Guidance on the Inactivation or Removal of Select Agents and Toxins for Future Use: Procedure Development

When following an inactivation procedure, the limits of detection of the viability testing procedures and expected run-to-run variation may prevent demonstrating full sterility of inactivated material. These sources of error must be considered to achieve the regulatory requirement of developing a validated inactivation procedure. FSAP expects the entity to determine the risk of live agent in inactivated BSAT material. This risk should take into account the safety and security consequences associated with failure of inactivation as well as the margin of error associated with the inactivation procedure. The variability in the nature of the specific pathogens (including ID50), the use of the inactivated material, and the inactivation procedures prevents establishing a specific national standard for measures of uncertainty associated with inactivation procedures.

If there is any deviation or modification to an inactivation or select agent removal procedure that was previously validated on-site, the entity should revalidate the procedure under the new conditions to ensure the procedure still achieves the appropriate result. For example, if the procedure was validated with Ebola virus and you want to use the procedure to inactivate Chapare virus or if the procedure was validated with a starting concentration of 10⁶organisms and you want to inactivate 10⁸organisms, these modifications would warrant the reevaluation of an existing procedure or redevelopment and revalidation of the procedure.

Entities can either develop their inactivation procedures in- house (see procedure development section below) or they can 1) use exact conditions of a commonly accepted procedure that has been validated in-house, such as autoclaving or 2) use a published procedure and adhere to the exact published conditions that have been validated in-house as applied. For more information on validation of previously developed procedures see the validation and verification sections below.

Procedure development via a kill curve

Entities that develop their inactivation procedures in-house also need to validate the procedures in-house and determine the measures of uncertainty with their particular procedure. This can be accomplished by developing empirical kill curves based on specific conditions that will be used to prepare samples. Note: The development of a kill curve is not a regulatory requirement. This information provides guidance for those entities that choose to develop their procedures via kill curves. Kill curves are appropriate for inactivation procedures using heat, chemicals, or irradiation. For a traditional kill-curve, the y-axis represents concentration of organism and the x-axis represents an experimentally-controlled variable, such as time or irradiation dose, for a given treatment. For each kill curve, only one factor, the x-variable, should vary. For example, initial burden and chemical concentration could be held constant while time of treatment is varied. Alternatively, all other factors might be held constant while chemical concentration is varied to produce a kill curve in terms of disinfectant concentration for fixed time.

The efficacy of inactivation treatment can be affected by many variables, so those variables must be held constant. Inherent variability in these factors (like the ability to control temperature precisely) can produce variation in results, and the entity should account for this in the validation process (like more experimental replicates to establish confidence in the results). The following are examples of factors that may influence the inherent variability in an inactivation procedure:

• Exposure or incubation time with inactivation treatment
• Nature of starting material (virus, bacteria, spores, tissue)
• Matrix material (serum, plasma, water, media)
• Dose, temperature, and concentration of inactivation treatment
• Concentration or volume of starting material (virus, bacteria, spores, tissue)
• Incubation temperature
• Process controls
• Container that holds select agent during inactivation (like tubes used during irradiation)

The entity needs sufficient empirical data developed in-house with test microorganisms to determine the inactivation kinetics with a particular procedure, starting with a high concentration of organism to the LOD/LOQ. (For those inactivation procedures where the LOD/LOQ is reached so quickly that multiple data points cannot be collected to adequately determine kinetics, see the Bioburden Reduction section below). Use multiple replicates at each treatment point to assist with fitting the curve that determines the inactivation kinetics. The number of replicates will vary depending on the procedure and agent. It is the entity’s responsibility to determine a sufficient number of replicates. For development of the inactivation procedure, use all (100%) of the sample from each treatment point in viability testing to generate data for the curve. If working with large volume cultures 100% of the sample can be filtered and then the filter can be cultured.

Kill Curve Development Steps

1. Decide what the x-variable will be, such as time or irradiation dose, while keeping other conditions, like agent concentration, fixed.
2. Using your knowledge of the organism and inactivation procedure, select initial dosage steps (time duration, intensity of treatment, or what you chose for the x-variable) and number of replicates (typically three) at each dosage. Replicate experiments enough times to demonstrate reproducibility. For example, select triplicates at 2 minutes, 4 minutes, 6 minutes, 8 minutes, and 10 minutes for a particular organism and treatment. The goal is to produce a wide range of data with measurable, incomplete inactivation to identify (x,y) points on the kill curve.
3. Perform the experiments and collect data.
4. Calculate a safety margin to incorporate into the inactivation procedure (e.g., additional treatment to achieve a 2 log further reduction in bioburden, a target SAL like 10^-6, etc.).

