Keywords

1 The BCG Vaccine: History and Approval

Mycobacterium bovis BCG (Bacillus Calmette-Guérin) was generated over a century ago for protection against infection with Mycobacterium tuberculosis and is one the oldest vaccines still in use [1]. BCG originated from a virulent strain of Mycobacterium bovis isolated from the milk of a cow suffering from tuberculous mastitis, collected during studies of bovine tuberculosis. Around 1901, this isolate was transferred to the Institute Pasteur in Lille, France, and used by Albert Calmette and Camille Guérin in their investigations of culturing mycobacteria and potential development of a vaccine against tuberculosis (TB). The organism was difficult to culture and prone to clumping, making controlled subculture, and organism management challenging. To optimise bacterial preparations for their experiments, Calmette added ox bile, a detergent, to the glycerol-soaked potato slices, on which the M. bovis was cultured. Within a few months, a distinct and attenuated isolate with unusual colony morphology appeared. Calmette and Guérin continued the serial in vitro passaging of this M. bovis strain for the next 13 years (1908–1921). During this time, experiments with diverse animal models, including guinea pigs, rabbits, dogs, cattle, horses, chickens and non-human primates, established both the safety and efficacy of BCG, thereby consolidating these observations in support of vaccine development. When administered at different doses and by different routes, BCG was well-tolerated and failed to produce tuberculous lesions. Moreover, BCG vaccination provided protection against challenges with virulent M. tuberculosis strains. The first human trial occurred in July 1921 [2, 3]. An infant was given three 2 mg doses (6 mg total; ~2.4 × 108 colony-forming units) by the oral route. There were no deleterious side effects, and, most importantly, the child did not develop TB, despite the fact the infant’s mother had died of TB shortly after giving birth. Over the next year, additional newborns were vaccinated, and no ill effects were reported. For the first time, a safe and apparently effective vaccine was available for the prevention of human TB. These clinical studies predate the regulatory framework we are now familiar with, and in general, the acceptance and use of vaccines used to be a considerably more informal process. As early as 1924, cultures of BCG were distributed by the Institute Pasteur to laboratories around the world consolidating its position in the fight against TB. The production and application of BCG in public health measures varied from country to country. Because BCG is a live vaccine, it was necessary to transfer cultures to fresh medium every few weeks. Despite standardisation efforts, different passaging conditions were used in different production laboratories; the in vitro evolution of BCG was observed and has since continued to this day. Dozens of distinct daughter strains emerged, including four that are currently in major use: BCG-Pasteur (1173P2), BCG-Japan (Tokyo-172), BCG-Danish (Copenhagen-1331) and BCG-Glaxo (1077) [4, 5].

2 How Is BCG Vaccine Cultured and Manufactured: Fundamental Limitations?

Current production of BCG that is licensed for parenteral administration to humans and veterinary use is by growth of M. bovis BCG in liquid medium in stationary flasks and a pellicle forms at the liquid-air interface [6]. The bulk drug dry substance is prepared by harvesting of the wet pellicle and ball-milling it to generate a suspension of live bacteria, which is then lyophilised. Ball-millers consist of a rotating drum, containing ceramic balls, which grind the BCG and are used for grinding materials such as coal, pigments and feldspar for the pottery industry. The complexity of multiple culturing and processing steps has been challenging to standardise [7]. Current production times are lengthy (~21 days) and labour-intensive, leaving production centres unable to effectively respond to changes in demand or shortages [8,9,10]. There has also been a reduction in the number of BCG manufacturers, over time, particularly in the World Health Organisation (WHO) pre-qualified supplier programme, which has placed BCG availability for global public health mass vaccination programmes, at risk. Even obtaining BCG for recent clinical trials of new vaccines against TB has been a fraught process. A global shortage of BCG lasting months, occurred in 2014–2015, when technical issues halted production at a single site. The pellicle production method also causes quality control issues, where bacterial aggregation means that reliably assessing cell titres is challenging and will often produce highly variable results [9, 11, 12].

