Clinical development of Ebola vaccines

Abstract

The ongoing outbreak of Ebola virus disease in West Africa highlighted the lack of a licensed drug or vaccine to combat the disease and has renewed the urgency to develop a pipeline of Ebola vaccines. A number of different vaccine platforms are being developed by assessing preclinical efficacy in animal models and expediting clinical development. Over 15 different vaccines are in preclinical development and 8 vaccines are now in different stages of clinical evaluation. These vaccines include DNA vaccines, virus-like particles and viral vectors such as live replicating vesicular stomatitis virus (rVSV), human and chimpanzee adenovirus, and vaccinia virus. Recently, in preliminary results reported from the first phase III trial of an Ebola vaccine, the rVSV-vectored vaccine showed promising efficacy. This review charts this rapidly advancing area of research focusing on vaccines in clinical development and discusses the future opportunities and challenges faced in the licensure and deployment of Ebola vaccines.

Introduction

The current outbreak of Ebola virus disease (EVD) has ravaged West Africa with over 27,000 cases and 11,000 deaths recorded since the outbreak began in December 2013 [WHO, 2015]. Despite a slowdown in the outbreak, cases continue to emerge in Sierra Leone and Guinea while Liberia, declared Ebola free in May 2015, has reported the re-emergence of a few new cases [WHO, 2015]. Multiple outbreaks of EVD, predominantly in Central Africa, have occurred since the virus was first discovered in the Democratic Republic of Congo in 1976. The current outbreak in West Africa is the largest recorded in history and has affected the widest geographical area. The scale and persistence of the current outbreak has refocused attention on the medical and public health strategies available to combat this disease. Outbreaks of EVD are characterised by high mortality rates ranging from 50% to 90%; treatment options are entirely symptomatic and supportive with no specific antiviral drugs available. Neither is a licensed vaccine available and therefore Ebola control has relied on case identification, early medical care, minimizing transmission and protecting those in close contact with cases, particularly healthcare workers. The ongoing outbreak has underlined the urgent need for a vaccine and has led to the expedited development of vaccines against EVD. This review charts this rapid clinical development of Ebola vaccines and the challenges faced in deployment and licensure of an Ebola vaccine.

Ebolavirus

Genus Ebolavirus is one member, along with genus Marburgvirus and Ceuvavirus, of the family Filoviridae. The genus Ebolavirus is comprised of five distinct species: Bundibugyo (BDBV), Reston (RESTV), Sudan (SUDV), Taï Forest (TAFV) and Ebola virus, formerly designated Zaire ebolavirus (EBOV). The first species discovered were EBOV and SUDV from two separate outbreaks in Southern Sudan and the Democratic Republic of Congo in 1976 [WHO, 1978a, 1978b]. All species except RESTV cause disease in humans, with SUDV and EBOV the most pathogenic and the predominant cause of most EVD outbreaks. The 2014 Guinea strain causing the current outbreak in West Africa has 98% sequence homology to the Zaire ebolavirus species (EBOV).

Ebolavirus is an enveloped virus that contains a 19 kb single-stranded, negative-sense RNA genome encoding 7 proteins. The virion core consists of the RNA genome, a nucleoprotein (NP) that binds the genomic RNA, and the nucleocapsid viral protein 30 (VP30), which is surrounded by a lipid envelope with surface projections that are comprised of a glycoprotein (GP). The surface GP is a multimer of a single structural GP; it functions in cell attachment, fusion and cell entry, helps in immune evasion and likely plays a role in pathogenesis of disease [Sullivan et al. 2003].

This central role of the viral GP makes it a key antigenic target for development of Ebola vaccines and monoclonal antibodies. Analysis of the GP sequences between species of Ebola virus showed a high degree of diversity at the nucleotide and amino acid level (~60–65% conservation) [Sanchez et al. 1996, 1999]. This implies that GP targeting vaccines will need to be multivalent, encoding GPs specific for each species. In contrast, across strains within a species, there is high conservation (~97–98%) in the GP nucleotide sequence [Sanchez et al. 1996, 1999] and sequence of key GP epitopes [Ponomarenko et al. 2014]. This would suggest that vaccines using older strains of EBOV might provide cross-protection against the 2014 strain and future outbreaks of EBOV, although whether this is the case is a key question being currently evaluated.

Epidemiology and disease pathogenesis

EVD is a zoonotic disease in which humans and nonhuman primates (NHPs) are accidental hosts. Humans become infected when they come in close contact with infected animals such as monkeys, gorillas and bats found in forests during hunting, butchering or preparing meat. Although the natural host and reservoir of Ebola virus remains unknown, sequences of Ebola virus genome recovered from fruit bats and epidemiological data during an outbreak tracing human infection to fruit bats suggest that fruit bats are the most likely host reservoirs [Leroy et al. 2005, 2009]. Human-to-human transmission occurs through contact with the bodily secretions, the blood or skin of infected individuals, and fomites (clothing, surfaces, bedding) contaminated with the bodily fluids of infected individuals. The likelihood of transmission is dependent on the type of bodily fluid and amount of virus contained, with blood, vomit and faeces being the most infectious. Infected individuals only transmit virus once symptomatic, which occurs between 2 and 12 days after infection, and infected individuals remain contagious as long as they are viremic or virus persists in bodily fluids. Virus transmission has been identified through semen and breast milk even in the absence of viremia, and virus has been isolated from urine, tears and aqueous humour several weeks after the end of viremia [Bausch et al. 2007; Christie et al. 2015; Sonnenberg and Field, 2015; Varkey et al. 2015].

