Ebola: Lessons on Vaccine Development

Abstract

The West African Ebola virus (EBOV) epidemic has fast-tracked countermeasures for this rare, emerging zoonotic pathogen. Until 2013–2014, most EBOV vaccine candidates were stalled between the preclinical and clinical milestones on the path to licensure, because of funding problems, lack of interest from pharmaceutical companies, and competing priorities in public health. The unprecedented and devastating epidemic propelled vaccine candidates toward clinical trials that were initiated near the end of the active response to the outbreak. Those trials did not have a major impact on the epidemic but provided invaluable data on vaccine safety, immunogenicity, and, to a limited degree, even efficacy in humans. There are plenty of lessons to learn from these trials, some of which are addressed in this review. Better preparation is essential to executing an effective response to EBOV in the future; yet, the first indications of waning interest are already noticeable.

INTRODUCTION

Emergence and reemergence of infectious diseases have increased over the last 50 years, challenging local, regional, and global public health. Many of these incidences were caused by zoonotic pathogens such as filoviruses. Marburg virus (MARV) was discovered in 1967 as the cause of an outbreak of hemorrhagic fever among workers of a German vaccine company. The workers were exposed to MARV through preparation of primary cell cultures from organs of African green monkeys imported from Uganda (76). Ebola virus (EBOV) and Sudan virus (SUDV), both members of the Ebolavirus genus, were discovered in 1976 during simultaneous outbreaks of hemorrhagic fever in southern Sudan (now South Sudan) and northern Zaire (now Democratic Republic of the Congo, DRC) (127, 128). Two further species of African ebolaviruses were described with the discovery of Taï Forest virus (TAFV) in Côte d’Ivoire in 1994 (67) and Bundibugyo virus (BDBV) in Uganda in 2007 (119). Outside the African continent one finds Reston virus (RESTV), which was discovered in macaques imported to the United States in 1989 (59); today there is evidence for a wider distribution of RESTV in Asia (89). In 2008, RESTV was found in pigs in the Philippines, raising the concern of food safety, a matter that is still under investigation, as is RESTV pathogenicity in pigs (4). In Spain the RNA from Lloviu virus (LLOV) was isolated from bats; however, no virus has been recovered yet (95). The emergence of EBOV in West Africa in 2013–2014 has caused an unprecedented epidemic, with almost 30,000 cases and approximately 11,000 deaths (129) (for filovirus taxonomy see the sidebar titled Family Filoviridae Taxonomy).

VACCINE LICENSURE.

The US Food and Drug Administration (https://www.fda.gov) has implemented the animal rule, allowing licensing of countermeasures based on efficacy data in animal models combined with safety and immunogenicity studies in humans (108). In general, licensing agencies require different clinical trials evaluating safety, immunogenicity, and efficacy of a product. Clinical trials are classified into four categories: phase 1, product safety and dosage in a small cohort; phase 2, side effects and immunogenicity in larger cohorts; phase 3, efficacy data and monitoring adverse effects in large cohorts; and phase 4, long-term product safety and efficacy in larger cohorts (typically after product licensure).

Filoviruses are zoonotic pathogens that are occasionally transmitted to humans, nonhuman primates (NHPs), and perhaps other wildlife. The field currently follows the hypothesis of bats being potential reservoir species in nature. To this day, however, only MARV has been isolated from fruit bats (Rousettus aegyptiacus) (118). This important link is missing for all of the ebolaviruses. Wildlife or domestic mammalian species, such as duiker antelope or domestic/feral pig, might play a role as amplifying or intermediate hosts, but strong evidence has yet to be provided for this hypothesis (30). In the future it will be important to close these knowledge gaps, as wildlife vaccination might be considered for containing EBOV and other filoviruses (30, 94).

Filoviruses are filamentous, nonsegmented, negative-stranded, enveloped RNA viruses that replicate in the cytoplasm of the cell (5, 30, 35). The genome is approximately 19 kb and displays this gene order: 3′ trailer, nucleoprotein (NP), virion protein 35 (VP35), VP40, glycoprotein (GP), VP30, VP24, polymerase (L), 5′ trailer. Transcription of the genome leads to the expression of seven structural proteins, including NP, VP35, VP30, and L, which make up the nucleocapsid complex (RNP) and drive transcription and replication. VP40 is the matrix protein and important for particle morphology and budding. VP24 bridges between the RNP complex and matrix, with GP being inserted into the virus envelope as the only transmembrane protein mediating virus entry. All ebolaviruses express two additional nonstructural proteins encoded on their GP gene. The primary translation product is soluble GP (sGP) that gets released from infected cells. The transmembrane GP and the small soluble GP (ssGP) are the result of RNA editing (30). No defined functions have been assigned to sGP and ssGP. The transmembrane GP mediates particle binding to cellular surface entities and fusion with the endosomal membrane by interaction with the receptor Niemann-Pick 1 (NCP-1) (14, 18). Besides potential contribution by NP, VP40, and VP24, GP is considered the key immunogenic viral protein and the most important immunogen for vaccine development (for further reading on the molecular biology and evolution of filoviruses, see Reference 30).

Ebolaviruses and marburgviruses cause a clinical syndrome known as viral hemorrhagic fever. To reflect the increased variation in clinical symptoms and disease progression as most evidently observed during the West African EBOV outbreak, the filovirus diseases are now more commonly referred to as Ebola virus disease (EVD) and Marburg virus disease (MVD) (5, 34). So far, all reported African filoviruses are human pathogens, with case fatality rates up to 90% for MARV, Ravn virus (RAVV), and EBOV. Lower case fatality rates in the range of 25–50% are associated with SUDV and BDBV; the public health importance of TAFV is uncertain, with only a single clinical case. RESTV has infected humans but is not known to cause clinical disease. There are no data on human pathogenicity for Lloviu virus (95) (see the sidebar titled Family Filoviridae Taxonomy) or the putative new Chinese filoviruses, which have yet to be classified (50). The West African EBOV epidemic has changed the way we respond with regard to diagnosis, triage, and case patient management; altogether, this and/or unknown factors (e.g., biological features of EBOV Makona) have lowered the case fatality rate below 50%—a first for an EBOV strain. Given the high numbers of survivors during this epidemic, we also have a better understanding on sequelae leading to post–ebolavirus disease syndrome (15). In addition, it was discovered that EBOV temporarily persists in immune-privileged body sites, potentially leading to infrequent transmission through sex, and perhaps other routes (16) (for further reading on EBOV disease and pathogenesis, see Reference 5).