Variation in BCG batches, between and even within production centres, has been highlighted as a key issue [13,14,15,16]. The WHO has noted a need for standardising the production and quality assessment of BCG to not only mitigate the variable efficacy seen but to also allow for easier evaluation in clinical trials of “new” and “old” BCG vaccines [8, 11, 17]. If a new vaccine were to include expression of heterologous proteins in recombinant BCG vaccines, this might be a good time to move to a method of production that is more reproducible and defined to achieve standardisation. The current approach leads to clumpy bacteria that must be disrupted by ball-milling, which subsequently results in an ill-defined proportion of dead bacteria that will vary between batches. This in turn, could be impacting the types of immune/protective responses and resulting vaccine efficacy. The slower growth rate of BCG, prepared in this way, has previously been associated with larger scars, post-vaccination, and the increased prevalence of positive PPD (purified protein derivative) skin responses via a Mantoux test, providing further support for the involvement of BCG phenotype as a variable that impacts vaccination outcomes [13]. The quantitative measure for the variations between batches has been limited to colony-forming units, which does not provide full information about culture viability or the proportion of dead bacteria. How much of an impact this has in the context of other factors that potentially affect vaccine efficacy (BCG strain, exposure to environmental mycobacteria and different host-immune profiles) is not known [18,19,20], but a more consistent and quantitative method of production would almost certainly provide a foundation for the assessment of some of these factors. The introduction of different culture methods and medium composition will inevitably impact the physiological state of the organism and subsequent immune/protective immune responses. Our recent study showed that mycobacterial pellicles produced less alpha glucan than planktonically grown bacilli, it followed, that pellicles and their extracted sugars showed a reduction in C3 deposition compared to planktonic bacteria [21]. A previous study explored the impact of growing M. bovis BCG in flasks, on the protective efficacy of BCG in mice, compared to pellicle-grown, ball-milled BCG. The two approaches resulted in a similar level of protection against M. tuberculosis infection [22]. In a further study, inclusion of surfactant in the growth medium reduced encapsulation on BCG. Higher levels of encapsulation promoted a more potent immune response including higher interferon-γ responses, higher polysaccharide-specific capsule antibody, higher IL-17 responses in the spleen and more multifunctional CD4+ T cells [23]. These immune responses correlated well with a reduced bacterial burden in the lung and spleen. Although, as described here, the culturing method will impact the immune response and might alter protective efficacy of BCG, and once these changes are defined and well-established, a more reproducible and consistent approach could be adopted, particularly in controlled systems such as bioreactors or fermenters.