EVD begins with acute onset of fever, headache and myalgia that rapidly progresses to multi-organ failure with internal and external bleeding. Symptoms of EVD are a result of rapid viral infection of multiple cell types and unusually high rate of viral replication in infected cells. High levels of viremia, greater than 10 million copies per ml are associated with fatal human infections in comparison with those who survive, who have much lower levels of viremia [Schieffelin et al. 2014; McElroy et al. 2015]. At the same time, immunosuppression and uncontrolled inflammation are the causes of disease pathogenesis. Infection of endothelial cells causes widespread destruction of epithelial barriers and uncontrolled coagulopathies, while infection of monocytes and macrophages result in secretion of proinflammatory mediators setting up a cytokine storm [Baize et al. 2002]. Suppression of the innate immune response particularly the type I interferon pathway and inhibition of maturation of dendritic cells [Wauquier et al. 2010] results in reduced induction of the adaptive immune response, particularly T cells. In contrast to this lack of an adaptive immune response, patients who survive show an unprecedented lymphocytosis and T-cell activation, particularly CD8⁺ T cells, that can last up to 1 month after onset of symptoms alongside a rapid induction of serum antibodies [Baize et al. 1999] and a 10–50 fold increase in plasmablasts [McElroy et al. 2015]. Thus, fatal ebolavirus infection is characterized by broad immunosuppression with dysregulated innate immune response and little or no adaptive response causing uncontrolled viral replication with resultant pathology leading to multi-organ failure.

Immune correlates of protection

An immune correlate of protection may be an immune marker related to protection against disease or infection but which is not necessarily the immune mechanism conferring protection [Plotkin, 2013]. Retrospective opportunistic studies of survivors studying immunological differences between fatal cases and survivors have been instrumental in furthering our understanding of potential immune mechanisms of protection, albeit that they do not allow identification of a correlate of protection. Activated T cells, particularly CD8⁺ T cells, and rapid induction of antibodies have been observed in nonfatal Ebola patients suggesting a role for these immune mechanisms in protective immunity. Studying protective mechanism in EVD has depended on developing relevant animal models that can mimic all the clinical parameters of disease in humans. Rhesus and cynomolgus macaques have significant advantages over murine and rodent models of EVD. The symptoms and course of disease in NHPs reflects human disease and strains that infect humans can be used without laboratory adaptation. However, insights into protective mechanisms are best studied in murine models, particularly inbred mouse strains that allow manipulation of experimental systems.

A protective role for antibodies against ebolavirus was suggested by significantly higher and more rapid induction of virus-specific immunoglobulin (Ig) M responses in survivors compared with nonsurvivors [Baize et al. 1999]. Monoclonal antibodies isolated from patients showed the ability to neutralize virus in vitro [Maruyama et al. 1999] and passive transfer of monoclonal antibodies conferred pre and post exposure protection against Ebola in different animal models [Parren et al. 2002; Dye et al. 2012; Qiu et al. 2012a, 2012b]. DNA vaccination encoding Ebola virus GP induced high titres of GP-specific IgG antibodies in mice who survived an Ebola virus challenge. However, although DNA vaccine-induced protection in murine models was associated with high titre GP-specific IgG antibodies, passive transfer of hyperimmune sera did not protect naïve animals. This, in concert with evidence of protection with monoclonal antibodies, suggests that the quality and epitopes targeted by antibodies are critical in protective immunity. Nevertheless, IgG antibody titre to virus GP provided a quantitative correlate of protection following vaccination. This correlation between high titre IgG antibodies pre challenge and survival was observed with Ebola vaccines based on rAd5, rAd5-DNA, replicating vesicular stomatitis virus (rVSV) and virus-like particles (VLPs) encoding Ebola virus GP both in murine and NHP models [Sullivan et al. 2000; Wong et al. 2012; Jones et al. 2005; Warfield et al. 2007]. Thus, IgG antibodies to the Ebola virus GP measured by enzyme-linked immunosorbent assay (ELISA) are the best currently available quantitative correlate of protection.

Cellular immune responses also play an important role in mediating protection [Sullivan et al. 2009]. Adoptive transfer of Ebola virus NP-specific CD8⁺ T cells in mice was able to protect naïve mice against lethal viral infection [Wilson and Hart, 2001]. Depletion of GP-specific CD3⁺ and CD8⁺ T cells induced after rAd5 vaccination encoding viral GP abrogated vaccine-induced protection in NHPs highlighting the role of antigen-specific CD8⁺ T cells in protective immunity induced by viral-vectored vaccine [Sullivan et al. 2011]. Interestingly, the role of CD8⁺ T cells was also observed with other vaccine platforms including rVSV [Wilson and Hart, 2001; Wong et al. 2012]. Despite the experimental evidence for T-cell mediated protection from animal models and a correlation between frequency of interferon-γ (IFNγ) secreting T cells measured by enzyme-linked immunospot (ELISPOT) or flow cytometry with protection [Wong et al. 2012], a quantitative T-cell correlate of protection remains to be determined.

