The utility of human challenge studies in vaccine development: lessons learned from cholera

Abstract

Experiments in which virulent infectious organisms are administered to healthy adult volunteers with the intent to deliberately induce infection have been practiced for centuries. Many useful applications have developed from these experiments such as the provision of evidence of microbial pathogenicity and the identification of key virulence factors. Challenge studies have also played an important role in the evaluation of preliminary efficacy of potential vaccine candidates. Over the past 40 years, these experimental human challenge studies have found particular utility with regards to the development of both living and nonliving attenuated cholera vaccines. This review highlights some of the important contributions made by these challenge studies to cholera vaccine research.

Challenge studies

Volunteer challenge studies involve the intentional induction of infection by the administration of virulent organisms to healthy, consenting volunteers under carefully controlled conditions. Challenge studies may at first seem to be a direct violation of one of the sacred maxims of the Hippocratic oath, “I will keep them from harm …,” promised by physicians across the world. These studies, however, can be ethically justified when there is a compelling rationale to investigate infections that are self-limited or that can be easily and fully treated.¹ The studies must be conducted by competent investigators who abide by rigorously developed protocols with meticulous attention to safety. Volunteers must be fully informed of the risks and anticipated discomforts and freely provide consent before being allowed to participate.¹ In the appropriate setting, challenge studies can save time, money, and resources, and have proven to be a valuable tool in recent vaccine development.

Challenge studies can be applied to prove microbial pathogenicity, confirm host factors that contribute to the acquisition of infection and the severity of disease, define microbial virulence factors, and identify potential vaccine candidates capable of inducing protective immunity.² Perhaps one of the most useful applications of challenge studies, though, is the assessment of preliminary vaccine efficacy.³ Challenge models can prevent the unnecessary exposure of thousands of human subjects in large Phase III field trials by eliminating vaccines that do not demonstrate preliminary evidence of protection. In addition, challenge studies can be used to refine the formulation and schedule of a vaccine that will be further evaluated in field trials.²

Limitations

Results from challenge studies may not always be fully generalizable and careful consideration is needed before extrapolating data obtained from these studies, as demonstrated by cholera vaccine challenge studies. The challenge population, which has traditionally consisted of healthy adults from developed countries, may have many differences from the population at risk for natural disease, which usually consists of children residing in developing countries where the infection of interest is endemic, nutrition may be suboptimal, and coinfection with other intestinal bacteria and parasites is common. Indeed, there has been a recent call for the need to study strategies to overcome this “intestinal barrier,” the poorly understood phenomenon of diminished responses to oral vaccines seen in populations from developing countries.⁴

Another limitation of challenge studies is that they are often designed to assess short-term protection.² This may be suitable for vaccines that are being developed for use predominantly in travelers, but may be less applicable when the intent is for use in endemic areas where long-term immunity is the desired goal. The experimental challenge model is often modified by increasing the virulence of the strain, inoculum, or vehicle in which it is administered to manipulate outcomes such as the attack rate of illness in volunteers. Hence, experimental infection may differ from natural infection.² For all of these reasons, challenge studies still require correlation through the conduct of large field trials in the population for which the vaccine is ultimately being developed. In addition to the evaluation of efficacy, large field trials also allow for continued evaluation of the safety of candidate vaccines in a more natural setting.

Challenge studies and cholera

Volunteer challenge studies with Vibrio cholerae have been a useful way to study many aspects of cholera. Challenge studies involving cholera date from 1892, with the first recorded intentional infection in humans with V. cholerae an attempt to fulfill Koch’s postulates.⁵ Over the past 40 years, challenge studies have served as a unique research tool with many useful applications in the development of cholera vaccines.

Cholera

Cholera is an acute gastrointestinal illness caused by the ingestion of food or water contaminated with the Gram-negative bacillus V. cholerae. Infection is associated with profuse, toxin-mediated, watery diarrhea and can result in rapid and fatal dehydration if untreated. There are over 200 serogroups of cholera, based on the polysaccharide O-antigen. Epidemic cholera is associated with the O1 and, more recently, O139 serogroups. The O1 serogroup is further classified by biotype, classical or El Tor, and within this biotype by serotype, Ogawa or Inaba.⁶ The 2009 World Health Organization annual cholera report identified 221,226 cases of cholera in 45 countries, resulting in 4946 deaths with a case fatality rate of 2.24%; but disease burden is grossly underreported and the true incidence is more likely to be in the millions.⁷ Cholera is endemic in Asia and Africa. Since the early 19th century, there have been seven world pandemics with V. cholerae O1 of the El Tor biotype. The seventh pandemic is still ongoing. In 1993, epidemic cholera due to serogroup O139 was reported in India and Bangladesh and spread rapidly raising the concern of an eighth pandemic. However, the incidence quickly fell and infection was only seen in South East Asian countries. Attention is currently being paid to an ongoing outbreak of V. cholerae serogroup O1 of the El Tor biotype in Haiti, which began in October 2010 with an initial reported 7% case fatality rate, one of the highest recorded in recent history.⁸