3 Could Fermentation Be an Approach for BCG Vaccine Production?

There has been increasing interest in the possibility of producing the BCG vaccine by growing it in bioreactors, which could address some of the issues around variation between batches and increase BCG yield. Post-genomic studies have adopted the use of fermentation of M. bovis BCG in chemostats under defined and controlled conditions, combined with the screening of transposon mutant libraries to determine the genetic requirements of adaptation to slow growth rate (69 h mean generation time (MGT)) or fast growth rate (23 h MGT) [24]. These studies were later extended to determine an inventory of macromolecules under these different growth rates, and the proportion of lipids in the mycobacterial cell was found to change at the different growth rates, possibly reflective of the amounts of storage lipids [25]. This, and related studies of M. tuberculosis, using defined culture conditions in fermenters, have demonstrated the benefits of controlled conditions during fermentation in reducing heterogeneity and an increase in culture to culture reproducibility [26,27,28]. There have been two key studies that have observed the impact of fermenter growth on vaccine protection in animals infected with M. tuberculosis. The first study, by Dietrich et al. [9], showed that bioreactor-grown M. bovis BCG exhibited similar protection in BALB/c mice compared to pellicle-grown ball-milled BCG [9]. It should be noted that pellicle growth in Sauton minimal medium was required for preparation of the inoculum in the Dietrich study [9], followed by fermentations in the same medium. In contrast, Pascoe et al. [29] reported that a more consistent inoculum could be generated through growth, in a few days, in highly aerated shaking flasks [29]. Curiously, they observed that M. bovis BCG would not grow in Sauton medium using this approach, or in Sauton medium in a fermenter. It is possible that for M. bovis BCG to grow in Sauton medium, it first needs to be passaged and adapt in a clumpy biofilm-like state in static cultures prior to growth in stirred or shaking cultures. This raises concerns about the impact of multiple ill-defined culturing steps on the reproducibility of the resulting biomass and the genetic stability of the strain. The second notable BCG vaccination study in animals from Lesellier et al. [30] demonstrated consistent induction of protection when the efficacy of bioreactor-grown BCG (delivered via the oral cavity) was evaluated in TB-infected badgers [30]. BCG was grown reproducibly whilst aerated at a controlled oxygen level of 10% dissolved oxygen tension, in Middlebrook 7H9 medium, harvested in late exponential phase after 7 days of growth, at a titre of 4.64 × 109 ± 2.20 × 108 colony-forming units (cfu) mL−1. This study neatly demonstrated the advantage of fermentation over conventional production, in obtaining a higher BCG yield, and therefore subsequent increases in the availability of doses of BCG for distribution.

Equally of relevance is the development of new technologies (besides cfu mL−1) for improved BCG characterisation, to ensure consistent outcomes. Fluorescence-labelling and flow cytometry allow for rapid determination of viability and interrogation of population heterogeneity. This type of characterisation has limited application in clumpy, ill-defined cultures and benefits from defined fermentation growth, if only to reduce blockages in the flow cytometer. Dietrich et al. [9], used Fluorassure reagent to quantify viable bacteria in their bioreactor growth. The reagent is taken up by metabolically active cells and cleaved by esterases to become fluorescent, with subsequent intracellular accumulation. Similarly, and more recently, Gweon et al. [31] assessed a method to measure viability, using fluorescein diacetate, which is also cleaved by intracellular esterases, used flow cytometry, compared to viable counts, and showed a strong positive linear relationship between the two methods [31]. There was significantly reduced covariance between replicate batches of BCG using fluorescence as a measurement. Esterases can still be detected in dead cells, allowing for false positive results: therefore, inclusion of a second dye that simultaneously measures the dead population would help to negate these issues. Hendon-Dunn et al. [32, 33] developed a flow cytometry method, using the two dyes, Calcein Violet-AM (labels metabolically active bacteria) and Sytox Green (labels the DNA of dead/compromised bacteria), which allows for the proportions of live or dead bacteria to be quantitatively determined and for non-culturable bacteria to be captured. This flow cytometry method was applied for the characterisation of bioreactor growths during studies to optimise BCG fermentation at Public Health England, Porton Down (Fig. 9.1) [29]. These studies identified Roisin’s medium as an optimal medium for fermentation, as there was comparable growth compared to Middlebrook 7H9. Roisin’s medium has the advantage of being a defined medium for which quantitative substitutions can be made for individual medium components [24, 25, 34]. This creates an opportunity for further optimisation of the medium for BCG production by fermentation to determine the impact of larger scale fermentation on the cultivation of BCG and protection/virulence studies in animal models. Other conditions for consideration are aeration and the inclusion of detergent in culture. In any bioreactor culture, detergent is necessary for generating single cells or smaller clumps of bacilli in a more reproducible and homogenous manner. However, consideration needs to be given to the presence of detergent because of changes to the cell wall and capsular composition of BCG, which in turn will impact the efficacy of BCG vaccination [23].