The unforeseen and unprecedented West African epidemic, caused by EBOV strain Makona, has highlighted the urgent need for effective and rapidly deployable countermeasures. The epidemic created a clear necessity to allow fast-tracked antiviral/therapeutic (for further reading see Reference 19) and vaccine testing, providing insight into the feasibility of such approaches for future epidemics. Unfortunately, clinical trials could only begin toward the end of the epidemic and suffered from, among other issues, a lack of preclinical data, questionable designs, low number of patients, and controversial politics. Despite those drawbacks, the vaccine trials have offered valuable data on safety and immunogenicity, with one trial even providing the first data for efficacy in humans (https://www.clinicaltrials.gov; http://www.pactr.org). However, almost two years after the epidemic, only Russia and China (8) have licensed an EBOV vaccine, whereas efforts in other countries are still held up in production and licensing. In this review, we address vaccines focusing on EBOV, as most vaccine developments have been targeted toward this virus. We review current promising vaccine candidates and human clinical trials and then discuss challenges and lessons learned for future development.

VACCINE OVERVIEW

Vaccine development against EBOV (and to a lesser degree other filoviruses) has been hampered in the past by the need for high-level biocontainment, general lack of interest among pharmaceutical companies, and lower priority in the public health sector. EBOV was considered a rare zoonosis with infrequent transmission events into the human species that either were self-limiting or were stopped with basic public health interventions (109). This view started to change with the EBOV-Kikwit outbreak (DRC, 1995), and definitely with the classification of EBOV as a biothreat pathogen and then a U.S. Select Agent and Tier 1 Pathogen (33). EBOV became of increasing interest for biodefense research with the goal to produce countermeasures such as diagnostics, therapeutics, and vaccines. It is interesting to note that despite attention and funding, and except for a few diagnostic tests (99), no product, whether antiviral, therapeutic, or vaccine, is licensed for human use today. Particularly promising vaccine candidates were stalled between the preclinical and clinical milestones on the path to licensure. Of course, the move into clinical trials is an expensive step in product development that often requires partnerships with industry, and there did not seem to be interest then, largely due to an uncertain market. Also, the public health sector seemed hesitant to give high priority to products for rare emerging viruses, such as EBOV, as other infectious and chronic disease issues were burdening the system. Thus, the unprecedented West African EBOV epidemic caught the academic, industrial, and government sectors by surprise, forcing them to respond immediately, even to use unfinished products in humans.

Animal Models

Traditionally, vaccine candidates for EBOV and other filoviruses are screened in rodent models, including mouse, hamster, or guinea pig (132) (Table 1). Unfortunately, all clinical filovirus isolates cause no or very limited disease in these rodent species. Therefore, typically serial adaptation is required to produce uniform lethality in rodents. Adaptation is associated with genome mutations occurring often in, but not limited to, those genes that encode interferon antagonists such as VP24 for ebolaviruses and VP40 for marburgviruses. Rodent models utilizing adapted filoviruses do not always closely mimic human disease manifestations and progression, especially not the mouse models. Nevertheless, rodents are preferred screening models, with the drawback that they may not predict overall efficacy in NHP models and humans (42, 101). NHP models are still considered the gold standard for filoviruses, due to disease presentation similar to what is observed in humans (42). Cynomolgus macaques are preferred for prophylactic vaccine studies, whereas rhesus macaques are often utilized for therapeutic studies because their prolonged time to death allows for an extended window for intervention. Today, most EBOV vaccine candidates have gone through rodent screening followed by NHP confirmation testing prior to clinical trials (see the sidebar titled Vaccine Licensure).

Table 1.

Animal models for Ebola virus

Animal	Virus	Inoculation route	Human disease display	Lethal	Ease of handling	Tools and reagents	Cost	Use/predictive value
Laboratory mouse	MA-EBOV	IP, IN, AR	Low	Yes	Easy	Yes	Cheap	Screening model/low
Humanized mouse	WT-EBOV	IP	Medium	Yes	Easy	Yes	Expensive	Not well established/unknown
CC-RIX mouse	MA-EBOV	IP	Dep.	Dep.	Easy	Yes	NA	Not well established/unknown
Hamster	MA-EBOV	IP	Medium	Yes	Easy	Limited	Cheap	Screening model/medium
Guinea pig	GPA-EBOV	IP, SC, AR	Low	Yes	Easy to moderate	Limited	Cheap	Screening model/medium
Ferret	WT-EBOV	IN	Medium	Yes	Moderate to difficult	Limited	Moderate	Not well established/unknown
Nonhuman primate	WT-EBOV	Various	High	Yes	Difficult	Yes	Expensive	Confirmatory model/high

Abbreviations: AR, aerosol; Dep., depends on CC-RIX strain (asymptomatic, disease, lethality); GPA, guinea pig adapted; IN, intranasal; IP, intraperitoneal; MA, mouse adapted; NA, not applicable (no commercial source); SC, subcutaneous; WT, wild type.