The diarrhea of cholera is caused by cholera toxin. This toxin is composed of one A subunit and five B subunit polypeptide chains. It acts on the target intestinal cell through activation of adenylate cyclase, leading to increased intracellular levels of cyclic adenosine monophosphate (cAMP) with subsequent decreases in luminal sodium uptake and increases in chloride and bicarbonate export. More detailed information regarding the microbiology, epidemiology, pathogenesis and clinical features of cholera can be found in the excellent review by Kaper et al.⁹

Treatment

The mainstay of cholera treatment involves rehydration and, in some cases, the use of antibiotics. Parenteral cholera vaccines have been available since 1885 and were first evaluated by Jaime Ferrán, but are no longer used.^10,11 After decades of work, the search for an ideal cholera vaccine is still ongoing. Two types of oral cholera vaccines are currently available: Dukoral® (Crucell-SBL Vaccines, Stockholm, Sweden) is a monovalent-killed whole-cell V. cholerae O1 vaccine (classical and El Tor, Inaba and Ogawa) with recombinant cholera toxin B subunit. It can be administered to adults and children over the age of 2 years, and has been used mostly for travelers to developing countries. An efficacy of up to 90% after two doses has been demonstrated in the first 6 months, falling to 60% after 2 years; mORCVAX (National Institute of Hygiene and Epidemiology, Vietnam) and Shancol (Shantha Biotechnics, Hyderabad, India) are closely related bivalent vaccines that contain O1- and O139-killed whole cells but no cholera toxin B subunit. These can be used in adults and children over the age of 1 year. A booster is recommended 2 years after the primary vaccination. Protective efficacy after two doses was shown to be about 60% but remained at about 50% 3–5 years after vaccination. Use has been in the endemic setting. An oral live attenuated vaccine (CVD 103-HgR) was licensed in the 1990s in several countries but is no longer available.¹²

Widespread use of the currently available vaccines has been limited by cost, incomplete protection, difficulty predicting when and where epidemics will occur, and distribution barriers to populations that would likely benefit the most from vaccination in endemic settings. The current cholera outbreak in Haiti is a stark reminder of how devastating cholera epidemics can be when access to safe drinking water is limited and adequate sanitation has been compromised, highlighting the need for effective prevention strategies to control this infection under emergency situations.

Studying cholera

There are several challenges that make studying cholera difficult. Humans are the only natural host of V. cholerae. Although there is a passive protection model set up to challenge young mice¹³ and rabbit models have also been used,¹⁴ there is really no good model for immunization and challenge with cholera vaccines in animals. In addition, there is no absolute marker of vaccine efficacy. Both antibacterial and antitoxin immunity exist and act synergistically. The serum vibriocidal antibody is the best correlate of antibacterial immunity currently available. These antibodies mediate bacterial killing in vitro, in the presence of complement. They are primarily directed to the V. cholerae lipopolysaccharide (LPS)¹⁵ but may also recognize outer membrane proteins.¹⁶ Epidemiologic studies have shown that vibriocidal antibody titers increase with age in endemic areas and the risk of disease is inversely proportional to titer. These antibodies do not seem to provide protective immunity, but rather are a marker for intestinal mucosal immunity. Secretory immune globulin A intestinal antibodies are thought to mediate the actual protection. Antitoxin titers do not correlate well with protection.⁹

Challenge models

Early challenge models

Field trials performed in Bangladesh, India, and the Philippines during the 1960s established the protection of the parenteral whole-cell bacterial cholera vaccine at about 60%. Protection was short lived, lasting only 3–6 months in endemic areas, and there was an unfavorable adverse event profile.¹² The search for new, more tolerable, cholera vaccines that induced better and longer protection began while parenteral vaccines fell out of favor and became unavailable. However, parenteral cholera vaccines have been recently reviewed and they may have been more effective and better tolerated than realized.¹⁷

In challenge studies dating back to 1969, Cash et al administered classical Inaba strain 569B and classical Ogawa strain 395 in escalating doses to 111 subjects.¹⁸ At least 10⁸ colony-forming units (cfu) were required to induce diarrheal disease in humans. If administered with 2 g of sodium bicarbonate, which neutralizes gastric acid, this infecting dose could be reduced to 10⁶ organisms to induce diarrhea in 80% of volunteers. This early study provided the basis for the use of 10⁶ organisms administered with sodium bicarbonate in fasting volunteers in many subsequent challenge studies to evaluate cholera vaccine efficacy and immunogenicity.