Fig. 9.1
Six scatter plots. a and b plot C F U m l power negative 1 versus time in a day. c plots O D subscript 600 nanometer versus time in a day. d plots bacterial per m l versus time in a day. e plots median sytox green fluorescence versus time in a day. f plots median Cacien violet fluorescence versus time in a day. a and b give values for 7 H 9 and Roisins, respectively. c and d gives values for both 7 H 9 and Roisins. e and f give values for 7 H 9, Roisins, and heat-killed.

The total viable counts of fermenter-grown M. bovis BCG in 800 mL of either Middlebrook 7H9 (a) or Roisin’s minimal medium (b) over 12 days of culture in batch fermenters. The cultures were sampled daily for total viable counts (CFU mL−1) (a, b) and turbidity (c). Samples were taken on days 0, 4, 6, 7 and 12 for total bacterial counts (bacteria/mL) determined using fluorogenic beads and flow cytometry (d). Samples were taken on the same days for analyses of cell death and metabolic state using flow cytometry and fluorescent dyes, Sytox Green (e) and Calcein Violet-AM (f), respectively. A heat-killed M. bovis BCG control was included. Data represent the mean average of three independent culture repeats ± standard error bars. (Source: Pharmaceutics 2020 Sep 22;12(9):900 doi: https://doi.org/10.3390/pharmaceutics12090900 [29])

Full size image

4 BCG Product Characterisation

The Bill of Testing for the BCG vaccine, particularly, the batch release testing requirements, is limited compared to any of the more modern live vaccines in use or in clinical study (Table 9.1). This reflects the long history of the BCG vaccine and its passage from pre-National Regulatory Authority (NRA) regulations, into the International Committee of Harmonisation coordination of regulatory advice and regulations. In the last decade, improved technologies have provided additional avenues for BCG product characterisation, some of which have gained some traction with regulators and have been included in guidance documents (Table 9.2). Although progress has been made from both technology and assay development perspectives, hurdles remain in bringing some of these newer assays into the regulatory and batch release vaccine testing spaces. To do so, it requires significant investment and development time to ensure new approaches provide relevant information regarding the quality and functionality of the BCG vaccine. The International Committee of Harmonisation (ICH, www.ich.org) has developed a series of guidelines that provide the laboratory and data package framework and is required to bring new assays into the regulatory environment (ICH Q2(R1); Q2(R2)/Q14 EWG; Q5C; Q5E). An aspect of BCG characterisation that remains unresolved are the assessment and measurement of potency that reflect clinical efficacy in clinical studies and in normal use. No agreed correlate of immunity has yet to be agreed, although some biomarkers of relevant biological activity have been studied for potential application [14]. The only accepted potency readouts for BCG, currently used by manufacturers and the NRA, are viable counts and relative viability against total bacterial numbers, using a variety of technologies [31, 35, 36]. This has been shown to have some association with clinical outcome, but no straightforward dose response curve has been determined, and relative (to a reference BCG vaccine) potency is difficult to quantitate. In vivo assessments (mouse protection studies, guinea pig protection studies) are complex, long-term and not amenable to manufacturing or batch release activities. A new assay that can determine in vitro bacterial inhibition has been studied extensively, measuring the reduction in bacterial replication by leucocytes in vitro culture. The assay has proven to be highly variable, and although much has been done to standardise protocols and measurement, there remains some way to go in showing sufficient specificity, reproducibility, and sensitivity, for routine use. Moreover, there is no clear evidence of validations of assay outcomes and clinical effects (a requirement of modern potency assays for biologicals) [37,38,39].