Past and Current Vaccine Efforts

In the 1970s and 1980s, inactivated virus preparations were tested first in animal models, with varying success (41, 74). The strategy was abandoned because its efficacy was limited and there were safety concerns. This was followed in the 1990s with subunit and DNA vaccination. Vaccine development was accelerated around the beginning of the new millennium, with a variety of nonreplicating and replication-competent viral vector platforms mostly targeting EBOV, the prototype filovirus associated with the highest case fatality rates (40). The leading platforms included alphavirus replicons (55, 56, 98), human adenoviruses (37, 115, 116), chimpanzee adenovirus (110), paramyxoviruses (10, 11), rabies virus (6, 125), and different strategies with recombinant vesicular stomatitis virus (rVSV) (61, 86). In addition, virus-like particles (124), a biologically contained EBOV lacking VP30 (82), DNA (43), cytomegalovirus (CMV) vectors (83), modified vaccinia virus Ankara (MVA) (27), and several combinations of DNA, MVA, and adenoviruses have also demonstrated the ability to completely protect NHPs (110, 115). The main viral immunogen for vaccine approaches is the filovirus transmembrane GP, used in various versions (full length, transmembrane anchor deleted, or mucin domain deleted), which is known to be the sole target for classical neutralizing antibodies. Some vaccine approaches include additional immunogens such as VP40 and, less frequently, NP, and VP24 [i.e., DNA and virus-like particle (VLP) vaccines]. The following paragraphs introduce those EBOV vaccine candidates with promising efficacy in NHPs (Figure 1), some of which have proceeded to human clinical trials (Figure 2; Table 2).

Figure 1.

Vaccine approaches with complete and partial efficacy in nonhuman primate models. Successful single platforms in a prime or prime-boost approach are pictured on the left, whereas combined platforms in a prime-boost approach are depicted on the right. Abbreviations: Ad, adenovirus; CMV, cytomegalovirus; EBOVΔVP30, Ebola virus lacking the gene for VP30; HPIV3, human parainfluenza virus type 3; MVA, modified vaccinia virus Ankara; RABV, rabies virus; VEEV, Venezuelan equine encephalitis virus; VLP, Ebola virus–like particles; VSV, vesicular stomatitis virus.

Figure 2.

Vaccine platforms in clinical trials. Completed and ongoing phase 1–4 clinical trials are shown for each vaccine platform, with various antigen combinations. The monkey symbolizes the first publication demonstrating complete protective efficacy of the vaccine approach in nonhuman primates. Abbreviations: Ad, human adenovirus; cAd, chimpanzee adenovirus; GP, glycoprotein; HPIV3, human parainfluenza virus type 3; MARV, Marburg virus; MVA, modified vaccinia virus Ankara; NP, nucleoprotein; SUDV, Sudan virus; TAFV, Taï Forest virus; VSV, vesicular stomatitis virus.

Table 2.

Human clinical trials of Ebola virus vaccine candidates

Start	Status	Phase	Location	Enrollment 01/2018	Study ID	References
VSV, 100% protection in NHPs (61)
10/2014	Completed	1	USA	39	NCT02269423	100
10/2014	Completed	1	USA	39	NCT02280408	100
11/2014	Completed	1	Guinea	30	NCT02283099	2, 22, 57
11/2014	Completed	1	Canada	40	NCT02374385	29
11/2014	Completed	1, 2	Switzerland	115	NCT02287480	2, 57, 58
12/2014	Active, not recruiting	1	Kenya	40	NCT02296983	2, 57, 58
12/2014	Completed	1	USA	512	NCT02314923	54
01/2015	Active, not recruiting	2	Liberia	1,500	NCT02344407	63
03/2015	Completed	3	Guinea	8,851	PACTR201503001057193	51, 52
04/2015	Completed	2, 3	Sierra Leone	8,651	NCT02378753 PACTR201502001027220
08/2015	Completed	3	Canada, Spain, USA	1,198	NCT02503202	47
11/2015	Completed	1	Gabon	155	PACTR201411000919191	1, 2, 57
12/2015	Completed	1	USA	38	NCT02718469
01/2016	Enrolling by invitation	2	USA	18	NCT02788227
11/2016	Enrolling by invitation	1	Switzerland	100	NCT02933931
03/2017	Recruiting	2	Guinea, Liberia	5,500	NCT02876328 PACTR201712002760250
08/2017	Recruiting	2	Canada	200	NCT03031912
12/2017	Suspended	3	Uganda	500	NCT03161366
Ad26 and MVA, 100% protection in NHPs (37)
12/2014	Completed	1, 2	UK	87	NCT02313077	88, 126
01/2015	Completed	1	USA	164	NCT02325050
03/2015	Completed	1	Kenya	72	NCT02376426
04/2015	Completed	1	Uganda, Tanzania	78	NCT02376400
06/2015	Active, not recruiting	2	France, UK	408	NCT02416453
09/2015	Recruiting	3	Sierra Leone	1,019	NCT02509494 PACTR201506001147964
09/2015	Completed	3	USA	525	NCT02543567
09/2015	Completed	3	USA	329	NCT02543268
11/2015	Recruiting	2	Burkina Faso, Côte d’Ivoire, Kenya, Uganda	1,056	NCT02564523
12/2015	Recruiting	2	Kenya, Mozambique, Nigeria, Tanzania, Uganda, USA	575	NCT02598388
05/2016	Recruiting	3	Burkina Faso, Côte d’Ivoire, France, Kenya, Mozambique, Nigeria, Sierra Leone, Tanzania, Uganda, UK, USA	1,417	NCT02661464
03/2017	Recruiting	2	Guinea, Liberia	5,500	NCT02876328 PACTR201712002760250
05/2017	Recruiting	1	UK	56	NCT03140774
	Active, not recruiting	1	USA	60	NCT02891980
cAd3 and MVA, 100% protection in NHPs (110)
09/2014	Active, not recruiting	1	UK	92	NCT02240875	31, 66
10/2014	Completed	1	Mali	91	NCT02267109	117
01/2015	Completed	1	Uganda	90	NCT02354404
03/2015	Completed	1	USA	64	NCT02408913
03/2015	Completed	1	Mali	60	NCT02368119
04/2015	Active, not recruiting	1	UK	38	NCT02451891
07/2015	Completed	1	Senegal	40	NCT02485912
cAd3, 100% protection in NHPs (110)
08/2014	Completed	1	USA	50	NCT02231866	69, 117
10/2014	Completed	1, 2	Switzerland	120	NCT02289027	23
01/2015	Active, not recruiting	2	Liberia	1,500	NCT02344407	63
07/2015	Completed	2	Cameroon, Mali, Nigeria, Senegal	3,024	NCT02485301
11/2015	Completed	2	Mali, Senegal	600	NCT02548078
	Active, not recruiting	1B	Uganda		PACTR201412000957310
Ad5, 100% protection in NHPs (112)
09/2006	Completed	1	USA	48	NCT00374309	68
12/2014	Completed	1	China	120	NCT02326194	71, 133
03/2015	Completed	1	China	61	NCT02401373	130
07/2015	Completed	1	China	110	NCT02533791	71
10/2015	Completed	2	Sierra Leone	500	NCT02575456 PACTR201509001259869	134
VSV and Ad5, 100% protection in NHPs (26)
09/2015	Completed	1, 2	Russia	84	No. 495 (Russia)	26
10/2016	Completed	1	Russia	60	NCT02911415
10/2016	Completed	1	Russia	60	NCT02911428
08/2017	Recruiting	4	Guinea, Russia	2,000	NCT03072030 PACTR201702002053400
11/2017	Recruiting	1, 2	Russia	120	NCT03333538
DNA, 83% protection in NHPs (43)
10/2003	Completed	1	USA	27	NCT00072605	75
01/2008	Completed	1B	USA	20	NCT00605514	107
02/2010	Completed	1	Uganda	108	NCT00997607	64
Ad26 and cAd3
09/2015	Active, not recruiting	1	UK	32	NCT02495246
HPIV3, 100% protection in NHPs (87)
08/2015	Completed	1	USA	30	NCT02564575
Subunit
02/2015	Completed	1	Australia	230	NCT02370589