Procedure development via Bioburden reduction

For some procedures or agents, developing inactivation procedures via a kill curve is not practical and bioburden reduction is preferred. Examples of this include when inactivation is too efficient to allow for the collection of many data points on a kill curve, or for extracts that involve multiple processing steps. Bioburden reduction studies for viable organisms involve the deliberate addition of the specific agent or surrogate (see below for more information on surrogates) and measuring the extent of removal or inactivation during subsequent processing, inactivation, or removal steps. The entity should evaluate critical parameters that influence the effectiveness of inactivation/removal steps (bioburden reduction) such as the factors listed above in the kill curve section. Validation studies should establish the reduction achieved by inactivation and removal steps and be in excess of the greatest possible viable bioburden that may be found in the test materials undergoing inactivation/removal validation.

The procedure of bioburden reduction relies on these steps:

1. Determine inactivation effectiveness of a fully-specified protocol/procedure, e.g. 5 logs.
   1. Create a spiked sample with an organism concentration well above the expected inactivation effectiveness of one application (or one unit of treatment).
   2. Apply the inactivation procedure to the spiked sample and quantitatively measure the effectiveness by counting the remaining viable organisms.
   3. Replicate the experiments enough times to demonstrate reproducibility.
2. Apply as many applications as possible to achieve “overkill” in an amount consistent with the observed variability and the acceptable likelihood (presumed to be a very low probability) of survival of the particular organism.
   1. For example, consider a starting concentration of organisms of 10⁸ and an inactivation procedure that leads to a 5-log reduction in organisms using one unit of treatment.
      1. The resultant specimen after one unit of treatment should contain a concentration of 10³organisms, which can be counted via viability testing.
      2. Applying another unit of treatment for an additional 5-log reduction in organisms would be considered overkill and provide a safety margin of 10^-2 or an approximate probability of ¹⁄₁₀₀ of a live organism, which can’t be counted via viability testing.
      3. An additional unit of treatment would provide a safety margin of 10^-7 and give an approximate probability of ¹⁄_10000000 of a live organism, which also can’t be counted via viability testing.
   2. The amount of overkill should be commensurate with the potential consequences for inactivation failure of the particular organism and the need to retain characteristics for future use.

Limitations of the Bioburden procedure

It is important to note that the above bioburden methodology provides only a point-estimate, or the expected value, of the probability of survival. As such, these estimates do not include variance estimate or estimates of assurance. Further, bioburden methodology assumes that increasing the inactivation dose results in the same rate of reduction of viable organism that was seen above LOD/LOQ. Further, this procedure assumes spiked samples represent a naturally infected sample which, for some matrices (tissue), may not be true. As mentioned above, a variety of factors must be considered when developing an inactivation or select agent removal procedure. For example, when inactivating agent in tissues, the spiking of agent at concentrations that are above what would be expected in a natural infection would be performed in vivo and the affected organs would be removed and subjected to the inactivation procedure. For initial procedure validation, the tissue would be processed to detect the remaining viable organisms. Once the procedure is validated, one could take an immediately adjacent sample of the affected tissue in subsequent inactivation experiments to verify inactivation of the agent (see verification section below).

Surrogates

Ideally, an inactivation procedure is developed with the exact organism it is intended for. However, there may be occasions when this is not possible. Consider an entity that identifies a select agent in a small diagnostic sample and wants to inactivate the agent for future use. The entity wants to extract nucleic acids from the select agent. Or, consider an entity that wants to fix tissues from a select agent infected animal. In these situations, procedure development studies would leave very little or no sample for experimental purposes. Therefore, surrogate strains that are known to possess equivalent properties with respect to inactivation can be used to develop an inactivation procedure. If there are known strain-to-strain variations in the resistance of a select agent to an inactivation procedure, then an inactivation procedure must be developed using the more resistant strain. If the entity uses surrogates to develop an inactivation procedure, it is recommended that the final inactivation procedure be validated on the select agent it is intended for. However, in some cases, such as with Variola virus, validation with the intended select agent is not possible. In these situations, it is recommended that entities include risk mitigation steps during inactivation procedure development, validation, and verification such as validation of the procedure with all (100%) of the surrogate, inclusion of process controls, a sampling strategy for subsequent inactivation of samples, and/or the incorporation of a safety margin.

Bacteria from the same genus can be suitable surrogates for select agent bacteria, such as Yersinia pseudotuberculosis for assay validation for Yersinia pestis, but it does depend on the method of inactivation and it cannot be guaranteed that it would work all the time with a similar bacteria. Viruses from the same family can be suitable surrogates for select agent viruses, such as using Ebola as a surrogate for Marburg. For regulated nucleic acids, any positive single stranded RNA can be suitable surrogates for regulated positive single stranded RNA, such as VEE genome as a surrogate for Omsk hemorrhagic fever virus genome. It is the entity’s responsibility to research possible surrogates and determine the risk associated with using a surrogate strain.