Developing and Manufacturing Attenuated Live Bacterial Vaccines

ABSTRACT

There are a number of specific characteristics to be considered when developing and manufacturing vaccines. This article discusses specific requirements to be fulfilled for three attenuated live bacterial vaccines (LBVs) including Mycobacterium bovis BCG vaccine against tuberculosis, Salmonella typhi Ty21a vaccine against typhoid fever, and Vibrio cholerae CVD 103-HgR vaccine against cholera. Special characteristics for these vaccines comprise the appropriate level of attenuation, the balance between safety and immunogenicity, the genetic stability of the organisms combined with environmental risk assessment, the challenge of old-fashioned upstream and downstream methods in combination with quality control of the final product, and the release requirements.

Vaccination has proven to be the most efficient, cost-effective means for preventing a wide variety of infectious diseases. Vaccine development and manufacturing, however, poses several challenges. Inherent to all biological systems is the difficulty to achieve robust production processes guaranteeing reproducible efficacy and safety of the products. Highly automated bioprocess systems or advanced analytical systems used for closed-loop control may be a solution to overcome this.^1–3 Yet these techniques are still not standard in the industrial environment or cannot be applied to vaccine production processes because of historical reasons, as some of the vaccines have been developed early in the previous century.

(Crucell NV)

Because vaccines are administered to healthy people, the efficacy and safety of vaccines are most important. Regulatory cGMP requirements are increasing to guarantee the safety of vaccines being developed and produced. Using raw material of animal origin is avoided in the early-stages of research and development. Only few techniques such as nanofiltration and chromatography steps are applicable to viral removal. For viral inactivation, only formaldehyde treatment and beta-propiolactone inactivation are approved. The worldwide ban of preservatives such as thiomersal raised the standard for sterile production and supported the trend to monodose presentations.

To date, vaccines based on three different technologies are registered for human use: (1) whole inactivated vaccines containing entire killed bacteria or viruses, (2) subunit vaccines, containing only the relevant antigens of the pathogens in a highly purified form, and (3) live attenuated vaccines. The quality and safety requirements are even higher for live attenuated vaccines than for the killed and subunit vaccines. In this article, we will describe how vaccine developers and manufacturers solve the challenges related to manufacturing live attenuated vaccines.

Live Attenuated Vaccines

Live attenuated vaccines are among the most widely used vaccination technologies. Attenuated vaccines consist of bacterial or viral strains, which are weakened by stable mutations that allow the bacteria or viruses to infect humans only transiently. This transient infection elicits immune responses, while the vaccine strains are designed in such a way that they will not cause the symptoms of natural infection by the wildtype pathogen. There are a number of advantages of live attenuated vaccines in comparison to killed and subunit vaccines: (a) they mimic natural infection, therefore eliciting immune responses that are highly specific, effective, and long-lasting (b) they can prevent infection by the pathogen, not just disease symptoms, (c) in comparison to highly purified subunit vaccines, they are relatively cheap to produce and administer, and do not require sophisticated downstream processing or formulation with adjuvants, and (d) several live attenuated vaccines can be administered orally, which has a higher acceptance and better safety profile than injection with syringe and needle, and mimics natural infection better.

Three attenuated live bacterial vaccine (LBV) strains are currently licensed for human use: Mycobacterium bovis strain Bacille Calmette-Guerin (BCG), Salmonella typhi Ty21a, and Vibrio cholerae CVD 103-HgR. M. bovis BCG is the vaccine strain in the parenteral vaccine against human tuberculosis, S. typhi Ty21a represents the only oral vaccine against typhoid fever, and Vibrio cholerae CVD 103-HgR is the orally administered live vaccine against cholera (Table 1).^4–6