Abbreviations: Ad, human adenovirus; cAd, chimpanzee adenovirus; HPIV3, human parainfluenza virus type 3; MVA, modified vaccinia virus Ankara; NHP, nonhuman primate; VSV, vesicular stomatitis virus.

Virus-like particles.

VLPs are a noninfectious, protein subunit–based vaccine platform and can be produced from cells transfected with plasmids encoding EBOV VP40 and GP, which leads to the release of filamentous particles similar in shape to authentic EBOV particles (123) (Figure 1). Coexpression of additional EBOV proteins, such as NP and VP24, can improve VLP production and immunogenicity (72). The VLP platform is highly immunogenic, and vaccination has been shown to induce innate, humoral, and cellular immune responses (122). The humoral immune response, particularly to GP, seems important for protection; T cell–mediated immunity is likely enhanced by including NP in the particle preparation. The VLP platform is considered safe, but production can be difficult and expensive. The baculovirus-based expression technology may partially solve this issue.

DNA plasmids.

DNA vaccines are readily adaptable and noninfectious and can be produced in large quantities, making them a fast, safe, and cheap vaccine approach (Figure 1). DNA vaccines have been shown to induce cellular as well as humoral immune responses, but they regularly require administration of several doses or combination with other vaccine approaches to achieve the desired immunity (44). EBOV DNA vaccine approaches based on GP have shown efficacy in rodent models (103, 120), but efficacy in NHP models has yet to reach 100%. Recently, 83% protection was reported for a codon-optimized GP-DNA vaccine (43). Given new and more powerful delivery technologies for DNA plasmids, this approach should be continued and optimized. In addition, combination approaches with other platforms should be considered, as they have shown promising results.

Venezuelan equine encephalitis virus replicons.

Alphavirus replicons are well-established vaccine vectors (Figure 1). The foreign antigen is cloned to replace the alphavirus structural genes. These replicons transcribe mRNA and replicate the genome, but they do not make virus particles, due to the absence of alphaviral structural proteins. Venezuelan equine encephalitis virus (VEEV) replicons expressing EBOV-GP, EBOV NP, or both have been assessed in NHPs with inconsistent results (41, 55). However, increasing the vaccination regimen by two or more injections might improve protection.

Recombinant EBOVΔVP30.

Reverse genetics for EBOV have enabled the establishment of a recombinant biologically contained ebolavirus (46). The viral genome lacks the VP30 gene (EBOVΔVP30), which is essential for transcription, and thus the virus is no longer replication competent. Providing VP30 in trans allows for easy and safe propagation of EBOVΔVP30 in cell lines, whereas no progeny virus is produced in the absence of VP30 (45). This vaccine vector was shown to confer complete protection of NHPs against EBOV challenge in a single dose, i.e., a prime-boost approach; a prime-boost regimen using hydrogen peroxide–inactivated vaccine also completely protected NHPs (82). The inactivated prime-boost vaccine strategy is likely to go forward, as it addresses remaining safety concerns with EBOVΔVP30 (Figure 1).

Recombinant cytomegalovirus-based vectors.

A new promising vaccine approach is the use of CMV vectors. A recently published study using rhesus CMV (RhCMV) in rhesus macaques established the efficacy of CMV as an EBOV vaccine platform (83) (Figure 1). In contrast to previous studies (48, 49), this is the first one showing an unexpected induction of high antibody levels against a CMV-expressed foreign antigen, suggesting a way to bias immunity toward either antibody or T cells through selective transgene promoter usage. This has far-reaching potential for many vaccine and immunotherapeutic platforms beyond CMV vectors. High preexisting immunity is not an issue, as CMV reinfects individuals despite existing immunity. Therefore, development of CMV-EBOV vaccines for use in humans, and perhaps wildlife, should continue.

Modified vaccinia virus Ankara.