A subset of these volunteers were rechallenged with classical Inaba strain 569B or classical Ogawa strain 395 and protection was compared with cohorts of volunteers vaccinated with Inaba whole-cell parenteral cholera vaccine, Inaba whole-cell vaccine administered orally with bicarbonate, parenteral toxoid vaccine, and control volunteers.¹⁸ Twenty-one volunteers who developed clinical cholera with first infection were completely protected against diarrhea when rechallenged 4–12 months later with the homologous organism, and vibrios were recovered from only 1/21 (4.8%) volunteers. Diarrhea developed in 4/6 (66.7%) volunteers challenged with a heterologous organism and vibrios were recovered from 5/6 (83%) volunteers. Vaccine efficacy against protection of diarrhea ranged from 81% (parenteral whole-cell vaccine) to 47% (toxoid vaccine). Several lessons were derived from this group of challenge experiments. Prior clinical cholera in volunteers due to classical Inaba strain 569B conferred immunity to rechallenge with the homologous strain for up to 1 year. There was no correlation in protection between an individual’s vibriocidal and antitoxin titers with infection or diarrhea. Immunity induced by whole-cell vaccine appeared adequate for use in short-term visitors but would likely lack the long-term protection needed for residents of endemic areas. Lastly, protection by infection was more complete and longer lasting than vaccination. Rationale for the continued development of cholera vaccines and a volunteer model to produce clinical illness in North American volunteers comparable to natural disease as a way to test vaccine efficacy was thus established.

A challenge model for El Tor strains, which by now had replaced classical strains in both endemic and epidemic settings, was also developed.¹⁹ Escalating doses of 10³ to 10⁶ cfu of El Tor Inaba strain N16961 with sodium bicarbonate were given to 26 volunteers. Infection, diarrhea, and serologic responses were seen with doses as low as 10³ vibrios, and the severity of illness was directly proportional to the inoculum size.²⁰ Additional challenge studies were able to show that ingestion of 10⁶ El Tor Inaba vibrios with sodium bicarbonate produced similar results to ingestion with a meal and thus the bicarbonate model was able to mimic natural infection to a reasonable extent.²⁰

Later challenge models

A model for El Tor Inaba strain N16961 in volunteers from Thailand has been validated. Cholera infection is endemic in Thailand and population differences such as gut flora composition and immunological background are expected compared with the North American population. Inoculation of 1.3 × 10⁷ organisms produced a diarrheal attack rate of 90%, though clinical illness appeared milder than that seen in the prior studies with North American volunteers.¹⁹ To improve consistency among challenge studies, a model of cholera with frozen challenge bacteria was also validated.²¹

An epidemic strain of V. cholerae O139 Bengal emerged in Asia in 1993. Immunity to the O1 serogroup conferred no immunity to O139 serogroup. A model was established for challenging volunteers with the new epidemic O139 Bengal strain using freshly harvested (Al 1837) and frozen (Al 4260B) bacteria in North American volunteers.^22,23 This was followed by validation of a challenge model using the frozen 4260B strain in Thai volunteers.²⁴

These early and late challenge models have been summarized in Table 1.

Table 1.

Summary of Cholera challenge Models

Biotype	Strain	Volunteer		Dose rangea	Clinical Responseb	Microbiological responsec	Dose at which response observed	No. of volunteers challengedd	Reference
Serotype		No.	Type		No. (%)	No. (%)
Serogroup O1
Classical Inaba	569B	67	NA	104–106	42 (81)	48 (92)	106	52	Cash et al, 197418
Classical Ogawa	395	25	NA	104–106	22 (88)	22 (88)	106	25	Cash et al, 197418
El Tor Inaba	N16961	26	NA	103–106	9 (90)	10 (100)	106	10	Levine et al, 197931
El Tor Inaba	N16961	26	T	104–107	10 (91)	11 (100)	107	11	Suntharasamai et al, 199219
El Tor Inaba	N16961e	40	NA	105	34 (85)	36 (90)	105	40	Sack et al, 199821
Serogroup O139
	Bengal AI 1837	13	NA	104–106	7 (78)	9 (100)	106	9	Morris et al, 199523
	0139 4260Be	25	NA	105–106	14 (93)	15 (100)	106	15	Cohen et al, 199922
	0139 4260Be	35	T	104–107	11 (73)	15 (100)	107	15	Pitisuttithum et al, 200124