Table 1. Attenuated live bacterial vaccines licensed for human use

Preclinical Development

While the advantages of LBVs are widely accepted, there are several key considerations associated with their successful development. Most importantly, researchers developing an LBV must generate a vaccine strain that meets a delicate balance. The strain must reach an appropriate level of attenuation to be safe and be sufficiently immunogenic to ensure protective efficacy. Traditionally, live attenuated vaccines were developed by passing the pathogens under in vitro conditions until they had lost virulence for humans. This empirical approach was taken in the case of the M. bovis BCG vaccine strain. The BCG strain was attenuated by Calmette and Guerin between 1908 and 1920 by 231 serial passages of a virulent M. bovis strain through bile salts.⁴ During in vitro passage, the M. bovis microbes became attenuated because of the loss of numerous gene complexes, as was demon-strated by a recent genome analysis of the vaccine strain.⁷

The second example, S. typhi Ty21a, underwent a more targeted attenuation approach. Germanier and FFCrer reasoned that a S. typhi strain, which is sensitive to galactose and could not express a polysaccharide coat (which protects the bacteria from immune responses), should be attenuated.⁸ They generated the vaccine strain Ty21a in the early 1970s by chemical mutagenesis of wildtype S. typhi using nitrosoguanidine and screening for clones that had a phenotype, which is negative in the enzyme galactose epimerase, resulting in galactose sensitivity, and which is also unable to express the Vi-polysaccharide capsule.⁸ As a result of the chemical mutgenesis method, the strain was also mutated in genes responsible for amino acid biosynthesis and stress resistance, making it auxotrophic and less resistant to environmental stresses.

Finally, Kaper and Levine used genetic engineering technology in the 1980s to generate the V. cholerae vaccine strain CVD 103-HgR.⁶ They reasoned that the main virulence factor of V. cholerae is the expression of the cholera toxin. CVD 103-HgR was derived from a wildtype V. cholera strain by the targeted deletion of 95% of both chromosomal copies of the ctxA gene, which encodes the toxic A subunit of the cholera toxin while keeping the expression of the nontoxic but immunogenic B subunit, leading to intermediary strain CVD 103. Subsequently, they inserted a mercury resistance marker into the genome to readily allow for identifying the vaccine strain and its differentiation from wildtype organisms on vaccination.⁶

The safety and immunogenicity of vaccines are tested in animal models before they enter clinical trials. However, in the case of LBVs this is only possible if appropriate animal models exist, which is the case only for some bacteria. We would like to refer interested readers to a detailed article by Passetti, et al., which describes the challenges when developing an animal model that allows meaningful vaccine testing for S. typhi-based LBVs.⁹

Clinical Development

There are general differences in developing therapeutic agents for vaccines. All vaccines licensed to date have a prophylactic effect against infectious diseases. For some infectious pathogens, clear correlates of protection are known, for example, for threshold antibody titers. In such cases, the clinical development may be quite smooth. However, for many diseases such correlates of protection do not exist. In such cases, vaccines are administered to healthy individuals in the pivotal Phase-3 trials, and researchers then have to wait for the trial participants to get infected by the pathogen the vaccine should protect against and for this pathogen to cause disease. Differences in the incidence of infection or disease between the vaccine and placebo group then allow the researchers to calculate the protective efficacy of the vaccine. For some infectious diseases with high incidence, such a clinical trial may take just a few months. However, for diseases like tuberculosis, for which clinical symptoms may occur only more than a decade after initial infection, Phase-3 trials may take 10–20 years. Furthermore, the incidence of a disease may be low, requiring enormous numbers of clinical trial participants to get significant data. Therapeutics are administered to ill people to cure them or control disease whereas vaccines are administered to healthy individuals to protect them from becoming ill. The acceptance of side effects and safety risks is therefore much lower for vaccines in comparison to therapeutics. Hence, vaccines have to undergo careful pre- and postlicensure safety studies.

Regulatory Hurdles

In addition to regulations governing all vaccines, LBVs face additional regulatory hurdles. First, they must achieve the right balance between safety and immunogenicity. Additional issues include the potential for genetic reversion to partial or full pathogenicity, gene transfer into and out of the vaccine cells, and the potential risks for humans and the environment. For many vaccine candidates, the appropriate balance of attenuation and immunogenicity could not be met, and development of such candidates was consequently discontinued at an early stage of clinical evaluation. For the trivalent oral polio vaccine, which has been licensed for decades, it is known that reversion to virulence occurs quite often in the gastrointestinal tract of vaccinated individuals. This reversion to wildtype is a common cause of disease when a virus is shed and transmitted to nonimmune household contacts. Hence, these safety issues have to be addressed very carefully during preclinical and clinical development.

