Optimization of a Human Bacille Calmette-Guérin Challenge Model: A Tool to Evaluate Antimycobacterial Immunity

The quest for an effective tuberculosis vaccine is a public health emergency [1]. There is a critical need for a vaccine that can provide greater and more-consistent protection against tuberculosis than that offered by the only licensed vaccine, bacille Calmette-Guerin (BCG). This is especially important in an era that sees an estimated 9.0 million incident tuberculosis cases and 1.5 million tuberculosis-related deaths each year, exacerbated by an increasing burden of antimicrobial resistance [1]. The lack of reliable and validated immunological correlates of protection hampers the development of tuberculosis vaccines. In the absence of measureable markers to predict candidate vaccine effectiveness, the field has so far relied on animal challenge models and in vitro functional assays that assess the ability of a vaccine to inhibit mycobacterial growth (such as the mycobacterial growth indicator tube [MGIT] [2]). It is unclear how reliably these models forecast in vivo human efficacy. Selecting the best or most-appropriate candidate tuberculosis vaccines for further investment, research, and development is difficult. A safe and relevant mycobacterial human challenge model to allow more-rapid early assessment of candidate tuberculosis vaccines could be a game changer. In general, large and expensive phase 2b and 3 field efficacy trials are required to demonstrate whether safety and immunogenicity results from small phase 1 trials translate into an impact on overall disease burden. Developers of vaccines, including those targeting pathogens responsible for malaria and typhoid [3, 4], have adopted human challenge models to more quickly and inexpensively assess vaccine effectiveness, helping to rationalize which vaccines progress to field efficacy trials. Successful use of human challenge studies has accelerated advancement in these vaccine development pipelines. Safety and ethical reasons preclude the use of Mycobacterium tuberculosis in a human mycobacterial challenge model. We have evaluated an alternative approach: the Mycobacterium bovis BCG challenge model. The model is based on the hypothesis that a tuberculosis vaccine that successfully reduces replication of M. tuberculosis should also reduce BCG replication. BCG is a potentially useful surrogate of M. tuberculosis for a challenge model. Years of use as a licensed vaccine verify its good safety record [5, 6]. BCG, when administered intradermally, causes a self-contained and limited infection. Importantly, the CD4+ T-cell mediated immune responses elicited by both mycobacteria are very similar [7]. Although there are some M. tuberculosis antigens not present in BCG, which, if selected as tuberculosis vaccine candidates, would not be expected to have an impact on BCG replication, the optimization of a BCG human challenge model can still establish the clinical parameters for subsequent challenge models using attenuated M. tuberculosis strains. We previously demonstrated that BCG-vaccinated mice that were later challenged with intradermal BCG had suppressed mycobacterial growth that mimicked observations following intranasal M. tuberculosis challenge, suggesting that a skin mycobacterial challenge may adequately reflect a vaccine effect in the lung [8]. We have recently applied this BCG challenge model to humans. Our pilot study of a human BCG challenge model demonstrated that the degree of growth suppression of BCG can be measured in a punch biopsy specimen the skin from the challenge site where BCG was intradermally administered. We showed that the live mycobacterial load can be quantified up to 1 month after challenge [9]. We found that the model can distinguish between BCG-naive and BCG-vaccinated groups [10], suggesting that prior BCG vaccination gives some protection against a subsequent challenge dose, in a population in which BCG has been shown to be protective [11]. We have also assessed this BCG challenge model in individuals administered the candidate vaccine MVA85A [12]. We found that MVA85A receipt prior to BCG challenge had no effect on the subsequent recovery of BCG, a finding that may be consistent with the results of the recent infant MVA85A efficacy trial or, alternatively, a reflection of the limitations in the model's sensitivity to date [10, 13]. However, a major limitation of the model to date has been low mycobacterial readouts that approached the lower limit of detection, reducing both the sensitivity and ability to detect inter-individual variation in BCG suppression [10]. In this study, we evaluated the effect of both BCG strain and dose on subsequent mycobacterial recovery, with a view to improving model sensitivity and the ability to discriminate between individuals with differing levels of vaccine-induced antimycobacterial immunity. We compare the use of 2 licensed strains of BCG at 2 different doses, to select the most-suitable conditions for BCG challenge for the future testing of tuberculosis vaccine efficacy.