Appendix XV M. Substitution of in Vivo Method(s) by in Vitro Method(s) for the Quality Control of Vaccines

Purpose

The purpose of this general chapter is to provide guidance to facilitate the implementation of in vitro methods as substitutes for existing in vivo methods, in cases where a typical one-to-one assay comparison is not appropriate for reasons unrelated to the suitability of one or more in vitro methods. This general chapter will not discuss the details of assay validation as such, since those principles are described elsewhere.

The general chapter applies primarily to vaccines for human or veterinary use, however the principles described may also apply to other biologicals such as sera.

Context

The test methods used for routine quality control of vaccines are intended to monitor production consistency and to ensure comparability of the quality attributes between commercial batches and those batches originally found to be safe and efficacious in clinical studies or, for veterinary vaccines, in the target species.

While the in vivo potency and safety assays described within Ph. Eur. vaccine monographs have historically played a central role in safeguarding the quality of vaccines, the inherent variability of in vivo assays can make them less suitable than appropriately designed in vitro assays for monitoring consistency of production and for assessing the potential impact of manufacturing changes. As a result, it is essential continually to challenge the scientific value and relevance of these in vivo test methods. When in vivo tests are found to be of limited or no value, it is imperative to eliminate them, given the ethical considerations and the obligations under the relevant conventions. In addition, there is a substantial effort to develop in vitro methods (including immunological, molecular and physico-chemical tests) to replace the animal tests. In several cases this has led to the successful introduction of new in vitro methods in vaccine monographs. The use of appropriate in vitro methods not only reduces the use of animals while maintaining or improving the scientific relevance of the assays involved, but also substantially reduces assay variability and the time and resources required, and enhances the predictability of the release of safe and effective vaccine lots for use.

In addition to the benefits resulting from the substitution of appropriate in vitro methods for existing in vivo methods, under the European Convention for the Protection of Vertebrate Animals Used for Experimental and Other Scientific Purposes, the Ph. Eur. Commission has committed to the reduction of animal usage wherever possible in pharmacopoeial testing. Under the convention, those associated with the work of the European Pharmacopoeia are encouraged to develop and/or implement in vitro procedures, and the General Notices support the introduction of alternatives to in vivo methods described in Ph. Eur. monographs.

General considerations

One of the consequences associated with the inherent variability of in vivo assays is the problem this poses with their replacement by the more-consistent in vitro methods, which typically requires one-to-one assay comparison. This can be a challenge in some cases as repeated efforts through multicentre international collaborative studies can fail due to the variability inherent in the in vivo methods. Another consideration is that many of the legacy in vivo safety and potency assays for vaccines were generally shown to be fit for purpose and have historically proven their value in ensuring the efficacy and safety of vaccines. However, this was in an era when validation requirements, such as ICH Q2 (R1) or VICH GL2 guideline, were not in place, making a formal one-to-one comparison challenging or even impossible in some cases. Since precision, reproducibility, limits of detection and quantification were not established for the in vivo method, the comparability of one method to another becomes difficult to evaluate. It should also be noted that, because Ph. Eur. methods are considered validated under the General Notices, it is not only impractical and excessively costly now to undertake a retrospective ICH/VICH validation of these methods, but it would also be unethical given the above-mentioned convention on animal use in pharmacopoeial testing.

When considering the transition from an in vivo-based to an in vitro-based quality control assay system, it is important to understand what in vivo assays can and cannot offer. Although properly established in vivo potency assays in laboratory animals have the potential to measure complex functional responses for demonstrating proof of concept, these do not necessarily predict the actual responses in the target population. In addition, in vitro bioassays have the potential to mimic specific elements of complex in vivo responses with generally lower variability and higher sensitivity.

Another key consideration is that when an in vivo test for a given product is to be replaced with an in vitro test, the quality attribute(s) of the product will likely be assessed differently. Examples include: the determination of antigen content or a functional response (e.g. virus or toxin neutralisation) in an in vitro bioassay instead of in vivo potency; molecular consistency instead of in vivo neurovirulence or attenuated phenotype; absence of the extraneous agent genomes using molecular methods instead of absence of micro-organisms through in vivo testing; and demonstration of toxin binding and enzyme activity instead of in vivo specific toxicity. As a consequence, a demonstration of agreement between the 2 methods is generally not scientifically justified and should not always be expected. Even where pass/fail results from the 2 test procedures are in agreement, the correlation between 2 quantitative methods across the assay range may still be low. Regardless, the in vitro method(s) or testing strategy must provide at least the same confidence that the key quality attributes, which are necessary to ensure the consistency of a product’s safety and effectiveness, are adequately controlled.

While the focus of this general chapter is on the replacement of existing methods for approved products, it is important to consider the use of in vitro methods for quality control during product development and to understand that the use of in vivo assays is not mandatory.

Alternative approaches for the substitution of in vivo methods

The primary focus for the implementation of any proposed in vitro methods within a quality control system should be the scientific relevance of in vitro assays for control of the relevant quality attributes. Additionally, any in vitro methods will have to meet the current validation requirements.

In the Ph. Eur., in vivo assays for vaccines are typically replaced by in vitro assays following multicentre collaborative studies, but this should not be a prerequisite for in vivo assay replacement initiatives for individual products. Additionally, while it may be desirable to have assays that are widely applicable to a class of products, this should not be a requirement.