^aAdministered with bicarbonate

^bWith diarrhea

^cWith positive stool culture

^dVolunteers challenged with dose at which response observed

^ePrepared from frozen vials

NA – North American, T - Thai

Lessons in pathogenicity

Challenge studies have helped to establish the role of cholera toxin in the pathogenesis of disease. The production of experimental cholera in a human volunteer was described in 1966 after Syncase cholerigen, a sterile filtrate of broth culture of V. cholerae classical Inaba strain 569 B, was directly introduced into the volunteer’s small intestine.⁵ This observation showed that diarrhea could be produced in the absence of viable cholera vibrios and suggested that a cholerigenic factor was responsible for clinical disease. Experimental cholera was later demonstrated in a dose–response fashion after purified cholera toxin was administered orally to volunteers.²⁵ There are probably other toxins expressed by V. cholerae that contribute to diarrhea, as volunteers challenged with genetically engineered strains exhibiting deletions of the CTX genes that encode for one or both cholera toxin subunits demonstrated milder forms of diarrhea.

ToxR, a regulatory protein of pathogenic V. cholerae O1 strains, controls the expression of cholera toxin and the expression of a rigid toxin-coregulated pilus structure known as tcpA. Through the administration of classical Ogawa strain 395 toxR and tcpA mutants to volunteers, it was shown that deletion of toxR resulted in decreased colonizing capacity and deletion of tcpA prohibited colonization.²⁴ This group of challenge studies provided evidence for the critical role of a specific pilus structure in colonization of the human intestine by V. cholerae and the importance of the toxR regulon in pathogenesis. The role of tcpA in colonization by V. cholerae O139 was also established, while another putative pilus expressed in V. cholerae O139 strains and O1 El Tor biotypes, the mannose-sensitive hemagglutinin (mshA), did not appear to assist in colonization when volunteers were given modified strains of CVD 112, a derivative of O139 strain Al 1837, altered by deletions in tcpA and mshA.²⁶

Lessons in immunity

Early challenge studies in North Americans led to some interesting observations about immunity in a population that was naïve to cholera. Serologic responses and relation to clinical or bacteriological protection were assessed without being confounded by prior infection. After 19 volunteers received monthly doses of purified glutaraldehyde-treated cholera toxoid orally or enterally at doses of either 2 mg or 8 mg for a total of 3 months, 6/10 (60%) volunteers who received a 2 mg dose and 7/9 (78%) who received an 8 mg dose had a fourfold or greater rise in antitoxin titers, but this did not correlate with clinical protection. Viable vibrios were rarely cultured from the stools of rechallenged volunteers, suggesting that antibacterial, rather than antitoxic, mechanisms play an important role in immunity, perhaps through the interference of mucosal colonization.²⁶ Animal studies and epidemiologic studies have also provided evidence that both antibacterial and antitoxic immunity are important.^27–29

Clinical and bacteriological protection was noted in volunteers with clinical cholera due to classical biotype strains when rechallenged with homologous and heterologous classical V. cholerae strains of either serotype, expanding on earlier observations.¹⁸ Of the volunteers who received 2 mg doses of toxoid, 6/10 (60%) were challenged with 10⁶ classical Inaba 569B vibrios, along with six unimmunized controls. Eight of the volunteers given the 8 mg dose and eight controls were challenged with 10⁶ classical Ogawa 395 vibrios. There were no significant differences in stool volume between the vaccine and control groups. A homologous Ogawa rechallenge study was performed using four of the volunteers who developed cholera while serving as controls in the Ogawa challenge. These volunteers were rechallenged 9 weeks later with 10⁶ Ogawa vibrios. Five controls were also given Ogawa vibrios. None of the rechallenged volunteers developed diarrhea or excreted vibrios. All five controls developed diarrhea and excreted vibrios. In heterologous challenge studies, seven control volunteers who developed clinical cholera with Ogawa 395 were rechallenged 10 weeks later with 10⁶ Inaba 569B vibrios. Eleven of twelve (92%) controls developed cholera, but none of the rechallenged volunteers did. Vibrios were cultured from the stools of one (14%) veteran and all controls. Similarly, five volunteers who developed cholera with Inaba challenge were rechallenged 8 weeks later with 10⁶ Ogawa organisms. Nine of ten (90%) controls developed diarrhea and excreted vibrios in the stool, but none of the five rechallenged volunteers developed diarrhea or excreted vibrios. Four volunteers who received 10⁶ Ogawa vibrios were rechallenged 3 years later; none experienced diarrhea, compared with 4/5 (80%) control volunteers. Vibrios were cultured from the stools of 1 of the 4 (25%) rechallenged volunteers and all of the control volunteers.³⁰ These homologous and heterologous rechallenge studies proved it is possible to induce immunity lasting at least several years against homologous and heterologous serotypes.