While MVA-based EBOV vaccines have been developed over the past decade, protective immunity against lethal EBOV challenge was only achieved in combination with an adenovirus-based vaccine (92). This strategy has been assessed with good immunogenicity in human clinical trials (Table 2). However, recently an MVA-based vector expressing EBOV-GP and VP40 leading to VLP release from the infected cell was shown to protect NHPs against lethal disease after a single-dose vaccination (27). This MVA-EBOV vaccine is moving toward human clinical trials.

Recombinant human parainfluenza virus.

Live-attenuated vectors based on human parainfluenza virus type 3 (HPIV3) are actively being investigated as vaccines for HPIV3 and other pediatric pathogens (28, 62). The HPIV3 system is replication competent; therefore, safety concerns require testing such as determining tolerance in immunocompromised individuals and neurovirulence testing. As with Ad5, preexisting immunity to this common childhood pathogen is a concern. However, it was demonstrated that vaccination with HPIV3 expressing the EBOV-GP under preexisting immunity conditions resulted in an immune response to the foreign antigen (13). A novel approach based on HPIV type 1 demonstrated good immunogenicity, but no challenge data were provided (73). Recently, an HPIV3 vector expressing the EBOV-GP has shown promising protective efficacy in NHPs following combined intranasal-intratracheal as well as aerosol vaccination (Figure 1) (87).

Recombinant rabies virus–based vectors.

The rabies-based vectors for EBOV have been developed with the idea of a dual vaccine approach against rabies virus (RABV) and EBOV infection, which overlap in many EBOV-endemic areas (Figure 1). The vector is based on the reverse genetics system for the RABV vaccine strain SAD B19, which is currently used for wildlife immunizations. A point mutation in the RABV glycoprotein attenuates this vector (7). A single dose of the live-attenuated RABV vaccine vector expressing the EBOV-GP (rRABV/EBOV-GP) has shown complete protection in the EBOV NHP model (6). In contrast, one dose of rRABVΔG/EBOV-GP and a prime-boost regimen with inactivated rRABV/EBOV-GP resulted in only 50% protection. Protection by RABV vectors was largely dependent on the quality of antibody responses to EBOV-GP (6). Recently, the inactivated RABV vaccine vector was improved using a codon-optimized antigen (125) and showed promising results in an adjuvanted prime-boost regimen in the EBOV NHP challenge model (60). Production of this vaccine for human clinical trials has been initiated in adherence to good manufacturing practices (60).

Recombinant adenovirus-based vectors.

Replication-defective adenoviruses, mainly human adenovirus serotype 5 (Ad5), have been used as vaccine vectors for viral diseases, including EVD (80) (Figure 1). However, a significant concern with this vaccine platform is the preexisting immunity in certain human populations (85). In vivo studies utilizing Ad5-based vaccines have shown that protective vaccine efficacy is significantly lowered with preexisting immunity to the vector (20, 102). Mucosal delivery of Ad5-based vaccines, however, can circumvent preexisting immunity without affecting protective efficacy; interestingly, this also led to a beneficial improvement of T cell responses (20, 102). A milestone vaccine approach that first protected NHPs from EBOV infection was a result of a DNA-prime and Ad5-boost vaccination (115) (Figure 1). Protection seemed associated with humoral immunity and T memory helper responses, while cell-mediated immunity was not a requirement (115). Subsequently, it was shown that a single injection of high-titered Ad5 expressing the EBOV-GP alone resulted in complete protection against high-dose intramuscular EBOV challenge (112). As a nonreplicating vector, the Ad5 platform is considered safer than replication-competent vectors. The need for high-titered (>10¹⁰ infectious units) vaccine, however, might cause problems with vaccine production and safety in immunocompromised people. The preexisting immunity issue pushed the search for replacement vectors based on different human serotypes, such as Ad26 and Ad35 (37), or to primate-specific adenoviruses such as chimpanzee adenovirus [(c)Ad] and simian Ad21 (106, 110). Varying success was achieved with the less-prevalent human adenovirus vectors as vaccines for ebolaviruses, but optimization of vaccine approaches is still ongoing. More promising was the cAd3 vaccine vector approach expressing EBOV-GP, which provided 100% protection with a single dose (110).

Recombinant vesicular stomatitis virus–based vectors.

Since the mid-1990s, the Rose laboratory has spearheaded the use of recombinant vesicular stomatitis virus (VSV) as live-attenuated and replication-incompetent vaccine vectors (104, 105). The main vector used for the filovirus vaccines is an attenuated version of the VSV Indiana serotype expressing the filovirus GP instead of the VSV glycoprotein (G), which normally mediates VSV entry and contributes to neurotropism and pathogenicity of the virus (36, 105) (Figure 1). While this vaccine virus is attenuated, it can be propagated to high titers and is highly immunogenic (2, 58, 61, 100). To date, VSV-EBOV has proven to be one of the more successful candidates (detailed in Reference 111). Briefly, a single intramuscular or mucosal vaccination of NHPs elicited complete protection against a high-dose challenge of homologous EBOV given 28 days later (61). Depletion studies have identified antibodies as a key mechanism of protection (78). This platform has been used to generate vaccines for all known ebolaviruses and several members of the marburgvirus species (81). Recently, it was shown that in NHPs the VSV-EBOV vector conferred 100% protection when administered within a week of a lethal EBOV challenge (84). These data are intriguing as they suggest the vaccine can be used as an emergency vaccine. To address safety, pigs representing a susceptible livestock species and SIV-infected NHPs were infected with VSV-EBOV; the animals showed no signs of disease, and VSV-EBOV shedding was minimal (25, 38). Interestingly, the simian HIV-infected macaques were protected against lethal EBOV challenge and protection correlated with the level of the humoral immune response, again supporting the importance of antibodies elicited by this vaccine for protection (38). An alternative VSV vaccine vector is VSV-N4CT1-ZEBOV-GP, which carries a mutated VSV-G and expresses VSV-N and the EBOV-GP at positions four and one in the VSV genome, respectively. This vaccine has recently shown protective efficacy in NHPs (93).