Determining Safety and Efficacy

In the case of Ty21a, safety and efficacy of the vaccine were demonstrated in a large number of clinical trials, with over 500,000 vaccinated children and adults in the US, Europe, Africa, Latin America, and Asia. Excellent tolerability and an overall protective efficacy of 67–80% were demonstrated for up to seven years in large field trials.⁵ The safety and tolerability profile of Ty21a was further confirmed in more than 200 million vaccinees during its more than 25-years of use worldwide. Recent postmarketing surveillance has identified only mild and infrequent adverse events associated with Ty21a. From 1990 to 2000, more than 38 million people were vaccinated with Ty21a with only 743 spontaneous reports of adverse events, an incidence of 0.002%. The most common adverse events reported with Ty21a were mild and transient gastrointestinal disturbances, followed by general symptoms such as pyrexia.

As mentioned above, the most important safety feature is to demonstrate that the vaccine strain is unable to revert to a virulent phenotype during production, inside the human body and after potential excretion of the vaccine strain, if applicable. For Ty21a, reversion to virulence has not been observed in vitro or in vivo during the more than 30 years since the strain was developed. No mutations were found in master and working seed lots produced over a 25-year period in genetic stability studies, nor was any reversion found in clinical trials. Clinical trials have also shown either a limited and transient level or a complete lack of shedding in the stools of volunteers depending on the administered dose of Ty21a. With a 10-fold overdose, mainly on day one post-vaccination, a low rate of excretion was observed. Further studies showed a lack of fecal excretion of Ty21a upon administration of the commercial formulation. The reason for the low excretion rate is probably the limited ability of the vaccine strain to proliferate in vivo. However, if there is excretion, additional studies need to be performed. In the case of Ty21a, these included an analysis of vaccine strain transmission to household contacts, which was not observed for Ty21a. Furthermore, Ty21a demonstrated a limited ability to survive in the environment because of its auxotrophy and reduced stress resistance, therefore not posing environmental risks.

The clinical program was somewhat different for CVD 103-HgR. For V. cholerae, a human challenge model exists. This means that after vaccination, humans can be infected intentionally with wildtype, fully virulent V. cholerae bacteria to determine whether the vaccine is indeed efficacious against infection and disease. In the volunteer challenge model, significant efficacy of CVD 103-HgR against wildtype V. cholerae challenge infection was demonstrated in North American adults as early as eight to 10 days after vaccination and for at least six months.¹⁰ Hence, oral vaccination with CVD 103-HgR is highly suitable to elicit protective immunity in North Americans, as shown in trials that are comparatively easy to perform. However, challenge trial results in North American adults may have limited relevance for the vaccines' protective efficacy in cholera-endemic areas of the world. Hence, a field trial was performed in Indonesia with CVD 103-HgR. Unfortunately, protective efficacy of CVD 103 HgR could not be demonstrated in this field trial. More than 60,000 children and adults were given either vaccine or placebo. However, cholera was virtually absent in Indonesia after initiating the clinical trial, and therefore, there were almost no cholera cases in the vaccine and the placebo group. This made it almost impossible to calculate a protective efficacy for the vaccine in this particular setting. This exemplifies the challenges vaccine field trials can pose.