Challenge studies have also provided important insight into disease severity in relation to host factors. Epidemiologic observations of increased cholera severity among people with the blood group O were confirmed through challenge studies³¹ and Tacket et al went on to establish a model of South American cholera that could be used to predict field efficacy of candidate vaccines among a population with a high prevalence of blood group O.³² Diarrhea resulting from the ingestion of V. cholerae was also found to be more severe in challenge volunteers with low stomach acid.³³

Preliminary vaccine efficacy trials

A number of live and nonliving oral vaccines against V. cholerae have been developed and tested using the volunteer challenge method – several of these are summarized in Table 2 and will be discussed further. This list is not exhaustive, as many challenge studies have been performed with cholera vaccine candidates. Additionally, many volunteer studies that investigate immunogenicity without subsequent challenge have been performed. Studies that highlight specific benefits or pitfalls of challenge studies have been included for illustrative purposes. Two recent Cochrane Database reviews cover the spectrum of both oral and parenteral cholera vaccines.^17,34

Table 2.

Summary of cholera vaccines tested using challenge studies

Vaccine Strain	Parent Strain	Protective Efficacy	Reference(s)
Non-Living Oral Vaccines
Whole vibrio	V. cholerae O1 Classical Inaba strain Cairo 48, Classical Ogawa strain Cairo 50, El Tor Inaba strain Phil 6973	56%	Black et al , 198735
Whole vibrio-B subunit	V. cholerae O1 Classical Inaba strain Cairo 48, Classical Ogawa strain Cairo 50, El Tor Inaba strain Phil 6973 + purified cholera toxin B subunit	64%	Black et al, 198735
Live Attenuated Oral Vaccines
Texas Star	V. cholerae O1 El Tor Ogawa	61%	Levine et al , 198439
Peru-15	V. cholerae O1 El Tor Inaba strain isolated in Peru in 1991	93%	Cohen et al, 200240
638	V. cholerae O1 El Tor Ogawa strain C7258	100%	Garcia et al , 200541
JBK70	V. cholerae O1 El Tor Inaba strain N16961	89%	Levine et al , 198842
CVD-111	V. cholerae O1 El Tor Ogawa strain N16117	80.90%	Tacket et al , 199745
CVD-103 HgR	V. cholerae O1 Classical Inaba strain 569B	65–100%	Levine et al , 198843 Losonsky et al , 199351 Tacket et al, 199952 Tacket et al , 199248
Bengal-15	V. cholerae O139 strain MO10	83%	Coster et al , 199556
CVD-112	V. cholerae O139 strain AI1837	84%	Tacket et al , 199546

Nonliving oral vaccines

The immune response and protective efficacy of two oral nonliving cholera vaccines were tested in volunteers.³⁵ One of the vaccines contained heat-killed classical Inaba and Ogawa strains and formalin-treated El Tor Inaba (whole vibrio vaccine). The other contained the same whole vibrios plus purified subunit B of the cholera toxin (whole vibrio-B subunit). The vaccines were administered orally at 2 week intervals for a total of three doses. There was no reactogenicity reported in North American volunteers following immunization with either vaccine. There was a significant rise in serum vibriocidal antibody titers in 10/14 (71%) volunteers who received the whole vibrio vaccine and 17/19 (89%) volunteers who received the whole vibrio-B subunit vaccine. Four weeks after completing immunization, volunteers were challenged with El Tor Inaba strain N16961. Protective efficacy was 56% in volunteers who received whole vibrio vaccine and 64% in those that received whole vibrio-B subunit.³⁵ These results were validated in a field efficacy trial in Bangladesh in which the whole vibrio and whole vibrio-B subunit vaccines elicited 58% and 85% protection, respectively, 6 months after vaccination.³⁶ Levels of protection elicited by the whole vibrio and whole vibrio-B subunit vaccines at 12 months were 53% and 62%, respectively.³⁷ Unfortunately, the efficacy in children aged ≤5 years of age was 31% for whole vibrio and 38% for whole vibrio-B subunit at 12 months after vaccination.³⁷

These studies highlight the utility of challenge studies prior to initiating large-scale field trials in endemic areas, while also illustrating the differences in short-term protection measured during challenge studies and the long-term protection desired of a vaccine for use in endemic regions. Additionally, the difference in sustained efficacy between adult volunteers and young children is shown. These preliminary studies led to further efficacy trials and eventual licensure of the whole vibrio-B subunit vaccine as Dukoral. Other volunteer studies of nonliving oral vaccines in multiple countries have been reviewed.³⁸