Human Clinical Trials

Human clinical studies (see the sidebar titled Vaccine Licensure) are costly, and before the 2013–2016 West African EBOV epidemic only four EBOV phase 1 vaccine trials had been conducted evaluating either DNA- or Ad5-based vaccines (64, 68, 75, 107) (Figure 2; Table 2). However, since the year 2005 several other vaccine platforms have shown promising efficacy in NHPs (Figure 1), and during and after the 2013–2016 EBOV epidemic several of these vaccine candidates were assessed in phase 1–3 clinical trials (Figure 2; Table 2). These trials included, but were not limited to, VSV-EBOV (1, 2, 22, 29, 47, 51, 52, 54, 57, 58, 63, 100), Ad5-EBOV (71, 130, 133, 134), rVSV-EBOV combined with Ad5-EBOV (26), cAd3-EBOV with or without MVA-EBOV (23, 31, 66, 69, 117), and Ad26-EBOV combined with MVA-EBOV (88, 126). Most of the conducted trials analyzed immunogenicity of the EBOV-GP antigen; however, some evaluated immune responses to GP and/or NP from different ebolavirus species (Figure 2). In general, the EBOV vaccines in these phase 1 and 2 trials showed no serious adverse effects and elicited good immunogenicity against the immunogen. Several VSV-EBOV phase 1 trials, however, reported adverse effects including temporary arthritis with high-dose vaccination (2, 58), something that was less documented in the later trials with VSV-EBOV (1, 29, 51, 52, 54, 63). A more recent review summarizes most aspects of VSV-EBOV clinical trials (111).

Although there are over 50 phase 1 and 2 trials registered (completed or ongoing) with the US National Library of Medicine (https://www.clinicaltrials.gov) (Table 2) or the Pan African Clinical Trials Registry (http://www.pactr.org) (Table 2), there were only a few opportunities to conduct phase 3 trials as the West African EBOV epidemic ceased. Only one of those phase 3 trials reported results on vaccine efficacy—the ring vaccination, open-label, cluster-randomized trial in Guinea utilizing VSV-EBOV (51, 52). This study showed statistically significant protection against EBOV infection, with no new cases among immunized individuals from ten days after vaccination in randomized and nonrandomized clusters (51). However, despite the promising results, VSV-EBOV is still not licensed today, two years after the first trial (8). Instead, Russia licensed a combination vaccine approach based on a VSV-EBOV prime and Ad5-EBOV boost, with limited preclinical data and a single phase 1/2 trial (26). Similarly, China licensed an Ad5-based EBOV vaccine based on limited preclinical data (131) and more robust phase 1 and 2 clinical trials (71, 130, 133, 134). Intriguingly, both products are very similar to the two leading earlier and well-studied EBOV vaccine approaches, VSV-EBOV and Ad5-EBOV (Figure 1), developed and preclinically evaluated in North America/Europe and tested in phase 1–3 trials at multiple sites before and during the West African epidemic (Figure 2; Table 2). It will be interesting to see if the World Health Organization (WHO) and national regulatory authorities will promote the use of the two licensed products, and when those other promising vaccine approaches will finally be approved for human use.

CHALLENGES FOR VACCINE DEVELOPMENT

Despite the progress in EBOV vaccine development, plenty of challenges remain to be addressed. The following discussion raises some of those issues and is meant to motivate the reader rather than provide a complete list.

Animal Models

Animal modeling is being critically discussed these days. EVD and MVD are not uniformly fatal in humans, and several investigators are speaking out about the importance of having animal models that reflect this. The challenge strain is a matter of discussion. EBOV-Mayinga was the reference strain in the early days but was replaced by EBOV-Kikwit after the 1995 outbreak; nowadays, the EBOV-Makona strain seems favored. Of note, growth of ebolaviruses on Vero cells favors the conversion of a wild-type virus with seven uridine residues at the GP editing site to an eight-uridine-residue mutant. This mutant strain is still lethal but slightly attenuated in NHP models and may lower the bar in efficacy studies (42). The challenge dose in NHP models has been high (1,000 plaque-forming units, ~1,000 LD50). Lower challenge doses are being discussed and in fact have already been used. Furthermore, the EBOV challenge route is arguable. Most previous NHP studies used intramuscular challenge, but mucosal challenge is thought to mimic natural infection more closely. The aerosol route is favored for biodefense countermeasure development.

It will be difficult to reach consensus among researchers on the challenge strain, and it may not matter much which EBOV strains are used, as they all are uniformly lethal in NHPs upon intramuscular inoculation (77, 79). This is not the case with rodent models, where adapted strains are needed. For many years, only mouse-adapted EBOV-Mayinga was available (9), but a few additional viruses/strains have been adapted recently. Lower challenge doses might be more applicable for efficacy testing of antivirals and therapeutics than for vaccine efficacy testing. As we already have multiple vaccine platforms that protect against the high challenge dose, lowering the bar might result in suboptimal products. The mucosal route of challenge does not result in uniform lethality, something that might be favored by some investigators because it would be more similar to infection outcome in humans. However, lack of uniform lethality will increase the number of animals needed to obtain meaningful results, unless one changes the parameters for efficacy (i.e., viral load), which has so far largely been survival. The Filovirus Animal Non-Clinical Group (FANG) is trying to define parameters on animal work that should ideally be applied to filovirus countermeasure development (https://report.nih.gov/crs/View.aspx?Id=2323). Such parameters could be helpful, and many argue they are needed, but they should not impede science because important things might be missed.

Immunogen (Antigen) Choice

Most approaches today utilize the EBOV-GP as the immunogen (antigen). Other antigens that have been utilized largely in combination with GP are NP, VP40, and VP24. So far, only vaccine approaches based on GP have shown complete protection in the NHP models (101). Based on this, currently only GP-based vaccine approaches have entered clinical trials and advanced for licensure (121) (Figure 2; Table 2).