In many controlled studies and after a decade of postmarketing experience, CVD 103-HgR has been shown to be safe and well tolerated. In diverse populations, no adverse reactions occurred significantly more often in vaccinees than in placebo recipients.¹⁰

Genetic Stability and Environmental Risks

Several tasks were performed to determine the genetic stability and environmental risks of CVD 103-HgR including (1) global and local genetic characterization and stability of the vaccine strain with emphasis on the modified gene loci, (2) identifying natural cryptic prophages and plasmids, (3) confirming the absence of DNA sequences from the various plasmids used during construction, and (4) evaluating biosafety aspects pertaining to the rate of excretion from vaccinees, the potential for vaccine survival in various ecosystems, and the potential for acquisition or export of genetic material. The strain was shown to be stable in a number of studies. For stability studies, two vaccine production lots manufactured three years apart were compared to the CVD 103-HgR master seed lot (MSL)after prolonged storage in a lyophilized state and outgrowth for 16 to 17 generations. In addition, fecal isolates from immunized volunteers were also compared to the MSL. This is important because DNA rearrangements may conceivably occur during transit of the vaccine organisms through the human gastrointestinal tract.¹¹

Studies on American adult volunteers revealed that CVD 103-HgR is only excreted at low levels (2 x 10² CFU/g of feces) by 20 –30% of the volunteers for a maximum of seven days with a peak on day four. In people living in endemic countries, the excretion rate is even lower. However, since it is excreted, CVD 103-HgR had to be compared to wildtype V. cholerae for its ability to survive under various conditions reflecting an actual environmental microcosm. The data demonstrated that CVD 103-HgR does not differ from its wild-type parent strain in terms of survival for example in estuarine water or soil. Overall, the studies showed a short-term survival of the vaccine strain in environmental microcosms and emphasized the fact that it presents no selective advantage over wild type V. cholerae, an important feature in the context of an environmental-risk analysis.

The BCG vaccine gives rise to a long-lived cellular immune response and causes a limited incidence of notable side-effects.¹² Worldwide, a single dose of BCG in newborn children significantly protects against the development of severe forms of childhood tuberculosis. However, protection by BCG against pulmonary tuberculosis in adults, the most prevalent form of the disease, is highly variable. As evaluated in large numbers of clinical trials, and in various populations and geographic regions, the calculated protective efficacy of BCG varies between zero and 80% and the reason for this variation remains unknown. Concerning BCGsafety and stability, reversion to virulence was never observed and recent genome analyses have shown that the vaccine strain exhibits a striking genetic stability over the 80 years it has been in use.⁷ While it was initially administered orally, BCG is currently administered parenterally. Because of this, shedding is not expected to occur and consequently, shedding studies are not required.

Manufacturing Processes and Release Testing

The manufacturing process for the BCG vaccine has remained virtually unchanged since the 1920s. The conventional BCG vaccine is produced as a surface pellicle in culture flasks. These cultures require months and allow only limited opportunities for recording and correcting culture parameters. Furthermore, the bacteria grow in large aggregates rather than as single cells. To make things even more complex, the initial inoculation steps of working seeds occur on potato slices, posing even greater challenges to process robustness and reproducibility. On the other hand, downstream-processing of the BCG vaccine is quite simple. After harvesting the cultures, the bacteria are filled into ampoules and lyophilized. The exact quantification and quality control of the final product is hampered by the formation of bacterial aggregates during cultivation as well as an undefined proportion of bacteria that are nonviable in intermediates and final product. These factors may influence the efficacy of the BCG vaccine and could also account for variations in the occurrence of side effects.

Recently, technologies for dispersed cultures of M. bovis BCG in synthetic media in small-scale bioreactors were developed.¹³ These cultures allow recording and adjusting of culture parameters and give rise to single bacilli with a high degree of live bacteria. In mouse studies, bioreactor-grown M. bovis BCG exhibited slightly stronger replication and persistence than the vaccine produced under the classical conditions. The protective efficacy in the mouse against challenge with M. tuberculosis was identical for both vaccine preparations. However, the novel manufacturing technique is unlikely to be implemented in commercial production of the BCG vaccine against tuberculosis since such a change would require a clinical trial that may take over a decade.