Live attenuated oral vaccines

V. cholerae O1-

Texas Star, a derivative of El Tor Ogawa strain 3083, was attenuated using chemical mutagenesis with nitrosoguanidine. This strain produces the B but not the A subunit of cholera toxin. The vaccine was given at doses of 10⁵ – 5 × 10¹⁰ organisms as either one or two doses 1 week apart. Sixteen of the 68 (24%) vaccinees developed diarrhea following vaccination that was not dose dependent. All doses induced serum vibriocidal antibodies in 63/68 vaccinees (93%), but antitoxin antibodies were elicited in only 11/42 (26%) and 9/26 (35%) after one or two doses, respectively. Eight vaccinees that received a single dose of either 10⁸ or 10¹⁰ organisms and four unvaccinated controls were challenged with 10⁶ El Tor Ogawa strain 3083 organisms 4–6 weeks after vaccination. None of the control volunteers developed diarrhea after challenge, although they excreted the strain and had serological responses, suggesting the 3083 strain had diminished pathogenicity. Vaccinees who received a single dose of 5 × 10¹⁰ Texas Star organisms were challenged with 10⁶ virulent El Tor Ogawa strain E7946 organisms to determine the efficacy against challenge with a homologous serotype. To determine whether Texas Star was protective against challenge with El Tor vibrios of heterologous serotype (Inaba), volunteers that received two doses of 10⁹ or 2 × 10¹⁰ Texas Star organisms were challenged with 10⁶ El Tor Inaba strain N16961 5–7 weeks after completion of vaccination. Texas Star had an overall efficacy of 61% against homologous or heterologous serotypes in challenge models that induced clinical disease in 70%–80% of control volunteers. Although this vaccine elicited vibriocidal antibody responses and was moderately protective against homologous and heterologous challenge, the random nature of nitrosoguanidine mutagenesis makes identification of the precise genetic mechanisms of attenuation difficult to establish. Without knowledge of the genetic mechanisms of attenuation, reversion to virulence is a theoretical possibility.³⁹ Due to these uncertainties, the vaccine has not been pursued. However, these challenge studies established that live attenuated oral cholera vaccines can elicit protection in volunteers and provided the groundwork for development of live attenuated oral cholera vaccines using recombinant DNA technology.

Recombinant DNA technology has been used to develop a number of vaccines against V. cholerae, many of which have been tested by volunteer challenge studies.^40–49 Vaccine strain JBK70 was derived from El Tor Inaba strain N16961 by deletion of the genes for both A and B subunits of the cholera toxin. Volunteers received 10⁶, 10⁸, or 10¹⁰ JBK70 organisms with 1/4 (25%), 2/5 (40%), and 4/5 (80%) developing diarrhea, respectively.⁴³ Fourfold or greater rise in vibriocidal antibody titers was observed in all 14 (100%) vaccine recipients. One month after vaccination, volunteers were challenged with parental strain N16961 and the vaccine had a protective efficacy of 89%. Despite high levels of protective efficacy, significant reactogenicity limited the further utility of this strain as a vaccine.⁴³ Evaluation of this vaccine in challenge studies indicated that genetically engineered strains could be designed that would confer protection; however, these studies also provided evidence that other factors besides cholera toxin significantly contribute to disease⁴³.

Peru-15 is a live attenuated strain derived from El Tor Inaba strain N16961. The mechanisms of attenuation include deletion of the cholera toxin gene element, defective motility, and inability to recombine with homologous DNA. In a randomized, double blind, placebo-controlled trial, volunteers received either 2 × 10⁸ cfu of Peru-15 or placebo (buffer alone). Following a single dose of Peru-15, 39/40 (97%) vaccinated volunteers had a fourfold or greater rise in vibriocidal antibody titers. Volunteers were challenged 3 months after vaccination with El Tor Inaba strain N16961 and Peru-15 demonstrated protective efficacy of 93% against any diarrhea and 100% against severe or moderate disease.⁴⁰

V. cholerae 638 is a derivative of El Tor Ogawa strain C7258 attenuated by deletion of cholera toxin and disruption of the hemagglutinin/protease coding sequence by insertion of Clostridium thermocellum endoglucanase A gene. Following vaccination, 96% of V. cholerae 638 recipients demonstrated fourfold or greater rise in vibriocidal antibody titers.⁴¹ Two challenge studies were performed one month after vaccination, the first with attenuated strain El Tor Ogawa 81. After V. cholerae 81 challenge, there was excretion of V. cholerae 81 in the feces of only 2/5 (40%) vaccinated volunteers and 5/5 (100%) controls. None of the vaccinated volunteers and 3/5 (60%) controls had diarrhea after challenge. Based on these results, a second study using challenge with virulent El Tor Ogawa 3008 was performed. The virulent strain caused diarrhea in 7/9 (78%) controls and none of the vaccine recipients. V. cholerae 638 was 100% protective against diarrhea in this group of 12 volunteers.⁴¹