As explained in the guidance below, in some cases an existing method may need to be substituted by more than 1 in vitro test, in order to characterise the key qualitative and quantitative attributes measured by the existing test.

Potency tests

When it is not possible to show agreement between the in vitro and in vivo methods due to low discriminating power and/or high variability of the in vivo assay, the following approach can be used. It is assumed that the product under consideration has a well-established safety and efficacy profile, with consistent manufacturing.

The in vitro test(s) should be able to detect differences that are relevant to the control of the production process as justified scientifically. This should be supported by data demonstrating the capability of the proposed assay(s) to control key quality attributes of the vaccine and maintain the link between the quality of the batches to be released and those batches found to be safe and efficacious through clinical studies or routine use. With the setting of appropriate specifications, the consistency of manufacturing with the in vitro method(s) will be maintained.

The design of an assay/assay system for vaccine quality control needs to reflect both antigen content and functionality. If a single method is used, it should preferably measure the content and integrity of the antigen by targeting epitope(s) relevant to the protection offered by the vaccine. An example of this would be a monoclonal antibody or monoclonal antibodies against an epitope or epitopes as the main target for generating neutralising antibodies. The epitope or epitopes should preferably be conformational in order to have a stability-indicating assay (as is the case for rabies vaccine). In some cases, a single in vitro method may not adequately reflect the content and functionality. This can be remedied through the use of multiple assays, as is the case with conjugate polysaccharide vaccines, where molecular size, conjugate integrity, and total and free polysaccharides are examples of relevant measures.

To establish quantitative measurements with an in vitro method, samples that differ in the magnitude of the response will be needed. In most cases, samples that are below the minimum approved potency specification with the in vivo method will not be available because production consistency is generally well maintained, and potency between batches does not differ significantly and/or the precision of the in vivo assay is such that it cannot discriminate between batches unless the difference is very large. Therefore, initial assay evaluation should be performed with samples at different concentrations, which could be followed by testing of samples subjected to different types of stress conditions to assess further the stability-indicating potential of the new method. The inability to demonstrate agreement between an in vitro and an in vivo method does not necessarily mean that the in vitro method is not suitable/relevant. In many cases, an in vitro method will detect changes in the product profile that would not be detected by the in vivo method. In such cases, the in vitro method may be considered superior for monitoring the consistency of production and may be more relevant to assess the impact of manufacturing changes.

Safety tests

Specific toxicity

An in vitro method for detection of residual toxic components should be specific and at least as sensitive as the existing in vivo method. Where possible, a fully functional in vitro system should be used (e.g. toxin-sensitive cell line). Where no functional in vitro system is available, an in vitro testing strategy could be based on the detection/measurement of more than 1 parameter, sequentially where relevant, that together reflect the mode of action for the toxic components in question. Examples include the use of assays with immunological and biochemical steps to detect receptor binding and enzyme activity. In most cases, where an in vivo assay is to be replaced there will be data available on the sensitivity of that model for detection of the toxin in question. Therefore, new in vitro methods can be characterised to demonstrate that they are sufficiently sensitive using spiking experiments and referring to historic data for the in vivo assay. Such assays, in conjunction with the appropriate time and temperature conditions, could also be used to demonstrate the absence of reversion of a specific toxoid.

Molecular consistency by deep sequencing versus the neurovirulence test

An in vitro genotypic method to assess the molecular consistency of a viral vaccine has the potential to replace an existing in vivo neurovirulence test. A prerequisite for any in vitro genotypic method is an in-depth knowledge of the molecular markers responsible for the attenuation of the live viral vaccine (as is the case for oral poliovirus vaccine, for example). In such a case, monitoring the consistency of the vaccine lots would be achieved by confirming the presence of the required molecular attenuation markers and percentage of mutants with methods such as deep sequencing.

Detection of viral extraneous agents by novel molecular methods

Detection of viral extraneous agents in cell banks, seed lots and cell culture harvests is currently conducted using a panel of in vivo and in vitro methods at different stages of the manufacturing process. Novel, sensitive molecular techniques with broad detection capabilities are available, including deep sequencing or high-throughput sequencing methods, degenerate polymerase chain reaction (PCR) for whole virus families or random-priming methods (associated or not with sequencing), hybridisation to oligonucleotide arrays and mass spectrometry. The use of these new molecular methods has highlighted gaps in the existing testing strategy by identifying previously undetected viral contaminants in final product, the cell banks from which it was produced and intermediate manufacturing stages. These new molecular methods (e.g. deep sequencing or high-throughput sequencing) detect genomes while the existing in vivo methods are based on observations of the effects viruses have on experimental animals. The implementation of such new molecular methods as substitutes for in vivo methods requires a comparison of the specificity (breadth of detection) and the sensitivity of the new and existing methods. For this purpose, an appropriate panel of representative, well-characterised model viruses should be used to assess the ability of the new method to detect viruses that are (or are not) detected by the in vivo methods, and to determine if the sensitivity is at least equivalent to the sensitivity of the in vivo methods. This last element is particularly complex since these new molecular methods do not detect the same characteristic of the viral contaminant (genome for molecular methods versus infectious virus for in vivo methods) and also since no or limited validation data exist for the in vivo methods. It should also be emphasised that the outcome of the new molecular methods is not the final result since the detection of a genome or fragments of a genome does not necessarily indicate the presence of an infectious virus.