It might be helpful to study the contribution of other EBOV antigens to protection, in particular T cell–mediated protective responses. This might be more important regarding cross-protection and durability, as these antigens are usually more conserved and potent T cell responses may elicit stronger, longer-lasting, and broader immune responses.

Short Time to Immunity (Fast-Acting Vaccination)

An important feature of vaccines against rare, high-consequence pathogens such as EBOV is rapid protection with a single dose, because outbreaks, epidemics, and biothreat events usually do not allow for multiple-injection strategies. Several filovirus vaccines prevent infection in NHPs when administered as a single injection (6, 55, 61, 82, 87, 110, 112) (Figure 1). In particular, VSV-EBOV is very potent in this regard and can completely protect NHPs when administered only a week prior to high-dose homologous challenge (84). The ring vaccination trial in Guinea is a prime example of the success of such an approach (51, 52).

Given the epidemiology and ecology of rare zoonotic diseases such as EBOV, the development and optimization of single-dose or prime-boost approaches with shortened immunization intervals should continue. The current interest is shifting toward long-lasting immunity, which is likely easier to achieve and more reliable with prime-boost approaches. However, it is not even clear what target populations will be addressed with these time-consuming, labor-intensive, and expensive vaccination approaches over the use of emergency (ring) vaccination.

Cross-Protection

There is little to no cross-protective immunity between natural infections with marburgviruses and ebolaviruses, nor with vaccines targeting these distinct viruses. Limited cross-protection may occur among different species within a genus such as Ebolavirus. For example, a single injection of VSV-EBOV provided partial protection against BDBV in NHPs (32). One way to achieve cross-protection is to blend individual vaccine vectors. This was successful for VSV vaccines expressing several filovirus GPs, showing the utility of this platform as a single-dose, multivalent vaccine (39). DNA-prime and Ad5-boost with plasmids/vectors expressing the EBOV and SUDV GPs protected macaques against BDBV challenge (53). Additionally, combining VSV-SUDV and VSV-EBOV vectors using either a single-dose blended vaccination approach or a prime-boost regimen against heterologous BDBV challenge in NHPs resulted in 100% protection (91). Solid cross-protection, however, does occur within a single species. For example, within the species Marburg marburgvirus, several vaccines protected NHPs against both MARV and RAVV challenges (21, 116). The same held true for heterologous vaccine/challenge studies with distinct EBOV isolates. For example, VSV-EBOV is based on the EBOV-Kikwit glycoprotein and shows good protection against EBOV-Makona (51, 52, 84).

The overlapping endemicity zones of filoviruses, arenaviruses, and other pathogens in parts of sub-Saharan Africa raise the need for cross-protective vaccine approaches. Current research is actively addressing multivalent vaccines for filoviruses and/or other related pathogens with overlapping endemicities (for example Lassa virus). However, to our knowledge there are no reports of dual infections with different filoviruses or filoviruses and arenaviruses; thus, the medical and public health importance seems questionable. In addition, immune responses to multivalent vaccines might be complicated, favoring certain immunogens over others and thus affecting efficacy. As appealing as a multivalent, single vaccine might be, an easier, less expensive, and more effective alternative might already be possible by blending successful monovalent vaccines or combining vaccines in a prime-boost strategy. However, coinfections with bacteria and parasites (such as Plasmodium spp.) certainly occur (24). Multivalent vaccines that also target pathogens that are more prevalent and more important to public health might be worthwhile, considering that people are more likely to comply with vaccination against rare pathogens if the vaccines provide simultaneous protection against more common pathogens.

Durability

The durability of filovirus vaccines is of concern, as nearly all past preclinical studies have focused on short-term protection. A promising result was achieved when NHPs were vaccinated with a single dose of VSV-MARV 14 months prior to challenge (90). Less potent was a vaccination using a cAd3 vector expressing the EBOV and SUDV GPs that completely protected macaques against EBOV challenge 5 weeks after vaccination but failed to protect when challenge occurred 10 months after vaccination (110). However, complete protection against EBOV was observed in the same study when animals received a cAd3 EBOV/SUDV vaccine prime followed by an MVA EBOV/SUDV boost. The data suggest that long-term protection may require strong memory CD8 T cells to reestablish effector and memory CD8 T cell responses after challenge. A small study using an Ad5 codon-optimized EBOV vaccine delivered by the respiratory route showed protection against homologous EBOV challenge 5 months after the final vaccination, suggesting that durability may also be improved by certain routes of vaccination (17).

Durability is certainly an important feature of any vaccination approach. However, given that filoviruses rarely strike twice in the same region so as to cause reinfection of humans with the same or a related virus, the importance of the durability of a vaccine is questionable. Current promising vaccine candidates providing solid short-term protection should not be dismissed for limited durability. The final goal, of course, is to achieve both fast and long-lasting immunity.

Mechanism of Protection

The mechanism of protection against natural EBOV infection is not understood and might differ from that of vaccination (97). It was shown for the Ad5-EBOV vaccine that CD8⁺ T cells appear to be critical for protection against lethal infection (113). In contrast, the VSV-EBOV vaccine elicits antigen-specific IgG responses that are essential for protection from lethal disease (78). Thus, the mechanism of protection may depend on the vaccine platform, vaccination route, and immunogen and may differ from natural infection.

Understanding the mechanism(s) of protection of any vaccine candidate would be highly desirable and should be pursued in preclinical studies and clinical trials. As mentioned above, the mechanism(s) may be platform and/or antigen dependent. In the absence of known mechanisms of protection, correlates of protection might help to define the likelihood of a protective immune response in an individual. This has been studied for several vaccine candidates, in particular Ad5-EBOV (114). If phase 3 efficacy trials cannot be performed in the future due to lack of cases, we may need more animal studies to define at least potential correlates of protection. These data can then be utilized to correlate findings on immune responses from the many phase 2 clinical trials to define those correlates in humans.