Vaccine development

The first wave of Ebola vaccine development, beginning soon after the discovery of the virus, focused on attempts to inactivate the virus. Since then, preclinical development of a variety of different platforms including DNA vaccines, recombinant viral vectors, recombinant proteins, subunit proteins and VLPs have been progressed. Vaccines are generally advanced to clinical development when efficacy in preclinical models or promising immunogenicity at putatively protective levels is observed.Clinical trials seek to evaluate vaccine reactogenicity and immunogenicity to identify optimal doses and schedules for generating protective immunity in the target population. There are eight vaccines in clinical trials (reviewed below), all targeting the Ebola virus GP but differing in the predominant immune response induced, the manner in which the antigen is delivered and vaccine reactogenicity (Table 1).

Table 1.

Ebola vaccine strategies in clinical development.

Vaccine	Multivalent/ monovalent	Antigen and strain	Strategy	Phase of clinical development
ChAd3.EBOZ	Monovalent/ bivalent	GP from Mayinga 1976 EBOV, Gulu 1977 SUDV strain	One-shot high dose/ prime-boost with MVA	Phase IIb/III
rVSV.EBOZ	Monovalent	GP from Kikwit EBOV strain	One-shot	Phase III
Ad26.ZEBOV-	Monovalent	GP from Mayinga strain EBOV	Prime-boost with MVA	Phase IIb
MVA BN Filo	Multivalent	GP from EBOV, SUDV, Marburg and NP from TAFV	Prime-boost	Phase IIb
MVA.EBOZ	Monovalent	GP from Mayinga EBOV	Prime boost	Phase I
GP VLP	Monovalent	GP from Makona 2014 EBOV	One shot	Phase I
rAd5.EBOV	Monovalent	GP from Guinea 2014 EBOV	High dose single shot	Phase Ib
DNA plasmid (EBODNA023-00-VP)	Bivalent	GP from Mayinga 1976 EBOV, Gulu 1977 SUDV strain	Prime-boost	Phase Ib

EBOV, Ebolavirus Zaire species; GP, glycoprotein; MVA, Modified Vaccinia Ankara; NP, nucleoprotein; SUDV, Ebolavirus Sudan species; TAFV, Taï Forest virus; VLP, virus-like particle.

DNA vaccines

DNA vaccines, where plasmids are used to encode the antigen against which an immune response is needed, are a very attractive choice for the development of vaccines against diseases causing sporadic outbreaks. DNA vaccines can be rapidly adapted for new strains or viruses, clinical grade product can be manufactured in large quantities, they are noninfectious and the production of target antigen in situ results in induction of both humoral and cellular immune responses. Candidate DNA vaccines have been evaluated for a number of diseases and have an excellent safety profile, although they are relatively poor immunogens requiring high doses and repeated vaccinations to achieve strong immune responses.

Two DNA vaccines encoding Ebola GP have been tested in phase I clinical trials. The first DNA Ebola vaccine to be tested in a clinical trial encoded a transmembrane-deleted GP sequence from the Zaire strain of EBOV and Gulu strain of SUDV along with the NP from the Zaire strain of EBOV. This vaccine, administered by intramuscular injection, was tested in 21 individuals in a dose-escalation study in a 3-dose regime [Martin et al. 2006]. Although the vaccines induced mild-to-moderate side effects, repeated vaccinations with a high dose was necessary to achieve high antibody titres. A second DNA vaccine in clinical trial was modified to use the full length wildtype (WT) GP and contained two DNA plasmids in equal proportion expressing the WT GP of EBOV and SUDV. This vaccine, also administered by intramuscular injection, was tested in 10 healthy adults in the US at a dose of 4 mg administered 3 times 4 weeks apart with a subset of participants administered a booster dose 32 weeks after the last vaccination [Sarwar et al. 2015]. Although, the vaccine generated antibody titres in 70% (7/10) of individuals, T-cell responses were poorly induced. Immune responses also waned rapidly and a fourth booster vaccination was needed to improve the longevity of immune responses [Sarwar et al. 2015]. This DNA vaccine was also tested in a phase Ib trial in 60 healthy adults in Uganda and included a group where this vaccine was co-administered with another DNA vaccine encoding Marburg virus GP [Kibuuka et al. 2015]. This study demonstrated two key findings. First, vaccine-induced antibodies were comparable between African and North American participants. This has significant implications for evaluating vaccines in Europe and North America in phase I studies before advancing them for use in the population of interest [Sridhar, 2015]. Second, concomitant administration of Marburg and Ebola DNA vaccines did not interfere in the induction of immune responses to either GP, which provides proof-of-concept for co-administration of filovirus vaccines.

These studies, albeit promising, highlight one of the major disadvantages of using DNA vaccines, namely the need for multiple doses and booster vaccinations to maintain durable immune responses. Further work continues on improving DNA vaccines through electroporation techniques, which allow administration of larger quantities of vaccine, and employing DNA vaccines in a prime-boost approach with other vector delivery platforms.

