The Safety and Immunogenicity of a TB Vaccine; MVA85A, in Healthy Volunteers Who Are Infected With HIV

The need for new vaccine against tuberculosis Tuberculosis (TB) kills about three million people annually. It is estimated that one third of the world's population are latently infected with Mycobacterium tuberculosis (M.tb) (Dye, 1999). These latently infected individuals are at risk of reactivation of infection, should they become immunosuppressed. Worldwide, coinfection with HIV is the commonest cause of immunosuppression and increases the chances of reactivation from a 10% lifetime risk to a 10% annual risk (Corbett, 1996). The currently available vaccine, M. bovis BCG, is largely ineffective at protecting against adult pulmonary disease in endemic areas and it is widely agreed that a new more effective tuberculosis vaccine is a major global public health priority (Colditz, 1994). However, it may be unethical and impractical to test and deploy a vaccine strategy that does not include BCG, as BCG does confer worthwhile protection against TB meningitis and leprosy. An immunisation strategy that includes BCG is also attractive because the populations in which this vaccine candidate will need to be tested will already have been immunised with BCG. Given the high prevalence of infection with M.tb, a vaccine that could be administered to latently infected individuals and eradicate latent infection would have an enormous impact on the mortality and morbidity from TB. M.tb is an intracellular organism. CD4+ Th1-type cellular responses are essential for protection and there is increasing evidence from animal and human studies that CD8+ T cells also play a protective role (Flynn, 2001). However, it has generally been difficult to induce strong cellular immune responses in humans using subunit vaccines. DNA vaccines, recombinant viral vectors and protein/adjuvant combinations all induce both CD4+ and CD8+ T cells, however none of these antigen delivery systems induce high levels of antigen specific T cells, when used alone.

Heterologous prime-boost immunisation strategies involve giving two different vaccines, each encoding the same antigen, several weeks apart. Using a DNA prime-recombinant modified vaccinia virus Ankara (MVA) boost induces higher levels of antigen specific CD4+ and CD8+ T cells than using homologous boosting with the same vector in a number of different disease models (Schneider, 1998; McShane, 2001). Given the protective efficacy of BCG in childhood, ideally BCG would be the priming immunisation in such a prime-boost strategy. In order to do this, we have focused on antigen 85A as a candidate antigen. Antigen 85A is highly conserved amongst all mycobacterial species and is present in all strains of BCG. Antigen 85A is a major secreted antigen from M. tuberculosis which forms part of the antigen 85 complex (A, B and C). This complex constitutes a major portion of the secreted proteins of both M.tb and BCG. It is involved in fibronectin binding within the cell wall and has mycolyltransferase activity. Antigen 85A is immunodominant in murine and human studies and is protective in small animals (Huygen, 1996).

Recombinant modified vaccinia virus Ankara (rMVA). Many viruses have been investigated as potential recombinant vaccines. The successful worldwide eradication of smallpox via vaccination with live vaccinia virus highlighted vaccinia as a candidate for recombinant use. The recognition in recent years that non- replicating strains of poxvirus such as MVA and avipox vectors can be more immunogenic than traditional replicating vaccinia strains has enhanced the attractiveness of this approach. MVA (modified vaccinia virus Ankara) is a strain of vaccinia virus which has been passaged more than 570 times though avian cells, is replication incompetent in human cell lines and has a good safety record. It has been administered to more than 120,000 vaccinees as part of the smallpox eradication programme, with no adverse effects, despite the deliberate vaccination of high risk groups (Stickl, 1974; Mahnel, 1994). This safety in man is consistent with the avirulence of MVA in animal models. MVA has six major genomic deletions compared to the parental genome severely compromising its ability to replicate in mammalian cells (Meher, 1991). No replication has been documented in non- transformed mammalian cells. Viral replication is blocked late during infection of cells but importantly viral and recombinant protein synthesis is unimpaired even during this abortive infection. The viral genome has been proven to be stable through a large series of passages in chicken embryo fibroblasts. Replication-deficient recombinant MVA has been seen as an exceptionally safe viral vector. When tested in animal model studies recombinant MVAs have been shown to be avirulent, yet protectively immunogenic as vaccines against viral diseases and cancer. Recent studies in severely immuno-suppressed macaques have supported the view that MVA should be safe in immuno-compromised humans (Akira, 2001; Stittelaar, 2001). There is now safety data from a number of recombinant MVAs that are currently in Phase I/II trials in both the UK and Africa. Useful data on the safety and efficacy of various doses of a recombinant MVA vaccine comes from clinical trial data with a recombinant MVA expressing a number of CTL epitopes from Plasmodium falciparum pre-erythrocytic antigens fused to a complete pre-erythrocytic stage antigen, Thrombospondin Related Adhesion Protein (TRAP). To date MVA ME-TRAP has been administered to over 600 healthy volunteers (adults and children) in Oxford and Africa (The Gambia and Kenya) without any serious adverse events (Adrian Hill, unpublished, personal communication). Volunteers have received one to three doses of from 3 to 15 x 107 pfu per dose of intra-dermal vaccine at three-week intervals. All subjects have temporary local redness with typically a 5mm central red area with a paler pink surrounding area that ranges in size from about 1 -7cm in diameter and peaks at 48 hours post vaccination. At seven days post vaccination generally only the central red area remains. This fades over the next few weeks and is usually not apparent at 2 months after vaccination. The emerging safety profile of recombinant MVA vaccine is excellent and supported by data from clinical studies of three other MVA recombinants made in Oxford and currently in clinical studies using MVAs for HIV, HBV and melanoma. To date these vaccines have been administered to over 600 people with no serious adverse events (Hill; personal communication) 40 HIV positive individuals treated with highly active anti-retroviral therapy (HAART) have been vaccinated with at least 5 x 107 PFU MVA containing HIV antigens with no serious adverse events (Dorrell, unpublished data; Cosma et al 2003; Harrer et al, 2005). 7 HIV positive individuals have been vaccinated with MVA containing malaria antigens with no serious adverse events and no significant or sustained rise in viral load (Bejon at al, CID 2006, in press).

