced the level of the organism in a biological system known to cause its transmission in vivo—e.g., reduction of the malarial parasite P. berghei in red cells following countermeasure administration—that could be a useful demonstration of efficacy based on an understanding of the mechanism of the disease. Such an approach also allows for dose-dependent efficacy studies, which are always useful in seeking licensure of a new countermeasure, as they can be used to estimate a dose of countermeasure that would cause a similar response in humans. In addition, the route of exposure is of less importance than the appearance of the agent in a biologically important site, so the ability to test a countermeasure against multiple routes of exposure is enhanced. Similar examples can be found for the development of countermeasures to viral infections.

Most of the preceding discussion has assumed that standard models of exposure are available for aerosolized agents—i.e., a fixed method for generation of an aerosolized agent and a set route of administration. Validation of such models across species is needed so that one is not comparing intranasal administration in one animal model with aspiration in another. This challenge, previously discussed, is important to address. Otherwise it will be difficult to convince the regulatory agency of comparable levels of agent and the ability of a countermeasure to provide an efficacious outcome that is understood at the pharmacological level in at least two species.

Although a variety of techniques exist for determining outcomes that either predict efficacy or measure it directly through the survival of animals following antidote administration, emerging technologies may offer more precise and less invasive ways to determine positive outcomes. The ability to noninvasively measure antidote concentrations via new imaging technologies for biodistribution studies, or the application of new human clinical technologies to measure efficacy of a drug at the cellular level (e.g., toxicogenomic or proteomic studies of lung tissue or lavage fluid as a predictor of a new antidote to prevent injury), offers great promise in the design of new antidote-evaluation studies.