Recombinant viral vectors

Viruses engineered to serve as antigen delivery platforms have become increasingly popular because of their ability to induce cellular immune responses as antigen are expressed and processed intracellularly from cells infected by the recombinant virus. Viral vectors can be either replication-competent or replication-deficient. Replication-deficient viruses are engineered by deleting viral genes essential for replication and using cell lines complemented with these viral genes to grow these recombinant viruses. These advantages of viral vectors make them a popular choice especially for diseases in which cellular immune responses may play a role in protection, although the quality, magnitude and durability of immune responses induced depends on the type of viral vector. The most advanced Ebola vaccines in clinical development are viral vectors encoding the full length GP of Ebola virus; recombinant human adenovirus, recombinant simian adenovirus, recombinant vaccinia virus (VV) and a live vesicular stomatitis virus (VSV).

Recombinant human adenovirus vectors

Adenoviruses are nonenveloped double-stranded DNA viruses, a vast number of which have been isolated from mammalian species. Adenoviruses are classified into serotypes based on the neutralizing antibodies to the hexon protein on the surface of the adenoviral capsid and these serotypes are divided into six subgroups or species (A–F) based on their sequence homology and haemagglutination properties [Colloca et al. 2012]. Human adenoviruses of serotype 5 (Ad5), serotype 1 (Ad1) and serotype 2 (Ad2) are ubiquitous and cause mild upper respiratory tract infections. Other serotypes can cause conjunctivitis and gastroenteritis, and rarely can cause fatal infections like pneumonia and meningoencephalitis in children and immunocompromised individuals.

Adenoviruses are used as recombinant vectors by deleting the E1 region to make the virus replication-deficient. Recombinant adenoviral vectors of human Ad5 are very popular vaccine vectors because of their ease of manipulation, ability to grow to high titres, and the strong cellular and humoral immunity induced against the encoded antigen [Small and Ertl, 2011]. Replication-deficient recombinant Ad5 (rAd5) vector expressing Ebola virus GP has shown 100% efficacy in NHPs as a single dose [Sullivan et al. 2000, 2006, 2011] as well as when used in a DNA-rAd5 prime-boost vaccine regime [Sullivan et al. 2000]. The first clinical trial of a rAd5 vaccine encoding an optimized GP from historical strains of the SUDV and EBOV species evaluated a dose of 2 × 10⁹ viral particles (vp) and 2 × 10¹⁰ vp in 24 healthy volunteers. The higher dose (2 × 10¹⁰ vp) induced Ebola GP-specific IgG antibodies in all participants and the antibodies induced were more durable than those induced by the lower dose. Detectable neutralizing antibodies were induced only in 1/24 participants and, surprisingly, T-cell responses induced by this vaccine were poor. Vaccine-induced antigen-specific T-cell responses measured by IFNγ ELISPOT were induced in 25–45% of participants across the study to either SUDV or EBOV GP. Using flow cytometry to characterize GP-specific CD8⁺ and CD4⁺ T-cell responses, CD8⁺ T cells above a positive threshold were found in only 10% of vaccinated participants with no difference in individuals given the higher dose, whereas CD4⁺ responses were observed in up to 80% of individuals with higher proportion of responders in those administered the higher dose [Ledgerwood et al. 2010]. As with DNA vaccines, rAd5 vaccines encoding GPs from EBOV and SUDV co-administered together showed no interference [Kibuuka et al. 2015].

More recently, an rAd5 vaccine encoding the full length WT GP from the 2014 Guinea strain of EBOV was tested in a phase I clinical trial in 80 healthy adults in China using higher vaccine doses of 4 × 10¹⁰ vp and 1.6 × 10¹¹ vp [Zhu et al. 2015]. Side effects were mild-to-moderate in severity and, as expected, a greater proportion of individuals in the high dose group experienced side effects. A single shot at 1.6 × 10¹¹ vp induced antibodies which reached protective titres (based on studies assessing efficacy in NHPs) in 100% of volunteers 1 month post vaccination, but not at earlier time points. T-cell responses induced by the doses in this study were different to previous trials with rAd5 vectors [Ledgerwood et al. 2010], peaking by day 14 and decreasing in magnitude by day 28.

The major drawback in using rAd5 as a vector is the issue of pre-existing immunity, which impacts the efficacy of rAd5 vaccines and limits its use in humans [Harro et al. 2009; Frahm et al. 2012]. In both trials of rAd5 vectored Ebola vaccines, individuals with anti-Ad5 antibodies prior to vaccination had significantly lower GP-specific humoral and T-cell responses after vaccination [Ledgerwood et al. 2010; Zhu et al. 2015]. Encouragingly, in the Chinese study, the high dose of rAd5 (1.6 × 10¹¹ vp) was able to mitigate the impact of pre-existing rAd5 immunity, although this may need to be balanced against the increased reactogenicity caused by high dose vaccines. Respiratory and sublingual delivery of rAd5 vectored Ebola vaccines to overcome pre-existing immunity to Ad5 has shown some efficacy in NHPs [Richardson et al. 2013; Choi et al. 2014]. An alternative strategy that has been developed to circumvent the issue of pre-existing immunity is the use of adenovirus serotypes that rarely circulate in humans, such as Ad35 and Ad26 and a range of chimpanzee-adenovirus (ChAd) serotypes [Barouch et al. 2011; Colloca et al. 2012].