The production of the S. typhi Ty21a vaccine is based on a master and working seed lot system. During the production process of the Ty21a vaccine, bacteria derived from working seed lot ampoules are inoculated in shake flask cultures, followed by growth in medium-and large-scale bioreactors. Bacteria were harvested by centrifugation (Figure 1). For downstream-processing, the bacteria are mixed with a stabilizer containing sucrose, ascorbic acid, and amino acids, and then lyophilized. The lyophilized bacteria are subsequently mixed with lactose and magnesium stearate as excipients and filled into gelatine capsules that are coated with an organic solution to render them resistant to dissolution in stomach acid. The enteric, coated capsules are then packaged into blister packs for distribution. Each capsule contains 2–10 x 10⁹ (2–6.8 x 10⁹ in the US) lyophilized live bacteria and they are administered orally.¹⁴ Alternatively, a double chambered sachet formulation has been developed, with one sachet containing the lyophilized vaccine and the other containing a bicarbonate buffer to neutralize stomach acidity. The contents of the two sachets are dissolved in 100-mL water and ingested by the vaccinee. The production process is similar for CVD 103-HgR. However, for this vaccine, only the double-chambered sachet formulation was developed, with 2–10 X 10⁸ live bacteria per sachet for travellers from nonendemic regions and 2–10 X 10⁹ for residents of endemic regions.

Figure 1. Ty21a vaccine production: During the production process of the Ty21a vaccine, bacteria derived from working seed lot ampoules are inoculated in shake-flask cultures, followed by growth in medium-and large-scale bioreactors. The bacteria are harvested by centrifugation. For downstream-processing, the bacteria are mixed with a stabilizer and lyophilized.

The release of commercial batches of both vaccines is based on microbiological and biochemical tests. The potency assay for both vaccines relies on determining the live bacteria, and therefore, special attention has to be given to this assay. Another critical test is determining the vaccine purity. Contaminating microorganisms may not be easily detected in a vaccine dose containing more than one billion live vaccine bacteria, and hence, special purity assays needed to be developed. Finally, the attenuated phenotype of the bacteria has to be demonstrated for each vaccine batch.

The manufacturing, quality-control, and release testing of all vaccines have to follow the guidelines issued by regulatory authorities covering cGMP requirements for pharmaceuticals, biologicals, and vaccines. For example, Annex 2 of the PIC-guide to good manufacturing practice for medicinal products, clearly specifies that dedicated facilities should be used for producing the BCG vaccine.¹⁵ There are additional challenges posed by LBVs in comparison to other vaccines. When working with lyophilized live bacteria in large quantities, special cleanroom design and monitoring procedures are required to maintain appropriate cleanroom conditions. Also, relevant biosafety guidelines have to be followed for large-scale manufacturing of live bacteria.

Pharmacopoeia monographs are in place for the release testing of Ty21a and BCG.^16,17 However, after internal testing and release by the vaccine manufacturers, each vaccine batch must undergo additional quality-control testing by regulatory authorities before it can be commercialized.

Conclusion

Live attenuated vaccines have numerous advantages over killed and subunit vaccines. However, they also have higher requirements regarding safety and quality. We have highlighted the challenges faced during preclinical and clinical development, as well as manufacturing and release testing for three attenuated live bacterial vaccines registered for human use—M. bovis BCG, S. typhi Ty21a, and V. cholerae CVD 103-HgR.

When generating attenuated vaccines, attention must be given to the appropriate balance of attenuation and immunogenicity. We have demonstrated the progress in attenuation approaches from the empirical approach of the M. bovis BCG strain to the targeted attenuation of Ty21a and CVD 103-HgR.

In the absence of correlates of protection of the vaccine, large Phase-3 field trials need to be conducted. Apart from ethical considerations caused by exposure of the trial participants to pathogens, potential side effects and safety issues are most prominent topics during the development and clinical trials of a vaccine. The orally administered Ty21a vaccine exhibits an excellent tolerability with an incidence rate of only 0.002% in addition to the absence of reversion to wildtype during the 30 years since its development. This clearly demonstrates the advantage of oral vaccination with a live attenuated vaccine. On the other hand, the issues of performing field studies in absence of the pathogen were highlighted with the example of CVD 103-HgR.

The challenges of manufacturing vaccines originating from the last century were demonstrated with the example of the BCG vaccine. Although novel manufacturing techniques may exist, their implementation will probably require new field trials, and hence, vaccine manufacturers are likely to stick to the historical technologies.