Derived from wild-type El Tor Ogawa strain N16117, CVD 111 was also tested using the volunteer challenge model. The N16117 strain was used as a parent strain due to lower virulence than strain E7946, the parent of CVD 110 that was found to be overly reactogenic.⁴⁷ CVD 111 was attenuated by deletion of the virulence cassette, which includes the genes for cholera toxin (ctxAB), core-encoded pilus (cep), zonula occludens toxin (zot), and accessory cholera enterotoxin (ace). Additionally, ctxB, the B subunit of cholera toxin, and mercury resistance gene, were introduced. CVD 111 was given in a single oral dose and was noted to be mildly reactogenic with 3/25 (12%) volunteers developing diarrhea after vaccination. Ogawa vibriocidal antibodies were detected in 23/25 (92%) vaccinees. Thirty-five days after vaccination, volunteers were challenged with El Tor Ogawa strain 3008. Vaccine efficacy was 81% and stool volume was significantly less in vaccinees who developed diarrhea than in controls.⁴⁵ Additional volunteer trials were undertaken to determine the safety and immunogenicity of combination CVD 111 (El Tor Ogawa) with CVD 103-HgR (classical Inaba).^49,50

CVD 103-HgR was engineered from the classical Inaba strain 569B. It is attenuated by deletion of 94% of the cholera toxin A subunit and insertion of a mercury ion resistance gene into the hemolysin A locus. Multiple studies in North American volunteers have shown excellent immunogenicity and efficacy after a single dose.^42,48,51,52 Vibriocidal antibody increases of fourfold or greater were seen in 39/43 (91%) CVD 103-HgR recipients.⁵² Protective efficacy of 100% against homologous challenge has been demonstrated as early as 8 days and as late as 6 months after vaccination.⁴⁸ CVD 103-HgR is also protective against challenge with biotype-heterologous O1 El Tor Inaba or Ogawa (it has about 65% protective efficacy for as long as 3 months after immunization).^42,52 CVD 103-HgR was licensed in the 1990s for use in several countries based on the results of these challenge studies.⁵³

When CVD103-HgR was tested in Indonesian children (aged 5–9 years), a tenfold higher dose was required to obtain similar seroconversion rates to those identified in North American volunteer studies.⁵⁴ Potential causes of this reduced immunogenicity in endemic areas include background intestinal immunity, which could potentially interfere with the vaccine’s ability to infect and elicit vibriocidal responses, and the presence of small bowel bacterial overgrowth, which may inhibit the vaccine strain.^54,55 The role of small bowel bacterial overgrowth on vibriocidal antibody response to CVD 103-HgR was evaluated in Chilean schoolchildren and increased peak H₂ (measurement of bacterial overgrowth) was associated with decreased seroconversion.⁵⁵ Additionally, the V. cholerae–specific cellular antibody responses to cholera toxin and LPS for volunteers vaccinated with CVD 103-HgR followed by challenge with classical Inaba strain 569B were evaluated but did not correlate with protective immunity.⁵¹

V. cholerae O139-

Bengal-15 is a nonmotile derivative of vaccine prototype Bengal-3. To create Bengal-3, O139 strain MO10 was attenuated by deletion of virulence genes ctxAB, zot, ace, and cep as well as disruption of recA and insertion of ctxB.^56,57 Bengal-15 was given in a single, oral dose to volunteers and was well tolerated without causing diarrhea. Vibriocidal titers were detected in 3/4 (75%) volunteers receiving Bengal-15. Challenge with O139 was performed 1 month after vaccination and Bengal-15 demonstrated protective efficacy of 83%.⁵⁶

Attenuation of O139 strain AI1837 was performed by deletion of the genes ctxAB, zot, ace, and cep along with insertion of cholera toxin B subunit and a mercury resistance gene into the hemolysin A gene creating vaccine candidate CVD 112. To determine the optimal dose, a single oral dose of either 5 × 10⁶ or 5 × 10⁸ cfu CVD 112 was given to volunteers.⁴⁶ None of the volunteers that received the lower dose, and only 3/6 (50%) volunteers who received the higher dose, developed diarrhea. No systemic symptoms were reported. There were no vibriocidal antibodies detected after vaccination with either dose. Five weeks after vaccination, volunteers were challenged with O139 AI1837. CVD 112 elicited a protective efficacy of 84% despite the absence of detectable vibriocidal antibodies.⁴⁶ V. cholerae O139 serogroup strains differ from the O1 serogroup strains in that they produce a capsule of polymerized O-antigen molecules that are not covalently linked to the core polysaccharide.^58,59 It has been hypothesized that this capsule competitively interferes with binding of antibody to the core linked O-antigen resulting in ineffective complement fixation and decreased serum bactericidal activity.^58–60 The differences in detection of vibriocidal responses following vaccination with O139 strains with similar protective efficacy (Bengal-15 83% and CVD 112 84%) suggest that, unlike for O1 serogroup strains, vibriocidal antibodies may not be good predictors of protection against the O139 serogroup. Alternatively, differences in the vibriocidal methods used could account for the lack of responses described in some studies.^46,56 It has been shown that, unlike O1 strains, assay conditions, including diluents, level of complement, and concentration of indicator bacteria, may significantly affect O139 susceptibility to antibody and complement-mediated killing.⁶⁰ Later studies, however, also failed to correlate vibriocidal antibody titers with protection against V. cholerae O139, despite detection of robust responses.⁶¹