The replication-deficient Ad35 [Keefer et al. 2012; Ouedraogo et al. 2013; Kagina et al. 2014] and Ad26 [Baden et al. 2013, 2015] vectors have been used for malaria, tuberculosis (TB) and HIV, and shown to be safe and immunogenic. A vaccine regime combining Ad35 and Ad26 encoding EBOV and SUDV GP showed 33% efficacy in NHPs against a potent WT EBOV challenge, but 100% efficacy was obtained when a combination of three Ad26 vaccine vectors encoding EBOV, SUDV and Marburg GP was boosted with a multivalent modified vaccinia Ankara (MVA) vector encoding multiple filovirus GPs [NIH, 2014]. This regime of using Ad26 encoding the EBOV GP followed by a MVA vaccine (MVA BNFilo) is among the candidate vaccines being tested. Four phase I clinical trials in the UK, USA, East and West Africa are evaluating the safety of these two vaccines in a prime-boost combination, assessing optimal interval (2, 4 or 8 weeks) between vaccinations, doses and whether Ad26/MVA or a MVA/Ad26 regime is more immunogenic. Preliminary results from these trials have been promising and identify the Ad26/MVA regime as more immunogenic than MVA/Ad26. These results have led to the initiation of a phase Ib trial in Africa and a phase II trial in the UK.

Recombinant simian adenovirus vectors

An alternative to rare human adenovirus serotypes-based as vectors is the use of chimpanzee adenoviruses to which there is a very low prevalence of pre-existing immunity in the human population [Dudareva et al. 2009; Zhang et al. 2013]. A number of different serotypes of ChAd are in preclinical development and a few have been advanced into clinical trials [Zhou et al. 2006; Roy et al. 2007; Ewer et al. 2013; Sharma et al. 2014]. As early as 2006, Kobinger and colleagues used a ChAd serotype 7 as a vector for Ebola GP and demonstrated protection in mouse and guinea pigs [Kobinger et al. 2006]. Chimpanzee adenovirus serotype 3 (ChAd3) encoding Ebola GP from strains of EBOV and SUDV has undergone extensive preclinical development and is now among the leading Ebola vaccine candidates. A single high dose of 10¹¹ vp of ChAd3 encoding the full length GP of EBOV protected 100% (4/4) macaques against an Ebola challenge infection 5 weeks after vaccination, although protection waned to 50% when animals were challenged 10 months post vaccination. Interestingly, a lower dose of 10¹⁰ vp did not provide any protection 10 months after vaccination. To obtain durable protection, the ChAd3 vaccine was boosted with an MVA Ebola vaccine 8 weeks later and protected all the animals challenged 10 months after the last vaccination [Stanley et al. 2014].

The ChAd3 Ebola vaccine, both as a single shot and in combination with MVA as a prime-boost regime, is in advanced stages of clinical development. Two phase I clinical trials have reported preliminary results on the use of a single dose of ChAd3.EBOZ. The first trial in USA tested a bivalent vaccine using two ChAd3 vaccines encoding GPs from the Mayinga strain of EBOV and the Gulu strain of SUDV at doses of 10¹⁰ vp and 10¹¹ vp. This bivalent vaccine was well tolerated with self-resolving mild to moderate side effects, although an asymptomatic prolongation of activated partial thromboplastin time (aPTT) was observed in 3/20 recipients. Investigation of this biochemical abnormality revealed that it was associated with an induction of antiphospholipid antibody causing a false positive rise in aPTT [Ledgerwood et al. 2014]. Compared with the Ad5 vector, ChAd3 was more immunogenic, inducing Ebola GP-specific antibodies and T cells in a greater proportion of vaccinated individuals. As expected, the higher dose induced significantly higher magnitude of antibodies and T cells compared with the lower dose, with responses peaking 4 weeks after vaccination. The authors compared and found that antibody titres in individuals vaccinated with the higher dose were within the range associated with protection in NHP models. In a separate study in the UK, a monovalent ChAd3 encoding GP of EBOV was tested in a dose-escalation study at doses of 1 × 10¹⁰, 2.5 × 10¹⁰ and 5 × 10¹⁰ vp. The safety profile was similar to that observed in the US study, with no vaccine-related serious adverse event and an asymptomatic prolongation of aPTT observed in 4/60 volunteers. However, in contrast to the US study, GP-specific antibody titres were lower, mean titres did not reach levels associated with protection in NHPs, a dose-dependent increase in immune response was not observed and T-cell responses peaked at 14 days rather than 28 days [Rampling et al. 2015]. Interestingly, in the UK study, a subset of volunteers were boosted with a multivalent MVA vaccine with preliminary results suggestive of a significant increase in T cells and increase in antibody titres to putatively protective levels [NIH, 2014].

Although publication of data from other studies evaluating this vaccine is awaited, the results were promising enough to lead to the initiation of a phase III efficacy trial in Liberia. However, this trial was stopped as no further Ebola cases were being reported in Liberia, underlining the challenge in obtaining efficacy data for Ebola vaccines. In addition to regimens using a single dose of ChAd3, three trials (two in the UK and one in Africa) are evaluating a prime-boost regimen of ChAd3 followed by MVA vaccines.