Target Population

Unlike AIDS or tuberculosis, EVD is not a disease with global distribution. Therefore, population-based vaccination may be unjustified and unwanted. However, the West African epidemic gave a taste of how quickly EBOV can spread among a naive human population when introduced into large urban settings. Now with identified and evaluated vaccine candidates available, it is time to develop vaccination strategies and to define the target population for vaccination.

Emergency immunization in the form of ring vaccination might be one workable approach. Prophylactic vaccination of certain high-risk exposure groups should certainly be considered. Among those groups are first responders, health care workers, and perhaps military personnel. Family members of patients, another high-risk exposure group, can likely be protected more easily with ring vaccination. Distinct approaches may apply to these different scenarios. Therefore, it is necessary to not focus on a single vaccine candidate in order to have options. For emergency vaccination, a single-shot, fast-acting vaccine is needed; for prophylactic vaccination a more broad and durable immune response may be beneficial and can likely be better achieved with a prime-boost approach.

Licensing

China and Russia have gone ahead and licensed EBOV vaccines, based on concepts principally developed in other countries that have yet to license their own products (8). These vaccines have yet to be evaluated by WHO for use in humans. It remains to be seen if or when licensing of other products will occur.

The licensing pathway of a vaccine is lengthy and cumbersome. Unfortunately, the intent can change over time for marketing or other reasons. Let’s hope that this won’t be the case with some of the prime EBOV vaccines.

LESSONS LEARNED

The West African EBOV epidemic has demonstrated the limitations of our public health response to rare emerging, highly pathogenic infectious diseases. It also has shown that public health measures compete and interfere with human behavior, religion, traditions, and politics (70). This is not unusual, but in this case, it has had a disastrous outcome. Public health intervention came late, but not unusually late when compared to previous outbreaks. The interventions simply failed when applied to an extremely mobile population with intensive cross-border movement and fast and easy access to metropolitan areas. Overall, one must admit that local/regional public health and global response systems were ill prepared for such an overwhelming and catastrophic event. Nevertheless, the response was certainly justified and will likely be better prepared in the future should such a devastating epidemic ever happen again. On a positive note, the epidemic has allowed the utilization of new and sophisticated state-of-the-art diagnostic and surveillance tools in the field that had never been applied to outbreak management before; this relates to assays diagnosing patients in minutes or hours rather than days or months and on-site sequencing that allowed tracking of transmission events (3, 96). In addition, clinical trials tested safety, immunogenicity, and in rare occasions even efficacy of countermeasures (121).

Since the West African EBOV epidemic, new approaches have been added and promising existing approaches have advanced to phase 1, 2, and even 3 clinical trials (121). The Coalition for Epidemic Preparedness Innovations (http://cepi.net/) has been founded to speed up the development of vaccines for rare and emerging pathogens with epidemic/pandemic potential, such as EBOV, MARV, Lassa virus, Nipah virus, and Middle East respiratory syndrome coronavirus. This initiative is highly welcome and globally supported by governmental, academic, industrial, and other nongovernmental entities, a coalition that is unique in its constellation and support. On the other hand, it is disappointing to hear about the delay in licensing of the VSV-EBOV vaccine (8). Let’s hope that this is just a temporary delay for a plausible reason and not an indication of further lowering prioritization and waning interest in countermeasure development, similar to what we have experienced following previous outbreaks.

Stalling countermeasures at the level of preclinical work was a mistake that hopefully will not be repeated. At the very minimum, we need to move into phase 1—or even better, phase 2—clinical trials. Limited stockpiles of vaccines should be available for immediate use in humans and stored at locations that are not vulnerable to financial and political constraints when it comes to immediate release. Licensure would be preferred, but use on compassionate grounds would be sufficient.

A plan for administering vaccines should be developed, and appropriate protocols should be prepared or in place for immediate ethics approval. Discussion on how to best run clinical trials should occur ahead of time and not when vaccination is required, as happened during the West African outbreak. Regardless of disagreements over clinical study designs in emergency situations, one would think that public health and ethics would always prevail.

Currently, one can observe a shift from emergency to prophylactic vaccination, which may require different types of vaccines and immunization concepts. This raises the question of the target population, which is better defined for emergency vaccination. The ring vaccination trial in Guinea can serve as a prime example for that approach (51, 52). It will be tough to argue for population-based vaccination against EBOV. The region of EBOV endemicity is not well defined, and outbreaks have usually not recurred at the same location. The unknown durability and mechanisms of protection as well as potential adverse effects of most/some vaccine candidates do not yet favor a wide distribution. Thus, we are likely limited to vaccination of certain high-risk groups such as medical personnel and aid workers from entities that support outbreak management, laboratory workers, and medical personnel in countries or regions that have experienced outbreaks. However, this will only work if regular follow-ups and perhaps booster immunizations are offered and people comply with recommendations.

From a public health point of view, we would see vaccines as the priority EBOV countermeasure that will hopefully prevent infection and consequences such as post–ebolavirus disease syndrome and temporary EBOV persistence. Nevertheless, the development of antivirals and therapeutics needs to continue for certain applications and intervention strategies (19), but therapy might have limitations when it comes to larger public health responses.

FAMILY FILOVIRIDAE TAXONOMY (12, 65, 95).

The family Filoviridae comprises three genera.

Marburgvirus

Marburg virus (MARV) and Ravn virus (RAVV) belong to the species Marburg marburgvirus.

Ebolavirus

Five species are recognized in the genus Ebolavirus: Bundibugyo ebolavirus [Bundibugyo virus (BDBV)], Reston ebolavirus [Reston virus (RESTV)], Sudan ebolavirus [Sudan virus (SUDV)], Tai Forest ebolavirus [Taï Forest virus (TAFV)], and Zaire ebolavirus [Ebola virus (EBOV)].

Cuevavirus

Lloviu virus (LLOV) is assigned to the species Lloviu cuevavirus.