Hybrid vaccines

Using the Salmonella typhi vaccine strain Ty21a as a backbone, a typhoid-cholera hybrid vaccine (EX645) was developed and tested in volunteer challenge studies. Ty21a is a live attenuated strain of S. typhi that has been extensively studied.^62–65 A plasmid containing the genes for LPS O-antigen of O1 Inaba was inserted into a rifampin-resistant (to facilitate selection of vaccine strain from stool), thymidine-dependent (to maintain the plasmid) strain of Ty21a. To allow expression of the cholera O-antigen on the surface, the rfa region of Ty21a was replaced with the homologous region from Escherichia coli K-12. Volunteers received three doses (on days 0, 2, and 4) of 10¹⁰ viable organisms with the plasmid. Following vaccination, 6/14 (43%) EX645 recipients had a fourfold or higher rise in vibriocidal antibody against O1 Inaba. Four weeks after vaccination, challenge was performed with El Tor Inaba strain N16961. The vaccine efficacy was only 25%. Although the efficacy was not significant, these results provide valuable information indicating that response against LPS O-antigen, the only V. cholerae antigen in the vaccine strain, contributes to protection.⁴⁴

Conclusion

With careful attention to detail, it is possible for experienced investigators to safely and ethically perform volunteer challenge studies that can significantly aid vaccine development. There are many ethical considerations involved in the undertaking of challenge studies. Volunteers are subjected to potential harm and discomfort with no personal benefit. As with any volunteer study, it is critical to obtain informed consent and to ascertain that the participant fully understands the implications of their consent. Furthermore, the investigators are obligated to minimize risk as well as design appropriate studies so that the information obtained contributes to the scientific community and society at large.

There are many components that are required to carry out an ethical challenge study for vaccine development. Foremost of these is a pathogen that produces a self-limited or treatable disease.² It is of critical importance that both the investigators and facilities for these studies allow for appropriate management of the disease, including infection control (to prevent spread of the disease outside the setting of the trial) and treatment of the induced illness and potential complications.

Cholera provides an excellent example of a model system in which valuable information has been obtained through the use of challenge studies contributing to the development of several vaccine candidates and the rejection of vaccine candidates that had excessive reactogenicity, poor immunogenicity or poor protective efficacy. Cholera possesses many of the features desired of a disease to be studied in this manner: there is a well-established model of infection that induces reproducible rates of disease, there are no long-term sequelae of disease, and treatment with hydration and antibiotics are readily available in the experimental setting.

Studies to determine the protective efficacy of the whole vibrio B subunit vaccine provide a prime example both of the utility and some of the pitfalls of vaccine challenge studies. Challenge studies in North American volunteers identified a vaccine candidate that induced sufficient protection to warrant further investigation in large field trials.³⁵ However, there were distinct differences in efficacy between North American and Bangladeshi volunteers.³⁶ Additionally, the inclusion of children in the field trials highlighted the fact that studies carried out in healthy adults are not necessarily applicable to young children in endemic areas.³⁷ Multiple studies involving volunteers in endemic regions supported the initial findings of challenge studies.^36–38 The live attenuated vaccine CVD-103HgR was licensed in many countries in the early 1990s and the initial licensure was based exclusively on the results of challenge studies. The safety, immunogenicity, and practicality of this single-dose vaccine were validated post-licensure in numerous studies. Specifically, the practicality of a single-dose vaccine supported the use of CVD-103HgR during outbreaks.^66–68 In Micronesia, a retrospective analysis following outbreak intervention with single-dose CVD 103-HgR found a 79.2% efficacy in the target population.⁶⁷ It is important to consider differences in the volunteer population and the target populations in endemic areas as evidenced by field trials in endemic areas that showed lower rates of seroconversion and lower efficacy than North American trials;^36,37,54,55 however, challenge studies provide a cost-effective preview of the vaccine candidate before undertaking large-scale field trials and allow for the rejection of vaccine candidates with excess reactogenicity or poor immunogenicity.^39,43