Non-replicating vaccinia virus vectors

Vaccinia viruses (VV) of the genus Orthopoxvirus in the family Poxviridae are large, double-stranded DNA viruses that replicate in the host cytoplasm followed by lysis of the infected cell within 12–24 hours of infection. These viruses were among the first viral vectors to be developed. A large genome that allows stable insertion of antigenic constructs up to 25 kb in size, wide host range, lack of chromosomal integration, ability to induce strong cellular response and stability in lyophilized form make VVs an attractive vaccine vector. A live VV was used in the global eradication of smallpox. Although relatively safe, side effects in children during the vaccination programme [Lane et al. 1969] and the risk of disseminated vaccinia in immunosuppressed individuals led to the development of replication-incompetent attenuated forms of the virus. MVA was derived from 570 serial passages of WT chorioallantois vaccinia virus Ankara (CVA) in chicken embryo fibroblasts (CEFs). MVA had 6 major deletions in its genome, totalling approximately 31 kb and includes numerous mutations affecting host interactive genes leading to a restricted host range and an inability to replicate in mammalian cells. MVA was administered to over 120,000 people in the latter part of the smallpox eradication era with no major safety concerns. More recently, safety data on recombinant MVA have been generated from clinical trials of MVA vaccines encoding antigens from plasmodium, influenza, Mycobacterium tuberculosis and HIV [Cosma et al. 2003; Lillie et al. 2012; Ewer et al. 2013; Tameris et al. 2013; Antoine et al. 1998].

MVA is particularly useful in boosting immune responses following initial priming with a different vaccine. A bivalent MVA, expressing EBOV and SUDV GPs, when administered to NHPs 8 weeks after a bivalent ChAd3 provided durable efficacy and was more efficacious than using a heterologous ChAd3/ChAd63 or homologous ChAd3/ChAd3 prime-boost regimen [Stanley et al. 2014]. Two MVA Ebola vaccines are being tested in clinical trials: a multivalent filovirus MVA vaccine (MVA-BN Filo) that encodes the GPs from EBOV, SUDV and Marburg virus and a nucleoprotein from TAFV Ebola species. MVA-BN Filo used to boost macaques primed with an Ad26 showed 100% protection against an Ebola virus challenge [Van Hoof, 2015]. Three phase I trials using different combinations of Ad26 and MVA-BN Filo are underway in the US, UK and Africa and another phase I trial in the UK has evaluated boosting of responses induced by ChAd3.EBOV with MVA-BN Filo. Preliminary data show that boosting of responses with MVA-BNFilo 3–10 weeks after ChAd3.EBOV vaccination can improve the humoral and cellular immune responses 10–20 fold [NIH, 2014]. Whether the ChAd3 or Ad26/MVA prime-boost regime induces stronger and more durable immune responses than a single dose of ChAd3 or rVSV Ebola vaccines will need to be assessed as these results are published.

In addition to the multivalent vaccine, a monovalent MVA encoding EBOV GP is also being tested in a prime-boost regime following the monovalent ChAd3 at the University of Oxford, UK. This MVA is unique as it is the first MVA vaccine to be produced in a cell line, the duck retinal cell line AGE1.CR.pIX, which increases production capacity and circumvents issues with egg allergy seen with MVAs produced in CEFs. The key parameters being tested in these prime-boost clinical trials are vaccine dose and interval between first and second vaccinations, and how these impact the immune response. Animal models suggest that longer intervals (up to 8 weeks) are most immunogenic, although one clinical trial is testing the very short interval of 7 days between the prime and boost. If this short interval between vaccinations can generate the same levels of immune response as longer intervals, it would facilitate deployment of such prime-boost regimes in outbreak settings.

Replicating vesicular stomatitis virus vector

Vesicular stomatitis virus (VSV) is an enveloped, single-stranded RNA virus of the Rhabdoviridae family that infects cattle. The virus causes a zoonotic disease in humans which is usually subclinical or a febrile flu-like illness. However, pre-existing immunity to this virus in humans is limited enabling its use as a vaccine vector.

A recombinant replicating VSV Ebola vaccine (rVSV-ZEBOV) is among the leading Ebola vaccine candidates and has been rapidly progressed to a phase III efficacy trial in Guinea. The key determinant of VSV pathogenicity is the surface GP which is also the predominant target for immune responses. The rVSV vaccine is designed to exploit this by replacing the VSV-GP with a GP from an EBOV strain. This chimeric design with switching of GPs attenuates the pathogenicity of the virus while allowing the vaccine virus to replicate using the Ebola GP to attach and enter cells. At the same time, the absence of VSV GP mitigates any potential impact of anti-vector immunity. rVSV vaccines encoding GPs from EBOV and SUDV have demonstrated preclinical efficacy in animal models. In NHPs, rVSV vaccine conferred protection against intramuscular and aerosol Ebola virus challenge 4 weeks after vaccination and protection was afforded even when various routes of vaccine delivery (intranasal, intramuscular or oral) was used [Geisbert et al. 2008a, 2008b, 2009].