Table 9.1 European Union pharmacopeia; current BCG testing recommendations
Full size table
Table 9.2 Novel or recently included BCG vaccine characterisation tests
Full size table

5 Regulatory Considerations

BCG is an authorised vaccine that is licensed for use, based on a range of evidence provided in the registration dossier, which include detail on the process of manufacture, in process control testing (and required outcomes), along with a full quality and Bill of Testing data package, for the product. As BCG has been used as a vaccine for a long period of time, the information contained in the regulatory documents accumulated prior to current regulatory frameworks and legislation came into place. Therefore, much of the data, quality control, and manufacturing information, that is, available for other current live vaccines, have not been produced for BCG. Existing product data for BCG has simply been incorporated into more recent legislative and operational frameworks, which presents challenges in the compatibility of the available data with regulatory requirements. Making significant changes in the manufacturing process, such as a switch to fermentation, may change the nature of the product, particularly with a live vaccine such as BCG [47]. Major redesigns of the manufacturing process are likely to trigger reviews by the NRA and a recommendation to consider fermented BCG as a new product, which would negate the existing licensure for use [48, 49]. A new product may require new evidence of bioequivalence and may attract a requirement for preclinical studies as well as clinical trials from Phase I through to efficacy studies [50]. The WHO recommendations to assure the quality, safety and efficacy of BCG vaccines (Annex 3 [51]) state that production of new BCG vaccine (i.e. a modified strain of BCG) is a prerequisite for initiation of clinical studies [52]. This places a large developmental burden on existing BCG manufacturers but may allow new products into the manufacturing space, as they will be undergoing testing to the current standards [53]. It should also be remembered that BCG is a major player in global health, and it remains a low-cost vaccine for both the user and consumer. The latter negatively impacts the economic pressure to change or improve the existing manufacturing processes. Because of these competing pressures, loss of manufacturing flexibility and the previously noted constraints of existing BCG availability, the WHO supported a discussion (Montreal, October 2015) on how to move to a more standardised and modern manufacturing process for BCG. Representatives from NRAs, manufacturers, and leading mycobacterial experts, attended. No clear way forward was agreed, which epitomises the complexities around establishing major change in BCG manufacturing processes, further restricting the options for the improvement of BCG manufacture and availability [8].

New live TB vaccines, including those based on BCG, will be travelling a conventional registration pathway and as such will suffer none of the manufacturing constraints described here. No new live TB vaccine has achieved approval and the time frame for such approvals remains unclear. Therefore, the concerns surrounding BCG manufacture, yield, and availability, remain significant considerations for all involved. What remains unclear is the degree of bioequivalence of BCG between differing manufacturing processes, as well as the functional comparisons of the new live TB vaccines. Comparisons of new vaccine candidates with BCG remain part of the product development pathway and are incorporated into the preclinical and clinical studies designs already undertaken or planned. The effect of manufacturing on functional bioequivalence in terms of the regulatory Bill of Testing for release of BCG vaccines for human use is unknown. The WHO meeting supported the development of a study plan to directly compare classical pellicle production with bioreactor-grown BCG, in the battery of tests commonly used for the batch release of BCG, as well as a more exploratory preclinical efficacy study [22]. This study showed bioequivalence between these two manufacturing processes. Interestingly, in the exploratory preclinical efficacy studies, some minor differences were observed, the significance of which remain unclear. Further studies expanding on these observations are needed to inform the NRA, WHO and public health bodies, on the benefits and possible risks in changing manufacture of BCG, or to develop a strategy for BCG replacement, if deemed appropriate.

6 Concluding Remarks

Considerable work has been done in refining and optimising fermentation technologies and approaches for BCG production, including the growth medium, the supplements, and the methods of BCG characterisation, all of which have been subject to review and assessment. The resulting published studies have informed potential manufacturers and vaccine developers of the approaches that are available and the issues of yield, comparability, and consistency of manufactured BCG characteristics. Despite much discussion at the highest levels, there seems to be no clear path to the improvement of current licensed BCG manufacturing processes though adoption of fermentation/bioreactor technology. There are multiple considerations, as noted here, ranging from regulatory constraints to the cost of changing the technology and the impacts of obtaining authorisation for a major change in the manufacturing procedures. There may be new approaches to these issues available with the development of a more flexible and pragmatic regulatory landscape, which has developed with the pressures and timelines of the COVID-19 vaccine development efforts, which have been so effective in delivering effective and safe vaccines in record time.