Recombinant MVA encoding antigen 85A MVA85A induces both a CD4+ and a CD8+ epitope when used to immunise mice. When mice are primed with BCG and then given MVA85A as a boost, the levels of CD4+ and CD8+ T cells induced are higher than with either BCG or MVA85A alone, and this regime is more protective than either vaccine alone (Goonetilleke et al, 2003). In the more sensitive guinea pig model, guinea pigs vaccinated with BCG, and then MVA85A, and then a second viral vector, fowlpox expressing antigen 85A, 6/6 guinea pigs are alive at the end of the experiment, compared with 2/6 guinea pigs vaccinated with BCG alone, and 0/6 control animals (Williams et al, 2005).

In rhesus macaques, this BCG prime-MVA85A and Fowlpox85A boost is more immunogenic than any of the vaccines alone, and is more protective than BCG alone (verrek et al, unpublished data).

Clinical studies using MVA85A MVA85A (at a dose of 5 x 107pfu) has been administered to 48 healthy volunteers in the UK, 21 healthy volunteers in The Gambia and 18 in South Africa, with no serious adverse events. We have designed our Phase I studies to allow for a vaccination of volunteer groups sequentially with a step-wise increase in mycobacterial exposure, in order to minimize the possibility of a Koch reaction. A Koch reaction describes the development of immunopathology in a person or animal with tuberculosis, when an exaggerated immune response to M.tb is stimulated. It was described in patients with TB disease when Koch performed his original studies employing mycobacteria as a type of therapeutic vaccination. It has now been demonstrated in the mouse model of therapeutic vaccination (Taylor, 2003). Available animal data suggest that these reactions do not occur in mice latently infected with M.tb, suggesting that such reactions may correlate with high bacterial load and that the Koch phenomenon may not pose a problem for vaccination of healthy albeit latently infected humans. We started these studies in healthy volunteers who were as mycobacterially naïve as possible. They were skin test negative and Elispot negative for PPD, ESAT 6 and CFP10, and had not had previously been vaccinated with BCG. We have now completed studies in the UK vaccinating volunteers previously vaccinated with BCG (McShane, Nature Med, 2004) and in The Gambia.

There are two ongoing studies in Oxford. The first is a phase I trial of the safety and immunogenicity of MVA85A in individuals who are latently infected with M.tb. We have now vaccinated all 12 volunteers with no serious adverse events. The second is a dose selection study evaluating the safety and immunogenicity of 2 different doses of MVA85A (1x10^8 and 1x10^7 pfu), in healthy adult volunteers who have previously been vaccinated with BCG. Here we have vaccinated 11 individuals with the higher dose (1x10^8) with no serious adverse events, the main side effect being a fever at 24-48 hours post-vaccination, which completely resolves over 24 hours. Results from the low dose arm will be collected over the next few months.