Multi-centric phase I clinical trials across Europe, the UK and Africa with rVSV-ZEBOV were initiated to evaluate the safety and immunogenicity of the vaccine as well as identify doses and regimes that may be taken into phase II and phase III trials. Preliminary results from dose-escalation studies evaluating doses of 300,000, 3 million, 10 million, 20 million and 50 million plaque-forming units (pfu) have recently been reported [Agnandji et al. 2015; Regules et al. 2015]. A transient viremia of viral RNA assayed by polymerase chain reaction (PCR) 1 day after vaccination and reaching undetectable levels by 5–7 days was observed in most individuals. Associated with this viremia, mild-to-moderate adverse events occur in 90% of individuals that resolved within 24–36 hours after vaccination. The severity of adverse events was dose-dependent between the low dose of 300,000 pfu and 3 million pfu, but there was no difference in severity at higher doses. Transient spontaneously resolving asymptomatic leucopoenia, lymphopenia and neutropenia are observed within the first 3 days after vaccination. At the Geneva trial site where participants were dosed with 10 million or 50 million pfu of rVSV-ZEBOV, 11/51 participants developed arthralgia 9–13 days after vaccination [Agnandji et al. 2015]; 3/11 individuals also developed a rash and vesicles on fingers and toes that lasted 1–2 weeks. Arthritis was confirmed on magnetic resonance imaging (MRI) in 9/11 participants in Geneva and self-resolving arthritis was also observed in 1 volunteer at Kilifi, Kenya, and one volunteer in Hamburg, Germany. Knee arthrocentesis found rVSV in the aspirate and VSV was also detected from local skin lesions, although no viremia was observed in these patients. Thus, although the mechanism of this arthritis remains unclear, it is likely that local VSV replication may play a role. It will be critical to confirm the frequency of these side effects in the large phase III efficacy trial being run in Guinea to enable a more informed evaluation of the risk of live rVSV vaccination.

EBOV GP-specific IgG antibody titres measured by ELISA were observed to peak 28 days after vaccination and titres were significantly higher using 3 million pfu dose compared with the lower dose of 300,000 pfu. Of particular note was that mean antibody titres in the group administered 3 million pfu reached levels associated with protection in NHP models [Regules et al. 2015]. It was noteworthy that, in the multi-centric study undertaken in Europe and Africa, there was no difference in titres of IgG antibodies between groups administered 3 million, 10 million, 20 million or 50 million pfu but higher doses did increase the titres of neutralizing antibodies to EBOV. Further follow up from these studies to determine the durability of antibody responses is awaited. One important unanswered question is whether rVSV induces any cellular immunity and how these immune responses correlate with protection.

One advantage of rVSV over other viral vector vaccines for Ebola is the potential for use for post exposure protection. rVSV encoding SUDV GP administered after Ebola virus infection was able to prevent 4/4 NHPs from succumbing to the disease [Geisbert et al. 2008c] and showed partial post exposure protection against EBOV species [Feldmann et al. 2007]. In humans, one case of post exposure use of rVSV has been reported. A healthcare worker was administered 100 million pfu of rVSV.EBOV vaccine 43 hours after a needlestick injury while working in Sierra Leone. A transient VSV viremia with self-limited febrile illness was observed after vaccination along with induction of innate and Ebola virus-specific immune responses. This individual did not develop an Ebola infection. Although this suggests the potential for use in post exposure settings, whether the lack of Ebola virus infection in this individual was due to vaccine-induced protection or insufficient exposure is unclear [Lai et al. 2015].

A phase III efficacy trial in Guinea using a ring vaccination approach, similar to that used with smallpox vaccination, is underway and has recently reported very encouraging preliminary results following a planned interim analysis [Henao-Restrepo et al. 2015]. In this ongoing trial, 90 clusters (7651 individuals) consisting of contacts or contacts of contacts of an index case received 20 million pfu of rVSV either immediately or 21 days after identification of the index case. A total of 16 cases of EVD, with symptom onset at least 10 days after vaccination, were observed in the delayed vaccination group compared with no cases in the immediate vaccination group; a vaccine efficacy of 100% [95% confidence interval (CI) 74.7–100]. No cases in individuals receiving vaccination were observed after day 6 post vaccination. This study enables some cautious preliminary conclusions and offers some intriguing opportunities to further our understanding of EVD. The data suggest that rVSV confers protection between 6–21 days after vaccination; how much longer vaccine-induced protection lasts is unknown. Vaccine failures within the first 6 days would suggest that vaccine-induced protection needs at least a week to reach effective levels and raises some doubt on the claim that rVSV could work rapidly and provide post exposure prophylaxis. A nonsignificant indirect protective effect among unvaccinated individuals was also observed. The magnitude of this effect and whether it was mediated through a reduced viral load and transmission remains to be determined. Lastly, this study offers a unique opportunity to identify vaccine-induced correlates of protection. In phase I trials, the antibody titres induced by rVSV vaccination reached titres associated with protection in NHPs [90% effective concentration (EC90) titres between 650–9000] only by day 28 post vaccination. The efficacy observed in this trial would argue that significantly lower titres may correlate with protection. This analysis can now be undertaken by combining immune responses from ongoing phase I and phase II studies with clinical efficacy data.

At the same time, work continues on finding ways to further attenuate VSV for use as a vaccine vector. One vaccine vector has attenuated VSV through mutations in its genome and by expressing antigens from a distinct transcription unit without replacing the VSV GP [Cooper et al. 2008]. This vector encoding a HIV antigen [Cooper et al. 2008] has been shown to be safe and immunogenic in phase I clinical trials and recent work coding an Ebola GP showed efficacy in NHPs [Matassov et al. 2015].