There are several requirements to be fulfilled during the development and cGMP manufacturing of live attenuated vaccines. We have listed the most important challenges and provided examples of how they can be met.

At the time when this article was written, GUIDO DIETRICH was vice president of operations at Berna Biotech AG, Berne, Switzerland. He currently is the director of bulk technologies at GlaxoSmithKline Biologicals, Wavre, Belgium, ANDRE COLLIOUD is the vice president of industrial development at Crucell NV, Berne, Switzerland, and SIMON A. ROTHEN is the CEO of Sartec GmbH, Thoerishaus, Switzerland, +32-10-85-3625, guido.dietrich@gskbio.com

References

1. Sonnleitner B, Locher G, Fiechter A. Automatic bioprocess control. 1. A general concept. J Biotechnol. 1991;19:1–18.

2. Locher G, Sonnleitner B, Fiechter A. Automatic bioprocess control. 2. Implementation and practical experiences. J Biotechnol. 1991;19:127–144.

3. Rothen S A, Saner M, S Meenakshisundaram, Sonnleitner B, Fiechter A. Glucose uptake kinetics of Saccharomyces cerevisiae monitored with a newly developed FIA. J Biotechnol. 1996;50:1–12.

4. Calmette A, Guerin C, Negre L, Bocquet A. Sur la vaccination prE9ventive des enfants nouveau-nE9s contre la tuberculose par le BCG. Ann Inst Pasteur. 1927;3:201.

5. Levine MM, Ferreccio C, Black RE, Germanier R. Large-scale field trial of Ty21a live oral typhoid vaccine in enteric-coated capsule formulation. Lancet. 1987;1:1049–1052.

6. Levine MM and Kaper JB. Live oral vaccines against cholera: an update. Vaccine. 1993;11:207–212.

7. Behr MA, Wilson MA, Gill WP, Salamon H, Schoolnik GK, Rane S, Small PM. Comparative genomics of BCG vaccines by whole-genome DNA microarray. Science. 1999;284:1520–1523.

8. Germanier R, FFCrer E. Isolation and characterisation of Gal E mutant Ty21a of Salmonella typhi: a candidate strain for a live, oral typhoid vaccine. J Infect Dis. 1975;131:553–558.

9. Pasetti MF, Levine MM, Sztein MB. Animal models paving the way for clinical trials of attenuated Salmonella enterica serovar Typhi live oral vaccines and live vectors. Vaccine. 2003;21:401–418.

10. Tacket CO, Cohen MB, Wasserman SS, et al. Randomized, double-blind, placebo-controlled, multicentered trial of the efficacy of a single dose of live oral cholera vaccine CVD 103-HgR in preventing cholera following challenge with Vibrio cholerae O1 El tor inaba three months after vaccination. Infect Immun. 1999;67:6341–6345.

11. Viret JF, Dietrich G, Favre D. Genetic aspects relevant to the biosafety of the recombinant live oral Vibrio cholerae strain CVD 103-HgR. Vaccine. 2004;22:2457–2469.

12. Bloom BR, Murray CJL. Tuberculosis: commentary on a reemergent killer. Science. 1992;257:1055–1064.

13. Dietrich G, Mollenkopf HJ, Weber H, Diehl KD, Knapp B, Kaufmann SHE, Hundt E. Cultivation of Mycobacterium bovis BCG in bioreactors. J Biotechnol. 2002;96:259–270.

14. Dietrich G, Griot-Wenk M, Metcalfe IC, Lang AB, Viret JF. Experience with registered mucosal vaccines. Vaccine. 2003;21:678–83.

15. Pharmaceutical Inspection Convention. Guide to good manufacturing practice for medicinal products. Annex 2 manufacture of biological medicinal products for human use. Available from:www.picscheme.org.

16. Monograph: Typhoid Vaccine (Live, Oral, Strain Ty21a). European Pharmacopiea 5.0/1055

17. Monograph: BCG Vaccine (lyophilized). European Pharmacopiea 5.0/0163.