Virus like particles

This is a protein-based approach where multiprotein structures are generated that mimic the organization and conformation of authentic native viruses but lack the viral genome, potentially yielding safer and cheaper vaccine candidates. VLP vaccines are currently commercialized worldwide for hepatitis B virus and human papilloma virus. VLP-based vaccines encoding the GP and NP of Ebola virus have been in preclinical development for some time with encouraging results [Warfield et al. 2007; Marzi and Feldmann, 2014]. In response to the current outbreak, an EBOV GP VLP vaccine has been rapidly produced and progressed to being tested in a phase I clinical trial. Encouraging safety and immunogenicity results using this VLP-based vaccine administered with a saponin-based adjuvant Matrix M have been reported at a World Health Organization (WHO) meeting [Marzi and Feldmann, 2014; NOVAVAX, 2015].

Perspectives on vaccine policy

The current 2014 Ebola virus outbreak has galvanized the development of filovirus vaccines. An extraordinary response has brought together academia, health regulatory bodies, funders, governments, pharmaceutical companies and nongovernmental organizations in a concerted effort to develop and evaluate vaccines to control the ongoing outbreak and for use in future outbreaks. Since June 2014, in just over a year, an unprecedented 7 different vaccines (ChAd3, MVA-BNFilo, Ad26, MVA-EBOZ, rAd5, rVSV and a VLP-vaccine) have been expedited into clinical development. Remarkably, real-time monitoring of safety and immunogenicity data has enabled the initiation of two phase III efficacy studies with two different vaccines and demonstration of a high level of efficacy and effectiveness with the rVSV vaccine. As data continue to flood in, critical issues remain in deciding vaccine policy and strategies to combat the ongoing and future Ebola outbreaks.

The first issue is to clearly articulate the clinical purpose and objective of an Ebola vaccine to define strategies around clinical needs. Do we want an outbreak response vaccine that can be rapidly deployed? If so, a vaccine that can be stockpiled, manufactured rapidly in large quantities, protective as a single shot and capable of inducing protective immune responses within 7–14 days would be an ideal candidate. Single doses of rVSV and ChAd3.EBOZ would be ideal candidates for such a purpose and the phase III efficacy data from the rVSV ring vaccination trial provides encouraging preliminary data for the efficacy of such a strategy. Would a prophylactic vaccine administered to populations living in high risk areas be preferable? Such a vaccine would need to generate durable efficacy, be easy to integrate into existing health delivery services, and is safe and efficacious in children. In such cases, a prime-boost vaccination strategy like Ad/MVA or VLP followed by a viral vector or multiple booster vaccinations of rVSV may be a more likely strategy. Alternatively, post exposure vaccination might be more desirable. This would be particularly useful as outbreaks of Ebola are highly unpredictable and control strategies are affected by the high fatality rate in healthcare workers caring for Ebola patients. However, studies to demonstrate the efficacy of such vaccines in humans are almost impossible to design and will need innovative trial designs or very clear efficacy signals from multiple case reports. It is more likely that monoclonal antibodies or antiviral therapies rather than a vaccine would be licensed and deployed for this purpose.

The second issue that needs to be resolved and underpins the development and evaluation of vaccines is to identify immunological correlates of protection. No human immunological correlate of protection exists. The recent rVSV efficacy trial offers a unique opportunity to address this question by combining immunological observations of the rVSV vaccine with efficacy data from the current outbreak. However, it is equally conceivable that immunological correlates might vary depending on the type of vaccine or the immune mechanisms to prevent infection after exposure versus limit viral replication in a post exposure setting.

The third key issue is how to license vaccines for which traditional models to demonstrate efficacy may not apply. For diseases like Ebola which appear as unpredictable outbreaks, traditional phase III clinical trials to measure efficacy are difficult to undertake. This is a problem that needs to be tackled both by global regulatory agencies in consultation with vaccine manufacturers and clinical trial experts. Although, the current efficacy data for rVSV are likely to enable licensure of this vaccine, the process for other vaccines remains unclear. Will an efficacy trial be initiated with the ChAd3 vaccine? Will vaccines that reach antibody titres similar to that induced by rVSV be eligible for licensure? The US Food and Drug Administration (FDA) allows, in exceptional circumstances, licensing of vaccines under the ‘animal rule’ [Sullivan et al. 2009]. In these cases, a vaccine must show efficacy in relevant animal models in combination with phase I and phase II clinical trials demonstrating safety in humans and ability to generate immune responses at levels that would confer protection in animals to be licensed. It is very likely that Ebola vaccines currently being tested would be licensed under this particular rule, although what mechanisms and data might need to be generated for submission continues to remain an ongoing discussion.

Conclusion

The field of Ebola vaccine development has advanced at an exponential rate with multiple candidates in advanced stages of clinical development. Trials being undertaken in the USA, Europe and Africa have already shown that most of these vaccines are safe and well-tolerated. Although a human correlate of protection remains unknown, a single dose of rVSV and high dose of ChAd3.EBOZ is able to generate putatively protective antibody titres. These immune responses are significantly enhanced in prime-boost regimes using MVA-based virus vectors as a boosting vaccination, although the optimal interval between the priming and boosting vaccination is not known. Although the promising efficacy of rVSV vaccine is encouraging, challenges remain in improving vaccines to provide durable efficacy and identifying optimal ways for vaccine deployment. The next steps would be to link the immunological readout with efficacy data to determine correlates of protection, translate available epidemiological and clinical trial data into optimal vaccine policy and develop pathways to license and potentially stockpile vaccines for future Ebola outbreaks.